Deverbal Adjectives at the Interface: A Crosslinguistic Investigation into the Morphology, Syntax and Semantics of -ble 9781614510659, 9781614510642

This volume explores the syntax, semantics, and morphology of -ble adjectives within Distributed Morphology. It presents

230 108 6MB

English Pages 356 Year 2013

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Deverbal Adjectives at the Interface: A Crosslinguistic Investigation into the Morphology, Syntax and Semantics of -ble
 9781614510659, 9781614510642

Table of contents :
Acknowledgements
Content
List of abbreviations
1 Introduction
2 -BLE
3 Case Study 1: V todo lo Vble
4 Case study 2: N-ble
5 Conclusions and directions for further research
References
Index

Citation preview

Isabel Oltra-Massuet Deverbal Adjectives at the Interface

Interface Explorations

Editors Artemis Alexiadou and T. Alan Hall

Volume 28

Isabel Oltra-Massuet

Deverbal Adjectives at the Interface A Crosslinguistic Investigation into the Morphology, Syntax and Semantics of -ble

ISBN 978-1-61451-064-2 e-ISBN 978-1-61451-065-9 ISSN 1861-4167 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.dnb.de. © 2014 Walter de Gruyter, Inc., Boston/Berlin Typesetting: PTP-Berlin Protago-TEX-Production GmbH, Berlin Printing: Hubert & Co. GmbH & Co. KG, Göttingen ♾ Printed on acid-free paper Printed in Germany www.degruyter.com

Per a tu, Lluís (Amb el permís dels meus pares)

Acknowledgements This book is a revised version of my doctoral dissertation, defended in January 2010 at the CLT-UAB. I am truly grateful to my thesis supervisor, Joan Mascaró, and to the members of my thesis committee, Eulàlia Bonet, Carlos Piera, and, in particular, to Artemis Alexiadou for encouraging me to submit the proposal for this Mouton series. I am especially indebted to the members of the CLT-UAB for advice, support, and extremely helpful observations during the writing of my dissertation, most especially to Anna Bartra, Maria Teresa Espinal, Jaume Mateu, Carme Picallo, and Gemma Rigau. The revision of this book owes an enormous debt to Patricia Cabredo-Hofherr, Ora Matushansky, and Gereon Müller for their generous help in the form of detailed reviews of my dissertation with invaluable comments, significant observations, constructive criticisms, and instructive suggestions, which have helped me out to substantially improve the original. I am also extremely thankful to Antonio Fábregas, Alec Marantz, and Michal Starke for their insightful observations. I wish to express my most sincere gratitude to Violeta Demonte for her support, encouragement, and friendship, as well as for placing her trust in me and offering me the possibility of a postdoctoral JAEDoc position at the CCHSCSIC in Madrid, where I was able to dedicate my time to research and was provided with financial support to attend an important number of conferences. I would like to extend my warmest thanks to all the members of the LyCC group at the CCHS-CSIC for two extraordinary and intense years. Special thanks to Elena Castroviejo, Isabel Pérez-Jiménez, Olga Fernández Soriano, and Norberto Moreno Quibén for their help, their support and, most of all, for their friendship. Parts of this work have been presented at a number of conferences, workshops and seminars since 2008. I wish to deeply thank the audiences at the following meetings for their generous help and feedback: Seminari del Centre de Lingüística Teòrica of the UAB, 10th Seoul International Conference on Generative Grammar, 39th Poznań Linguistic Meeting, 10th Encuentro de Lingüística en el Noroeste, 33rd Aedean Conference, 42nd Annual Meeting of the Societas Linguistica Europaea, 20th Colloquium on Generative Grammar, IX Congreso Internacional de Lingüística General, CLS 47, GLOW 34, CASTL Workshop Categorization and c­ ategory change in morphology, 50 years of Linguistics at MIT – A scientific reunion. Many thanks to the editors at Mouton de Gruyter, Emily Farrell, Lara Wyson, and especially to Angelika Hermann, for her patience and help with production and copy-editing. I would like to express once more my most heartfelt thanks to all my infor­ mants, linguists and non-linguists, colleagues and friends, for their invaluable help with data.

viii   

   Acknowledgements

Needless to say, all remaining errors and loose ends are my own. The research underlying this volume has been partially supported by grants to the projects HUM2005-00312 (MCT), FFI2009-07114 (MICINN), and FFI201022181-C03-01 (MICINN).“ On a more personal side, infinite thanks to my family and friends, they have been more than helpful, even without knowing it. Gràcies infinites a la meva família i amics, que m’han ajudat fins i tot sense saber-ho. A tots els que d’una manera o altra m’heu fet costat, gràcies. Molt especialment als meus pares, Joaquim Oltra Amat i Dolors Massuet Nicolau. Last, but certainly not least, I want to express my love and immeasurable thanks to Lluís Macià Pica, my everything, el meu tot. Thank you for your endless support, encouragement, understanding, and patience, for taking care of me at every moment. This book is dedicated to him.

Content Acknowledgements  vii List of abbreviations  xiii 1 Introduction  1 1 Derivation in morphology  1 2 The framework: Distributed Morphology  3 2.1 Uncategorized roots  4 2.2 Romance theme vowels in DM  6 3 Deverbal adjectives  8 3.1 Participles versus deverbal adjectives  8 3.2 Participles in DM: Embick (2003, 2004)  11 4 Book overview  12 2 -BLE  24 1 Introduction  24 1.1 Main claims and chapter overview   25 2 The morphology of -ble  28 2.1 Aspects of the morphology of English and German -ble  28 2.2 Aspects of the morphology of Catalan and Spanish -ble   33 2.3 English vs. Romance: Theme vowels and beyond  38 2.4 Summary  42 3 The external syntax of -ble  42 3.1 The base  43 3.2 Being deverbal  46 3.2.1 The expression of arguments   47 3.2.2 Adverbial modification  55 3.2.3 Negated -ble adjectives and eventivity  59 3.2.4 Interim conclusion  64 3.3 Being passive  64 3.3.1 The external argument of the adjective  66 3.3.2 The passive component   69 3.3.2.1 Non-passive -ble adjectives  76 3.3.3 The external argument of the verb  80 3.3.3.1 Constraints on the expression of the verb’s external argument  82 3.3.3.2 The implicit external argument  86 3.3.4 Two main types of -ble  93 3.4 Summary   96

x   

   Content

4 The semantics of -ble  98 4.1 The modality of -ble crosslinguistically  101 4.2 Two interpretations for modal -ble  105 4.2.1 Two interpretations of -ble across languages  110 4.3 Type of modality  114 4.4 The modal context of -ble   121 4.4.1 Licensing free choice any  121 4.4.2 On subjunctive and specificity  122 4.5 Interim summary   127 5 Previous approaches  128 5.1 Based on passive  129 5.2 Based on middle  134 5.2.1 The middle   134 5.2.1.1 Properties of middles  135 5.2.1.2 Middles and -ble  139 5.2.2 -ble derived from middle  142 5.3 -ble in Distributed Morphology  145 6 Towards an analysis of -ble  148 6.1 The internal syntax of -ble  149 6.2 The structure for high -ble  151 6.3 The structure for low -ble  153 6.4 The Vocabulary item -ble crosslinguistically  155 6.4.1 The case of German -ble   156 6.4.2 On multiple theme vowels  157 6.4.3 Morphological evidence  159 7 Conclusions and remaining issues  162 3 Case Study 1: V todo lo Vble  165 1 Introduction  165 2 Main claims and chapter overview  167 3 V todo lo Vble. General Properties  171 3.1 Pleonastic structure  173 3.1.1 Pleonastic structure with allomorphic verbal roots  174 3.2 Quantification  176 3.3 Aspectual shift  176 3.4 The interpretation of V todo lo Vble  181 4 Unergative verbs  187 4.1 todo lo Vble as a cognate object  188 4.2 lo as D  199

Content   

   xi

Unaccusative verbs  206 Cognate objects with unaccusative verbs  213 Unaccusative COs in English  214 Unaccusative COs in English and Spanish: A crosslinguistic parallelism  224 5.2 On degrees   226 5.3 lo as Deg   232 5.4 Case assignment  245 6 Unergative and unaccusative -ble  246 7 Conclusions  250

5 5.1 5.1.1 5.1.2

4 Case study 2: N-ble  253 1 Introduction  253 1.1 Main claims and chapter overview  255 2 Denominal -ble adjectives  256 2.1 Previous approaches  256 2.1.1 English Nble  256 2.1.1.1 Analyzing English Nble   258 2.1.2 Catalan and Spanish Nble   261 2.2 General properties  262 2.2.1 Romance Nble  262 2.2.2 The productivity of Nble in Romance   267 2.2.3 Productive Nble: Romance versus English  268 2.2.4 Properties of the N-base  269 3 Towards an analysis of Nble  282 3.1 Predictions of the analysis  302 4 Conclusions  303 5 Conclusions and directions for further research  305 1 Summary and further research on V todo lo Vble   305 1.1 Crosslinguistic variation   306 1.2 Cognation across languages   307 1.3 Degree cognates and resultatives as incremental themes  309 1.4 todo lo Vble as an amount relative?   311 2 Summary and further research on Nble   311 2.1 Post-denoting -ble across languages   312 2.2 The case of Italian Nble  315 2.3 Intralinguistic variation: German -bar/-wert/-abel  315 3 Concluding remarks  316

xii   

   Content

References  319 Index  339

List of abbreviations 1, 2, 3 1st, 2nd, 3rd Person acc Accusative agr Agreement Alg Alguerese aor Aorist Ar Arabic aspr Aspect (resultative) asps Aspect (stative) aux Auxiliary Cat Catalan caus Causative clloc Clitic (locative) CO Cognate Object cond Conditional dat Dative dim Diminutive DM Distributed Morphology Dut Dutch ECM Exceptional Case Marking Eng English epv Epenthetic vowel f Feminine FP Functional Phrase Fr French fut Future Geo Georgian Ger German Ger Gerund HO Hyponymous Object Heb Hebrew Hun Hungarian Ice Icelandic ind Indicative inf Infinitive ipfv Imperfective It Italian m Masculine mod Modal

xiv   

   List of abbreviations

MS Morphological Structure n Neuter Nah Nahuatl part Participle pst Past pass Passive pl Plural Por Portuguese pref Prefix prs Present refl Reflexive sg Singular sbjv Subjunctive Sp Spanish sub Subordinator suf Suffix sup Superlative th Theme Position tr Transitivizer Tur Turkish VI Vocabulary Insertion

1 Introduction 1 Derivation in morphology The dichotomy between what belongs to the lexicon and what is part of syntax has played a major role in Generative Grammar to explain a number of issues. Recent proposals within the framework of Distributed Morphology (Halle and Marantz 1993, 1994; Marantz 1997, 2001; Embick 2000, 2003, 2004; Harley and Noyer 1998, 2000; inter alia) have argued that syntax proper is the locus of word formation, so that there is no lexicon as a separate module with generative capacity (Marantz 1995, 1997, 1999a, b; Embick 2000; Alexiadou 2001; Fábregas 2005; inter alia. See also Borer 2001, 2004, 2005 for a similar view). Lexicalist approaches to word formation (e.g. Aronoff 1976; Lieber 1980; Anderson 1992) endorse the view that the morphological properties of the base, e.g. their syntactic category or their argument structure properties, determine their syntactic behavior (endo-skeletal view). Therefore, words are created in a separate module, the lexicon, from which they can project into the syntax. That is, lexicalist models assume a division of labor between the lexicon (a generative module that is also a repository for the storage of idiosyncratic forms and meanings) and the syntactic component (a second generative component). The Distributed Morphology framework assumes a single generative mechanism instead, i.e. syntax, which is responsible for all structure-building operations, be they at the level of the word or at the level of the phrase. This means that the computational system does not manipulate the outcome of the lexicon, but syntax itself creates words out of roots unspecified for lexical category (Marantz 1997). Also with respect to meaning and issues pertaining to eventive structure, these are transferred to syntax and consequently different syntactic heads are held responsible for the varied thematic and event structural interpretations of arguments. Meaning aspects that were traditionally accounted for by the lexical projection of the base emerge now from syntactic structure building (Kratzer 1996; Marantz 1997; Borer 1998; and much related work) (exo-skeletal view). Postulating a single generative engine for all structure building raises a number of important questions, among them, whether, how and to what extent words are or can be related to more complex syntactic structures with similar morphological, syntactic or semantic properties in a non trivial way. The main goal of this book is twofold. First, this piece of research seeks to develop a crosslinguistic analysis of morphologically complex adjectives, which

2   

   Introduction

is based on the study of -ble,1 one of the most productive affixes. The suffix -ble has not only been examined to different degrees by many,2 if not all, linguists working in the area of morphology (Vendler 1968; Aronoff 1976; Val Álvaro 1981; de Miguel 1986; Anderson 1992; Gràcia 1995; DiSciullo 1997; Batiukova 2006; inter alia), but it has also been analyzed by some linguists in the realm of phonology (Steriade 1999), syntax (Chapin 1967; Kayne 1981, 1984; among others), and it has also been considered in the field of semantics (Lyons 1977; Chierchia and McConnell-Ginet 1990; or Krifka et al. 1995). The reasons for such popularity are related to its challenging properties. Among others, to be discussed in the next chapters, this suffix is one of the most productive suffixes crosslinguistically (e.g. Aronoff 1976; Val Álvaro 1981; Fabb 1988); it shows both regular and irregular morphophonological properties; it is selected by a variety of bases; it exhibits both verbal and adjectival properties; it parallels sentential passive constructions; and it shows unique modal properties among derivational affixes. The analysis to be developed is crosslinguistic in two different respects. On the one hand, it is crosslinguistic in the sense of being across languages, i.e. for the study, I essentially look at two different types of languages, Germanic (basically English, with extensive reference to German) and Romance (Catalan and Spanish), although data from other languages are also considered for morphological evidence. On the other hand, it is a crosslinguistic study in that the analysis involves a study of morphologically complex adjectives across different linguistic subdisciplines, from morphology itself to syntax and semantics (and reference to pragmatics where necessary). That is, focusing on the internal structure of these morphologically complex -ble adjectives and how their syntactico-semantic properties relate to other sentential structures, as well as on their appearance with different kinds of unexpected possible bases, I can investigate into aspects of the interaction of morphology with syntax and semantics, which is in fact the underlying primary goal of this book. The second aim of this volume is to contribute strong empirical arguments to the longstanding debate on the locus of word formation. On the one hand, chapter 3 provides new strong empirical evidence for a syntactic approach to word formation on the basis of a previously unnoticed structure of the type V todo lo Vble where essentially any verb, be it unaccusative or unergative or tran-

1 Throughout I will use the label -ble adjectives to refer to all kinds of -able or -ible adjectives in the different languages. I will only refer to some specific suffix, e.g. -bar in German, where this is necessary. 2 See for instance the book by Hall (1877), which is almost entirely devoted to the discussion of the adjective reliable in English.



The framework: Distributed Morphology   

   3

sitive, can appear. The existence of forms such as *llegable ‘arrive.ble’, *llovible ‘rain.ble’, or *estornudable ‘sneeze.ble’ in this construction characterizes a new empirical paradox on the nature of impossible words, i.e. words that are systematically created by the computational system despite their being systematically ungrammatical. On the other hand, chapter 4 also offers a strong case for the interaction of morphology with semantics in word-structure building by looking at a productive process of -ble derivation with a particular type of nominal bases, post-denoting nouns, in Catalan and Spanish, such as alcaldable ‘mayor.ble’ or papable ‘Pope.ble’ and its systematic absence in English.

2 The framework: Distributed Morphology The analysis is couched within the DM framework. In this theory, there is a Morphological Structure level (MS) between Spell-out (Halle and Marantz 1993, 1994, and related work), and Phonology as represented in (1). (1)

Syntax (← Narrow Lexicon: Syntactic features)

MS

(←Vocabulary)



(Encyclopedia→)

LF

PF In this syntactic approach to word-formation there is no lexicon where words can be generated, which means that we cannot have recourse to the distinction between lexical and syntactic derivation to account for differences in apparently similar structures. Instead, there are three lists containing information, each one associated with one level (Marantz 1997). Chomsky’s narrow lexicon refers now to the complexes of syntactic and semantic features that are bundled together into terminal nodes, here called morphemes, selected by each particular language from a UG inventory; the Vocabulary is a list of correspondence rules between phonological material and the morpho-syntactic features that it spells-out or its context of insertion. This Vocabulary is also associated with the conceptual meanings listed in the Encyclopedia, which also contains special meanings and extra­linguistic

4   

   Introduction

knowledge. Only information that is relevant for syntactic and semantic interpretation is present in the syntax. After the operations of the computational system, the syntactic structures enter into the morphology or Morphological Structure (MS) where morphological processes may further modify them before Vocabulary Insertion supplies the terminal nodes with phonological features (Late Insertion of underspecified Vocabulary items). Vocabulary insertion is a competition-based operation, subject to the Subset Principle (Halle 1997 :  428), according to which the phonological exponent of a Vocabulary item is inserted if it is the most highly specified item that matches all or a subset of the grammatical features specified in the terminal node. Various versions of Halle’s morphological Subset Principle have been known as the Specificity Condition, the Elsewhere Principle, the Proper Inclusion Principle or Condition, the Blocking Principle, or Panini’s Principle (e.g. Kiparsky 1973; Di Sciullo and Williams 1987; Noyer 1992; Stump 2001). Distributed Morphology assumes a strong version of Beard’s (1995) Separation Hypothesis, which allows syntactico-semantic features to be dissociated from its phonological realization. Such division of labor, together with the late insertion of phonological material into the structure will turn especially convenient to explain the kind of crosslinguistic variation we find in the realization of -ble. On the other hand, the basic assumption of having all structure-building in a single syntactic component will be crucial in allowing word-building to be the result of the interaction of different modules of the grammatical system.3

2.1 Uncategorized roots In addition, roots are not specified for category, but are merged in the syntax with category-assigning functional heads (Marantz 1997), i.e. “the syntactic categories  N, V, A are morphological categories created by the syntax” (Marantz 1999a : 1). This means that roots do not encode any information as for the category N, V, or A they belong to. Their category is determined by the syntactic functional heads that appear in the local environment of the root. In other words, syntactic categories are not primitives, but they are abstract morphosyntactic features which are post-syntactically realized. For instance, a root √mod becomes verbal when it merges with the verbalizing head little v realized by the phonological exponent -ific, but it does not become a verb unless it is further c-commanded by the functional head(s) Tense/Aspect/Mood, as in modificaries ‘you would modify’; that is, it may be verbal but end up as a noun modificació ‘modification’

3 See also Sproat (1985) and Baker (1985, 1988) for syntactic approaches to morphology.



The framework: Distributed Morphology   

   5

or an adjective modificable ‘modifiable’. This means that the syntactic category of the root is determined only after all syntax has taken place, i.e. roots are postsyntactically realized. According to Marantz (1999b), traditional lexical word formation (SPE ‘+’ affixation or Siegel’s (1979) Class I affixes) involves the semantics of the root, but not its syntactic argument structure; it is a case of direct affixation to the root, so that it cannot involve the external argument, which is assumed to be introduced by a functional head (Kratzer 1996). Traditional syntactic word formation (SPE ‘#’ affixation or Siegel’s (1979) Class II affixation) means affixation outside of a category-assigning functional head. Hence, it can involve the external argument as well as some operation on argument structure. This distinction is exemplified in (2). Whereas (2a) is an example of direct root affixation, (2b) shows that affixation above a little v may involve the external argument introduced in [Spec, vP] (cf. Kratzer 1996). (2) a. root affixation

b. affixation above v

v/a/n affix

v/a/n √

affix

… vP Spec

v v



This difference also correlates with possible distinctions in interpretation. Whereas affixation above a functional head results in compositional meaning, root semantics may be non-compositional. In recent work by Marantz (2001, 2005) and Arad (2003) this difference is associated to Chomsky’s (1999) notion of phase. That is, the first category-assigning functional head would define a phase (Marantz 1999b), so that when the affix attaches directly to the root the structure is spelled-out semantically and phonologically. This can give rise to special semantic and phonological outputs. However, once spelled-out, all subsequent word formation above the first category-assigning head is completely regular and transparent. To illustrate this point, whereas the -ble adjective amable ‘kind’ (from √am- ‘love’) contrasts with acceptable ‘acceptable’ in that the former but not the latter is semantically opaque, the semantics of their corresponding abstract -itat nominalizations amabilitat ‘kindness’ and acceptabilitat ‘acceptability’ are both compositional, because they both are regularly built above the first category-assigning functional head. Thus, in this model, all word formation

6   

   Introduction

takes place in the same module, syntax, i.e. there is a single mechanism that generates structure.

2.2 Romance theme vowels in DM As is well-known, verbs in Romance languages are idiosyncratically classified as belonging to a particular conjugation class, which is encoded in the choice of the so-called theme vowel (th), a sintactico-semantically empty formative that appears next to a (simple or derived) root. Both Spanish and Catalan have three different conjugation classes, exemplified in (3) for Spanish. (3) a. mod - ific - a - r, cant - a - r √ v th inf √ th inf ‘to modify, to sing’

(1st conjugation)

b. palid - ec - e - r, llov - e - r √ v th inf √ th inf ‘to become pale, to rain’

(2nd conjugation)

c. sufr - i - r √ th inf ‘to suffer’

(3rd conjugation)

For the analysis of Catalan and Spanish, I assume Oltra-Massuet’s (1999, 2000) proposal that the verbal theme vowel is inserted in the morphology as a result of a well-formedness condition on syntactic functional heads, as represented in (4) and implemented in (5) for verbs, nouns and adjectives, after all syntactic and morphological operations have taken place. That is, this condition requires that all syntactic functional heads are adjoined a theme position at Morphological ­Structure. (4) a. At MS, all syntactic functional heads require a theme position.4

4 Williams (1981a) makes a similar proposal for Latin, whereby roots and tense markers bear lexical diacritics that determine the class of the following ‘connective vowel’. As noted in Oltra-Massuet (1999, 2000), this is a stipulation in Williams (1981a), while, in Oltra-Massuet’s analysis, the existence of multiple theme vowels follows from a general property of the language. See chapter 2 § 2.3 and 6.4.2 for additional information on theme vowels.



The framework: Distributed Morphology   

b. Fo

Fo Fo

th

a. Verbs (5) a. Verbs5

b. Nouns and b.  Nouns andAdjectives Adjectives T T

v v

T th

a

n

v √

   7

T

√ Agr

n n

a



th

a

th

th

The strongest argument in favor of this hypothesis is that it unifies the notions of nominal class marker and verbal theme vowel, and allows for an explanatory account of their stress system (see also Oltra-Massuet and Arregi 2005 for the application and extension of this analysis to Spanish). If we apply this proposal to the present object of study, the suffix traditionally referred to as -able/-ible must be reanalyzed, in Romance, as containing the ­idio­syncratic theme vowel that corresponds to the verbal conjugation, /i/ for the second and third conjugations, as in (6a)–(6b) or with root-based adjectives in (6c); /a/ for the first conjugation, as in (7a), with noun-based adjectives in (7b) or with root-based forms in (7c). This is here signaled as th.ble for illustration. From now on, since the theme vowel is always present in Romance -ble, I will omit it from the glosses. (6) a. témer, temible (Cat) fear.th.inf fear.th.ble ‘to fear, fearsome’ b. resistir, resistible resist.th.inf resist.th.ble ‘to resist, resistible’

5 I assume that Agr is adjoined to T at MS as a result of a well-formedness condition. Thus, Agr being a morphological head, it is not subject to (4). See Marantz (1991) and Oltra-Massuet (1999, 2000) for details.

8   

   Introduction

c. assequible achieve.th.ble ‘achievable’ (7) a. dissenyar, dissenyable (Cat) design.th.inf design.th.ble b. alcaldable mayor.th.ble ‘that can {become / be made /appointed} N’ or ‘candidate for N’ c. potable √pot.th.ble ‘potable’ Next, I turn to a brief revision of the status of deverbal formations, whose study has essentially concentrated on the analysis of participles.

3 Deverbal adjectives Deverbal adjectives pose interesting questions, because they may involve problems of morphology, syntax and semantics. On the one hand, they can be used as a basis to study how morphological processes influence the syntactic and semantic properties of the resulting adjective; or to investigate into the relationship between the syntactic subcategorization properties of the source verb and those of the resulting deverbal adjective. On the other hand, they manifest properties of both adjectives and verbs. Thus, exploring the morphological, syntactic, and lexico-semantic properties of deverbal adjectives may involve aspects of the study of categorization and category change, issues on conceptualization, or general features of the architecture of grammar, how it is constrained and the ways in which the different modules can interact.

3.1 Participles versus deverbal adjectives Most prominent among studies of deverbal adjectives has been the analysis of participles and their alleged status as intermediate forms between adjectives and verbs, as well as, to a much lesser extent, the investigation into the differences between participles and deverbal adjectives.



Deverbal adjectives   

   9

Indeed, the study of deverbal adjectives has very often concentrated on the study of participles. Since Wasow (1977) many, if not all, of these studies have focused on the distinction between verbal and adjectival participles in the grammars of languages, e.g. Levin and Rappaport (1986); Bosque (1989), (1990); Kratzer (1996); Embick (2001), (2004); Varela (2002), (2003) for -ed participles; Bennis and Wehrmann (1990); Borer (1990); Alexiadou (2005a) for -ing participles, to mention just a few. The label deverbal adjective is supposed to cover all kinds of adjectives derived from verbs: -ing and -ed adjectival participles, but also -ble, -ive, and other ­adjectival suffixes. However, not all of them show the same syntactic and ­semantic properties.6 According to Bosque (1999a: 287), the Spanish participle keeps or inherits the verbal properties listed in (8), illustrated with Bosque’s ­own examples.7 (8) a. indirect objects: libros devueltos a la biblioteca ‘books returned to the library’

b. PP complements: periodistas comparados con otros periodistas ‘journalists compared with other journalists’

c. adverbials: una edición reducida a la mitad ‘an edition reduced to the half’

d. agentive complements: una novela escrita por Cervantes ‘a novel written by Cervantes’

e. predicatives: un diputado elegido senador ‘a deputy elected senator’

f. two interpretations for preposed degree adverbs: un asunto muy estudiado ‘1. a lot studied; 2. studied many times’

6 There is an important difference between these suffixes: whereas -ble is still very productive, the other non-participle suffixes are fossilized in English. In Spanish there is a wider variety of suffixes that are still productive to different degrees: -ble, -dizo, -nt, -dor, -dero, -ivo. The latter two show a strong tendency to fossilization, according to Lang (1990 : 208–209). See Rainer (1999) for a complete list of deverbal suffixes in Spanish. See also Gràcia et al. (2000) for a description and analysis of Catalan and Spanish derivational suffixes. 7 Here Bosque leaves aside so-called participios truncos, which exhibit other properties. See Bosque (1989, 1990), Borgonovo (1999) or Varela (2002, 2003) for the various types of participles in Spanish, and Embick (2004) for different types of participles in English.

10   

   Introduction

Deverbal -ble adjectives can inherit properties (8c) and (8d), according to Bosque (1999a), although as will be discussed in chapter 2, some of them can exhibit the verbal properties (8a)–(8d),8, 9 even if with some restrictions that do not hold for participles. Those -ble adjectives that do not allow (all of) them seem to build a conceptually more or less homogeneous set. And still other deverbal adjectives do not license any of these elements. All in all, it seems that we have at least three different types of deverbal adjectives, which show various degrees of adjectivehood (or verbhood). On the one hand, participles can essentially display all verbal properties, as the Spanish examples illustrate. On the other hand, there are at least two types of -ble adjectives: those that can exhibit most of the verbal properties with some restrictions deriving from their internal structure, as will be shown in the next chapter; and those that display just a few or none of them. The latter are closer to derived adjectives with other suffixes in English, mostly -ive, -nt.10

8 I have found only a couple of examples of Spanish -ble with predicatives, as the one in (i). I will also show that their appearance in English with resultatives is also unclear. (i) tampoco es elegible Diputado a la Asamblea Nacional ningún empleado … neither be.3sg.prs eligible Deputy at the Assembly National no employee … ‘No employee is eligible Deputy at the National Assembly, either …’ (extracted from the Panamanian Constitution 1941, Title V, Article 81) 9 Throughout I use the label (W) to mark those examples that I have taken or modeled (generally by shortening) on actual sentences retrieved from Google. These have been confronted, in all cases, with native speakers’ judgments. 10 The situation in Spanish is more complex. The suffixes -nte or -dor display some verbal properties, notably the argument structure of the verb, as illustrated in (i)–(ii). See Laca (1993) for these suffixes in Spanish, and Cano Cambronero (in progress) for a recent syntactic approach to the internal structure of -nte and -dor. (i) una empresa distribuidora de piezas a company distributor of pieces ‘a parts distributing company’ (Lang 1990: 208) (ii) coincidente con coincident with Other suffixes can sometimes inherit the arguments of the corresponding verb, according to Bosque (1999), e.g. -ivo or -orio. (iii) gestos representativos de una actitud peculiar gestures representative of an attitude peculiar ‘gestures representative of a peculiar attitude’ (iv) actitudes atentatorias contra la libertad de expresión(id.) attitudes attempt.ory against the freedom of expression ‘attitudes that attempt against freedom of speech’ (Bosque 1999 : 238)



Deverbal adjectives   

   11

3.2 Participles in DM: Embick (2003, 2004) Turning to participles, there is an extensive literature on participles in the Distributed Morphology framework that will not be reviewed here (see e.g. Anagnostopoulou 2003; Alexiadou and Anagnostopoulou 2008; inter alia). For the purposes of my analysis of -ble, I will just concentrate on Embick’s (2003, 2004) three-way structural distinction for participles. According to Embick (2003, 2004), in addition to eventive participles, there are two types of stative participles. Resultative participial formations in (9b) differ from simple state participles in (9a) – which pattern with simple adjectives – in that the former contain a verbalizing v. Both types of participles are stative “in a broad sense”. However, whereas Asps defines a simple state, the resultative requires an Aspr that expresses a state resulting from a grammatically represented prior event, the head v. See also Alexiadou (2001) for an also ternary distinction between perfect participle (have written), passive participle (were written by) and adjectival participle (well-written) with similar structures, but differences in the featural specification of the different nodes. 9.

a. Stative

b. Resultative

The door was open

The door was opened Asp

Asp AspS

√ open

AspR ed

vP DP the door

v v

√ open

Both functional heads v and Asp contain “basic information about the eventivity/ stativity, along with further aspectual information about the status of the event or state.” This is essentially the basic structure that we find for the eventive passive, though with different features, because it is the locus of participial morphology (Embick 2003, 2004).

12   

   Introduction

(10) Eventive

Asp Asp

vP v [AG]

√P √ open

DP the door

This tripartite division between an eventive participle, a resultative participle and a stative participle will be taken as the basis for the analysis of -ble adjectives. The next section closes this chapter with an overview of this book.

4 Book overview This book presents a syntactic analysis of the internal constituent structure of modal passive adjectives, i.e. adjectives ending in -ble, that is essentially based on their syntactico-semantic properties, i.e. on their modal (middle-)passive content to be precise.11 More specifically, in chapter 2 I show that -ble adjectives are to be analyzed as containing different layers of functional structure that relate them to other modal (middle-)passive constructions. Despite interesting attempts to relate these adjectives to the middle (Gràcia 1992, 1995; Azkarate and Gràcia 1995; Gràcia et al. 2000), a number of issues remain unexplained, essentially because -ble adjectives do share some properties with middles, but at the same time they differ from them in a number of aspects. In chapter 2, I show that -ble adjectives cannot possibly have been built on a middle verb as has been proposed in the literature, though they share the same kind of circumstantial modality (Kratzer 1981) - and may share part of their syntactic configuration. On the other hand, I will agree with previous work relating the internal structure of

11 The label middle has been and is still used in the Romance tradition to refer to a number of constructions that are related because of the appearance of the clitic se. These include reflexives, reciprocals, ergatives, inherent reflexives, etc. In this book, I use this label to refer to the property reading we find in English middles of the type This book reads easily. See Kemmer (1993), Mendikoetxea (1999b), Bartra (2002).



   13

Book overview   

-ble adjectives to the passive (Chapin 1967; Kayne 1981, 1984; Williams 1981b; or Fabb 1984, inter alia), and will show that previously unexplained puzzles arising form the comparison of the two structures can be accounted for by appealing to the internal structure of -ble and its semantics. The claim is, thus, that only a syntactic approach to word formation can capture the fine-grained syntacticosemantic differences underlying these structures, in such a way that they can still be related in a principled way. I will pave the way for a development of a principled account of the relation between middles and passives and -ble, although a detailed contrastive analysis of the syntax-semantics of these constructions is a project in itself that will not be undertaken in this volume. Thus, the goals of this book are limited to providing a unified analysis of the various types of -ble adjectives we can find in the languages under study, namely Catalan, Spanish, English and also German. Different types of -ble adjectives have been mentioned in the literature, active (11a) or passive (11b); denoting possibility (12a), expressing some evaluative judgment (12b), or being causative (12c).12 (11) a. agreeable, perishable ‘that V’ b. modifiable, realizable ‘that can be Ved’ (12) a. modifiable, realizable ‘that can be Ved’ b. admirable, enviable

‘that should be Ved’

c. horrible, terrible

‘that causes N’

The difference in the type of modality displayed by the adjectives in (12) correlates with interesting contrasts. For instance, the examples in (13) allow the aspectual modifier to refer to the underlying event, i.e. (13a) expresses that the subject of the adjective has the property that it is possible to carry it out within a time span of an hour. (13) a. (És) una feina realitzable en una hora. be.3sg.prs a job realizable in an hour ‘(It is) a job realizable in an hour.’

(Cat)

12 According to Vendler (1968), (12b) and (12c) are grouped into the same set of causative adjectives (e.g. that causes envy or horror), which are different from those expressing possibility. Note, however, that they show a formal difference. There are no such verbs *to horr(or), *to terr(or), but to horrify and to terrify.

14   

   Introduction

b. (És) una feina realitzable a la primavera. be.3sg.prs a job realizable at the spring ‘It is a job realizable in the spring.’ On the other hand, the adjective in (14) is either anomalous, as in (14a), or it can only mean that a state holds at the time expressed by the temporal adjunct, i.e. (14b) means that in the spring, this town has a beautiful landscape, not that it is possible to admire the landscape in the spring.13 The examples are from Catalan, though similar facts hold in the other languages. (14) a. ??* (És) un paisatge admirable en una hora. (Cat) be.3sg.prs a landscape admirable in an hour b. (És) un paisatge admirable a la primavera. be.3sg.prs a landscape admirable at the spring ‘It is an admirable landscape in the spring.’ In chapter 2, I show that a syntactic approach to word formation allows us to distinguish between such a variety of syntactico-semantic types of -ble adjectives. I will argue that differences must be derived from the kind of functional heads that merge with the root, and their attachment site. Evidence for the internal syntactic structure of the various types of -ble adjectives will come from various sources, including their morphophonological properties and their external syntactic properties. Among the latter, mainly from an examination of their argument structure properties; more specifically, from the way they behave with respect to the expression of the arguments selected by their

13 The sentence is much better with an intonation change after the adjective, of the type found with right dislocation. Thanks to Eulàlia Bonet for clarifying this issue. Notice also the contrast in (i). Whereas (ii) predicates the property of being a beautiful landscape, attributing a job the property of being realizable sounds pragmatically weird, according to native speakers. Note that a modal sentential parallel would equally sound strange, as in (ii). (i) a. ?? A la primavera, és una feina realitzable. (Cat) at the spring be.3sg.prs a job realizable ‘In the spring, it is a realizable job.’ b. A la primavera, és un paisatge admirable. at the spring be.3sg.prs a landscape admirable ‘In the spring, it is an admirable landscape.’ (ii) ?? A la primavera, la feina pot ser realitzada. (Cat) at the spring the job can.3sg.prs be.inf realize.part ‘In the spring, it is a realizable job.’



Book overview   

   15

underlying verbal base. I will show that, in a sense, we can speak of different degrees of verbality, because the expression of verbal arguments interacts with the type and content of the functional heads present in the structure, thus providing further support for such a syntactic analysis. For instance, the examples in (15) show that the expression of the external argument is restricted by some notion of specificity or genericity, i.e. only those that are less specific seem to be correct. (15) a. una novela adaptable al cine por un buen guionista (Sp) a novel adaptable at.the cinema by a good scriptwriter ‘a novel adaptable to the cinema by a good scriptwriter’ b. * una novela adaptable al cine por Pablo a novel adaptable at.the cinema by Pablo Again, different verbs are constrained in different ways, as illustrated in (16), where not even a generic by-phrase can be licensed.14

14 The examples above show that this constraint on the genericity of the external argument is in principle irrelevant in the case of Subject-Experiencer psychological verbs, i.e. external arguments are banned with these verbs. However, I have found examples like (i)–(iii) that apparently contradict this assertion. (i)  la política comunitaria sobre Turismo “ni the policy communal (= from the European Community) on tourism “neither existe ni es deseable por la mayoría de los Estados” (W) (Sp) exist.3sg.prs nor be.3sg.prs desirable by the majority of the states ‘The [European] community policy on tourism does not exist and is neither desirable that it exists for most of the states’ (ii)  Detestable por la sociedad sería que un agente que observa la detestable by the society be.3sg.cond that an agent that observe.3sg.prs the transgresión de una norma … (W) (Sp) infringement of a norm ‘It would be execrable by society that an agent that observes the infringement of a norm …’ (iii) supera el nivel de monóxido de carbono tolerable por el ser exceed.3sg.prs the level of monoxide of carbon toleratable by the being humano (W) (Sp) human ‘it exceeds the level of carbon monoxide that can be tolerated by the human being’ Notice that the verb in (iii) is no longer used as a psychological verb, so that it would be analysed as containing a regularly derived high -ble adjective (hence toleratable and not tolerable in the English glosses). My speakers and I actually do not accept (i), and (ii) appears as very hard to process. In both cases there seems to be an implicit verb consider missing.

16   

   Introduction

(16) a. * una novela admirable por un buen editor (Sp) a novel admirable by a good editor b. * una novela admirable por Pablo a novel admirable by Pablo Examples (17) and (18) illustrate that the same type of constraint that restricts the appearance of the external argument seems to be at stake with respect to the expression of other arguments, as well. The restrictions on the expression of arguments with -ble adjectives and their semantic characterization will be discussed at length in chapter 2, where I will argue that they derive from general properties of the internal grammar of -ble that relates them to statives (Dowty 1979) by means of the presence of a modal (generic) operator that arbitrarily-binds the external argument, and defines a modal context. (17) a. Any good theory should be explainable to a barmaid.

b. *Any good theory should be explainable to John.

(18) a. *The glass is breakable into six pieces.

(Aronoff 1976)

b. Glass is fragile and easily breakable into sharp pieces. (W) Eventually, all these issues can be reduced to a major question: whether all these facts are characteristic of individual lexical units or whether they can be derived from some more general property, or from a cluster of properties. On the basis of well-known tests, I will discuss the main properties of -ble adjectives and will establish their internal constituent structure. The analysis develops a basic insight in Vendler (1968), who linked the rule of -ble adjectival formation to past participle formation by proposing two transformational rules, both based on passive content, that differ only in that one has a modal component whereas the other lacks modality. That is, informally, -ble adjectives are analyzed as modalized passive participles, which corresponds to the interpretation in (19).15 (19) that can be V-ed Mod Pass Part

15 More recently, Gómez Fernández (2006) also establishes the relationship between -ble adjectives and the passive participle. She notes that there are cases where Spanish has a -ble adjective, which have been conceptualized with a passive participle in French.



Book overview   

   17

On the other hand, the analysis also builds on Embick’s (2003, 2004) and ­Alexiadou’s (2001) dual analysis of participle forms within a syntactic approach. I will propose a similar distinction for -ble adjectives. That is, whereas an adjective like modifiable must be understood as predicating the possibility of reaching a resultant state (‘that can be modified’), adjectives like admirable express a modalized stative property, in the sense that in a sentence like They did an admirable job, the adjective admirable means something like ‘very good’ with no direct reference to an underlying active verbal component. As for the morphological properties of -ble adjectives, previous approaches, both syntactic or lexical, have already proposed that there are (at least) two types of -ble adjectives in English, a more idiosyncratic and often lexicalized one, e.g. edible, reparable, and another regular and transparent -ble, e.g. eatable, repairable. This has been considered to be indicative of the existence of two suffixes, a root-based -ble and a word-based suffix (Aronoff 1976). For syntactic approaches, regular formation would happen in the syntactic component, whereas non-transparent -ble adjectives would be created in the lexicon (Chapin 1967; Kayne 1981; Fabb 1984; among others). This duality has been translated into the Distributed Morphology framework as a difference on the level of syntactic attachment: word formation at the root level, which permits special behavior of derived words, e.g. with respect to stress assignment or non-compositional meanings, versus word formation above the first category-giving functional head, with totally transparent results (Marantz 1999. See also Arad 2003). This proposal has been implemented for -ble adjectives by Volpe (2005), who suggests that whereas regular -ble adjectives contain a verbalizing head, this is absent in those lexicalized -ble adjectives.16 While this analysis might a priori be valid for a language like English, I show that it is problematic for Romance languages like Catalan or Spanish, which display a more complex morphology. Specifically, all -ble adjectives in Catalan, whether transparently derived from a verbal base or not, display the so-called verbal theme vowel. Assuming the Distributed Morphology analysis of theme vowels proposed in Oltra-Massuet (1999, 2000), I show that Volpe’s account of the internal structure of these adjectives is insufficient for the language-particular requirements of Catalan or Spanish. Moreover, I argue that the structure he proposes cannot capture all syntactico-semantic

16 There are two other Distributed Morphology works dealing with -ble adjectives. I came across Nevins (2002) right at the point of completing the dissertation this book is based on; and discovered McGinnis (2010) quite recently, in 2012. As their discussion parallels mine with respect to the distinction of two -ble adjectives in English that is based on their syntactic argument structure properties, I will incorporate their findings as I proceed, given that, except for my awareness of Volpe (2005), all of us had independently worked on -ble adjectives.

18   

   Introduction

properties that are displayed by this type of adjectives across different languages. The semantic interpretation of -ble adjectives suggests that they encode a far richer (syntactico-semantic) internal structure than the morphology of languages like Catalan/Spanish (Romance) or English/German (Germanic) can capture. We expect that morphologically richer languages like Hungarian or Turkish can offer some insight into the identification of the underlying components of -ble adjectives. Based on a number of morphophonological and syntactico-semantic tests applied to English and Romance, I propose the two basic structures in (20) and (21). On the one hand, potential or high -ble adjectives are those that are regularly derived from a transitive verb, such as modifiable or translatable. These adjectives show eventive properties in that they can license a number of verbal-related phrases. Their configuration, thus, contains more functional structure. Specifically, there is a passive little v head that introduces eventivity by categorizing the root as verbal. This head can also license a by-phrase and other agent-related elements, even though it cannot have an explicit agent in [Spec, vP].17 This structure is further c-commanded by an AspP, which quantifies over event variables (Hacquard 2009). Since it is a resultative AspP (e.g. Embick 2004), it creates a resultant state out of a prior event, which is then selected by a possibility operator in ModP that creates a modal (generic) environment. The whole configuration finally merges with a little aP that turns the eventive structure into a property. This is summarized in (20). (20) High -ble: It expresses a generic property according to which it is possible for some originator to achieve a resultant state, e.g. Catalan modificable, traduïble, publicable ‘modifiable, translatable, publishable’ aP a

it creates a property

Mod ◊

creates a modalized resultant state with a circumstantial modal base

ModP AspP Rc

Aspres v [pass]

creates a resultant state out of a prior event vP

event interpretation / implies external argument √

17 My analysis follows proposals in Chomsky (1995), Marantz (1997) or Harley (1995, 2005) that assume that a single projection vP serves two functions, to verbalize the root and to introduce the external argument, instead of assuming a v/Voice split structure (Kratzer 1996).



Book overview   

   19

On the other hand, low -ble adjectives are those that show idiosyncratic properties, be these semantic or morphological. In this group, we may find adjectives expressing a modality that differs from the regular possibility found with high -ble adjectives, as in deplorable or admirable. We can also find adjectives showing some morphophonological irregularity, such as a stress shift, as in to compáre – cómparable,18 or root allomorphy, as in to apply – applicable. These irregularities are accounted for by having an uncategorized root merged with a stative aspectual head and a modal head before being finally categorized as an adjective. That is, this corresponds to the process of root-attachment by which a root may individually negotiate its meaning with some functional head, before it is categorized. In this case, the stative root can arrange its meaning with the possibility modal before little a, a category-giving functional head, categorizes the root. (21) Low -ble: It expresses a (sometimes idiosyncratically) modalized property, e.g. admirable, cómparable, applicable aP a

it creates a property

Mod ◊

modalizes a stative root

ModP AspP Rc

Aspstat

stativizes the root √

I further assume a single underspecified Vocabulary item /bl/ in Romance with a contextual information as for its context of insertion specifying that it is inserted in a little a node c-commanding a ModP, possibly specified as a possibility modal, as in (22). The underspecification of this item will ensure that it is inserted in all cases of low or high attachment of -ble. (22) [bl] ↔ a / Mod For English, given that both -ible and -able are pronounced exactly the same [əbəl] (Wells 2000; Roach et al. 2006), I would like to claim that there is a single vocabulary item, even though one must note that low attachment -ble may have to be linked to a list of adjectives spelled as -ible in writing.

18 Accent is marked in English for clarity whenever necessary.

20   

   Introduction

After establishing the internal syntax of -ble in chapter two on the basis of its external syntax, its semantics and its morphophonological properties, the next two chapters are devoted to the analysis of two special -ble constructions in Romance. Chapter 3 presents an interesting and challenging contrast found in Spanish and illustrated in (23)–(24), where an unaccusative verb ocurrir ‘happen’ and a purely intransitive verb dormir ‘sleep’ have a corresponding -ble form that can only appear in a particular type of construction with the structure V todo lo V-ble. That is, even though there are no -ble adjectives *ocurrible ‘happenable’ or *dormible ‘sleepable’, these forms are systematically created to appear in particular environments of the type V todo lo Vble. (23) a. * un accidente ocurrible (Sp) an accident occur.ble b. Ocurrió todo lo ocurrible. occur.3sg.pst all lo occur.ble ‘It happened everything that could happen.’ (24) a. * una siesta dormible a siesta sleep.ble b. Durmió todo lo dormible. sleep.3sg.pst all lo sleep.ble ‘She slept as much as one can sleep.’ In both cases, the semantic interpretation of the derived forms is essentially the same as that of any deverbal -ble adjective. I show that these striking data pose a new challenge for theories of morphology, as they are a priori ungrammatical and therefore impossible words. I argue that these data cannot be explained in a model based on the lexicon as a generative component. If words are built in the lexicon prior to insertion in the syntax, it is unclear why apparently non-existing adjectives like *ocurrible ‘happenable’ can be created in the lexicon but cannot be used as such, i.e. they are prevented from modifying an underlying object as is generally the case with -ble adjectives, e.g. un amor perdurable ‘a long-lasting love’. Likewise, if unaccusative and purely intransitive verbs are not admitted as bases for -ble derivation, forms like ocurrible ‘happenable’ or dormible ‘sleepable’ in (23b) and (24b) should not exist at all.19 Such a situation is in fact anticipated

19 There is an additional problem here, namely the existence of -ble adjectives derived out of a few unaccusative bases (cf. durable ‘lasting’, perdurable ‘long-lasting’ but *ocurrible



Book overview   

   21

in a syntactic model like Distributed Morphology, where the only component that can generate structures is syntax proper and words are phonologically realized in the post-syntactic component, i.e. a model that predicts the existence of words that result from the interaction between syntax, morphophonology, semantics, and language-particular rules. In the analysis of these forms, I establish an interesting and previously unnoticed correlation between Spanish, in (25b), and a construction of alleged cognate objects with unaccusative verbs in English of the type in (25a). (25) a. The tree grew a century’s growth within only ten years. (Kuno and Takami 2004) b. El árbol creció todo lo crecible en tan solo diez años. (Sp) the tree grow.3sg.pst all lo grow.ble in so only ten years ‘The tree grew to its highest possible degree of growth in only ten years.’ Such unaccusative cognate object constructions are especially problematic, given the well-established generalization that only unergatives can take a cognate object (e.g. Keyser and Roeper 1984; Levin and Rappaport-Hovav 1995). I present evidence against previous analyses of the English construction that regard it as an example of a standard argumental or adverbial cognate object structure and suggest a joint treatment of a century’s growth and todo lo crecible as degree or measure phrases (cf. Ojeda 1984; Gutiérrez-Rexach 1999, 2009 on degrees). Specifically, I argue that unaccusatives do not license standard cognate objects, but the cognate elements in (25) essentially correspond to the overt materialization of the otherwise implicit abstract scale with these verbs (Piñón 2000, Kratzer 2002, Rappaport Hovav 2008) that is used to evaluate their progress (Tenny 1994), so that the event can establish a relation of measurement with its theme argument, by assigning some value to it on that scale (e.g. Kennedy 1999; Kennedy and McNally 1999; Kennedy and McNally 2005). The function of this DegP is similar to Kennedy and Levin’s (2008) proposal that an explicit extent phrase with degree achievements plays the role of an incremental theme in that it determines the telicity of the predicate. I present a number of predictions stemming from the idea that the cognate object in the examples above is the overt manifestation of a degree argument that expresses the resultant extent of the event denoted by the

‘occur-ble’, *llegable ‘arrive-ble’). To my knowledge, there is no principled account of the contrast among unaccusative bases with respect to -ble derivation. I will deal with this topic in chapter 3, where I will suggest an explanation for this distinction.

22   

   Introduction

verb. Among other things, I analyze the nature of this DegP with verbal predicates, how we can account for its cognate nature, and their internal structure. In short, I show how syntax and semantics conspire with morphology to the effect of building such a priori aberrant forms. Chapter 4 deals with another set of apparently deviating forms, productively derived denominal -ble adjectives in Romance of the type in (26) in Catalan. (26) alcaldable, bisbable, ministrable, papable, presidenciable, rectorable (Cat) mayor.ble, bishop.ble, minister.ble, Pope.ble, presidency.ble, rector.ble ‘that can {become / be made /appointed} N’ or ‘candidate for N’ These data raise a number of questions related to the absence of the corresponding verbs, and the special temporal-aspectual properties of the relatively reduced set of nouns that make them appropriate as -ble bases. Additional problems have to do with the source of crosslinguistic variation, i.e. the absence of productive denominal -ble formation in English. If we take as a starting point the assumption that -ble formation targets essentially the same kind of roots for regular deverbal adjectival derivation in English and Romance, it is striking that they do not behave the same in this respect. The question is whether this divergence must be included in the specification of the context of insertion of the Vocabulary item -ble, or whether there is some feature of these nominal bases that varies across languages. Crucially, these data are also problematic for lexicalist approaches, since, as suggested in Val Álvaro (1981) the suffix behaves in exactly the same way with both nouns and verbs, both semantically and syntactically. I show that a syntactic model like Distributed Morphology can deal with such cases, since they result from the interplay between the syntactic features of -ble described in chapter 2 and the syntactico-semantic properties of post nouns. Crosslinguistic variation arises from differences in the syntactic behavior of the base nouns with respect to their event-temporal properties, and the semantic status of the lexical class of social roles across languages. Specifically, I develop the idea that the divergent behavior of -ble affixation in the nominal domain in English and Romance is related to a more general feature of the lexical class posts are a subset of: their ability to become the head of a bare NP. Romance post-nouns, like all other nouns within the lexical class of social roles, can be used either referentially or predicatively, as has been proposed in Bosque (1989, 1996), and more recently in Alexiadou (2005b), Beyssade and Dobrovie-Sorin (2005), or Depréz (2005), whereas in English uniquely identified posts like president must be analyzed as concealed definites, as advocated in Munn and Schmitt (2005) or Winter (2005). This trans-



Book overview   

   23

lates as a crucial distinction in the internal structure of the bases in the different languages, so that only in Romance the root can license the relevant functional heads (v and Asp) that allow -ble insertion. Chapter 5 summarizes the main contributions of this thesis and points to remaining issues and future prospects. It also includes additional information and predictions on crosslinguistic variation. More specifically, I suggest two main directions for future research on degree cognate objects that relates them to the resultative construction and to the so-called amount relatives. I further identify a new theoretical problem that the status and realization of degree or measure phrases as incremental themes pose for aspectual theories. As for Nble, on the one hand, I point to some predictions of my proposal and suggest an analysis of the intralinguistic variation exhibited by the corresponding German -ble suffix that provides strong support for the Distributed Morphology model. Besides, I point to some problematic issues that an extension of the analysis to other languages like Italian and Greek may involve. Thus, the next pages offer a micro-comparative piece of research into the behavior and general syntactico-semantic properties of morphologically complex -ble adjectives, and how these interact with the syntax and semantics of the configurations in which they appear.

2 -BLE 1 Introduction The main goal of this chapter is to describe the main properties of adjectives ending in -ble, mainly on the basis of data from English, Catalan, and Spanish, but also incorporating German at various points. The fact that -ble adjectives have been studied at length is not surprising once we look at the general properties that they display in the different languages. These are listed in (1). (1) a. They are deverbal – essentially formed from transitive verbs.

b. Their subject corresponds to the related verbal direct object.



c. They are interpreted as passive.



d. They express potential modality.



e. They are stative, i.e. they express a (non-trivial) property.



f. They express a generic statement.

These properties refer basically to regularly and transparently derived -ble adjectives of the type illustrated in (2) for different languages. (2) a. recomanable ‘advisable’

(Cat)

b. doable (Eng) c. trinkbar ‘drinkable’

(Ger)

d. divisible ‘divisible’

(Sp)

However, there are some -ble adjectives that do not exhibit one or more of the properties in (1). A first problem we are faced with is to establish whether there is just a single suffix -ble or we need to postulate more than one. Some adjectives are apparently built on non-verbal bases (violating (1a)), so that their subject may not correspond to the direct object of some underlying verb (violating (1b)). For instance, Cat. alcaldable ‘mayor.ble’ has no corresponding verbal base *alcaldar ‘to mayor’, but a noun alcalde ‘mayor’; or Eng. credible has no corresponding verbal base *to cred. Moreover, some -ble adjectives do not seem to convey a passive meaning (1c), e.g. Cat. agradable ‘agreeable’, or Eng. perishable; and still

Introduction   

   25

others may apparently express a different type of modality, necessity rather than possibility, as in admirable or deplorable in English, Catalan, or Spanish, or they have been said to lack the expected modality altogether, e.g. suitable. Properties (1e–f) are closely interrelated, and they seem to be challenged by the argument structure possibilities exhibited by -ble adjectives, which argue for some degree of eventivity. Hence, the first thing that needs to be done is to clarify all these issues by providing a thorough description of -ble adjectives, their morphological properties, their external syntax as well as their semantic properties. While describing the main properties of -ble adjectives, I will review the most relevant aspects and contributions of previous approaches, pointing at their strengths as well as their weaknesses. This should allow us to tackle the analysis of these morphologically complex forms in the last part of this chapter.

1.1 Main claims and chapter overview The basic assumption underlying this research is that -ble adjectives share a basic configuration in the languages under study, crosslinguistic differences being located in the vocabulary, e.g. the availability of various phonological exponents that realize a single morpheme or syntactic position, as in German, where three different pieces, {-bar, -abel, -wert}1 compete for insertion into a -ble syntactic morpheme, or just stemming from language particular constraints or requirements, such as the presence of a theme vowel in Romance. Sections 2 to 4 are devoted to the description of the main characteristics of -ble adjectives and, together with the revision of previous approaches in § 5, provide empirical support for the analysis of their internal constituent structure developed in section 6. I start in § 2 with the main morphological characteristics of English -ble that have led linguists to distinguish between two types of -ble, a lexicalized or low -ble and a regular, transparently derived high or potential -ble. These two types of -ble are assessed against the Catalan and Spanish data, and I claim that the kind of morphological complexity shown by the corresponding lexicalized -ble adjectives calls for a more complex structure.

1 A full analysis of the German data is beyond the scope of this volume, since it would imply the internal syntax of -wert, which involves still unsettled issues on its status – as a compound or as a derivational suffix, the function of the linking morpheme -s- (so-called Fugenmorphem), as well as considering other semantically related and thus competing suffixes, such as -fähig and -lich. Throughout, I restrict my discussion to the three suffixes listed in the main text, which will be argued to stand, at least to some extent, in complementary distribution.

26   

   -BLE

Section 3 considers the empirical evidence related to the type of base showing that -ble imposes two requirements on its base: (i) it must have an internal theme argument and (ii) (at least the implication of) what I will call an initiator or originator, following Borer’s work, which must be understood as agent, cause, experiencer or even natural force. The next subsections in § 3 are devoted to the description of the main syntactico-semantic properties of -ble adjectives, which also serves the purpose of providing a revision of some aspects of the most relevant analyses of -ble. First, I consider the external syntax of -ble adjectives. Based on well-established tests mainly related to the expression of the underlying verb’s arguments I conclude that there are two syntactic types of -ble adjectives that parallel the morphologically-based partition discussed in § 2. That is, the empirical evidence argues for the presence of passive and aspectual structure in high -ble but not in low -ble. On the other hand, the examination of the data also supports the presence of a possibility modal component in -ble that interacts with its passive structure. § 4 is devoted to the semantics of -ble adjectives, showing that there are two main interpretations of -ble that also correlate with the division into high and low -ble. Whereas high -ble exhibits a regular potential meaning, the meaning of low -ble adjectives is not homogeneous. The description of -ble adjectives closes with a brief revision of the main previous approaches to -ble in § 5 and summarizes the main morphological, syntactic and semantic properties of -ble. In particular, section 5.1 is concerned with proposals that treat -ble adjectives as being built on a passive structure. I show that the empirical problems they pose are only apparent, so that striking differences between the two constructions can be accounted for by appeal to general properties of the grammar of passives, statives, and resultatives. Having established that it is correct to assume an underlying passive configuration for -ble adjectives, I proceed to show that an analysis of -ble as being lexically derived from a middle verb creates a number of empirical problems, in 5.2, even though I agree with previous work that middles and -ble share a number of properties that would be worth exploring in future research. Section 5.3 reviews the treatment that -ble adjectives have received in the Distributed Morphology model. I move on to the analysis of these forms in section 6. I assume the basic division between what has been called root attaching -ble and an above-the-firstcategory-head attachment -ble that has been proposed in the literature since Marantz (1997). However, the empirical evidence shows that in both cases there is more structure than meets the eye. I propose that there is an underspecified Vocabulary item that is in all cases inserted in a node little a that stands in a local relation of c-commanding to a possibility modal head, i.e. -ble is specified for

Introduction   

   27

insertion in little a in the context of a possibility modal. The structure must have an internal theme available and the – syntactic, semantic, or conceptual – implication of an originator. The main difference among -ble adjectives lies in whether the originator is syntactically present, because we are dealing with a modal state resulting from a syntactically present previous event; or it is just conceptually implied because we are concerned with a non-eventive configuration, i.e. with a modal state. The two configurations are made explicit in section 6.1, where I show that this analysis captures intralinguistic variation, and can also accommodate morphologically more complex languages. Section 7 summarizes the main contribution of this chapter and points to some remaining puzzles. Before proceeding, two caveats are in order. First, it is not the purpose of this chapter to survey and fully review the extensive literature on -ble adjectives. Rather, building on their main insights I develop an analysis of the internal syntax of -ble adjectives that can suit the English and Romance data – and also German, and can serve as a basis to analyze the derivation of the special cases discussed in chapters 3 and 4. Second, -ble adjectives can be, and have actually been semantically associated to other structures. On the one hand, Bosque (1999:  255) already suggests that -ble adjectives may also be related to the so-called tough-construction, exemplified in (3), for they are both interpreted as passive and modal. Indeed, the tough-construcion has already been related to middles in Massam (1992), and middles have been associated with -ble adjectives in Gràcia (1992, 1995). (3) This book is easy to read (cf. This book is {readable/reads well}) On the other hand, Mendikoetxea (1999b) distinguishes between the three seconstructions in (4) on the basis of a variety of tests applied to Spanish: an impersonal se-construction in (4a), a se-passive in (4b) and a middle-se in (4c). (4) a. Se agasajó a los invitados. (Impersonal se) (Sp) se fête.3sg.pst to the guests ‘The guests were treated kindly.’ b. Se pasaron los trabajos a ordenador. (Se passive) se pass.3pl.pst the works to computer ‘The assignments were computer-typed.’ (Mendikoetxea 1999b : 1635)

28   

   -BLE

c. Esta camisa se lava muy bien con lejía. (Middle se) this shirt se wash.3sg.prs very well with bleach ‘This shirt washes well with bleach.’ (Mendikoetxea 1999b : 1641) However, according to Bartra (2002), there are no strong formal arguments to distinguish middles from the other pronominal se-constructions with nonspecific subjects in Catalan. For Bartra, the middle is in fact an interpretation that obtains from the interaction of a bunch of semantic properties. As the relation between -ble and middles has already been established in the literature, this will be briefly reviewed in § 5.2, though only to the extent that both structures share a cluster of properties – probably the same cluster of properties that Bartra attributes to the middle interpretation of a Romance passive se. Though tempting and challenging, a project to examine the formal relations between -ble, tough-construction, Romance se-passives and English middles lies well beyond the scope of this book. There is an extensive literature on these constructions, and many issues are still under debate, so that more work is needed to understand each of these constructions individually before a unified treatment can be undertaken.

2 The morphology of -ble The morphological properties of -ble have received extensive attention in English, as in Aronoff (1976), Anderson (1992), Steriade (1999), or Nevins (2002), and to a much lesser extent in the other languages that are the object of this study, e.g. Gràcia (1995) for Catalan, Val Álvaro (1981) for Spanish, or Riehemann (1993) for German. In this section, I concentrate on aspects of the morphology of -ble mainly in English and Romance, although I will show that similar facts can also be found in German.

2.1 Aspects of the morphology of English and German -ble As described in Anderson (1992:  186–195), who follows Aronoff (1976:  121–129), different morpho-phonological issues are involved in the derivation of -ble in English. There are minimal pairs that differ in the position of stress, illustrated in (5)–(6) with orthographical accent for the sake of clarity, which sometimes correlates with a difference in meaning, as illustrated in (7). Interestingly, semantically transparent forms correspond to those with regularly assigned stress, i.e.



The morphology of -ble   

   29

the same pattern of stress we find in the related verbs, whereas semantically non-compositional forms may have stress patterns that are unrelated to the corresponding underlying verb. For instance, according to Aronoff (1976: 128), while the meaning of (7a) is equivalent, this special meaning is absent from (7b), which has only the expected deverbal meaning capable of being compared.2 (5) a. préferable ← *préfer b. preférable ← prefér (6) a. réparable ← *répair b. repáirable ← repáir (7) a. cómparable = ‘equivalent’ b. compárable ← compáre When considering verbs derived through affixation, e.g. -ate verbs, it turns out that there are -ble adjectives that do not seem to be created on an existing -ate verb, but they are built on the underived form, i.e. the root, as in (8), whereas others do show the -ate suffix, as in (9). In other cases, both forms seem to coexist, as in (10). (8) a. exculpate – exculpable – *exculpatable b. vindicate – vindicable – *vindicatable (9) a. debate – *debable – debatable b. relate – *relable – relatable

2 See also Steriade (1999) for an analysis of the contrastive stress pattern in (i) which is argued to be an effect of lexical conservatism, a kind of output-output constraint, whereby “only an independently existing stress pattern from the same paradigm can replace that of the verbal base” (Steriade 1999: 176–177). Whereas the base rémedy has an additional allomorph appearing in words like remédial, párody has no second allomorph – forms like paródiable or paródic being unknown to speakers (Steriade (1999: fn4). (i) a. rémedy – remédiable b. párody – párodiable Nevins (2002) uses the contrast in (i) as additional evidence for the two syntactic attachment sites approach advocated for in the Distributed Morphology model.

30   

   -BLE

(10) a. demonstrate – demonstrable – demonstratable b. navigate – navigable – navigatable c. separate – separable – separatable (Aronoff 1976 : 124–125) Sometimes the optionality is only apparent, and the two forms exhibit different meanings. Crucially again, it is the shortened form that may display a more lexicalized meaning, as illustrated in (11)–(12).3 The suffixed -ate form will always express a regular potential meaning. (11) a. tolerable = ‘moderately good, fair’ b. toleratable = ‘capable of being tolerated’ (12) a. appreciable = ‘substantial’ b. appreciatable = ‘capable of being appreciated’ (Aronoff 1976 : 128) Another morphological issue concerns root allomorphy.4 All cases that exhibit allomorphy on the root correspond to marked Latinate roots showing the same type of allomorphy that they display before other Latinate suffixes like -ion, -ive, -ory, and -or, as illustrated in (13). In some cases, (14), the allomorphy rule is optional, so that we find at the same time both regular and allomorphic forms.

3 Notice that there are cases like eatable or readable that, as observed in Lyons (1977: 531), add to their regularly derived meaning some notion of absence of displeasure, which the idiosyncratic allomorphic forms edible and legible lack. The same facts apply to pairs like menjable ‘eatable’ and comestible ‘edible’ in Catalan or comible ‘eatable’ and comestible ‘edible’ in Spanish (see Fábregas 2011 for the semantics of the latter). However, the latter – which expresses the expected regular meaning – and not the former would be analyzed as irregular, for it has a morphologically bound root in the three languages. Lyons (1977: ­531– 534) treats adjectives like edible and legible as morphologically and syntactically regular with respect to the transformationalist -ble rule, and leaves only a few words like horrible, knowledgeable or reasonable as listed lexemes. 4 In the Distributed Morphology model, lexical roots showing allomorphy are derived through readjustment rules; if allomorphs are phonologically unrelated, the system treats them as contextually determined exponents of a functional morpheme that compete for insertion in a deterministic manner. I will not be concerned with the phonological derivation of these forms here. (See e.g. Halle and Marantz 1993; Marantz 1995, 1997; Embick and Halle 2005; Embick and Marantz 2008; Embick 2010; Bonet and Harbour 2012.)

The morphology of -ble   



(13) a. apply – applicable

   31

(cf. applicative)

b. multiply – multiplicable (cf. multiplication) (14) a. defend – defensible – defendable b. extend – extensible – extendable c. perceive – perceptible – perceivable (Aronoff 1976 : 128) Optionality may also be fake in these cases, as illustrated in (15a)–(15b), so that a sentence such as (15c) makes perfect sense. In this case, the -ble adjective selecting the special allomorphic root percept- receives an idiosyncratic interpretation, whereas the regularly derived -ble adjective has the expected modal passive meaning that can be Ved. Notice also that in all examples (13)–(15) the allomorphic adjectives take the more specific and non-productive suffix -ible, whereas transparent forms are built by attaching the English regular and productive -able suffix.5 (15) a. perceptible ‘appreciable, large enough to matter, significant’ b. perceivable ‘capable of being perceived’ c. A flaw may be perceivable, even if it is not perceptible. (Aronoff 1976 : 128) As pointed out in Anderson (1992 : 189), we also find -ble adjectives in English that do not correspond to any existing verbal base, listed in (16a), though they have the syntax and semantics of regular -ble adjectives. That is, they are passive and convey the potential modal meaning represented in (16b). Also Lyons (1977­: 530–534) does not consider them to be opaque, since their interpretation involves all the components characteristic of -ble adjectives. (16) a.  capable, credible, eligible, flexible, ineffable, inexorable, intelligible, possible, potable, probable, tangible b. that can be Ved

5 As I mentioned in the first chapter, -able and -ible are actually pronounced the same, [əbəl], which argues for a single Vocabulary item. However, I follow the tradition of taking this difference as evidence for the distinction between two types of -ble, since there is no single -ible adjective in English with the external syntactic properties of regular -able.

32   

   -BLE

Still, a few others cannot be related to any existing base and their meaning is totally idiosyncratic, as illustrated in (17). (17) formidable, hospitable Finally, there are some forms whose base is categorially inappropriate, in the sense that, assuming that -ble adjectives are deverbal, they have not been built on a verb, but they do contain a recognizable base. (18) comfortable, horrible, objectionable, peaceable, reputable Based on the above facts, Aronoff (1976:  48), following proposals in SPE, distinguishes two different affixes -able in English, a [+ abl] suffix with a weak morpheme boundary, and a [# abl] suffix with a strong word boundary, which would account for all the previously stated special morphological distinctions. This division has essentially remained in all subsequent work in morphology in order to account for the dichotomy between idiosyncratic forms and transparently derived adjectives (see Siegel’s 1979 Class I and Class II affixes; Selkirk’s 1982 Root-type affixes versus Word-type affixes; or the most recent Nevins 2002; Volpe 2005; Anagnostopoulou and Samioti 2009, 2010; or McGinnis 2010 above and below v within the Distributed Morphology framework). For instance, cases of truncation or with stress shifting, i.e. with stress different from the related verb, are cases with the [+ abl] affix. All regularly derived -able adjectives would be examples of [# abl] affixation.6 Interestingly, as noted in Aronoff (1976: 125), -ble adjectives target a different negative allomorph. Whereas [+able] selects /iN/, [# able] chooses /un/, as exemplified in (19). (19) a. inaplicable, inextensible, imperceptible, indivisible, incapable, incredible b. unextendable, unperceivable, undividable, untranslatable, unmodifiable To close this subsection, let me briefly mention that we can find similar distinctions among existing German -ble adjectives, thus suggesting that there are also two morphological classes of -ble adjectives in this language. As indicated in ­Riehemann (1993), there are cases of truncated -ig- contrasting with non-truncated forms, as in (20a); examples of -ble adjectives with inappropriate bases but

6 Note that this means that in Aronoff’s morphology model, there are in fact three different affixes, i.e. a denominal one, N # abl, and “two suffixes, + abl and # abl” (Aronoff 1976: 122).



The morphology of -ble   

   33

with the same modal passive meaning, as in (20b), such as non-existing verbal bases *abdingen, nouns like Sicht ‘vision’, or non-transitive verbs such as über etwas verfügen ‘have something at one’s disposal’; and still others that, according to Riehemann (1993), bear no relation to the other -ble adjectives, i.e. they are not built on an existing verb and do not express potential modality, as in (20c). See Riehemann (1993) for further details. (20) a. entschuldigen, entschuldbar vs. rechtfertigen, rechtfertigbar (Ger) excuse.inf excusable justify.inf justifiable b. unabdingbar, sichtbar, verfügbar indispensable visible available c. unmittelbar, offenbar immediate(ly) obvious(ly) In addition, there are cases of truncation where instead of the default German suffix -bar we get the specialized suffix -abel. The latter is generally described as a non-productive suffix traditionally reserved for bases of Latin origin (Fleischer and Barz 1995) with exactly the same potential meaning as the productive -bar, which shows no morphological restrictions. So, we can currently find doublets such as the ones in (21), which on the one hand, parallel the English cases of truncation, where we find doublets such as separable – separatable; on the other hand, they also parallel the English distinction -ible / -able. Like English -ible, German -abel is specified for appearing with a particular set of roots.7 (21) a. akzeptieren – akzeptabel – akzeptierbar accept.inf acceptable acceptable

(Ger)

b. diskutieren – diskutabel – diskutierbar debate.inf debatable debatable

2.2 Aspects of the morphology of Catalan and Spanish -ble Let us now turn to the morphological properties of the corresponding adjectives in Romance. In Catalan or Spanish there is no variation in the assignment of stress, nor cases of truncation. We find that, as in English, the same allomorphs

7 But see the discussion in section 6.4.1 and in chapter 5.

34   

   -BLE

that appear with other suffixes come out before -ble, as illustrated with Catalan in (22). Same facts obtain in Spanish, as shown in (23).8 (22) a. destruir – destructible – destructiu destroy.inf destroyable destructive

(Cat)

b. dividir – divisible – divisor divide.inf divisible divisor c. percebre – perceptible – perceptiu perceive.inf perceptible perceptive (23) a. acceder – accesible – acceso access.inf accessible access

(Sp)

b. admitir – admisible – admisión admit.inf admissible admission c. permitir – permisible – permisivo permit.inf permissible permissive Forms like percebible ‘perceivable’ are not possible in Catalan. However, forms like percibible ‘perceivable’ or dividible ‘dividible’ are not only possible in Spanish, but they are in fact the only ones admitted in the construction V todo lo Vble analyzed in chapter 3.9 Interestingly, some of these special forms are morphologically related to the associated irregular participle affix of the underlying verb in Catalan, as in (24). (24) a. admetre – admès – admissible admit.inf admit.part admissible

(Cat)

b. fondre – fos – fusible melt.inf melt.part meltable

8 Note that in Romance, unlike in English, there is no direct correlation between -able and -ible. Second and third conjugation theme vowels are neutralized in the context of -ble – as is often the case – so that there are cases of regular -ible, e.g. Spanish atribuible ‘attributable’ or enriquecible ‘enrichable’, and cases of -ible being affixed to allomorphic stems, e.g. admisible ‘admissible’ or visible ‘visible’. 9 See also Fábregas (2011) for an analysis of the root allomorphy appearing in alternations such as perceptible – percibible ‘perceptible’, visible – veíble ‘visible’ in Spanish within the syntactic framework advocated in Ramchand (2008).



The morphology of -ble   

   35

Although only some adjectives exhibit this type of allomorphy, I will connect it to the presence of some past-participial-related head in the constituent structure of -ble adjectives in section 6.4.3.10 As in English, we also find -ble adjectives in Catalan and Spanish that do not correspond to any existing verbal base, exemplified in (25)–(26), but whose meaning involves the semantic components that characterize these adjectives, i.e. they are opaque only with respect to their form, but not with respect to their meaning. (25) a. assequible but *assequir ‘attainable’ (= that can be achieved)

(Cat)

b. intel·ligible but *intel·ligir ‘intelligible’ (= that can be understood) (26) a. potable but *potar ‘potable’ (= that can be drunk)

(Sp)

b. fungible but *fungir ‘fungible’ (= that can be consumed) Likewise, there are a few really opaque adjectives, exemplified in (27) with Catalan/Spanish examples, which cannot be related to an underlying verb, or to any other forms in the language.11 They contain neither a passive meaning nor a modality component.12 (27) a. *afar, *formidar b. afable, formidable affable, formidable

(Cat/Sp)

10 There I show that there are languages like Hungarian where the modal adjective contains a participial suffix. 11 Though afabilitat ‘affability’, but *formidabilitat. 12 For Catalan, Mascaró (1986: 23) considers that the -ble suffix in (25)–(27) above is the same as the one appearing in regular forms, because of its morphological behavior, displayed in (i) and exemplified in (ii). (i) ble → bil in front of -íssim, -ista, -itar, -itat, etc. (ii) afabilíssim, formidabilíssim, assequibilitat, potabilitat possibilitar affable.sup formidable.sup affordability potability make.possible.inf ‘extemely affable, extremely formidable, affordability, potability, enable’

36   

   -BLE

Besides, there are a few adjectives apparently derived from nouns with a causative meaning.13 (28) a. horrible ‘that causes horror’

(Cat)

b. penible ‘that causes pity’ c. terrible ‘that causes terror’ In addition to all these cases, there is in Catalan and Spanish another set of interesting examples with -ble. These are adjectives productively derived from nouns that mostly refer to posts or positions, as exemplified in (29a–b) for Spanish. They have the meaning ‘that can {become / be made /appointed} N’ or ‘candidate for N’. Additional non-regular noun-based examples not referring to posts are given in (29c). As shown in (29d), there is no related verb for any of these -ble adjectives. Interestingly, only those in (29a–b) express a modal passive meaning, although they all contain the default theme vowel preceding the suffix -ble.14 Exactly the same process is available in Catalan  – and in other Romance languages like Italian or Portuguese. (29) a. alcaldable, ministrable, presidenciable (Sp) mayor.ble minister.ble presidency.ble ‘that can {become / be made /appointed} N’ or ‘candidate for N’ (Val Álvaro 1981: 197) b. papable, rectorable Pope.ble dean.ble ‘that can {become / be made /appointed} N’ or ‘candidate for N’

13 These would correspond to what Vendler (1968: 118) calls emotive adjectives, because they “attribute to the subject an emotional reaction evoked in a human”, as exemplified in (i) with his examples. (i) a. terrible event → event that terrifies N b. horrible event → event that horrifies N 14 Note that /a/ is the default theme vowel for verbs, not the default class marker for nouns or adjectives. The discussion in § 2.3 below illustrates the default realization of theme vowels with verbs. See Harris (1991) or Bermúdez-Otero (2006) for the realization of theme vowels or class markers in the nominal domain.

The morphology of -ble   



   37

c. favorable, saludable favorable health.ble ‘favorable, healthy’ (Rainer 1999: 4633) d. * alcaldar, *favorar, *ministrar, *papar, *presidenciar, mayor.inf  favour.inf  minister.inf  Pope.inf  presidency.inf *rectorar, *saludar15 rector.inf  health.inf e. alcalde, favor, ministro, papa, mayor favor minister Pope

presidencia, rector, salud presidency rector health

Finally, Spanish shows the very special but regularly derived -ble cases illustrated in (30a)–(31a), which represent a significant challenge for previous analyses of -ble, for they are transparently derived out of unergative (30) and unaccusative verbs (31), even though the -ble forms appearing in these examples are not ­possible when in isolation, as illustrated in the corresponding examples, (30b)– (31b). Interestingly, this construction is not found in Catalan, though, as the ungrammaticality of (32) shows. (30) a. Corrió todo lo corrible / Ladró todo lo ladrable /  run.3sg.pst all lo run.ble bark.3sg.pst all lo bark.ble Tosió todo lo tosible.  cough.3sg.pst all lo cough.ble

b. *corrible / *ladrable / *tosible

(31) a.  Apareció todo lo aparecible / Brotó todo lo appear.3sg.pst all lo appear.ble / sprout.3sg.pst all lo brotable / Llegó todo lo llegable.  sprout.ble / arrive.3sg.pst all lo arrive.ble

(Sp)

b. *aparecible / *brotable / *llegable

15 There is a verb saludar meaning ‘greet’, but not a verb *saludar related to health.

(Sp)

38   

   -BLE

(32) a. * Va bordar tot {lo/ el} bordable / Va aparèixer tot  aux.3sg bark.inf all  lo/ the bark.ble aux.3sg appear.inf all {lo/ el} apareixible. (Cat) lo the appear.ble

b. *bordable / *apareixible

2.3 English vs. Romance: Theme vowels and beyond Now, when we try to apply Aronoff’s bi-affixal proposal to Romance languages like Catalan or Spanish, we are confronted with a morphological problem: it does not seem possible to divide all affixation between root-affixation and word-affixation, to start with. In fact, in the case of -ble suffixation – as with most word formation, at least traditionally speaking, affixation is neither to the root, nor at the word-level, but it is attachment to the so-called stem, i.e. the root plus a (declension) class marker (see e.g. Harris 1991; Aronoff 1994). Recall that I have used -ble because the preceding vowel is clearly not part of the suffix in Romance, but it is the so-called verbal theme vowel, which has traditionally been considered as the element that builds the traditional stem (e.g. “elemento que se añade a la raíz para constituir un tema de flexión” Lázaro Carreter 1971: 388).16, 17 Thus, Aronoff’s root- versus word-level affixation analysis cannot be readily applied to Romance, although there is in Romance the same distinction between opaque cases and the more transparent ones, i.e. between high and low -ble.18 In Oltra-Massuet (1999, 2000) I proposed that the Romance theme vowel or class marker is a purely morphological item that signals the presence of a functional head. The generalization underlying this proposal is that whereas the absence of an overt theme vowel may be relevant or not to determine the existence of some functional structure, the presence of a theme vowel undoubtedly points to the presence of a functional head.19 This theme vowel is inserted in the morphological component as a result of a well-formedness condition on func-

16 “Element that is added to the root to compose the inflected theme” [author’s translation]. 17 As a matter of fact, the suffix should be just -bl, since the final vowel is epenthetic (see for instance Mascaró 1983: 45). 18 It should be clear that I am not suggesting that these vowels are also theme vowels in English. For simplicity, I use the label -ble to refer to all cases of -able or -ible in English. 19 See also Picallo (1991) for a proposal that treats the theme vowel as a word marker heading a functional VP projection. See also Bermúdez-Otero (2007) for a different view on theme vowels and class markers.



The morphology of -ble   

   39

tional heads, both in the verbal and in the nominal environment, as was stated in chapter 1. This simple proposal permits an account of a number of apparently unrelated facts (stress assignment, allomorphy of the theme vowel, and cyclicity problems), which provides strong support to this hypothesis.20 The phonological realization of the theme vowel is often contextually-determined under strict locality conditions, but it may also be inserted by default. To illustrate this point, consider the Spanish root dorm ‘sleep’. This root must be specified as a third conjugation class verb, which selects a theme vowel -i in the default case, as in the infinitive in (33a), where -i is the phonological realization of the theme position adjoined to a phonologically empty functional head v. However, when this verbalizing head v is not empty, as in (33b–c), where we find the verbal affixes -ecand -it-, then locality ensures that the theme vowel adjoined to v can no longer be determined by the root; so, we get the specific theme vowel -e selected by -ec- in (33b), but the default theme vowel -a for the affix -it in (33c). (33) a. dorm- i- r sleep- th- inf ‘to sleep’ b. a- dorm- ec- e- r pref- sleep- suf- th- inf ‘to numb’ c. dorm- it- a- r sleep- suf- th- inf ‘to snooze’ Interestingly, the theme vowel hosted by the functional head v in verbs is the same that appears in the corresponding -ble adjective, as illustrated in (34), with a first conjugation verb admirar and a third conjugation verb discutir. As seen in (35) for the Spanish verbs mover ‘move’ and creer ‘believe’,21 second conjugation

20 See Arregi (2000) for the application of this well-formedness condition to Spanish, and Oltra-Massuet and Arregi (2005) for an analysis of Spanish stress that is crucially tied to this condition. See also Embick and Halle (2000) for Latin and Guerzoni (2000) for Italian. 21 The same facts hold in Catalan for second conjugation class verbs, (i). They show additional phonological complexities that need not concern us here. (See Mascaró 1983, 1986; Wheeler 2002.) (i) a. mou-r-e, mov-i-ble, mov-i-ment (Cat) move-inf-epv move-th-ble move-th-sufn ‘to movev, movable, moven’

40   

   -BLE

verbs are neutralized to take the third conjugation theme vowel in many cases, including -ble adjectives. As exemplified in (36), newly-derived -ble adjectives not related to verbs select the default theme vowel a.22 (34) a. admir – a – r, admir – a – ble admire-th-inf admire-th-ble ‘to admire, admirable’

(Cat/Sp)

b. discut – i – r, discut – i – ble argue-th-inf argue-th-ble ‘to argue, arguable’ (35) a. mov – e – r, mov – i – ble, mov – i – da move-th-inf move-th-ble move-th-sufn ‘to movev, movable, moven’

(Sp)

b. cre – e – r cre – i – ble believe-th-inf believe-th-ble ‘to believe, believable’ (36) a. presid – i – r, *presidenci – a – r (Cat/Sp) preside-th-inf  presidency-th-inf ‘to preside’ b. presid – i – ble, presidenci – a – ble preside-th-ble presidency-th-ble ‘preside.ble, presidency.ble’

b. creu-r-e, cre-ï-ble believe-inf-epv believe-th-ble ‘to believe, believable’ 22 There is to my knowledge only one apparent exception to this statement, the adjective futurible, where there is neither a verb futurar, nor a verb futurir. However, we find the nouns futura ‘right to inherit an employment or benefit before being vacant’ (Latin futūra) and futuro ‘future’ (Latin futūrus). The former gives the derived form in (i), while the latter is related to the items in (ii). All definitions are from the DRAE. (i) futurario (Sp) ‘that belongs to a future inheritance’ (ii) futur-i-ble futur-i-ción (Sp) future-th-ble future-th-sufn ‘that can become real in the future, condition of being projected towards the future’



The morphology of -ble   

   41

Now, given that with all -ble adjectives in Catalan and Spanish the -ble suffix is preceded by a theme vowel, if Oltra-Massuet (1999, 2000) is on the right track, and the theme vowel signals the presence of a syntactic functional projection, then we expect to have some functional head that hosts this theme vowel. For regular and transparently derived adjectives, it is clear that this head will be v. Part of my job in the next subsections will be to show that this functional head is present in the structure and that it has indeed a syntactico-semantic function, i.e. it is not just a host for the morphologically projected theme position.23 The question that arises is what the syntactic head that will host this theme vowel can be in the case of those opaque adjectives, for which a v seems rather unlikely, e.g. Catalan/Spanish flexible ‘flexible’, formidable ‘formidable’. The question extends to whether adjectives like Catalan assequible ‘affordable’ or potable ‘potable’, which have no corresponding verb *assequir, *potar, contain a v layer in their structure because of the meaning they convey. Furthermore, since the head  v is related to the licensing of event-related arguments, and different adjectives behave differently in this respect, we must establish which functional head hosts the theme vowel where no v can be present. That is, unless we want to assume the existence of the additional suffixes -able and -ible for e.g. Catalan assequible ‘attainable’ and Cat./Sp. potable ‘potable’ or alcaldable ‘mayor.ble’ that would be different from the regular -ble suffix that attaches to existing verbal bases, we must elucidate how these forms differ in their internal constituent structure that allows us to determine the functional host of the theme vowel. I will discuss these and other issues related to the theme vowel in the analysis of these forms in section 6. There are more differences between English and Romance -ble adjectives. Besides the questions on the internal structure of -ble adjectives that have to do with the presence of theme vowels in Romance, the most surprising contrasts are related to the special -ble cases that we find in Romance, but not in English. Chapter 3 is entirely devoted to the analysis of the puzzling -ble forms derived from unergative and unaccusative verbs in the Spanish construction V todo lo Vble. The set of productive denominal adjectives is the sole topic of chapter 4, where I provide an analysis that accounts for this crosslinguistic variation. Therefore, the

23 Being just a morphological host is actually impossible in the Distributed Morphology model, for two main reasons. First, and more generally, non-syntactico-semantic relevant functional heads are not present in the syntactic structure. Second, Oltra-Massuet’s (1999, 2000) well-formedness constraint on adjunction of theme positions targets only syntactic functional heads.

42   

   -BLE

study of these forms will be postponed until then and will not be included in the analysis to be developed in this chapter.

2.4 Summary Although their internal structure must necessarily differ, due to the presence of the theme vowel, all English, German and Romance show two main types of -ble adjectives, whose main properties are summarized in Table 1. Following standard practice in the Distributed Morphology model (Marantz 1999), and specifically Nevins (2002) for -ble adjectives, I use the labels high -ble (or potential -ble) for those -ble cases that are semantically transparent and show a tighter relation to the verb, and low -ble for the rest.24 LOW -ble

POTENTIAL or HIGH -ble 1. 2. 3. 4. 5. 6. 7. 8.

always expresses possibility/ability meaning is compositional does not trigger stem allomorphy allows derived bases [English] does not shift stress [English] is always spelled as -able [German] is always spelled as -bar [English] selects negative /un/

1. 2. 3. 4. 5. 6. 7. 8.

may express no modality may convey idiosyncratic meaning may trigger stem allomorphy attaches only to underived (truncated) roots [English] may shift stress [English] may be spelled as -ible [German] may be spelled as -abel [English] selects negative /iN/

Table 1. Two morphological types of -ble

The next sections are concerned with the different syntactico-semantic properties of -ble that were listed at the beginning of this chapter in (1). Examining the external syntax of -ble adjectives as well as their semantic content will allow us to establish their internal syntactic structure, which will be the basis for the analysis of the data in chapters 3 and 4.

24 Recall that I have excluded Spanish/Catalan productive denominal -ble and Spanish V todo lo Vble from the discussion.



The external syntax of -ble   

   43

3 The external syntax of -ble This section deals with the main syntactic properties of -ble. After considering the various types of bases that can be selected by -ble and the problems that such variation pose, section 3.2 uses standard syntactic tests on the expression of internal arguments and adverbial modification to provide evidence for having verbalrelated functional structure within the internal syntax of potential or high -ble, as opposed to low -ble adjectives, which systematically show ungrammaticality in the application of the same syntactic tests. After considering the special behavior of negated -ble adjectives with respect to the eventivity of the -ble structure, the discussion turns to the examination of the passive component in -ble adjectives in section 3.3, where additional tests contribute further empirical evidence on the syntactic presence of a passive component within the structure of potential -ble adjectives. Section 3.4 summarizes the main findings on the external syntax of -ble.

3.1 The base It is traditionally assumed that -ble adjectives can be productively derived from any transitive verb (e.g. Aronoff 1976, Fabb 1984, Anderson 1992 for English; ­Riehemann 1993 for German) or some restricted subset of them (e.g. Roeper and van Hout 1999).25 For instance, Chapin (1967:  63) explicitly excludes the possibility of intransitive bases, *happenable, *sleepable, *talkable or *waitable; and Anderson (1992: 186) includes the transitivity of the verbal base as a condition in the word formation rule that derives -able adjectives. However, although most -ble adjectives do have a transitive verbal base, the suffix -ble apparently attaches to different types of verbs. And, as we have seen in the previous section, it may also be affixed to a root, or to a noun. In (37) I list and exemplify for the various languages the different types of possible syntactic bases that -ble attaches to.26, 27 (37) a. Transitive verb: Cat. defugible ‘dodge.ble’, Eng. readable, Ger. trinkbar ‘drinkable’, Sp. aceptable ‘acceptable’

25 See also Trips and Stein (2008) for an analysis of -ble in English and French that takes the event structure of the base into account. 26 This list does not include the special Spanish cases illustrated by the Spanish construction V todo lo Vble in (30)–(31) above. However, it includes denominal nouns for expository reasons. 27 Assuming an analysis of unergative verbs along the lines of Hale and Keyser (1993, 2002), the examples in (37b) would be included in any rule that applies to transitive bases.

44   

   -BLE



b. Intransitive with cognate NP: Cat. ballable ‘danceable’, Eng. danceable, Ger. tanzbar ‘danceable’, Sp. vivible ‘liveable’



c. Intransitive with selected PP: Cat. fiable ‘reliable’, Eng. reliable, Ger. verfügbar ‘available’, Sp. fiable ‘reliable’



d. Intransitive with Locative: Cat. transitable ‘passable’, Eng. skiable, Ger. schiffbar ‘navigable’ (Jung 1967, Fleischer and Barz 1995), Sp. esquiable ‘skiable’28



e. Unaccusative verb: Cat. durable ‘lasting’, Eng. perishable, Ger. brennbar29 ‘combustible’, Sp. perdurable ‘ever-lasting’



f. Noun: Cat. papable ‘Pope.ble’, Eng. cabinetable, Ger. sichtbar ‘visible’, Sp. alcaldable ‘mayor.ble’

g. Root:30 Cat. assequible ‘affordable’, Eng. credible, Ger. unabdingbar ‘indispensable’, Sp. fungible ‘perishable’ Some studies consider that verbs with an internal argument are the optimal bases for -ble derivation (Williams 1981b; Oshita 1994; Gràcia 1995). And still others have argued that the -ble rule requires verbs with a theme argument (de Miguel 1986). Gràcia (1995) and de Miguel (1986) intend to include those adjectives derived from unaccusative verbs. Gràcia (1995) can also account for those adjectives derived from verbs that select an internal argument different from theme. She would in principle explain (37a)–(37e).31 The examples in (37f)–(37g) are not accounted for by any of these approaches. As I will show in chapter 4, although it is true that examples in (37f) are not so

28 The word esquiable ‘skiable’ is perhaps a more newly coined word in Catalan, along the lines of navegable ‘navigable’ or transitable ‘passable’, although it already appears in Val Álvaro (1981: 193) for Spanish. 29 Riehemann (1993 : 11) classifies the verbal base brennen ‘burn’ as intransitive. Below I will make a distinction between purely unaccusative verbs and those verbs entering the causativeinchoative alternation that is significant for the formation of -ble adjectives. 30 This means that there is no corresponding verb with the intended meaning, as shown in (i). There is a verb fungir ‘act, serve’ with the meaning desempeñar un empleo o cargo ‘hold a job or a post’ (DRAE), unrelated to the meaning of the adjective fungible ‘perishable/expendable’ in e.g. material fungible ‘expendable material’. (i) Cat. *assequir, Eng *cred, Ger. *(un)abdingen, Sp. *fungir 31 Gràcia (1995: 100) takes (37d) as included in the set of verbs with an internal argument. It is unclear to me that this is so. They seem to me to be adjuncts, at least esquiable ‘skiable’, navegable ‘navigable’, transitable ‘passable’.



The external syntax of -ble   

   45

numerous, they are productive in Romance. Note that this is problematic for Aronoff’s (1976:  48) Unitary Base Hypothesis which states that “the syntacticosemantic specification of the base, though it may be more or less complex, is always unique. A WFR [Word Formation Rule] will never operate on either this or that.” As a consequence, Aronoff is forced to say that there are two different -ble adjectives, and hence two different word formation rules, one for verbs and one for nouns (recall footnote 6 above). I will discuss Aronoff’s evidence for positing the existence of a specific denominal -ble rule in chapter 4. For the time being, I will just point out that, as far as I can tell, since Aronoff’s proposal of a separate rule, denominal forms have been either left aside in the various analyses of -ble forms that I have found – with the exception of Di Sciullo (1997), to be discussed in chapter 4, or else, they have been grouped together with the non-transparent, idiosyncratic cases, i.e. essentially all cases in (37), except (37a–b), possibly also (37c), which can be said to be derived from a transitive structure. On the other hand, both types of -ble adjectives in (37f)–(37g) exhibit the same morphology and semantics, and in some cases even the same syntax as regular -ble adjectives, as illustrated in (38) for Spanish. Thus, any analysis of -ble adjectives should take them into account. (38) un libro fácilmente asequible por todo el mundo a book easily accessible by all the world ‘a book easily affordable by anyone’ Let us leave aside these two apparently non-verbal bases and consider the most relevant analyses that cover all other cases, (37a)–(37e), in some more detail. Analyses that suggest that -ble takes a verb with an internal argument as a base must face the following non-trivial problems. As pointed out in Oshita (1994) and Gràcia (1995), it is a mystery why there are so few unaccusative bases, if all that is required is an internal argument. As exemplified in (39) for Catalan and (40) for English, unaccusative bases are generally impossible as bases for -ble. (39) * tren arrivable, *un arbre creixible (Cat) train arrivable,  a tree growable (40) * emergeable truth, *vanishable civilizations (Oshita 1994 : 262) As noted in Gràcia (1995), pronominal verbs, generally analyzed as unaccusatives, do not allow -ble derivation, either, as illustrated in (41). Nor do bases expressing an internal process that cannot be caused by some Agent, as shown in

46   

   -BLE

(42a). However, if they can also express an externally-caused change of state, as in (42b), then they are fine as -ble adjectives. That is, verbs expressing only and always an internally-caused change of state are impossible as -ble adjectives, as illustrated in (43).32 (41) * agenollable, *penedible, *ruboritzable (Cat) kneel.down.ble,  be.sorry.ble,  blush.ble (Gràcia 1995 : 122) (42) a. * engreixable, *envellible, *enlletgible (Cat) fatten.ble,  age.ble,  make.ugly.ble (Gràcia 1995 : 121) b. aliatge no envellible alloy non ageable (43) * florible, *creixible33 (Cat) bloom.ble,  grow.ble Leaving aside the few unaccusative bases like English perishable or Catalan/ Spanish durable ‘lasting’ or perdurable ‘ever-lasting’, these data strongly suggest that some additional requirement is necessary for the insertion of -ble. Specifically, some notion of external argument, agent or cause seems to be at stake, as already advanced in Di Sciullo (1997: 80), together with the availability of an internal theme argument.34 The examination of additional data in the next sections will confirm this conclusion.

32 According to Roeper (1987) inchoative bases are not possible. But see de Miguel (1986) and Gràcia (1995) who claim that they are fine in cases like English breakable or Spanish oxidable ‘rustable’. 33 The verbs créixer ‘grow’ in Catalan or crecer ‘grow’ in Spanish do not have a transitive counterpart. To express the meaning of English transitive grow, we must resort to a different lexical item cultivar, as in Catalan Cultivo roses al jardí ‘I grow roses in the garden’. 34 Dryer (1985) argues against the hypothesis that the subject of -ble be a theme of the verb, and proposes that it takes an object instead. As I will argue below and in chapter 3, the requirement of -ble on having a theme must be extended so as to cover incremental themes of various types that Dryer’s object could not possibly cover.



The external syntax of -ble   

   47

3.2 Being deverbal A perhaps more important issue that I want to address in this section is related to the exact meaning of the notion being deverbal. Roots are not specified for category in the Distributed Morphology framework (Marantz 1997 and related work); they are merged in the computational system with a category-giving functional projection. Hence, being deverbal in this model amounts to having a root merge with some verbalizing functional head or heads in the syntax. As syntactic functional heads, they must have some syntactico-semantic properties. The point is that being a syntactic head, it cannot be vacuously present, i.e. there must be some evidence that supports its presence in the syntax. Thus, it is possible that what has been traditionally considered as ‘deverbal’ or ‘deadjectival’ does not necessarily contain a corresponding verbal or adjectival syntactic head, if this is not syntactically relevant.35 Indeed, Volpe’s (2005) analysis divides English -ble adjectives between those that contain a v head and those that do not. He does so on the basis of the morpho-phonological evidence described in Aronoff (1976). It remains to be seen whether this proposal correlates with a difference in the argument structure properties of these adjectives, and whether it can be applied to similar -ble adjectives in other languages. The linguistic literature provides a number of well-known standard tests to help determine the presence of event-related functional structure. In the remainder of this subsection, I investigate the internal syntax of -ble by concentrating on two kinds of tests: (i) the expression of verbal arguments, and (ii) adverbial modification. Section 3.3 will provide further evidence related to the verbal passive content of these adjectives and the implicit presence of an external argument. All tests support the presence of verbal and aspectual structure in the internal syntax of a set of -ble adjectives, which further reinforces the classification of -ble adjectives into two syntactically different types.

3.2.1 The expression of arguments Leaving aside the expression of the external argument of the verb, which will be tackled in the next section together with the passive content of -ble adjectives, the question is whether it is possible to overtly realize other internal arguments of the verb underlying -ble adjectives.

35 See for instance Arad (2003) for an analysis of derived verbs in Hebrew. This author provides evidence showing that alleged denominal verbs can be split into those that have been built directly on a root, and those that have an underlying noun in their internal structure.

48   

   -BLE

Consider the examples in (44)–(46) for Catalan, English, German, and Spanish, respectively. In all languages we find examples where there is a PP selected by the verb underlying the adjective, which clearly argues for the presence of some verbal functional head that should license them. It cannot be a mere coincidence that in all these languages a verb and a -ble adjective derivationally related to it share exactly the same highlighted complements. (44) a.  En l’univers clàssic, el vici no és derivable de in the=universe classical, the fault not be.3sg.prs derivable from la naturalesa humana. (W) (Cat) the nature human ‘In the classical universe, fault is not derivable from human nature.’ b. una afirmació aplicable al català a statement applicable to.the Catalan ‘a statement applicable to Catalan’ (Gràcia 1995: 103) c. una taula transformable en billar a table transformable in billiard ‘a table transformable into a billiard’ (Gràcia 1995: 103) d.  El valor afegit d’una trobada presencial no sembla the value added of=a meeting face-to-face not seem.3sg.prs fàcilment substituïble per les trobades virtuals. (W) easily replaceable for the meetings virtual ‘The extra value of a face-to-face meeting does not seem easily replaceable with virtual meetings.’ (45) a. Most failures are still attributable to specific circumstances. (W)

b.  Game Theory pretends that human actions are breakable into much smaller “molecules” called games. (W)



c.  Does anyone know if dehydrated meat is bringable to Canada?

(W)

d. The copy is barely distinguishable from the original. (W)



The external syntax of -ble   

   49

(46) a. Diese Problematik ist von den übrigen Fragen nicht this difficulty is from the further questions not abgrenzbar. (Ger) delimitable ‘This difficulty is not delimitable from the further questions.’ (Riehemann 1993: 17) b. … daβ diese Viren auf Menschen übertragbar sein könnten. … that these viruses onto humans transferable be could ‘… these viruses could be transferable to human beings.’ (Riehemann 1993: 17) c. Der Junge ist leicht mit seinem Bruder verwechselbar the boy is easily with his brother mistakable ‘The boy is easily mistakable for his brother.’ (Reiβe 2006: 44) d. [Er ist] für jeden Hinweis dankbar [he is] for every hint grateful ‘He is grateful for every hint.’ (Reiβe 2006: 13) (47) a. No he encontrado nada relacionable con el caso not have.1sg.prs find.part nothing relatable with the case que nos ocupa. (Sp) that us concern.3sg.prs ‘I have not found anything relatable to the case we are concerned with.’ b. No es una enfermedad notificable a la OMS. (W) not be.3sg.prs a disease notifiable to the WHO ‘It is not a disease notifiable to the WHO.’ c. una solución adaptable a las necesidades de la empresa a solution adaptable to the needs of the company ‘a solution adaptable to the company’s needs’ d. La ciencia es nítidamente separable de otras formas the science be.3sg.prs clearly separable from other forms de conocimiento. (W) of knowledge ‘Science is clearly separable from other forms of knowledge.’

50   

   -BLE

As has been extensively studied, for instance in Alexiadou (2001), in the case of nominalizations, only eventive nominals, which contain in her analysis a v layer, can license the expression of these inherited complements. The analysis of the corresponding verbal constructions is a much debated issue (see e.g. Larson 1988) and the details are not crucial to this study. The point is that, no matter which analysis is assumed for verbs with two internal arguments, it has been argued to contain some syntactic head to license the PP, which must be present in both the purely verbal structure and in the deverbal configuration. Note that all examples in (44)–(46) above are cases of regularly derived -ble adjectives, i.e. cases of potential or high -ble. The problem arises when we consider cases of low -ble, e.g. English -ble adjectives showing stem allomorphy, such as applicable or exhibiting stress shifting, like cómparable. It turns out that both can syntactically realize the verbal PP complement, as shown in (48). (48) a. The current situation is cómparable to the one in 1935. b. These are easily compárable to the others. c. Our approach is still applicable to a wide range of problems.

(low -ble) (high -ble) (low -ble)

Given that verbal inherited arguments can only be licensed by a verbal functional head, as proposed in Alexiadou (2001), these cases appear as problematic for a DM analysis. Whereas their morphology points to the absence of verbal functional projections, their syntax apparently supports the presence of a v layer. I will come back to these contrasts in section 3.4, after reviewing the application of all remaining tests. Randall (1985 : 44–45) argues that cases like (48) or divisible into must be listed, that is, they do not derive from a related verb by rule. In her proposal, this is a consequence of her Inheritance Principle, according to which, there are no -ble items that require anything else than just a single NP object. So, she predicts that there should be no cases like (48) or (49) below. Or else, they should all be listed as idiosyncratic. In other words, according to Randall, that the adjectives applicable, attributable or distinguishable select a PP complement that coincides with the PP complement selected by the verbs apply, attribute, distinguish, is as accidental as the cases afraid of or contrary to would be. (49) a. Most failures are still attributable to specific circumstances. (W)

b.  Trade names must be distinguishable from previous registrations on the record with the Secretary of State. (W)

The external syntax of -ble   



   51

Furthermore, Randall is forced to say that productively derived cases like the (a) examples in (50)–(52) are ungrammatical, like her example (65) repeated in (51c), though they are deemed right by the native speakers I have consulted. They seem to grow more and more common. They contrast with the more idiosyncratic (b) examples in (50)–(52). (50) a. events that are still deducible from general laws (W)

b.  The University will cover the $ 1000 deductible from the State’s self-insurance plan for damages to a vehicle. (W)

(51) a. a new 1000-seater Lecture Hall, dividable into smaller theatres (W) b. The pie is divisible into thirds.

(Randall 1985 : 44)

c. The pie is dividable (*into thirds).

(Randall 1985 : 44)

(52) a. Pandemic diseases are now rapidly transmittable to all parts of the globe. (W) b. These findings further confirm that TME is transmissible to raccoons. (W) As has been observed in Gràcia (1995: 103) for Catalan, the examples in (44)–(47) above are clear – and productive – counterexamples to Randall’s (1985: 63) Inheritance Principle, according to which -ble adjectives should be prevented from inheriting all but the “unmarked part of the base form’s entry.” These are cases of verbs selecting two internal arguments, often an internal direct object and an oblique argument, but also a dative plus an oblique argument as in (46d).36 According to Gràcia (1995: 106) -ble adjectives can inherit all internal complements that together with the verb form its maximal projection VP, because adjectives are [+ V] predicates. Notice that this proposal is very problematic; it would mean that other derived adjectives should also be able to inherit any verbal PP

36 Bosque and Gutiérrez-Rexach (2008: 268) provide the following example as an instance of a four-place predicate, where the predicate transportable ‘transportable’ has an underlined external argument, and three more arguments (Origin, Goal and Path). (i) [Esta mercancía] no es transportable [desde Cádiz] [hasta Barcelona] this goods not be.3sg.prs transportable   from Cádiz   until Barcelona [por carreteras secundarias].   through roads secondary ‘These goods are not transportable from Cádiz to Barcelona through secondary roads.’

52   

   -BLE

complements, because they would equally be [+ V], contrary to fact, as exemplified in (53)–(54) for Catalan and English. I present an additional argument against this proposal in section 3.2.3 below. (53) * aplicatiu a, *substitutiu per (Cat) applicative to,  substitutive for (54) *attributive to, *transformational into As regards the expression of non-PP internal arguments, linguists seem to disagree in this respect, probably because data are rather complex. Consider the English contrasts in (55). According to Aronoff (1976) and Kayne (1981), it is not possible to express the indirect object in English, whether in a dative construction, as in (55a), or in a double object construction, as in (55b). Nor can an indirect object be externalized, as shown in (55c). (55) a. * The film is showable to the children. (Aronoff 1976 : 49fn.) b. * Linguistic books are sendable prisoners. (Kayne 1981 : 155) c. ? Prisoners are sendable linguistic books under certain conditions. (Kayne 1981 : 155) Di Sciullo (1997: 81) distinguishes between double object verbs, such as give and tell, and verbs like explain or donate, which do not give rise to the dative alternation. (56) a. These books are donatable to the library. b. *These books are giveable to the library. In her account, the contrast in (56) follows from the fact that -ble cannot be inserted in a VP shell structure for double objects (Larson 1988). This, the argument follows, predicts that languages lacking double object verbs do not give rise to such contrasts, so that the corresponding sentence in (56b) is possible; French would be one such language. First, notice that (55b) is as deviant as the corresponding passive sentences in (57a), unless we insert the corresponding preposition, as shown in (57b). The presence of the preposition with the -ble form makes a substantial improvement,

The external syntax of -ble   



   53

so that in a context similar to (55c), (55b) would be just deviant or even fine for the native speakers I have consulted, as illustrated in (58). That would mean that the structural constraint on the double object construction is not particular of -ble, but rather of the passive structure (present in the internal configuration of -ble, as will be argued below). (57) a. *Linguistic books are sent prisoners. b. Linguistic books are sent to prisoners. (58) ?/Linguistic books are sendable to prisoners under certain conditions. Second, some of my informants report the example in (59a) as fine (even if peculiar to some instructional written style), which contrasts with the ungrammaticality of those in (55) above. Additional examples with alternating verbs like send and lend are given in (59b–c). (59) a. It should be givable to new users. (W)

b.  Do you also have any sounds of yourself playing this instrument that are sendable to us? (W)



c.  This license is not transferable or lendable to other institutions or campuses. (W)

I have found additional examples showing that -ble adjectives may allow the expression of an indirect object, even if with certain (perhaps dialectal) limitations. The contrast in (60) shows that some notion of specificity seems to be at stake in English. In section 3.3.3, I show that a similar constraint operates in the expression of a by-phrase with -ble adjectives, which will be related to the presence of modality in section 4. (60) a. Any good theory should be explainable to a barmaid (W)

b. *Any good theory should be explainable to John.

For German, Riehemann (1993: 17) cites the example in (61) from Toman (1987), who observes that dative objects can be inherited by the adjective. Nonetheless, Riehemann herself comments that there are hardly any inherited dative arguments in her corpora, noting, however, that non-existing forms do not seem impossible either. She concludes that, although matters are unclear, the notion “pragmatically possible” may be relevant in determining inheritance of oblique arguments.

54    (61)

   -BLE

Die Tat war dem Angeklagten nicht nachweisbar.  the deed was the.dat accused not demonstrable ‘It was not demonstrable that the accused was guilty of the deed.’ (Riehemann 1993: 17)

(Ger)

So, it seems that in both English and German, although dative objects are not generally expressed with -ble adjectives, they can be in particular cases, when the appropriate context is supplied. This could mean that there is a priori no linguistic constraint in this respect, and that the expression of oblique arguments is restricted to those cases expressing the right type of property. It is possible that examples like those in (55a) and (55c) are considered deviant because they cannot be understood as expressing that some characteristic of the object is relevant to express a non-trivial property, i.e. to say that a film can be shown to somebody is rather obvious.37 A Catalan sentence parallel to (55a) above is also deviant, as in (62a); however, it could be made more acceptable if for instance we had contrasting focus on the copula, to indicate that the film can be shown to the children, contrary to what one had thought. On the other hand, if we make the indirect object non-specific or indefinite, the sentence becomes completely fine, as shown in (62b). (62) a. * La pel·lícula és mostrable als nens. the film is showable to-the children b. menors de dotze anys.  La pel·lícula és mostrable a nens the film is showable to children younger than twelve years ‘The film is showable to children under twelve.’ According to Riehemann (1993:  12), a general principle of informativeness  – together with some notion of a “non-trivial property”, interacts with the specificity of the PP and constrains the expression of oblique arguments with -ble adjectives in all languages. As a matter of fact, -ble adjectives in general, leaving aside those that are more common and often lexicalized to some extent, seem to be pragmatically very constrained. In my view, this must be related to the fact that -ble adjectives express a property that is generally valid, that is supposed to hold as a general truth, i.e. it is linked to a more general restriction on characterizing predicates, of which -ble adjectives are an example of (Krifka et al. 1995: 7).

37 Though notice that Kayne marks (55c) as not so deviant, possibly because it would become fine given the right context.



The external syntax of -ble   

   55

As pointed out in Krifka et al. (1995: 13), characterizing sentences state properties that are “in some way essential”, as opposed to accidental, a distinction that “is real and manifests itself in striking results, but the underlying reason is not clear.” The kind of meaning conveyed by verbs that typically select a dative argument, e.g. give, transfer, etc and their compatibility with a generic or dispositional reading surely plays an important role. In section 4.4.2, I will argue that a specificity constraint can be derived from the syntactico-semantic presence of a modal component, which creates a stative modal context. If this is right, we can conclude from the data above that a verbalizing head is syntactically present in the internal structure of -ble adjectives, even though the licensing of verbal arguments is severely constrained by its interaction with other general principles of the grammar and general constraints on the semantics and pragmatics of -ble.

3.2.2 Adverbial modification A test that is generally used to evidence the presence of some verb-related functional head is connected to adverbial modification (see for instance, Borer 1994; Cinque 1999; Alexiadou 2001; Fu, Roeper and Borer 2001). I will leave aside the discussion of adverbial modification associated to the agent, i.e. agent-oriented adverbs and instrumental phrases; they will be considered together with the expression of the external argument in section 3.3.3 below. Here, I concentrate on non-agentive modification that provides information on the event, process or action expressed by the underlying verb (e.g. Alexiadou 2001).38 According to Bosque (1999: 281), adjectives do not license postponed manner adverbials in Spanish, though the same author observes that some deverbal adjectives like -ble may inherit some syntactic properties from the verb, specifically that of licensing a postponed manner adverbial and an agentive by-phrase. I repeat his examples in (63). See the next section for the licensing of by-phrases with -ble. Similar examples can be found in Catalan or English as well. (63) a. contratos renovables cada año (Sp) contracts renewable every year (Bosque 1999: 281)

38 The use of adverbial modification as a test with adjectives may seem tricky at first, since adverbs are certainly modifiers of adjectives. However, it seems possible to use it safely when other (individual-level) adjectives would not be possible in similar contexts.

56   

   -BLE

b. acuerdos revisables cada dos años agreements revisable every two years (Bosque 1999: 238) c. renovable a voluntad del consumidor renewable at will of.the consumer ‘renewable at consumer’s will’ (Bosque 1999: 238) Bosque also notes that adverbs that refer to the process or event in process, such as progresivamente ‘progressively’, paulatinamente ‘gradually’ or poco a poco ‘little by little’, are rejected by most simple adjectives. This is not so in the case of -ble adjectives, as the Spanish examples in (64) show. Notice that these adverbs appear postponed to the adjectives. (64) a. volante ajustable progresivamente, vertical y steering.wheel adjustable progressively vertically and horizontalmente (W) (Sp) horizontally ‘progressively adjustable steering wheel, vertically and horizontally’ b. empréstito reembolsable a la par amortizable progresivamente (W) loan refundable at the par redeemable progressively ‘loan refundable at par value redeemable progressively’ c.  el Gobierno presentará un proyecto de Seguro al the government present.3sg.fut a project of insurance to.the desempleo, aplicable paulatinamente … (W) unemployment, applicable gradually … ‘The Government will present a gradually applicable unemployment insurance project …’ According to the hypothesis put forth in Cinque (1999), aspectual adverbials are linked to an AspP. We expect that if these are licensed with our -ble adjectives, this will be indicative of the presence of such an AspP in their internal structure. As exemplified in (65) for Catalan, aspectual temporal adjuncts are fine with -ble adjectives. (65) a. un treball modificable en una hora (Cat) a work modifiable in an hour ‘a piece of work modifiable in an hour’

The external syntax of -ble   



   57

b. un llibre traduïble en una setmana a book translatable in a week c. activitats realitzables durant hores activities realizable during hours ‘activities realizable during hours’ None of these types of adverbial modification is possible with low -ble adjectives, as shown in (66) for Spanish. The same facts obtain in Catalan (67) or in English (68). (66) a. * un sonido perceptible durante una hora (Sp) a sound perceptible for an hour b. * proyectos permisibles cada año projects permissible every year c. * limitaciones admisibles progresivamente limitations admissible progressively (67) a. * un so perceptible durant una hora (Cat) a sound perceptible for an hour b. * projectes permissibles cada any projects permissible every year (68) a. * a sound perceptible for an hour

b. * gradually cómparable results

Likewise, locative expressions are fine with high -ble adjectives, as shown in (69a)–(71a) for Catalan, Spanish and English, as opposed to low -ble adjectives in (69b)–(71b). (69) a. una dansa ballable {en qualsevol lloc /  al menjador de a dance danceable  in any place/ at.the dining.room of casa} (Cat) home ‘a dance danceable in any place / in my home’s dining room’ b. * un experiment permissible a la universitat a experiment permissible at the university

58   

   -BLE

(70) a. un manuscrito publicable en una revista científica (Sp) a manuscript publishable in a journal scientific ‘a manuscript publishable in a scientific journal’ b. * un estado admisible en la ONU a state admissible in the UN (71) a. the amount of matter presently observable in the universe

b. *results cómparable in an experiment

German further supports this conclusion, since -bar, but not -abel can license the expression of temporal adverbials, as illustrated in (72). (72) a. Das ist stundenlang ergebnislos diskutierbar. (Ger) that be.3sg.prs hours.long unsuccessfully debatable ‘That can be unsuccessfully debated for hours.’ b. * Das ist stundenlang ergebnislos diskutabel. that be.3sg.prs hours.long unsuccessfully debatable According to Di Sciullo (1997:  89), temporal and spatial expressions cannot modify -ble adjectives, as exemplified in (73). (73) *There are chairs transformable at the inn/this morning. In view of the existence of -ble adjetives with adverbial modification provided in this section, my claim is that the ungrammaticality of Di Sciullo’s sentences is due to the general restriction on characterizing predicates, that of expressing an essential or non-accidental property, together with the general ban on individuallevel adjectives in the there-construction. Whereas being translatable in a week or being publishable in a scientific journal can be understood as essential predications deriving from some inherent characteristics of the book and the manuscript (say, style, difficulty, or quality), it seems to me imposible to view the property of being transformable at the inn or being transformable this morning as an essential feature of chairs.39

39 Note that the Catalan locative adverbial al menjador de casa ‘in my dining room’ in example (69) is clearly interpreted as a kind, i.e. any place as small/big as my dining room.



The external syntax of -ble   

   59

Thus, taking into account the existence of general restrictions on the kind of elements that can appear with characterizing predicates, the application of these adverbial modification tests further supports the classification of -ble into a low -ble and a high -ble where only the latter contains some aspectual functional projection that can syntactically license aspect-related phrases (Cinque 1999, ­Alexiadou 2001).

3.2.3 Negated -ble adjectives and eventivity Up to this point, I have exemplified all cases with positive -ble adjectives. Now, contrast the examples in (74)–(77) for Catalan, English, German and Spanish with the ones in (44)–(47) above on the expression of arguments. Negated -ble adjectives do not permit the realization of other selected complements.40 (74) * una trobada presencial insubstituïble per una de virtual (Cat) a meeting face-to-face irreplaceable for one of virtual (75) * a toy unbreakable into smaller pieces (76) * Karl ist mit seinem Bruder unverwechselbar. (Ger) Karl is with his brother unmistakable (Reiße 2006: 13) (77) * una solución inadaptable a las necesidades de la empresa (Sp) a solution unadaptable to the needs of the company As has been observed by Fabb (1984), when the adjective is negated the expression of oblique arguments is not possible.41 Note that if the explanation for the expression of internal arguments lay in the [+ V] feature of the adjective, as proposed in Gràcia (1995 : 134–136) these facts would remain a complete mystery. As a matter of fact, Gràcia’s (1995) suggestion wants to account for the eventiv-

40 The interaction between negation and the eventive properties of deverbal adjectives, e.g. how it affects the expression of the argument structure of the adjective, is a topic in its own right, for which I have not found any thorough study, apart from the references to participles cited in the main text. See Reiße (2006) for a corpus linguistic approach to the study of negation with -bar adjectives in German. 41 This assertion needs caution. See the discussion below.

60   

   -BLE

ity of -ble adjectives. However, it has been argued that the inability of negated adjectives to express the verbal arguments is precisely due to the non-eventivity of the negated adjective. Indeed, in the study of adjectival passive participles (e.g. Levin and Rappaport 1986; Varela 1990, 2002; Kratzer 2000; McIntyre 2011), there is an extensive use of the negative prefix to ensure that one is dealing with the adjectival form, and not with the verbal form. And it has been argued that it is not possible to realize the verbal arguments because of a change in category, from being a verb to becoming an adjective. Notice that in the case of the suffix -ble, this blocking cannot be due to the change of category in those frameworks, otherwise we could not explain the cases above where the adjective can be accompanied by all verbal arguments. That is, in principle, all cases above have an instance of a -ble adjective, so that the expression of arguments cannot be due to a variation in category. In the case of passive participles, the presence of additional arguments of the verb forces the dynamic, more eventive interpretation. The question is whether this is equally so in the case of -ble adjectives, i.e. whether we can talk of different degrees of adjectivehood or of verbhood in the case of -ble adjectives. It seems as if we had two kinds of adjectives: those that allow the realization of arguments are more eventive and would be closer to the verbal passive participle, whereas those that do not allow it would be closer to simple adjectives. Note that negated -ble adjectives do not allow the expression of adverbial aspectual adjuncts either, as the examples in (78)–(79) show. (78) un treball immodificable (*en una hora) (Cat) a piece.of.work unmodifiable   in an hour (79) un texto intraducible (*en una semana) (Sp) a text untranslatable   in a week For similar cases with participles, Varela (1990, 2002, 2003) suggests that negation is a way of canceling eventivity, so that the agent can no longer be expressed, and the verbal meaning disappears, turning the verbal participle into a simple adjective. Fábregas (2005: 119) has translated this role of negation in the expression of eventivity as either attachment at the v level or merging at the little a (adjectival) level. That is, when the negative prefix merges with the eventive head, i.e. with the eventive little v head, it cancels the eventivity of this head. The prefix can also merge at a higher level with deverbal adjectives, it attaches to a little a head, in which case it can manipulate the scale expressed by the adjective and so it can select a specific value. Only in this latter case can

The external syntax of -ble   



   61

the adjective become a noun. The two sites for attachment are illustrated in (80) with Fábregas’ examples. (80) a. * un ininterrumpido b. un intratable (Sp) an uninterrupted an untreatable ‘an unsociable person’ aP aP a…

vP negP in-

vP … interrumpido

aP

negP ina ble

vP … trat(a)

A potential problem for this otherwise very convincing proposal is the existence of examples such as those in (81) in Spanish.42 (81) a. un disparo imparable por el portero (W) (Sp) a shot unstoppable by the goalkeeper ‘a shot unstoppable by the goalkeeper’ b.  Los enchufes han de estar situados a una altura the sockets have.3pl.prs to be.inf situate.part.pl at a height inalcanzable por los pequeños. (W) unattainable by the little.pl ‘Sockets must be located at a height that is not attainable by children.’ c. una maniobra política irrealizable por un juez (W) a maneuver political unrealizable by a judge ‘a political maneuver not realizable by a judge’

42 As noted, the examples were retrieved from Google through a web search. When confronted with these examples, some of my native speakers judged them as marginal. For instance, Anna Bartra (personal communication) suggests the examples in (i) instead. The point under discussion is not affected, though, if I change the examples in (81) for those in (i). (i) a. un código indescifrable por cualquier arqueólogo (Sp) a code undecipherable by any archaeologist ‘a code undecipherable by any archaeologist’ b. una fortaleza inexpugnable por enemigos mal armados a fortress unassailable by enemies badly armed ‘a fortress unassailable by badly armed enemies’

62   

   -BLE

If the expression of the external argument means that eventivity has not been cancelled, then negation should merge with the adjectival head, which would permit a category change, from adjective to noun. But, as far as I can tell, there are no cases like *un imparable ‘an unstoppable’, *un inalcanzable ‘an unattainable’, *un irrealizable ‘an unrealizable’ that could be equated to the grammatical un intratable ‘an unsociable person’. There are similar examples in English as well, (82), which contrast with Fabb’s (1984) observation according to which negation often bars the expression of the verbal complements in English. (82)  This will help ensure that the data transferred within Bankline is indecipherable by anyone who may attempt to intercept it (W) As suggested by Ora Matushansky (personal communication), examples with an expressed external argument in the presence of negation may be possible not because they are in fact grammatical, but because they are interpretable due to the existence of some repair strategy, perhaps some process of coercion. A full account of these facts will have to await further analysis on the interaction of negation with eventivity, and the exact conditions under which such coercion or any other repair strategy can apply in the realm of non-participial deverbal formations. Nonetheless, before concluding this section, let me point to an additional test in Spanish, which relates negation to the eventive content of -ble adjectives and their syntactic position, whether attributive or potsponed. Bosque (1999: 299) points out that past participles do not generally appear in preposed position, as shown in (83a). When they do, as in (83b), this is taken as a clear sign that they have become adjectives. When we observe the behavior of -ble adjectives with respect to their syntactic position, it turns out that high -ble adjectives do not generally accept the preposed position, as shown in (84a), which would support their being considered as dynamic and verbal as participles.43 In contrast, low -ble adjectives may appear in attributive position, as in (84b). Interestingly, when high -ble adjectives are negated they become fine in preposed position, as exemplified in (84c).

43 Note that the position of -ble adjectives, attributive or predicative, cannot be related to their status as stage-level or individual-level predicates (Kratzer 1995) in Spanish, since all -ble adjectives predicate individual-level properties with the copula ser, as illustrated in (i). See e.g. Demonte (1999) or the NGRALE § 13.4k. (i) a. El manuscrito es publicable./ Su labor es admirable. (Sp) the manuscript be.3sg.prs publishable her job be.3sg.prs admirable ‘The manuscript is publishable./ Her job is admirable.’



The external syntax of -ble   

   63

(83) a. * la escrita carta, *el traducido libro, *un aceptado encargo (Sp) the written letter  the translated book  a accepted order b. el asombrado presentador, la anunciada entrevista the astonished presenter the announced interview (Bosque 1999: 299) (84) a. * un publicable manuscrito, *una traducible lengua, *un a publishable manuscript  a translatable language  a olvidable encuentro (Sp) forgettable meeting b. una amable chica, una admirable labor, un confortable asiento a friendly girl a admirable job a comfortable sit c.  un impublicable manuscrito, una intraducible lengua, un a unpublishable manuscript a untranslatable language a inolvidable encuentro unforgettable meeting Exactly the same facts can be found in Catalan. As the contrast in (85) shows, more verbal participles are banned from the attributive adjectival position, which parallels the behavior of -ble adjectives in (86). (85) a. * l’escrita carta, *el traduït llibre, *un acceptat encàrrec(Cat) the=written letter  the translated book  a accepted order b. l’atordit presentador, l’anunciada entrevista the=astonished presenter the announced interview (86) a. * un publicable manuscrit, *una traduïble llengua, *una a publishable manuscript  a translatable language a oblidable trobada (Cat) forgettable meeting b. una amable noia, una admirable tasca, un comfortable seient a friendly girl a admirable job a comfortable sit

b. *El manuscrito está publicable./ *Su labor está  the manuscript be.stative.3sg.prs publishable her job be.stative.3sg.prs  admirable.  admirable

64   

   -BLE

c.  un impublicable manuscrit, una intraduïble llengua, una a unpublishable manuscript a untranslatable language a inoblidable trobada unforgettable meeting As just mentioned, the contrasting behavior of participles in the presence of negation has been accounted for in lexicalist approaches by attributing negation the power of canceling the eventivity of the underlying participle. I have no better explanation at this point; it is unclear why this should be so, i.e. why negation in the word domain should have this powerful effect on the base; and, equally puzzling is how this should be performed in a syntactic approach to word formation. It seems to me that a structural explanation should be possible along the lines of Fábregas’ (2005) proposal, where differences derive from different structural attachments sites, but this will have to await a deeper investigation.

3.2.4 Interim conclusion Up to this point, all the evidence I have uncovered provides support for the presence of some verbal component within the structure of potential or high -ble adjectives, thus showing that the verbality of -ble adjectives has a syntactic effect. In fact, the data on adverbial modification argue for the presence of an AspP in the structure of -ble adjectives, whereas the argument structure facts support the presence of a v head (cf. Alexiadou 2001 for nominalizations). The observations on the special behavior of -ble adjectives with respect to negation provide further support for their having some eventive content. Apart from being deverbal, -ble adjectives are said to be passive. The next section investigates into this second general property of -ble adjectives. After reviewing the literature on this point, I will show that a passive component is syntactically present only in high -ble. Also the (implicit) presence of an external argument in high -ble adjectives has syntactic effects in their external syntax, which further distinguishes them from low -ble adjectives.

3.3 Being passive It has often been noted in the (descriptive or theory-oriented) literature on different languages that deverbal -ble adjectives are related to passives. Many have stated this relationship in semantic terms, in the sense that they have claimed that the interpretation of -ble adjectives generally corresponds to



The external syntax of -ble   

   65

a passive construction (Jung 1967; Marchand 1969; Aronoff 1976; Val Álvaro 1981; de Miguel 1986; Lang 1990; Rainer 1993; Fleischer and Barz 1995; Rainer 1999; Altmann and Kemmeling 2000; Huddleston and Pullum 2002; Albrespit 2009; inter alia).44 Essentially, what these works often mean is that a -ble adjective can be paraphrased using a passive construction. Others have suggested an analysis of these adjectives as being explicitly built on a passive structure in the lexicon (Chapin 1967, Vendler 1968, Abraham 1970, Lyons 1977, Fabb 1984). If we consider the syntax of -ble, it is generally acknowledged that, as in passive constructions, the grammatical subject of regular -ble adjectives is equivalent to a notional object in the corresponding sentence with a related verb. The correlation between a modal passive verbal structure and a -ble adjective is found in numerous general grammars in the form of paraphrases. For instance, Jung (1967) states that, in German, (87a) would correspond to (87b) or (87c). (87) a. Das Gerät kann vielseitig verwendet werden. (Ger) the.n device can.3sg.prs varied used be.pass ‘The device can be used in many different ways.’ b. Das Gerät ist vielseitig verwendbar. the.n device be.3sg.prs varied usable ‘The device is usable in many different ways.’ c. Das vielseitig verwendbare Gerät the.n varied usable device ‘the device usable in many ways’ In this section I review the empirical evidence for a treatment of -ble adjectives as related to the syntactic passive. First, I consider the projection of the external argument of the adjective. I continue with a revision of the empirical data that support the postulation of some passive component in the structure of -ble. The last subsection is devoted to the expression of the external argument of the verb, and its syntactic presence when not explicit. Both the passive component and the presence of the external argument are used to support the classification of -ble adjectives into high and low -ble.

44 It is important to note that not all works mentioned in the text provide complete analyses of these adjectives. Indeed, some of them describe (some of) their properties to different degrees of detail, whereas others provide more or less elaborate accounts. The reasons for mixing them all is to show how, in one way or other, this suffix has been a central target.

66   

   -BLE

3.3.1 The external argument of the adjective The first significant aspect related to the passive meaning of -ble adjectives that I would like to discuss is whether the subject of the adjective is a derived subject, so that it syntactically corresponds to the verbal internal argument, or whether the external argument of the adjective is directly projected as an external argument. The issue is not trivial and it is directly tied to the passive interpretation of the construction. Cinque (1990 : 36–37) has argued that the equivalent -bile adjectives in Italian are syntactically unergative, although they derive from passive/ ergative verbs. This, which seems rather contradictory in a syntax-based model (how can a predicate have its subject projected in an external position and at the same time be passive, if being passive involves having a complement that cannot be assigned accusative case?)45 is possible for him, because he assumes a lexical derivation for these adjectives that implies an internal morphological structure containing a change of category: -bile attaches only to verbs, resulting in a structure like (88).46 (88) [a [v ___ ]] He further assumes, following Chomsky (1986), that a category can select its object only under sisterhood. Applied to (88), this means that the verb can no longer theta-mark its internal object, which is thus syntactically projected via externalization, an operation not subject to this sisterhood condition. This is what happens with adjectival passive participles, and it is possible with -bile adjectives as well, because in both cases the verbal external argument is not projected. Thus, the change in the input theta-grid is a consequence of the morphological category change. As evidence, Cinque applies the test of ne-cliticization in Italian to prove that the external argument of the adjective is not an (underlying) object, as shown in (89).

45 The question is relevant, if we assume some version of Burzio’s generalization, which has been and still is in some way or other standardly presupposed in the analysis of unaccusative structures. 46 This assumption is already problematic. I have found examples like azzurrabile ‘azzurro. ble’ (that can become a member of the Italian soccer team, who is colloquially called azzurro), camionabile ‘truck.ble’, carrabile ‘carriage.ble’, carrozzabile ‘coach.ble’, ciclabile ‘cycle.ble’, papabile ‘Pope.ble’, rotabile ‘rail/track.ble’ or tascabile ‘pocket.ble’, all apparently derived from nouns. These forms have all a passive and a modal component, so that it is unclear where these features should be derived from if we assume Cinque’s (1990) lexical analysis in (88) above. See the discussion of these forms in chapter 5.



The external syntax of -ble   

   67

(89) a. * Ne sono confermabili/ condannabili/ desiderabili/etc. poche t Of.it are confirmable/ condemnable/ desirable few (di notizie) (It) (items of news) b. * Ne è giustificabile/ perseguibile/ truccabile/etc. la vendita t Of.it is justifiable/ prosecutable/ fixable the sale (Cinque 1990 : 36–37) It is unclear that this analysis can be extended to Catalan. When we apply the test of ne-cliticization, we get the corresponding Catalan sentences in (90).47 (90) a. ? En són confirmables/ condemnables/ desitjables poques t  of.them be.3pl.prs confirmable/ condemnable/ desirable few  (de notícies) (Cat)  (of news)  ‘Few of them are confirmable, condemnable, desirable (of news)’ b. ?N’és justificable/ perseguible la venda t  Of.it= be.3sg.prs justifiable/ prosecutable the sale  ‘The sale of it is justifiable / prosecutable’

47 The native speakers I have consulted are a bit uncertain about these sentences. For instance, Eulàlia Bonet (personal communication) considers them as rather forced – though not ungrammatical, and prefers a structure such as (i) instead. (i) N’hi ha poques de confirmables/ condemnables/ desitjables of.them=clloc have.3sg.prs few of confirmable/ condemnable  desirable (de notícies). (Cat) (of news) ‘There are few of them which are confirmable / condemnable / desirable (of news).’ Also Anna Bartra (personal communication) judges them as weird or at most very literary, but again not strictly ungrammatical. I must acknowledge that my informants and I also prefer (i) or (ii) to (90), although we do not consider the examples in the main text as ungrammatical as Cinque’s Italian examples are deemed. (ii) Se’n poden confirmar/ condemnar/ desitjar poques t (de notícies)(Cat) se=of.them can.3pl.prs confirm.inf condemn.inf desire.inf few of news ‘Few of them can be confirmed/condemned/desired (of news).’ Thanks also to Elena Castroviejo, Txell Granados, Jordi Jané, Xavier Jové, Lluís Macià, and JeanMarc Segarra for judgments on this structure. Neither of them (linguists and non-linguists) considered the sentences in (90) as ungrammatical.

68   

   -BLE

c. ?N’és acceptable una t (de solució)  Of.it=be.3sg.prs acceptable one (of solution)  ‘One of them is acceptable (of solutions)’ Although the evidence is rather weak, Catalan points to the derived status of the adjective’s external argument. When we consider other languages like English in an attempt to find support to decide on the syntactic status of the adjective’s external argument, arguments do not become much stronger either. According to Kayne (1981:  155) there is a trace in a sentence like (91), which signals “the presence of a syntactic representation for an object”, which makes -able “comparable to the passive”. No additional evidence is presented, though. (91) This booki is readable [NPi e] by a 10-year old. Also, Fabb (1984: 220), following Kayne (1981), suggests that -able behaves like passive participle -en on the basis of examples with a resultative construction, as in (92), a well-known context to determine the underlying position of the external argument of the adjective, because resultatives are predicated only of internal arguments. (92) a. Trout is hammerable [e] flat. b. Beef is eatable [e] raw. c. It is burnable [e] to ashes. Now, my English informants report all the examples in (92) as rather weird or even impossible. However, as Ora Matushansky (personal communication) suggests, assuming that productive -ble adjectives are built on the basis of a resultative passive, as I will propose in section 6, it is unlikely that two results can be structurally obtained.48 The examples in (92) may further be rejected on the grounds that, as pointed out in Williams (1983: 303), who attributes this suggestion to David Pesetsky, raising adjectives do not take adjective phrases as complements in English, as illustrated in (93). (93) *John is certain sick.

48 Note that this proposal would not account for Fabb’s reported examples in (92) above, nor for the very few weird though existing Spanish cases reported in footnote 6 in chapter 1.

The external syntax of -ble   



   69

Even though the empirical evidence is not very strong, for the analysis, I will assume that the adjective’s external argument is a derived argument, which is also intuitively in accordance with the passive nature of -ble adjectives. The discussion that follows will present enough evidence for a treatment of -ble as containing passive verbal functional structure, also along the lines of Cinque (1990), who suggests that -bile is lexically built on a past participial form.49

3.3.2 The passive component It has often been observed that -ble adjectives can only be formed from verbs that can also be passivized (Chapin 1967 or Riehemann 1993), as illustrated in (94) for English and in (95) for German. (94) a. Kate resembles her mother.

b. *Her mother is resembled (by Kate).



c. *Her mother is resemblable (by Kate).

(95) a. Der Vortrag dauert zwei Stunden. the talk last.3sg.prs two hours ‘The talk lasts two hours.’

(Ger)

b. * Zwei Stunden werden (von dem Vortrag) gedauert. two hours become.3pl.prs (by the talk) lasted c. * Zwei Stunden sind (von dem Vortrag) dauerbar. two hours be.3pl.prs (by the talk) lastbar (Riehemann 1993 : 13) The same is apparently50 true for Catalan or Spanish, as illustrated in (96) and (97) respectively.

49 See Meltzer-Asscher (2012) for a recent proposal according to which the adjective’s external argument is not generated in aP/AP but in a higher functional projection PredP. 50 See the discussion below for the meaning of ‘apparently’. For the moment I will just draw the reader’s attention to the relevant facts. Recall the existence of denominal adjectives in Catalan and Spanish, as well as the Spanish construction V todo lo V-ble, where apparently any verb may appear.

70   

   -BLE

(96) a. En Pau necessita els teus llibres. the Pau need.3sg.prs the your books ‘Pau needs your books.’

(Cat)

b. * Els teus llibres són necessitats (per en Pau). the your books be.3pl.prs need.part (by the Pau) c. * Els teus llibres són necessitables (per en Pau). the your books be.3pl.prs need.ble (by the Pau) (97) a. Pablo posee una gran casa. Pablo own.3sg.prs a big house ‘Pablo owns a big house.’

(Sp)

b. * Una gran casa es poseída (por Pablo). a big house be.3sg.prs own.part (by Pablo) c. * Una gran casa es poseíble (por Pablo). a big house be.3sg.prs own.ble (by Pablo) As shown in Chapin (1967), with transitive verbs that have two uses where only one of them allows passivization, only the passivizable one allows adjectivization. His examples are repeated in (98)–(99). (98) a. Rowena married Alfred.

b. *Alfred was married by Rowena.

(99) a. A rabbi married Alfred. b. Alfred was married by a rabbi. c. Alfred is marriable in a cathedral. Similar facts hold in Catalan and Spanish, as illustrated in (100)–(101) for Catalan.51 Whereas the stative reading of the verb mesurar ‘measure’ can be neither

51 As M. Teresa Espinal (personal communication) points out, the example in (i) is fine, which would be derived from a passive structure of the type in (ii), as expected. (i) El campanar és mesurable en metres. (Cat) the bell.tower be.3sg.prs measurable in meters ‘The bell-tower is measurable in metres.’

The external syntax of -ble   



   71

passivized nor merged with -ble, its dynamic counterpart is perfectly fine in both constructions. (100) a. El campanar mesura vint metres. (Cat) the bell.tower measure.3sg.prs twenty meters ‘The bell-tower measures twenty meters.’ b. * Vint metres són mesurats twenty meters be.3pl.prs measure.part c. * Vint metres són mesurables twenty meters be.3pl.prs measurable

pel campanar. by.the bell.tower pel campanar. by.the bell.tower

(101) a. En Pau va mesurar la taula. (Cat) the Pau aux.3sg.pst measure.inf the table ‘Paul measured the table.’ b. La taula va ser mesurada per en Pau. the table aux.3sg.pst be.inf measure.part by the Pau ‘The table was measured by Pau.’ c. La taula és mesurable. the table be.3sg.prs measurable ‘The table is measurable.’ Chapin further notes that restrictions found on the adjectival subject are the same as those on the corresponding verbal object, as exemplified in (102)–(103). Again, the other languages under investigation show the same pattern, as illustrated in (104)–(105) for Spanish. (102) a. to attain peace b. Peace is attainable. (103) a. *to read peace

b. *Peace is readable.

(ii) El campanar pot ser mesurat en metres. the bell.tower can.3sg.prs be.inf measure.part in meters ‘The bell-tower can be measured in metres.’

(Cat)

72   

   -BLE

(104) a. alcanzar la paz (Sp) attain.inf the peace ‘to attain peace’ b. La paz es alcanzable. the peace be.3sg.prs attainable ‘Peace is attainable.’ (105) a. * leer la paz (Sp) read.inf the peace b. * La paz es legible. the peace is readable Finally, Chapin also notes a certain parallelism between reflexives and self-compounds: the former are not possible with passives, the latter are not fine with -ble adjectives. He further observes that the passive transformation and the reflexive transformation cannot apply both to the same structure, as shown in (106). This author further argues that the reflexive transformation is involved in the formation of self- compounds. He then concludes that if -ble derivation involves the passive transformation, the absence of self- compounds follows straightforwardly without extra machinery, because of the incompatibility of reflexives and passives.52 (106) a. John shaved himself.

b. *John was shaved by himself.

But John was shaved by the barber.

52 It is not possible to find similar data for Catalan or Spanish, since there is no corresponding reflexive self- compound. As shown in (i), the particle auto- does not show the same restrictions in the formation of adjectival compounds. (i) a. forn autonetejable (Cat) oven selfcleanable ‘self-cleaning oven’ b. mecanismo autobloqueable (Sp) mechanism selflockable ‘self-locking mechanism’ Anna Bartra suggests a piece of evidence in (ii), if we use a more complex structure, which resembles the English contrast in the main text. (ii) a. * inflamable per ell mateix (Cat) inflammable by it self b. autoinflamable selfinflammable

The external syntax of -ble   





   73

c. *Himself was shaved by John.

(107) a. This door locks itself. b. lockable

c. *self-lockable

Another interesting contrast that could be taken as an argument for the presence of passive structure inside -ble adjectives comes from the interaction of passive with modality. Palmer (1979: 88) notes that subject-oriented modal can53 interacts with passive voice as shown in (108). According to this author, the passive seems unlikely when there is reference to the person who has the ability, because “ability is not strictly isolable from the circumstances in which that ability is exercised.” There is no such restriction when there is a non-specific or indefinite agent. (108) a. John can lift that weight.

b. ?That weight can’t be lifted by John.



c. That weight can’t be lifted by anyone / by one man.

Similarly, Quirk et al. (1972: 807) observe that whereas modal can may express either ‘capacity’ or ‘possibility’ in an active sentence, as in (109a), the corresponding passive structure only conveys a ‘possibility’ reading, as illustrated in (109b), even with the long passive, i.e. when the agent is specified in a by-phrase. (109) a. Anyone can do it.

→ Capacity / Possibility

b. It cannot be done by anyone.

→ Possibility

Similar facts obtain in Romance, as illustrated in (110) for Catalan.54 Whereas the modal can receive either a capacity or a possibility reading in the active sentence, only the possibility reading is available in the corresponding passive.

53 There are two types of dynamic possibility, according to Palmer (1979: 14), neutral and subject oriented. The latter is defined as “possibility that is linked to the subject,” a concept broader than ‘ability’, so as to include both animate and inanimate subjects. 54 Gavarró and Laca (2002: 2721) report a contrast between epistemic and deontic poder ‘can’ when they interact with passive. Only the epistemic active and passive are equivalents, as in (i). With deontic poder ‘can’, in (ii), active and passive are not paraphrases. I think that this parallels the ability versus possibility contrast described in Quirk et al. (1972) for English, and

74   

   -BLE

(110) a. Qualsevol pot justificar aquesta afirmació. anyone can.3sg.prs justify.inf this statement ‘Anyone can justify this statement.’ → Capacity / Possibility (Cat) b. Aquesta afirmació pot ser justificada per qualsevol. this statement can.3sg.prs be.inf justify.part by anyone ‘This statement can be justified by anyone.’ → Possibility My claim is that the same kind of interaction happens with -ble adjectives. If we now consider the corresponding -ble adjectives, only the possibility reading seems to be available. That is, whereas the sentence Anyone can do it in (109a) has both a capacity and a possibility interpretation, only the latter is available in (109b) and in the parallel -ble adjectival construction in (111a). Likewise for the Catalan example in (111b). (111) a. It is doable by anyone. b. Aquesta afimació és justificable per qualsevol. (Cat) this statement be.3sg.prs justifiable by anyone ‘This statement is justifiable by anyone.’ However, at least for Catalan, there are apparently cases where both readings, ability and possibility, are available. Crucially, these are cases with intransitive verbal bases that have no passive interpretation.55

also exemplified in (110) above for Catalan. (i) a. El lladre podia haver estripat el quadre. (Epistemic active) (Cat) the thief can.3sg.pst have.inf tear.part the painting ‘The thief could have torn the painting.’ b. El quadre podia haver estat estripat pel lladre. the painting can.3sg.pst have.inf be.part tear.part by.the thief ‘The painting could have been torn by the thief.’ (Epistemic passive) (ii) a. El metge podia visitar els pacients. (Deontic active) (Cat) the doctor can.3sg.pst visit.inf the patients ‘The doctor could visit the patients.’ b. Els pacients podien ser visitats (pel metge) (Deontic passive) the patients can.3sg.pst be.inf visit.part (by.the doctor) ‘The patients could be visited (by the doctor).’ 55 Bosque (1999: 255) suggests that the modal nature of the adjective could be the instigator of the passive interpretation in -ble adjectives. This should relate to the passive interpretation of the infinitival in tough-constructions, which would be triggered by the modality of the adjective, as well. This seems problematic for cases like durable, which are

   75



The external syntax of -ble   

(112) a. Aquest material pot durar this material can.3sg.prs last ‘This material can last.’

→ Capacity and Possibility (Cat)

b. Aquest material és durable → Capacity and Possibility this material be.3sg.prs durable ‘This material is durable.’ I will take this as evidence that -ble adjectives contain both some passive component and a modal syntactic layer (see section 4 below for additional evidence for the structural presence of such a modal operator). The modal operator interacts with the passive component, forcing a possibility interpretation of the predicate. As we will see in the discussion on the expression of the verbal external argument, only a non-specific external argument seems compatible with the absence of capacity. McGinnis (2010) has recently provided new evidence from English on the parallelism between passive constructions and -ble adjectives. She shows that in the same way that passives of exceptional case-marking (ECM) verbs like believe allow raising, as in (113a), it is possible to find cases of raised subjects with -able adjectives, as exemplified in (113b). (See also her appendix of Googled examples). Interestingly, as shown in (113c), raised subjects are only possible with high -ble, not with low -ble. (113) a. This film can easily be predicted [t to be a flop at the box office]. b. This film is easily predict#able [t to be a flop at the box office]. c. * This film is very predict+able [t to be a flop at the box office]. (McGinnis 2010: 7–8)56

modal but interpreted as actives. As I see it, the question is whether we are dealing with a passive interpretation of a non-passive structure or with a passive configuration. If the former, we would have to explain why the presence of modality does not always trigger a passive interpretation. If the latter, we should account for the absence of passive morphology in these constructions. All tests seen so far, as well as crosslinguistic variation seem to argue for the presence of some passive feature in the structure of regular high -ble. 56 McGinnis (2010: 7) uses SPE’s way of distinguishing between a [+abl] suffix with a weak morpheme boundary, and a [#abl] suffix with a strong word boundary, as exemplified in (i). (i) a. predict#able: capable of being predicted b. predict+able: dull, lacking originality

76   

   -BLE

To conclude this subsection, we find exactly the same restrictions on the kind of predicates and verbal objects that can target a syntactic passive and -ble formation.57 In addition, we see the same kind of syntactico-semantic effects deriving from the interaction of modality with the passive component. Finally, there is also a clear parallelism between reflexive passives and self- prefixed -ble adjectives, as well as in the availability of ECM structures in both constructions in English.

3.3.2.1 Non-passive -ble adjectives It has been generally observed that those -ble adjectives derived from intransitive bases are interpreted as active, but they are passive if the base is transitive. As noted in Gràcia (1995: 115), there are a few adjectives with active interpretation that derive from transitive bases. She lists the ones in (114) and analyzes them as deriving from unaccusative bases as in Belletti and Rizzi (1988).58 (114) a. delectable, espantable, espaordible, honorable (Cat) delightable, frightenable, scareable, honorable

b. ‘that V’

Interestingly, these adjectives all seem to express some causation (Pesetsky 1987, 1995; Arad 1998). A similar causative component could be said to be contained in other active -ble adjectives in Catalan, such as the ones in (115). Similar data can be found in Spanish. (115) agradable, convenible (Cat) pleasant, suitable The other active adjectives that can be found in Catalan, they all contain a modal component (see section 4 for this modal component).

57 See Di Sciullo (1997) for an account of (some of) these restrictions on -ble that does not take into account the passive nature of -ble. 58 The Catalan native speakers I have consulted and myself do not have the active readings for these -ble adjectives, even though the first three are given the interpretation in (114b) in the DIEC. If anything, I would interpret them as all other -ble adjectives, e.g. algú espantable ‘someone that can be frightened’. As for honorable, I would rather interpret it as ‘worth of honor’ rather tan ‘that honors’, i.e. with some modality. In this case, both meanings are listed in the DIEC.



The external syntax of -ble   

   77

(116) a. fal·lible, flotable, servible,59 transpirable (Cat) fallible, floatable, be.useful.ble, perspirable b. pot {fallar, surar, servir, transpirar} can.3sg.prs   fail.inf float.inf be.useful.inf perspire.inf ‘it can {fail, float, be useful, perspire}.’ Finally, two more adjectives are special in that their underlying verbal bases imply some ‘durative measure’, as illustrated in (117)–(118), i.e. in their positive use, they express that the entity they are predicated of has some (contextually specified) duration. The material in parentheses need not be uttered with some NPs, but it is certainly always understood. (117) a. durable lasting

(Cat)

b. La pel·lícula ha durat ?(molt de temps) the film have.3sg.prs lasted  (much of time) ‘The film has lasted a lot.’ c. Les bateries duren ?(molt de temps) the batteries last.3pl.prs  (much of time) ‘Batteries last a lot.’ (118) a. perdurable (Cat) everlasting b. El seu record ha perdurat (en el temps). the her memory have.3sg.prs lasted (in the time) ‘Her memory has remained for long (in time).’ Thus, active adjectives have either a causative component, a modal component, or a durative component. Similar facts obtain in English. The following data are reported as active adjectives in Adams (2001: 34–35) or Marchand (1969).60 In (119) we have adjec-

59 Servible could also be understood as que fa servei ‘that makes service’, i.e. as expressing causation. 60 I do not include here some of Adams’s (2001) “subjective” examples like changeable or Marchand’s (1969) passable, because they are at least ambiguous between a passive and an active reading, as these authors already observe.

78   

   -BLE

tives that can be said to incorporate some causation. The examples in (120) correspond to those active adjectives that convey a modal meaning. (119)  agreeable, comfortable, companionable, favourable, fashionable, serviceable (120) answerable, breathable, perishable The question that arises is what relates causation, modality and duration, and also how they are related to the passivity of -ble adjectives, i.e. why they are relevant to this research. I will deal with the relevance of the durative component at the end of chapter 3, where I will suggest an explanation for the existence of forms like durable ‘lasting’. As for the other two components, it is interesting to note that other languages use the same suffix to express both causation and potential modality, e.g. Chichewa -ka.61 On the other hand, Flury (1964: 102) observes that the interpretation of German active adjectives such as verwitterbar ‘weatherable’ derived from eventive intransitive verbs (Vorgangsverben), which are the majority among those derived from intransitive bases, are very close to a passive reading in that their adjectival argument would be best analyzed as an object on which some force of nature exerts some influence. Thus, we can try to relate those adjectives with a causative feature, e.g. agradable ‘agreeable’, delectable ‘delightable’ (and possibly also transpirable ‘perspirable’) and the regular ones with some agentive feature, by assuming that they all contain or imply some originator or initiator (in the sense of Borer 2008), which includes the natural force of eventive unaccusative verbs that is acting on some (underlying) object. I would like to argue that the requirement to become an adequate base for -ble derivation is not limited to having an internal argument, but it is also necessary that some kind of originator is (lexically,62 conceptually or syntactically) available, as was already advanced in section 3.1 above. That is, an appropriate base for -ble derivation must conform to the two prerequisites in (121), where originator includes agents, causers, natural forces, and alike. (121) a. An originator (syntactically, lexically, or conceptually available)

b. An internal theme (or locative) argument63

61 See the grammar on Intensive Chichewa (1969). 62 That is, ‘encyclopedically’. 63 It is unclear why this should be so, but Locative is apparently needed to cover examples (37d) above such as Catalan esquiable ‘skiable’, navegable ‘navigable’, passejable ‘walkable’,



The external syntax of -ble   

   79

This lexically or conceptually accessible causative or agentive feature needs explanation. I have already shown that verbs expressing only an internallycaused change of state are impossible as bases for -ble derivation,64 as illustrated in (122)–(123) for English and Catalan, respectively. (122) *arrivable, *di(e)able (123) * venible, *creixible (Cat) come.ble  grow.ble The point is that those verbs expressing both an internally or an externallycaused change of state, such as inflamar ‘burn’ or variar ‘change’, are fine with -ble (subject, to the general constraints for -ble derivation), even when the adjective conveys the internally-caused change of state. For instance, the adjective changeable is ambiguous between a meaning corresponding to the intransitive verb, as in (124a), and the interpretation related to the transitive base, as in (124b) (examples from Adams 2001). Examples like these have led previous approaches to the inaccurate conclusion that unaccusative verbs are appropriate bases. They are not, unless they can be related to a transitive variant that refers to a causative change of state.65, 66 (124) a. changeable weather

‘that can change’

b. changeable arrangements

‘that can be changed

transitable ‘passable’, vivible ‘liveable’, all unergative verbs with a locative argument. This could perhaps be solved if we take the notion theme as broadly defined in Anderson (1977: 367), as the entity that moves with motion verbs, the entity whose location is described with locative verbs, the patient of many transitive verbs or the entity that experiences the action expressed by the verb. Indeed, the notion of theme required by -ble will be extended so as to cover any argument that measures out the event denoted by the verb in chapter 3. Note that the kind of locative argument that can be selected by -ble is also a measurer (perhaps with the exception of vivible ‘liveable’ in una ciutat vivible ‘a liveable city’). 64 Though they may appear in the Spanish construction V todo lo Vble ‘V all lo Vble’. 65 A similar explanation is given in de Miguel (1992) for the analysis of unaccusative verbs as predicates of reduced participial clauses (construcciones de participio absoluto, CPA). An imperfective unaccusative verb cannot appear in the CPA unless it can be related to a transitive variant that refers to an active perfective event. 66 Another possibility for the analysis of change of state verbs would be to assume Hale and Keyser’s (1988) suggestion that the two uses of these verbs have the same structure, so that some notion of causation is present in both variants. It is indeed easy to see that the intransitive changeable can be said to imply at least some natural force.

80   

   -BLE

Now, I have suggested an additional requirement imposed by -ble adjectives on their bases: some implication of an Agent, Cause or natural Force is necessary to be selected as a base for -ble derivation. We must now determine whether this implied Agent or external Cause/Force is syntactically active or it is just conceptually implied by the encyclopedic meaning of the root. This is discussed in the next subsection together with other syntactic properties of -ble adjectives.67

3.3.3 The external argument of the verb Another issue related to the passive construction is the expression of the external argument. As is well-known, it is generally possible to express the notional subject with verbal passives. The various authors that have investigated -ble adjectives disagree with respect to the expression of the external argument with -ble adjectives, or else, they either exclude this possibility (Randall 1985) or they do not even mention this option (Aronoff 1976). Chapin (1967) considers that the Agent is deleted, although he acknowledges dialectal variation for some -able forms with very special by-phrases: by anyone and by any N are dialectally acceptable. On the other side, Kayne (1981, 1984), Fabb (1984), Roeper (1987) or Pesetsky (1995) observe that, as shown in (125) for English, it is possible with -ble adjectives to express what corresponds to the grammatical subject of the underlying verb. (125) a. The ball is retrievable by a dog. (Roeper 1987 : 303) b. That is deniable by any intelligent person. (Kayne 1981 : 156) One can also find analogous cases in German, as exemplified in (126).

67 Roeper and van Hout (1999: 204) state that the “empirical generalization is that -able requires a transitive verb with an Agent and a Theme”. Although similar, this statement and my proposal differ in crucial ways. I have already shown that the constraint on transitivity already excludes a number of (new) cases (e.g. skiable). In addition, the restriction on having an agent could not be extended to Romance, because it would leave the cases examined in chapters 3 and 4 unaccounted for.



The external syntax of -ble   

   81

(126) a. Der Kontostand ist von unseren Kunden jederzeit abrufbar. (Ger) the balance is by our customers anytime inquirable ‘The balance is at any time inquirable for our customers’ (Riehemann 1993 : 6) b. ein vom Benutzer (leicht) modifizierbares Programm a by.the.dat user (easily) modifiable program ‘a program easily modifiable by the user’ Similar facts hold in Romance as well, as illustrated in (127)–(128) for Catalan and Spanish, respectively. (127) a. ? un llibre fàcilment recomanable per qualsevol que l’hagi llegit (Cat) a book easily recommendable by anyone that it=has read ‘a book easily recommendable by anyone that has read it’ (Gràcia 1995 : 101)68 b. una opinió pública (fàcilment) manipulable pels mitjans de an opinion public (easily) manipulable by.the media of comunicació communication ‘a public opinion easily manipulable by the media’ (128) a. El envase es retornable por el interesado, sin the container is returnable by the interested.party, without cargo alguno. (Sp) charge any ‘The container is returnable by the interested party without any charge.’ (Duo de Brottier 2000) b. tapón auditivo moldeable por el usuario (W) plug auditory moldable by the user ‘earplug moldable by the user’ c. una novela adaptable al cine por un buen guionista a novel adaptable to.the cinema by a good scriptwriter

68 My informants and I find this sentence completely fine.

82   

   -BLE

3.3.3.1 Constraints on the expression of the verb’s external argument Passive differs from -ble adjectives in the type of restrictions that constrain the expression of the verbal arguments. When discussing the expression of the indirect object and adverbial modification, I have suggested that they are severely constrained by the genericity of the characterizing predicate and the specificity of the NP argument. In the case of the external argument, the contrast between the examples in (129)–(132) below with those in (125)–(128) suggest that the same kind of restriction is at stake. Note the non-specificity and genericity of the byphrases by a dog, vom Benutzer ‘by the user’, per qualsevol que l’hagi llegit ‘by anyone that has read it’, por el interesado ‘by the interested party’ in the grammatical examples above. (129) * un llibre recomanable per en Pau (Cat) a book recommendable by the Pau (130) * That is deniable by Kate. (131) * Der Kontostand ist von Klaus jederzeit abrufbar. (Ger) the.m balance is by Klaus anytime inquirable (132) a. * tapón auditivo moldeable por Pablo (Sp) plug auditory moldable by Pablo b. * una novela adaptable al cine por Pablo a novel adaptable to.the cinema by Pablo My contention is that the ungrammaticality of these examples is what has led a number of researchers to overlook the expression of the external argument altogether (Aronoff 1976) or to reject that possibility (Randall 1985). Also for Spanish, the suffix -ble has been said to generally lose the external argument (Piera and Varela 1999: 4396).69 On the other hand, some authors have mentioned that only when the by-phrase denotes a less specific or generic agentive is it possible to make the external argument explicit in the sentence (e.g. Roeper 1987 for English;

69 For Dutch, according to Booij (1992), the agent argument is lost with -baar adjectives, as well. Note, however, that the example in (i) that he provides as evidence is necessarily deviant. As will be discussed more extensively below, the expression of agentive arguments with these adjectives is subject to restrictions related to specificity. Although an investigation of Dutch -baar is well beyond the scope of this research, note that Googled examples like (ii) seem to confirm that such restrictions are also applicable to Dutch -baar.



The external syntax of -ble   

   83

NGRALE § 7.10g), however, they do not explicitly deal with this issue. Also Riehemann (1993) observes that it is in principle not normally possible to express the agent in a by-phrase in German, as illustrated in (133b), though it is acceptable if it is not specific, as we saw in (126) above with abrufbar ‘inquirable’ and modifizierbar ‘modifiable’. (133) a. Maria kann Deutschland mit Australien vergleichen. (Ger) Maria can Germany to Australia compare b. * Deutschland ist nicht von Maria mit Australien vergleichbar. Germany is not by Maria to Australia comparable (Riehemann 1993 : 6) Note that when the by-phrase is generic or non-specific, we find the expected contrast in German doublets selecting -bar and -abel. Only high potential -bar can license the expression of a non-specific external argument in a by-phrase, as exemplified in (134). (134) a. Die Themen waren von jedem diskutierbar. (Ger) the topics be.3pl.pst by everyone.dat debatable ‘The topics were debatable by everyone.’ b. * Die Themen waren von jedem diskutabel. the topics be.3pl.pst by everyone.dat debatable The property of genericity or specificity (Krifka et al. 1995; Enç 1991; see also Leonetti 1999; Brucart and Rigau 2002 for the notion of specificity as applied to Spanish and Catalan respectively) appears as a crucial factor in determining the properties of -ble adjectives, for it intervenes in a number of cases and has an influence on grammaticality judgments. In principle, the incompatibility of specific by-phrases with -ble adjectives is in fact expected. It seems unreasonable that an inherent-property-denoting construction can be thought as resulting from the intervention of a particular and specific agent. The question is whether this

(i) (ii)

Dit verschijnsel is verklaarbaar (*door mij) (Dut) this phenomenon is explainable by me ‘This phenomenon is explainable (*by me)’ voorspelbaar door de meteorologische dienst(W) (Dut) predictable by the meteorological service ‘predictable by the meteorological service’

84   

   -BLE

has to be represented in the structure, and if it has, how it is to be represented. That is, whether there is some specific functional head or some specific feature doing this job, or else this constraint results from the interaction of different features in a particular configuration. Consider the sentence in (135), which is perfectly grammatical with a specific external argument, despite being very similar to the ungrammatical (132b) above.70 (135) Una novela adaptable al cine por Pedro Almodóvar a novel adaptable to.the cinema by Pedro Almodóvar debe ser excéntrica. (Sp) must.3sg.prs be.inf excentric ‘A novel adaptable to the cinema by Pedro Almodóvar must be excentric.’ Although the by-phrase apparently refers to a particular and specific Pedro Almodóvar, the nominal is in fact interpreted as a kind, something like (136), not as denoting the individual named Pedro Almodóvar. (136) alguien como Pedro Almodóvar (Sp) somebody like Pedro Almodóvar I discuss this and similar examples in section 4.4.2, where I tie the restriction on the non-specificity of the external argument to the syntactic presence of a modal operator in the internal structure of -ble, which also conditions the non-specificity of other verbal arguments. For the time being, let me show that this restriction is not specific to -ble, but it has also been observed in Spanish passive sentences, which I will take as further support for the presence of a passive component in the structure of -ble. As noted in de Miguel (1992, 1999) and Mendikoetxea (1999a, b), by-phrases in periphrastic passive sentences with imperfective or stative verbs in Spanish are possible only when the notional subject is generic, (137a)–(137b), independently of grammatical aspect, whether perfective or imperfective. As shown in (137c), this restriction, whose origin, whether syntactic or semantic, is unclear according to Mendikoetxea (1999a : 1619), does not affect perfective verbs.

70 Thanks to M. Teresa Espinal for bringing this sentence to my attention.



The external syntax of -ble   

   85

(137) a. * El documento era / fue conocido por the document be.3sg.pst.ipfv be.3sg.pst.pfv know.part by Juan (Sp) Juan (de Miguel 1992: 208) b. El documento es conocido por todos. the document be.3sg.prs know.part by all ‘The document is known by everyone.’ c. La casa fue construida por Juan / por todos. the house be.3sg.pst build.part by Juan by all ‘The house was built by Juan / by everyone.’ (Mendikoetxea 1999a : 1619) Mendikoetxea (1999b) further points out that Subject-Experiencer verbs, which are generally lexically imperfective, form syntactically imperfective passives that require generic subjects. When the notional subject is specific, this cannot be expressed in a by-phrase.71 (138) a. Las nuevas movilizaciones anunciadas son temidas por the new protests anounced be.3pl.prs fear.part by toda la sociedad. (Sp) all the society ‘The new announced protests are feared by society.’ (Mendikoetxea 1999b : 1683) b. * Las nuevas movilizaciones anunciadas son temidas por the new protests anounced be.3pl.prs fear.part by mis padres. my parents (Mendikoetxea 1999b : 1684) Agent-subject verbs do not show this restriction, although they do seem to prefer generic subjects with se-passives, as well, i.e. they are more natural when

71 As is well-known, there are two kinds of passives in Spanish, pronominal se passives and periphrastic passives built with the auxiliary verb ser ‘to be’. As the corresponding se passives in (i) show, the type of passive does not affect the argument on the parallelism between sentential passive and -ble.

86   

   -BLE

they have a universal quantifier or are interpreted as generic. According to Mendikoetxea (1999a, b), this preference is favored by the more imperfective and less intentional character of the pronominal passive. (139) a. el edificio se construyó por carpinteros (Sp) the building se build.3sg.pst by carpenters (Mendikoetxea 1999b : 1684) b. se aceptó por todos que el acuerdo era papel se accept.3sg.pst by all that the agreement be.3sg.pst paper mojado wet ‘It was accepted by everybody that the agreement was worthless.’ (Mendikoetxea 1999a : 1587) c. *? el cuadro se pintó por Goya the painting se paint.3sg.pst by Goya (Mendikoetxea 1999b : 1684) To summarize this very important point, Spanish passives, whether periphrastic or se-passives, show restrictions on the expression of the external argument in a by-phrase on the basis of their genericity or specificity (cf. de Miguel 1992: 208). Crucially, such restrictions are tied to the stativity of the construction. Thus, we find exactly the same constraints on the expression of a by-phrase in both passives and -ble adjectives. In the case of -ble I will argue that the stativity of the -ble structure derives from the syntactic presence of a modal operator that scopes over the functional heads that license the verbal arguments.

3.3.3.2 The implicit external argument When the external argument is not expressed in a verbal passive, it is not only clearly implied, but there is extensive evidence showing that it is syntactically present. A variety of tests have been commonly used to prove this presence. First, the implicit agent of a passive clause can control into a purpose clause. Second, a passive verb can be modified by agent-oriented adverbials, which are licensed by an agent phrase. Third, passives can license an instrumental phrase as well, thus showing that there is a syntactically active agent. I will use these tests to prove that there is an implicit external argument in -ble adjectives, as well. Some authors have acknowledged that -ble adjectives have an implicit external argument, e.g. Keyser and Roeper (1984), Roeper (1987), Bhatt and Pancheva

The external syntax of -ble   



   87

(2005). Similarly, native speakers have a strong intuition in this respect as well, which gives rise to contrasts like (140).72 (140) a. #The pond is freezable.

b. The food is freezable.

According to Roeper (1987), there is an implicit agent in regular -ble adjectives, which can function as a controller, although this process is less productive than in passives, and speakers’ judgments show that it may also be subject to dialectal variation. This is illustrated in (141) with his examples. I provide additional examples in (142). (141) a. Some goods are insurable to decrease the profit risks. b. Goods are exportable to improve the economy. (Roeper 1987 : 280) (142) a. the light source is orientable to allow correction of the light range (W)

b. All seating, staging, masking etc is rapidly removable to leave a flat floor exhibition space of 775 square meters. (W)

Interestingly, as with the expression of the external argument, Roeper (1987: 303) mentions that “rationale clauses are less acceptable unless they are clearly generic”, as the contrast in (143) shows. In other words, the fact that it is a less productive process seems clearly tied to its contributing a characterizing interpretation of the entire structure. (143) a. ?A person is discardable to make money. b. People are discardable to make bigger profits. (Roeper 1987 : 303) Agent-oriented modifiers of the type carefully, cleverly, deliberately, voluntarily are generally not possible with -ble adjectives in English. I have only found a few examples in (144).

72 Note that (140a) would become fine if freeze is interpreted as a transitive verb, e.g. This pond is freezable with our new machinery (uttered in a sci-fi novel, for instance).

88   

   -BLE

(144) a. a cleverly usable object b. some deliberately modifiable design Some more examples of -ble adjectives with agent-oriented modifiers like manually, deliberately or skillfully seem possible in Catalan and Spanish, as illustrated in (145)–(146) respectively, although they are mostly infrequent and hard to find. (145) a. la bici més venuda d’aquesta marca, bloquejable manualment the cycle most sold of=this brand, lockable manually al manillar (W) (Cat) at.the handlebars ‘this brand’s most sold cycle, which is manually lockable at the handlebars’ b. una opinió pública controlable hàbilment per l’aparell de a opinion public controllable skillfully by the=machine of l’Estat the=State ‘a situation skilfully controllable by the State machine’ (146) a. una opinión pública manipulable deliberadamente por los medios (Sp) a opinion public manipulable deliberately by the media ‘a public opinion deliberately manipulable by the media’ b. una situación controlable hábilmente por el gobierno a situation controllable skillfully by the government ‘a situation skilfully controllable by the government’ The difficulty of finding examples with this kind of adverbs does not challenge the possibility of expressing the external argument or its implicit presence, though. Rather, it is related to the stativity of -ble, discussed in section 4. As pointed out in Dowty (1979: 55), the impossibility of the adverbs deliberately or carefully to occur with stative predicates is precisely one of the tests that is generally used to distinguish states, like *John deliberately knew the answer, from dynamic events, such as John ran carefully. A very specific manner adverb that appears frequently with -ble adjectives (as it does, more often than not, with middles), easily, deserves special attention. This adverb has been extensively analyzed in the context of middles, although authors disagree with respect to its syntactic function, often related to the licensing of



The external syntax of -ble   

   89

the implicit argument postulated for middles (see e.g. Roberts 1987; Hoekstra and Roberts 1993; also Condoravdi 1989 for a different view). I believe that there is a contrast in the interpretation of the sequences in (147) and (148), depending on the position of the adverbial, and the type of adjectival: more verbal adjectives, in the sense of being more dynamic or eventive, like realizable, allow both positions pre- and post-adjectival, whereas those that are, say, ‘more qualifying’, like admirable dislike or do not allow at all a post-adjectival easily adverb in Catalan. The contrast in (148) reveals that in (148a) the adjective receives a specialised meaning, which means that it has become ‘more adjectival’ and has, therefore, almost lost its verbal properties. These verbal properties are maintained in (148b), and as a consequence both positions pre- and post-adjectival remain available for the adverbial. (147) a. una obra fàcilment realitzable – una obra realitzable fàcilment a work easily realizable – a work realizable easily per qualsevol (Cat) by anyone ‘A piece of work that is easily realizable (by anyone)’ b. una actitud fàcilment admirable – *una actitud admirable an attitude easily admirable –  an attitude admirable fàcilment73 easily ‘An easily admirable attitude’ (148) a. una persona fàcilment accessible – ??una persona accessible a person easily approachable –   a person approachable fàcilment (Cat) easily ‘An easily approachable person’

73 This sequence would only make sense if we understood admirable as meaning ‘that it can be admired’ and not as ‘that deserves admiration’. Below in § 4.2 I will suggest that there are two semantically different types of -ble adjectives, which correlate with the two types of adjectives that I have already identified. Specifically, I will argue that admirable (in its meaning ‘that deserves admiration’ or ‘worth admiring’) belongs to the set of low -ble adjectives.

90   

   -BLE

b. una muntanya fàcilment accessible – una muntanya accessible a mountain easily accessible a mountain accessible fàcilment pel Sud easily through.the South ‘An easily reachable mountain – a mountain easily reachable from the South’ As for the third test, it is generally assumed that an instrumental phrase requires a transitive predicate and a subject that can be interpreted as an agent or causer to be licensed (see Gruber 1965; Kayne 1994; Rigau 2002; inter alia. But see also Hale and Keyser 1987 against this view). Instrumental phrases seem to be allowed with -ble adjectives, as exemplified in (149) for English and (150) for Catalan, which argues for an analysis of these adjectives as having some syntactically relevant agent or causer.74 (149) a. The surface is washable with regular dish soap and a damp sponge. (W) b. It’s removable with a 5 mm hex wrench to shorten the axle. (W) (150) a. El teixit és rentable amb un detergent neutre. (Cat) the fabric be.3sg.prs washable with a detergent neutral ‘The fabric is washable with a neutral detergent.’ b. Aquesta pàgina web és dissenyable amb Dreamweaver. this page web be.3sg.prs designable with Dreamweaver ‘This webpage is designable with Dreamweaver.’ However, Randall (1985: 42–43) maintains that -ble adjectives cannot appear with instrumental PPs modifying the verbal predicate, as the ones in (151), although they are allowed with sentential instrumental adjuncts, as shown in (152). (151) a. *The delicate sweaters are washable with Borax.

b. *The bolt was removable with a crescent wrench.

(152) a. Calculus is learnable with the right kind of instruction. b. This mountain is climbable with the right boots.

74  See Alexiadou and Schäfer (2006) for a recent proposal that views instrument phrases as being themselves agents or causers.



The external syntax of -ble   

   91

Let us check that our instrumental phrases in (149)–(150) above are indeed verbal modifiers and not sentential adjuncts. Rigau (2002: 2055 f ) observes that, in Catalan, a dislocated instrumental phrase can be recovered through a clitic hi only when it is a verbal modifier, but not when it is a sentential adjunct.75 The examples in (153) show the contrast. (153b) is a Catalan example analogous to (152b). (153) a. Amb Dreamweaver, s’hi pot dissenyar la pàgina with Dreamweaver se=clloc can.3sg.prs design the page web.(Cat) web ‘The web page can be designed with Dreamweaver.’ b. Amb les botes adequades, aquesta muntanya es (*hi) with the boots right this mountain se (clloc) pot escalar. can.3sg.prs climb ‘Wearing the right boots, this mountain can be climbed.’ According to Rigau, when dislocated constructions like those in (153) are negated, the sentential adjunct remains outside the scope of negation, whereas this is not so in the case of the verbal adjunct, due to the presence of the resumptive pronoun, which is in the scope of negation. The contrast in (154) confirms that the examples provided in (149)–(150) above are indeed of the right type, i.e. they are instrumental phrases whose function is modifying the verbal predicate not the entire sentence, thus providing support for the view that an implicit agent must be present in the structure in order to license this phrase. 76

75 I use here the corresponding verbal predicates instead of -ble adjectives, because with adjectives the clitic does not appear, or else judgments are rather hard to get. This should in no way weaken the force of the argument, for the instrumental phrase will be either verbal or sentential whether in a sentence with a verbal predicate or with an adjectival predicate. 76 Bosque (1990: 192) distinguishes between means and instrumental phrases (see also Marantz 1984), suggesting that the latter require an agent, whereas the former need just an event argument. If right, since the examples (149a)–(150a) in the main text are means phrases, this would mean that they cannot support the presence of an agent. Randall’s examples are means phrases, as well. It is harder to find with-phrases that are not means phrases, although the NPs in (i)–(ii) could be valid examples. (i) piscina cobrible amb una coberta de lona o plàstic (Cat) swimming.pool coverable with a cover of canvas or plastic ‘swimming-pool that is coverable with a canvas or plàstic cover’

92   

   -BLE

(154) a. Amb Dreamweaver, no s’hi pot dissenyar la pàgina with Dreamweaver not se=clloc can.3sg.prs design the page web.(Cat) web ‘The web page cannot be designed with Dreamweaver.’ b. Amb aquestes botes, aquesta muntanya no es pot with these boots this mountain not se can.3sg.prs escalar. climb ‘Wearing these boots, this mountain cannot be climbed.’ The question is why Randall’s examples in (151) are ungrammatical. The answer should be rather obvious for (151a). In this case, the instrumental refers to a very specific detergent, Borax, so that I would like to suggest that we are faced here with another effect of the above-mentioned restriction on the specificity of byphrases with -ble adjectives. As regards (151b), notice that the sentence is very similar to (149b), they only differ with respect to tense. I think that the sentence is deemed ungrammatical in Randall (1988), because it is assumed to correspond

(ii) un treball redactable amb el portàtil (Cat) a work writeable with the laptop ‘an assignment writeable with a laptop’ The argumentation in the main text is also applicable to these examples, as shown in (iii)–(vi). (iii) Amb una coberta de lona, la piscina s’hi pot cobrir.(Cat) with a cover of canvas the swimming.pool se=clloc can.3sg.prs cover.inf ‘The swimming-pool can be covered with a canvas cover.’ (iv) Amb una coberta de lona, la piscina no s’hi pot cobrir. (Cat) with a cover of canvas the swimming.pool not se=clloc can.3sg.prs cover.inf ‘The swimming-pool cannot be covered with a canvas cover.’ (v) Amb el portàtil, el treball s’hi pot redactar. (Cat) with the laptop the work se=clloc can.3sg.prs write.inf ‘The assignment can be written with the laptop’ (vi) Amb el portàtil, el treball no s’hi pot redactar. (Cat) with the laptop the work not se=clloc can.3sg.prs write.inf ‘The assignment cannot be written with the laptop’ Gemma Rigau (personal communication) observes that the examples in (iii)–(vi) are rather hard, and they would become better with the presence of an explicit agent, as in (vii). (vii)  En Pere hi podria cobrir la piscina, amb aquest plàstic.(Cat) the Pere clloc can.3sg.cond cover the swimming.pool with this plastic ‘Pere could cover the swimming-pool with this plastic.’ I certainly agree. If I left the impersonal forms is just to keep the parallelism with the -ble forms.



The external syntax of -ble   

   93

to a sentence that refers to a single event of removing in the past, Sue removed the bolt with a crescent wrench, which is not possible, because -ble adjectives predicate general properties. If (151b) is taken as predicating a property that was held to be true in the past, my informants judge it good.77 To summarize the discussion so far, all tests that have been examined clearly point to the syntactic presence of an implicit external argument that can be made explicit in a by-phrase if it is generic or non-specific.

3.3.4 Two main types of -ble Up to now, all standard tests that are commonly used to argue for the presence of verbalizing and aspectual functional heads have been applied to -ble adjectives that belong to the set of potential or high -ble adjectives. In the case of the expression of the external argument in a by-phrase, Kayne (1981, 1984), Fabb (1984) and Roeper (1987) distinguish between two groups of -able adjectives, depending on their compatibility with by-phrases, which correlate with high and low -ble. (155) a. *That is not available by minors b. French lessons are dispensable with by most people (Kayne 1981 : 156) (156) a. *irreversable by the judge78

77 Notice that if the examples we have examined have been correctly interpreted, this is a clear evidence against Randall’s (1985: 63) Inheritance Principle, according to which a derived form “inherits the full subcategorization of its base,” unless the relation between the two is distant, i.e. the two forms differ in two respects, category and meaning. In the latter case, the derived items “are treated as underived forms with respect to complementation, and will take an unmarked subcategorization, [ __ NP]” (p. 61). Since -ble adjectives have a distant relation to its base, this would mean that the external argument of the base verb, as well as the instrumental modifier, should not possibly be realized with -ble adjectives, contrary to the facts illustrated in the main text above. 78 Other examples that Fabb (1984) provides are: corruptable by bribes and digestable by babies with their corresponding deviant negated counterparts, *incorruptable by bribes and *indigestable by babies. I have a problem with these examples, for although they may exist for a reduced number of speakers, they are deemed weird by the native speakers I have consulted, because they should actually be corruptible and digestible. This may be related to the fact that both -ible and -able are pronounced exactly the same [əbəl] (Wells 2000, Roach et al. 2006).

94   

   -BLE

b. reversable by the judge (Fabb 1984) (157) a. *The view is defensible by anyone. b. This view is defendable by anyone. (Roeper 1987: 269) Likewise, when we try to apply all the other tests to low -ble adjectives we obtain ungrammatical results, as exemplified in (158)–(162) for English. (158) Control into a purpose clause * The sound is perceptible (in order) to determine its intensity. (159) Licensing of Agent-oriented modifiers * These fish are easily cómparable (except on the very high reading of easily, like That building is easily a mile away from here). (Nevins 2002: 6) (160) Licensing of instrumental phrases * These fish are cómparable with a measuring stick. (Nevins 2002: 6) (161) Licensing of aspectual/manner adverbials * The flaw was perceptible every two years / in one hour. (162) Licensing of raised subjects * This film is very predictable to flop. [predictable = dull, lacking originality] (McGinnis 2010: 8) There are still additional tests that support the classification of -ble adjectives into two types of -ble. Whereas low -ble adjectives generally admit further derivation with -ly, only a few potential or high -ble allow it, as shown in (163a–b) for English. According to Varela (1990:  84), only the “least verbal” -ble adjectives accept the corresponding Spanish suffix -mente (see also NGRALE § 7.10h). Among those that are “more verbal”, some of them admit -ly only after having been negated, as exemplified in (163c) for English. Exactly the same facts obtain in Catalan or Spanish, in (164)–(165) respectively. (163) a. *translatably, *modifiably, *compárably





The external syntax of -ble   

   95

b. applicably, tolerably (= ‘in a moderately good way’), cómparably, comfortably

c. *(in)evitably/*(un)avoidably, *(un)forgettably (164) a. * traduïblement, *modificablement translatably  modifiability

(Cat)

b. comfortablement, perceptiblement comfortably perceptibly c. * (in)evitablement, *(in)oblidablement inevitably  unforgettably (165) a. * traduciblemente, *modificablemente translatably  modifiability

(Sp)

b. amablemente, perceptiblemente friendly perceptibly c. * (in)evitablemente, *(in)olvidablemente inevitably  unforgettably In addition, as pointed out in the NGRALE § 7.10h, high -ble generally reject modification with the superlative suffix -ísimo, whereas low -ble tend to accept it. This is illustrated in (166). Same facts obtain in Catalan for the corresponding suffix -íssim, as shown in (167). (166) a. * traducibilísimo, *transportabilísimo (Sp) translatable.sup  transportable.sup b. amabilísimo, notabilísimo friendly.sup notable.sup (NGRALE § 7.10h) (167) a. * traduïbilíssim, *modificabilíssim (Cat) translatable.sup modifiable.sup b. amabilíssim, comfortabilíssim friendly.sup comfortable.sup There is no such superlative affix in English. However, both English and Romance -ble adjectives often show a consistent different behavior with respect to degree

96   

   -BLE

modification by very, which has generally been used as a test in the literature on adjectival passive participles to discuss their verbal or adjectival properties (e.g. Levin and Rappaport 1986 or Varela 1990, 2002). Whereas low -ble adjectives generally accept this modification, high -ble adjectives tend to reject it.79 (168) a. *very translatable, *very modifiable, *very obtainable b. very tolerable, very perceptible, very comfortable (169) a. * molt traduïble, *molt modificable, *molt obtenible (Cat) very translatable  very modifiable very obtainable b. molt comfortable, molt permissible, molt considerable very comfortable very permissible (= ‘tolerable’) very considerable (=‘substantial’) (170) a. * muy traducible, *muy modificable, *muy transportable very translatable very modifiable very transportable

(Sp)

b. muy amable, muy permissible, muy probable very friendly very permisible (= ‘tolerable’) very probable (= ‘likely’) I take all these facts as clear evidence for two different types of -ble adjectives that must have a syntactic reflection in their internal configuration. The external syntactic behavior of potential or high -ble adjectives argues for the presence of functional structure that can license agents, instruments and verbal modifiers in their internal structure. Low -ble adjectives on the other hand, which show neither eventivity nor syntactic agentivity cannot contain any such eventive internal functional structure.

3.4 Summary To summarize so far, the syntactic properties of -ble adjectives discussed in this section provide further support to the morphological classification of -ble adjectives into two main types: potential or high -ble and low -ble. Table 2 summarizes the various tests that have been used to build the empirical basis for this clas-

79 Note that the underlying verbal bases of high -ble adjectives do admit verbal degree modification, e.g. one can modify something more or less, etc.

The semantics of -ble   



   97

sification – in the same order that they have been introduced, which should be added to the seven morphological properties analyzed in section 2. POTENTIAL or HIGH -ble  1. allows the expression of other ­arguments  2. may license indirect objects  3. [Romance] allows postponed adverbials  4. licenses aspectual adverbials  5. [Romance] does not generally appear in attributive position  6. [English] allows raised subjects  7. licenses by-phrases  8. can control into a purpose clause  9. licenses (a restricted set of) agent-­ oriented adverbials 10. licenses instrumental phrases 11. does not generally license -ly affixation 12. [Romance] does not generally license -ísimo / -íssim affixation 13. does not generally allow degree ­modification by very

LOW -ble  1. does not allow the expression of other arguments  2. does not license indirect objects  3. [Romance] does not allow postponed adverbials  4. does not license aspectual adverbials  5. [Romance] may appear in attributive position  6. [English] does not allow raised subjects  7. does not license by-phrases  8. cannot control into a purpose clause  9. does not license agent-oriented ­adverbials 10. does not license instrumental phrases 11. does generally allow -ly affixation 12. [Romance] generally licenses -ísimo / -íssim affixation 13. generally allows degree modification by very

Table 2. Two types of -ble. Syntactic properties.

As for high -ble adjectives, I have provided ample evidence for a tight relationship between passive and -ble adjectives, which justifies the presence of a syntactic passive voice component inside these verbal -ble adjectives. In addition, although the expression of the external argument in high -ble adjectives appears to be more constrained than in the corresponding passive constructions, I have shown that a similar effect on the type of by-phrase is found with passives of stative verbs in Spanish. I have further claimed that the restrictions on the appearance of the verb’s external argument and some agent-oriented adverbials are due to the stativizing effect of the modalized structure of -ble adjectives, to be further discussed in §4 below. Moreover, I have surveyed empirical evidence supporting the syntactic availability of the external argument as a licensor of a number of agent-related elements. Nonetheless, I have maintained that the syntactic presence of the external argument is not necessary to create a -ble adjective; all that is required is the implication of an originator. In contrast, low -ble adjectives do not exhibit any of the verbal-related syntactic properties, but show regular adjectival properties, such as their ability to

98   

   -BLE

become the base for -ly affixation or accept adjectival degree modification. There is, however, an apparent exception. As noted in (48) above, repeated in (171), there are cases of morphologically low -ble adjectives, such as applicable or cómparable that can make their verbal PP complement syntactically explicit, which I have taken as a sign for the presence of a syntactic verbal component, given that in principle, only a verbal functional head can license an (inherited) verbal argument (e.g. Alexiadou 2001). Thus, in (171a) and (171c), we have a mismatch: morphophonologically these forms need a low -ble, but syntactically they seem to require a high -ble. (171) a. The current situation is cómparable to the one in 1935. b. These are easily compárable to the others. c. Our approach is still applicable to a wide range of problems.

(low -ble) (high -ble) (low -ble)

As they are consistently low -ble with respect to all the other (applicable) tests, I assume that in these cases, the meaning of these -ble words may carry the cognitive implication of the arguments of the verb, with which they share the root, but this does not entail that there is a verbal functional layer in their internal syntactic structure. For instance, in the case of cómparable to, the to prepositional complement would be treated as an adjectival complement in the same way that we treat the prepositional complement of the non-deverbal adjectives equivalent (to) or equal (to). The fact that we find the same preposition in both the verb compare and the adjective cómparable would not be accidental, though, if we assume that the selection of the preposition to could be related to the conceptual meaning of the root. Likewise for applicable to or divisible into. Now that we have studied the morphology of -ble adjectives and their external syntax, I turn to the semantic interpretation of -ble in section 4.

4 The semantics of -ble That -ble adjectives are generic or that they are characterizing predicates is uncontroversial (see e.g. Chierchia and McConnell-Ginet 1990; Krifka et al; 1995; Bosque 1999b). According to Krifka et al. (1995), there are two types of characterizing sentences, generalizations over events, and generalizations over characterizing properties of individuals. The former corresponds to the interpretation of habitual sentences. This means that -ble adjectives should be interpreted as expressing

The semantics of -ble   



   99

generalizations over characterizing properties of individuals.80 They are individual-level predicates and as such they are inherently generic (e.g. Carlson 1977a; Kratzer 1995; Chierchia 1995; inter alia). Individual-level predicates express permanent properties of individuals, like being tall or know French, as opposed to stage-level predicates, which express temporary properties of individuals, such as being available or reading a book. As pointed out in Di Sciullo (1997:  89), English -ble adjectives behave like individual-level adjectives in that they cannot occur in the there-construction, as illustrated in (172).81 (172) a. *There are people impressionable.

b. *There are chairs transformable.

In Spanish, as is well-known, the distinction between individual and stage-level adjectives is syntactically encoded in that the former are predicated with the copula ser ‘be’ and the latter with the copular estar ‘be.stative’ (e.g. Demonte 1999: 142).82 -Ble adjectives, low or high-level, potential or evaluative, systematically select the individual-level copula ser, as shown in (173). As most other individual-level adjectives, -ble adjectives, whether derived from active verbs (173a) or from stative verbs (173b), i.e. whether high or low -ble, can be coerced into stage-level adjectives in order to appear in non-stative contexts.

80 See Pylkkänen (1998) for the difference between individual-level and habitual predicates. 81 The characterization of -ble adjectives as individual-level predicates or characterizing predicates may appear as problematic for the view that only stage-level adjectives contain a Davidsonian event argument as opposed to individual-level adjectives and stative sentences (Kratzer 1995), since high -ble adjectives contain verbal projections that license eventmodifying adverbials (recall the discussion in section 3.2.2 on adverbial modification). Aspectual modifiers that refer to some underlying event are not permitted with individual-level predicates, as shown in (i). I will not go any further into the consequences this can have, but will leave it open for future research. (i) a. *John is tall at home / in an hour. b. *John knows French for hours / progressively. 82 The literature on Spanish ser and estar is extensive, and shows general agreement in their semantic classification. See, however, Camacho (2012) for a recent overview of the properties of both copulas, where it is argued that the stage versus individual-level distinction cannot capture the exact distribution of ser and estar, even though he agrees that they show a strong tendency to encode the permanent versus temporary contrast. See also Gumiel-Molina and Pérez-Jiménez (2012).

100   

   -BLE

(173) a. ser {comestible, confortable, admirable, lamentable}. (Sp) be.inf  edible comfortable admirable regrettable ‘be {edible, comfortable, admirable, regrettable} b. ser {modificable, realizable, traducible, transportable} be.inf  modifiable realizable translatable transportable ‘be {modifiable, realizable, translatable, transportable} (174) a. La ciudadanía está siendo cada dia más the citizenry be.stative.3sg.prs being every day more manipulable. (Sp) manipulable ‘The citizenry is gradually being more manipulable.’ b. Su comportamiento está siendo {lamentable / his behavior be.stative.3sg.prs be.ger deplorable / envidiable / admirable}. enviable / admirable ‘His behavior is being {deplorable, enviable, admirable}.’ On the other hand, the suffix -ble is generally assumed to express modality (Vendler 1968; Lyons 1977; Kratzer 1981, 1991; Bauer 1983; Gràcia 1995; Di Sciullo 1997; among many others). However, there is to my knowledge no study that reviews the relationship between the semantic properties of -ble and the external syntactic behavior of the adjective. The purpose of this section is to review all bits of information that I have found scattered in previous work and provide additional crosslinguistic evidence to make a case for the presence of a modal component in the internal constituent structure of -ble that has an effect in its external syntax. In the next subsections, I first provide some crosslinguistic examples that make explicit the presence of a modal component in the corresponding -ble forms. In § 4.2 I summarize the literature on the existence of two interpretations of modal -ble adjectives and show that they correlate with the two types of morphosyntactic -ble adjectives, low and high -ble, examined in the previous sections. I further present crosslinguistic empirical data showing that some languages translate this difference in interpretation into a variation in the Vocabulary items that express modality. I discuss in particular the intralinguistic variation displayed by German -bar versus -wert. Building on work by Kratzer (1981) and Hackl (1998), I continue the discussion on the type of modality of -ble adjectives in section 4.3, as well as the status of the modal component contained in -ble as the inductor of the



The semantics of -ble   

   101

stativity83 and genericity of these adjectives. I present in § 4.4 evidence that supports the intensional nature of -ble (Bosque 1999b). I show that -ble behaves as other modalizers in that it creates an opacity context that can license free choice items and subjunctive mood, a context that favors the non-specific interpretation of indefinite NPs. Hence, I claim that the modality of -ble is responsible for the specificity constraints on the expression and generic interpretation of the arguments of the underlying verb. Before addressing the issue of the different interpretations of -ble adjectives and the type of modality they express, I will first review some crosslinguistic data that supports the presence of a modal component in the structure of -ble across languages.

4.1 The modality of -ble crosslinguistically In section 3.3.2, I have already introduced some empirical evidence for the syntactic presence of a modal operator that interacts with the passivity of -ble adjectives. However, according to Bauer (1983:  155), the modality of -ble is better expressed as part of the lexical meaning of the suffix which is to be understood as an “unanalysable whole at the level of word-formation processes, although it might be lexically decomposed at deeper levels”, because it would be too complex and also strange to have a modality node that is either “empty or specified as can” appearing in processes of word formation, i.e. Bauer finds that the fact that only can would appear in word formation processes is a strong argument against an analysis of the suffix -ble as containing syntactic modal structure. First, it is unclear to me at this point why this should be so, but it turns out that there is a pervasive interaction between the passive structure and the expression of potential meaning in the languages of the world.84 I will illustrate this

83 According to Givón (1970: 834), -ble adjectives whose verbal base is active are themselves active, whereas those with a stative base would be stative as well, as the contrast in (i) shows. (i) a. He is being very objectionable. b. *It is being very regrettable. He adds that other -ble adjectives are different in that they are semantically interpreted as modalized. The modal can or be able to that apparently underlies these adjectives is itself stative, which would explain the stativity of -ble adjectives based on active verbs. Note, however, that he is forced to say that examples like (ia) are not modal. 84 It would seem that the passive expresses the fact that if an action has been realized on some object, it is interpreted that this is so because it is possible for this object to be acted upon, i.e. that some modality is always present or at least interpreted with the passive.

102   

   -BLE

interaction with languages as different as Hungarian, a Finno-Ugric language, Turkish, of the Turkic family, Chichewa and Swahili, of the Bantu family, Georgian, a South-Caucasian language, and Kanuri, a Nilo-Saharan language (see also Kemmer 1993 for a typological study of middle structures, or Siewierska 1984 for a typological study of passive structures).85 Second, languages that are morphologically complex like the ones just cited clearly contain a modal morpheme in the structure of these adjectives. Assuming a principle of language universality, this provides evidence for the presence of a modality node in the structure of these -ble adjectives. Since the meaning is essentially the same across languages, my claim is that this should be taken as an indication that a modality node is present with -ble adjectives in all languages.86, 87 In Turkish, there is a complex verbal suffix Il(y)Abil(A)r or (I)n(y)Abil(A)r, that is also used to form adjectives and has a similar property meaning as the corresponding -ble adjective in English. These suffixes can be decomposed into three

Pragmatic factors related to our knowledge of the world would interpret it in one way or other. (See e.g. Bartra 2002 for Catalan passives with a modal interpretation). 85 It seems that such a relation between passive and potential can also be found in Japanese, where the passive voice is signaled by the morpheme -(r)areru and the potential is -((ra)r)eru. (Material in parentheses refers to weak verbs only). And for Icelandic, Siewierska (1984: 131) reports that Icelandic become-passive expresses “an additional difference in introducing an extra modal meaning” as exemplified in (i)–(ii). (i) Ðeim var ekki bjargað (Ice) they were not save.part ‘They were not saved’ (ii) Ðein varð ekki bjargað (Ice) they become not save.part ‘It was not possible to save them’ The fact that become adds this modal meaning could be related to the fact that -ble adjectives expressing possibility seem to require a change of state (cf. Hackl 1998). 86 Anna Bartra (personal communication) points out that this could be related to the fact that the passive describes the resultant state, and therefore, that the object is affected. Indeed, I will suggest an analysis of -ble that treats it as a modalized predicate expressing the possibility of reaching a resultant state, which receives certain morphological support from Romance languages as well. 87 I wish to thank Meltem Kelepir for help with the Turkish data, and Zsuzsanna Gál for help with Hungarian.



The semantics of -ble   

   103

different pieces, as in (175) (see Cinque, to appear; Göksel and Kerslake 2005).88 Examples are provided in (176)–(177).89 (175) Il + (y)Abil + (A)r(Tur) Passive + Possibility Modal + Aorist (176) Bu oku-n-abil-ir this read-pass-mod-aor ‘This is a readable book.’

bir kitap.(Tur) a book

(177) Bu değiş-tir-il-ebil-ir bir program. (Tur) this change-caus-pass-mod-aor a program ‘This is a modifiable program.’ Also in Hungarian, the adjective used to express modality contains a clearly modal suffix -het / -hat that expresses possibility. As mentioned in Kratzer (1981: 304), who attributes it to Kiefer (1981), in its circumstantial reading, this suffix “can only be used for possibilities in view of the outside situation.” (178) gyógy-ít-hat-ó(Hun) cure-tr-mod-part ‘curable’ As reported in Dubinsky and Simango (1996: 760), the passive suffix -ik in Chichewa, a Bantu language, is ambiguous between a stative reading, ‘in a state of being V-ed’, and a potential interpretation, ‘able to be V-ed’. Interestingly, affixation of the stative -ik to the verb stem results in a verb, whereas affixation of the potential affix to the verb stem produces an adjective. Similar interpretations are possible with the analogous suffix in Swahili, another Bantu language, as reported in Sadler and Spencer (1998: 224). These authors suggest that “this is reminiscent of the meaning of the English middle”. Notice that in Romance, the middle construction has been argued to be an interpretation of the pronominal or se-passive (e.g. Kemmer 1993; Mendikoetxea 1999b and references cited therein;

88 I follow Göksel and Kerslake (2005) in using capital letters to represent variability in the quality of the vowel, and parentheses to indicate that the presence of the relevant item is morpho-phonologically restricted. 89 See Göksel and Kerslake (2005) for the allomorphy of the passive morpheme, and the generic contribution of the Aorist.

104   

   -BLE

see also Lekakou 2005 for a recent account), i.e. we would have an eventive sepassive, and a stative passive that expresses modality, similar to what happens in Bantu. I think that this can also be related to the formation of -ble adjectives, which will later be analyzed as a kind of modalized stative passive structure. Another completely unrelated language where we find a similar type of interrelation between passive and potential is Georgian. As reported in Asatiani (2001), the verbal prefix i- is poly-functional and “is a marker of functional disappearance of one of the main semantic roles”, so that it is used in the passive, the potential or the reflexive, among other constructions.90 The passive and the potential are illustrated in (179). (179) a. stumari svams ghvinos (Geo) ‘A guest drinks wine’ b. ghvino i-smeba stumris mier ‘The wine is drunk by the guests.’ c. ghvino i-smeba ‘The wine is drinkable.’ (Asatiani 2001: 4) In Kanuri, as described in Hutchison (1981: 145–148), the most common and productive use of the so-called Passive-Reflexive morpheme -t- is in the third person singular, which expresses a (human) passive, a reflexive and a reading comparable to that of English -ble adjectives; this third singular form is always possible with any transitive verb. Although Hutchison presents it as a ‘reading’ for this verbal form, and not as an adjective, I think that it is still valid as an example that shows the tight relation between the potential meaning and the expression of the passive. In fact, Hutchison (1981: 36) argues that there is no separate lexical category adjective in Kanuri, modification being realized by nouns, in predicative position by a verbal predicate. According to Dixon and Aikhenvald (2004), there are no languages without adjectives, though. I do not intend to discuss these issues here. My intention is merely to report that the same affix used to express passive and reflexive is also used, in third person singular, to express potentiality in Kanuri. Interestingly enough, this is exactly what happens in Georgian and in Romance (at least to some extent). Notice, however, that assuming a framework where words are built in the syntax implies that different words that have been

90 Again, notice that the Romance clitic se is equally used in the passive, the middle, and the reflexive constructions.



The semantics of -ble   

   105

categorized as belonging to the same lexical category may now have different configurations. In this context, to say that some word is or is not an adjective in such or such language is somewhat irrelevant. What is important is what their syntactic and semantic properties are and to what extent these are shared. More differences are expected in the realm of morphology and phonology. That is, I adopt the view advanced in Alexiadou (2001: 8–9) with respect to the distinctions between nouns and verbs that these are best captured in a model where these differences exist only at a superficial level. After this brief survey of empirical data showing the various ways in which modality interacts with passive across languages, let us move forward to the description of the two modal interpretations that have been proposed for -ble adjectives.

4.2 Two interpretations for modal -ble Various authors have distinguished between two different types of -ble adjectives depending on the modality they express. Although -ble adjectives are generally analyzed as expressing possibility, it has been noted by different authors that some of them convey either obligation or necessity (Lyons 1977; Rainer 1993; Huddleston and Pullum 2002) or an appreciative-evaluative meaning (Val Álvaro 1981; Gràcia 1992; Rainer 1999). Let us see some of these proposals in more detail. Vendler (1968: 118) distinguishes between those more regular adjectives that express “inclination, proneness, likelihood” and those that he calls “emotive adjectives”. As mentioned in the previous section, whereas the former are passive, the latter are represented in a transformational rule as expressing some type of causation, in the sense that they cause or attribute some “emotional reaction evoked in a human” to the adjectival subject. This is illustrated in (180)–(181) together with some of his examples. (180) a. Vble Ni ← Ni wh … [mod] be Ven – (by Nj) b. breakable glass, soluble salt, readable book (181) a. e(V)ble Ni ← Ni wh… [evokes, causes, etc] e (V) in Nj b. horrible face, terrible event, abominable monster Lyons (1977: 528–534) also observes that regular -ble adjectives express possibility or capacity. Although their meaning does not exactly correspond to a modalized sentence containing the underlying transitive verb, he suggests that they can be

106   

   -BLE

derived from this sentence through a transformational rule. The extra meaning added by the adjective would be listed in the lexicon. In addition, there is a set of adjectives derived from evaluative or judgment verbs like enviable or detestable that requires an interpretation of necessity or obligation instead of possibility. And still a few others express possibility or necessity depending on the context, as illustrated in (182) with his example. Finally, in those cases that lack a verbal base, he distinguishes between the ones that are morphologically and syntactically regular with same semantics, as in (183a), and those that are peculiar in these respects, in (183b). Only the latter would be treated as simple lexemes in the lexicon. (182) a. This bill is payable immediately. (It is necessary) b. This bill is payable at any post office. (It is possible) (183) a. feasible, legible, edible b. horrible, knowledgeable, reasonable For Spanish, Val Álvaro (1981) considers the suffix -ble as a way of expressing modality, as well. He distinguishes between a modality analogous to the Spanish verb poder, which expresses capacity, possibility and permission, and an appreciative modality, interpreted as x merece V-se “x deserves V-se”. The set of adjectives he uses to illustrate this second meaning, can be grouped together as the set of Subject-Experiencer psychological verbs, e.g. abominable ‘abominable’, admirable ‘admirable’, adorable ‘adorable’, deplorable ‘deplorable’, despreciable ‘despicable’, detestable ‘detestable’, envidiable ‘enviable’, etc. In addition, he claims that -ble adjectives can develop secondary semantic specifications, e.g. there is a set of adjectives that shows a secondary semantic specification of easiness (or “an intensive value of poder”), exemplified in (184), so that they can be paraphrased as ‘x is easy to V’. (184) desmoralizable, excitable, impresionable, irritable, pronunciable, demoralizable, excitable, easily.affected, irritable, pronounceable, manejable manageable (Val Álvaro 1981: 196)

(Sp)

Similarly, for Rainer (1999), Spanish regular passive -ble adjectives can be divided into potential adjectives, i.e. interpreted as ‘that can be V-ed’, and deontic adjectives, interpreted as ‘that must be V-ed’ or ‘that deserves being V-ed’.



The semantics of -ble   

   107

Likewise, the NGRALE § 7.10j–k points to the existence of two main semantic groups of -ble adjectives in Spanish depending on whether they can be paraphrased as a regular modal passive ‘that can be V-ed’, as in deducible ‘deducible’, intraducible ‘untranslatable’, or their meaning corresponds rather to ‘be worthy of’ or ‘that must be V-ed’, as in lamentable ‘regrettable’, deseable ‘desirable’, destacable ‘remarkable’. Chierchia and McConnell-Ginet (1990:  294) mention that -ble is a dispositional affix91 that is used to express the modal concepts of necessity and possibility. They first consider the validity of a word formation rule within the translational approach (using lambda abstraction). However, the meaning of many, if not all, -ble adjectives implies more than what the rule specifies, i.e. they are semantically non-transparent, so that the rule could not be part of their interpretation. They go on to propose a rule based on meaning postulates (Carnap 1947), as in (185), thus assuming that word formation rules only constrain but not determine semantic interpretation. It will be the context that will provide further cues for a complete understanding of the adjective. (185) a. If α is in Vt, then α + able is in Adj b. □∀x [[α + able]’ (x) → ◊∃y (α’ (y, x))]92 Where α = like, hate, wash, etc “A person who is V-ble can be V-ed by someone” For them, word formation rules like that of -ble create what they call “translucent formations,” i.e. items that are neither semantically transparent nor opaque. A question that some of these authors have raised (for instance, Lyons 1977 for English or Rainer 1999 for Spanish) is whether this difference in modality is just a contextual variation of a single modal use of -ble expressing possibility. That is, whether this variation is linguistically relevant or whether it is just a consequence of our knowledge of the world, in the sense that if something can be {admired / envied / detested} it is so because it deserves it or it is worth it. Intui-

91 Also Krifka et al. (1995: 50) mention that a predicate like be soluble in water, which expresses a lawlike sentence, is treated as a dispositional predicate (If x is put into water, it will dissolve) in the philosophical literature, and dispositionals are analyzed as modalized sentences, thus as containing modal quantification. Notice that this meaning connects these adjectives with the modal meaning expressed by the middle construction as understood in e.g. Lekakou (2005). 92 Notice in passing that they assume an underlying external argument in the meaning of these adjectives, as well as a passive meaning, in that the adjectival subject corresponds to an internal object.

108   

   -BLE

tively, the pragmatic transition from that can be admired to worthy of admiration seems to follow straightforwardly. If so, the question at issue should rather be why such pragmatic transition is found with a particular subset of adjectives and not with any -ble adjective. And, more importantly, whether this semantic distinction has a syntactic correlation. As for the latter issue, let us consider the syntax of the set of -ble adjectives that are generally considered non-potential, evaluative or as expressing necessity instead of possibility. In section 3, I have applied a number of standard tests to argue for the presence of functional verbal structure within the internal syntax of high -ble, which contrast with low -ble. When applying all those tests in the previous section, I have only made use of -ble adjectives denoting a regular potential passive meaning. If we now try to apply all those tests to these evaluative or nonpotential -ble adjectives we obtain ungrammatical results, as exemplified in (186) for Catalan. Note that the corresponding modal passive sentences in parenthesis are perfectly fine, which clearly points to the lack of event-related structure in non-potential -ble adjectives that could syntactically license all these elements. Exactly the same facts obtain in Spanish. (186) a. Control into a purpose clause (Cat) * L’atac és deplorable per calmar els ànims. (Cf. L’atac va poder ser deplorat per calmar  the=attack aux.3sg.pr can.inf be.inf deplore.part to calm.inf els ànims). the spirits ‘The attack could be deplored to calm things down.’ b. Licensing of by-phrase * L’atac és (fàcilment) deplorable per {la Maria, el govern}. (Cf. L’atac pot ser deplorat pel govern / the=attack can.3sg.pr be.inf deplore.part by.the government per la Maria). by the Maria ‘The attack can be deplored by the government / by Maria.’ c. Licensing of agent-oriented modifiers L’atac és deplorable *fàcilment. (Cf. L’atac va poder ser deplorat fàcilment). the=attack aux.3sg.pr can.inf be.inf deplore.part easily ‘The attack could be easily deplored.’



The semantics of -ble   

   109

d. Licensing of instrumental phrases * La situació és deplorable amb una carta. (Cf. La situació pot ser deplorada amb una carta). the situation can.3sg.pr be.inf deplore.part with a letter ‘The situation can be deplored with a letter.’ e. Licensing of aspectual/manner adverbials * L’acord és deplorable cada dos anys. (Cf. L’acord pot ser deplorat cada dos anys). the=agreement can.3sg.pr be.inf deplore.part every two years ‘The agreement can be deplored every two years.’ The corresponding sentences are equally ungrammatical in English, as shown in (187). The discussion on the parallel German adjectives will be deferred until section 4.2.1. (187) a. Control into a purpose clause * The attack is deplorable (in order) to calm things down. b. Licensing of by-phrase * The attack is (easily) deplorable by Mary / by the government. c. Licensing of agent-oriented modifiers * The attack is easily deplorable. d. Licensing of instrumental phrases * The situation is deplorable with a letter. e. Licensing of aspectual/manner adverbials * The agreement is deplorable every two years. Furthermore, whereas all necessity -ble adjectives admit further derivation with -ly, I have already shown that this is not so for the “more verbal” potential high -ble, as shown in (188a)–(188b) for English. Exactly the same facts are found in Romance, as illustrated in (189) for Catalan. (188) a. *translatably, *modifiably, b. admirably, enviably (189) a. *traduïblement, *modificablement b. admirablement, envejablement

(Cat)

110   

   -BLE

I take all these facts as clear evidence for the status of evaluative or nonpotential -ble as a subset of low -ble adjectives. That is, evaluative adjectives pattern syntactically with irregular -ble adjectives in that they do not show evidence of containing any verbal related structure. Just the opposite, all evidence points to its absence. In other words, whereas the external syntactic behavior of potential high -ble adjectives argues for the presence of functional structure that can license agents, instruments and verbal modifiers, necessity or evaluative -ble adjectives on the other hand, which show neither eventivity nor syntactic agentivity, cannot contain any such eventive internal functional structure and must therefore be classified as low -ble adjectives. The fact that they express an evaluative or necessity meaning will be argued to derive from the structure in section 6.1, specifically from a difference in the attachment site of a modal projection, and not from two different types of modals, say necessity and possibility. Now the question that still remains is about the modality of the two types of adjectives. In the next section I provide some crosslinguistic empirical data showing that different languages lexicalize these two meanings as different suffixes. This, which a priori could be taken as evidence for the existence of two different -ble adjectives, will be argued to follow from the structure, at least in the languages under study.

4.2.1 Two interpretations of -ble across languages I have just shown that there are two possible interpretations for -ble adjectives, potential and evaluative. When we consider analogous -ble adjectives in other languages, e.g. Hungarian or Turkish, or even German and Icelandic, we note that, interestingly, this variation in the modal interpretation of -ble involves essentially the same set of bases, which, in my view, provides further support for an analysis that does not treat the two meanings as deriving from two types of modals, but as deriving from the interaction between the modal and the specific semantics of this particular subset of bases. Hungarian and Turkish have in common that the suffix appearing in the adjectival structure is a modal suffix that expresses only potentiality. No other meaning can be expressed by the modal affix in both languages, so that their corresponding -ble adjectives do not show the kind of polysemy that is displayed by -ble adjectives in English or Romance. (190) a. -(y)Abil-

(Tur)

b. -het / -hat

(Hun)

The semantics of -ble   



   111

It turns out that the corresponding evaluative adjectives in Turkish and in Hungarian have a different suffix in Turkish, as shown in (191),93 or even a completely different structure in Hungarian, as exemplified in (192) (Érsek 1977, György 1992). (191) a. beğenilecek (cf. değiştirilebilir) admirable (cf. modifiable)

(Tur)

b. tapinilacak adorable (192) a. csodálatra méltó / csodálatos (cf. gyógyítható) (Hun) admiration worthy / admirable (cf. curable) ‘admirable / admirable’ b. imádatra méltó adoration worthy ‘adorable’ Similarly, although I cannot provide uncontroversial evidence that their -ble suffix has only a potential meaning, languages like German,94 Icelandic95, and Dutch have in common that they have a different suffix, e.g. German -wert ‘worth’, to express an evaluative judgment, as illustrated in (193)–(194). (193) a. beklagenswert, bewundernswert, empfehlenswert, wünchenswert (Ger) regrettable admirable recommendable desirable b. *beklagbar, *bewunderbar, *empfehlbar, *wünschbar (194) a. lastverður, markverður (Ice) reprehensible important

b. *lastlegur, *marklegur

(195) a. aanbevelenswaardig, bewonderenswaardig (Dut) recommendable admirable

93 Note, however, that these adjectives seem to contain the passive affix -il- as well. 94 Thanks to Hedda Franken for help with German. 95 See Riutort (1998). Thanks to Macià Riutort for his help with the Icelandic data.

112   

   -BLE

b. *aanbevelbaar, *bewonderbaar

Let me consider the intralinguistic variation in German in some more detail (see Oltra-Massuet 2012). Leaving aside a few verbal bases that allow both suffixes, as in (196),96 the suffix -wert ‘worth’ appears to stand in complementary distribution with -bar (Fleischer and Barz 1995: 228) for psychological verbs, exemplified in (193) above. (196) a. erwägbar, hörbar, lesbar (Ger) consider.ble hear.ble read.ble ‘that can be Ved’ b. erwägenswert, hörenswert, lesenswert consider.worth hear.worth read.worth ‘that is {worth / deserves} being Ved’ (Fleischer and Barz 1995: 228) Interestingly, the semantic dichotomy between -bar and -wert in German correlates with the syntactic distinction I have established in section 3 between high -ble and low -ble in English and Romance. Whereas the external syntax of -bar adjectives shows all the properties characteristic of internally eventive forms, displayed in (197), -wert does not allow the expression of verbal arguments or verbrelated modification, as shown in (198).97

96 Similarly, sichtbar ‘visible’ and sehenswert ‘worth seeing’, nennbar ‘nameable, specifiable’ and nennenswert ‘worth mentioning’. Note that these are not ‘emotive’ or psychological verbs. Likewise in Icelandic, aðdáunarverður versus aðdáanlegur ‘admirable’. Different languages are expected to conceptualize meanings in different ways. 97 The internal syntax of -wert is complex in that it involves unresolved aspects, such as the status of -wert, as either a derivational affix or a compounding base, or the nature and function of the so-called Fugenmorphem or linking morpheme -s-. Therefore, the fact that they pattern with low -ble with respect to their external syntax should not be understood as a claim that -wert is a low or root-attaching -ble. The internal syntax of German -wert adjectives will not be analyzed in this volume.



The semantics of -ble   

   113

(197) a. Control into a purpose clause  Ab sofort sind nun auch Lastminute-Reisen über  from immediately be.3pl.prs now too lastminute-trips over TravelSecure versicherbar, um das gesamte Angebot zur TravelSecure insurable in.order the.n entire offer to.the.dat Reiseversicherung abzurunden. (W) (Ger) trip.insurance round.off.inf ‘From right now last minute trips can be taken out through Travel­ Secure, in order to make the most of the entire travel insurance offer.’ b. Licensing of by-phrase ein vom Benutzer/ *von Maria (leicht) modifizierbares a by.the.dat user  by Maria easily modifiable.n Programm programm ‘a programm easily modifiable by the user’ c. Licensing of agent-oriented modifiers …, weil viele Aspekte vom Therapeuten absichtlich because many aspects by.the.dat therapist deliberately kontrollierbar sind. controllable be.3pl.prs ‘because many aspects can be deliberately controlled by the therapist’ d. Licensing of instrumental phrases mit kaltem Wasser und milder Seife waschbare Oberfläche with cold.dat water and mild.dat soap washable surface ‘surface washable with cold water and mild soap’ e. Licensing of aspectual/manner adverbials eine jedes Jahr verlängerbare Vereinbarung a.f every.n year renewable agreement ‘an agreement renewable every year’ f. Licensing of indirect objects Die Tat war dem Angeklagten nicht nachweisbar. the deed was the.dat accused not demonstrable ‘It was not demonstrable that the accused was guilty of the deed.’ (Riehemann 1993: 17)

114   

   -BLE

(198) a. Control into a purpose clause * Der Angriff ist beklagenswert um Nerven zu beruhigen. (Ger) the attack be.3sg.prs regrettable in.order nerves to calm b. Licensing of by-phrase * Der Angriff ist von {Maria / der Regierung} beklagenswert. the attack be.3sg.prs by Maria the government regrettable c. Licensing of agent-oriented modifiers * eine klug beklagenswerte Situation a cleverly regrettable situation d. Licensing of instrumental phrases * Das ist mit einem Brief beklagenswert. the be.3sg.prs with a.dat letter regrettable e. Licensing of aspectual/manner adverbials * eine jedes Jahr beklagenswerte Vereinbarung a.f every.n year regrettable agreement f. Licensing of indirect objects * Das Buch war dem Angeklagten nicht empfehlenswert. the.n book be.3sg.pst the.dat accused not recommendable Thus, the German data confirm that there are two main types of -ble, high and low, and that the special subset of evaluative adjectives derived from psychological verbs, which express a different modality, pattern with low -ble adjectives in not showing eventive properties. This behavior would support the empirical generalization that -ble contributes a unique type of modality in the languages under study, differences stemming from its combination with particular subsets of semantically related roots in a particular configuration. These examples can also be taken as crosslinguistic morphological evidence for the existence of a modal component in both types of -ble adjectives, those that express potentiality and those that express an evaluative judgment or necessity. Now I turn to the issue concerning the type of modality of -ble adjectives

4.3 Type of modality Up to this point, I have shown that most authors that have dealt with the semantics of these adjectives coincide in admitting that they apparently convey two



The semantics of -ble   

   115

different meanings: a clearly potential one, and another one that I have labeled evaluative, necessity, or non-potential, the latter being found only within a subset of low -ble adjectives. Different works suggest that one of the modal interpretations of -ble adjectives is the deontic meaning. However, the use of this notion from author to author seems unclear and differs. For Van der Putten (1997) the deontic meaning refers to expressing a property of the subject that makes the event possible, a typical meaning of root modality. In Huddleston and Pullum (2002), deontic modality is used to refer to permission, as opposed to dynamic modality used to express ability or capability. Varela (2005) uses this label for -ble adjectives expressing obligation, ‘that deserves’, as opposed to those denoting potential modality. More generally, according to Von Fintel (2005), deontic modality concerns what is possible, necessary, permissible, or obligatory given a body of law or a set of moral principles or the like. For Lyons (1977: 823) “deontic modality is concerned with the necessity or possibility of acts performed by morally responsible agents. […] There is a sense in which the sentence we utter can be said to express a proposition; but it is not a proposition which describes the act itself. What it describes is the state-of-affaires that will obtain if the act in question is performed”. Its truth value “is determined relative to some state of the world (wj) later than the worldstate (wi) in which the obligation holds.” This deontic necessity has generally an origin or derives from some cause that is understood as responsible. At first sight, applying Lyon’s definition, it seems that all -ble adjectives express deontic modality, in that they describe the state-of-affaires that will obtain if the event denoted by the underlying verb takes place, they denote either possibility (readable, modifiable) or necessity (admirable, regrettable), and imply the presence of some agent; however, the reference to morally responsible agents seems to me inappropriate. Rather, it is the subject of -ble adjectives that appears to have some characteristic that makes it responsible for being ascribed the adjectival property. In addition, there must be an originator (agent, causer, initiator, or alike) that is also involved in their meaning. That is, to say that this text is modifiable would roughly mean that this text has some properties (say style, language, quality, unpublished status, etc) that would make it possible for an agent to modify the text. The property expressed in the adjective, being modifiable, does not imply that the event described by the underlying verb has taken place, i.e. no morally responsible agent must or may have performed the action denoted by the underlying verb. In fact, the event of modifying the text may never be performed and the property of being able to be modified will still be truthfully attributed to the adjective’s external argument, i.e. there is no actuality entailment (Bhatt 1999). As proposed in Kratzer (1981: 313–314), this kind of modality is a type of

116   

   -BLE

circumstantial modality, specifically what she calls an in view of circumstantial modality. Let us see this in more detail. In possible world semantics, modal operators are defined by reference to three parameters: (i) a modal relation or quantificational force of the modal, which can be possibility or necessity; (ii) a modal base (conversational background in Krifka et al. 1995: 50), i.e. the function that maps the world of evaluation – the actual world – onto a set of accessible worlds, i.e. the set of worlds that are accessible from the world of evaluation; and (iii) an ordering source that gives us an ordering among the accessible worlds provided by the modal base. Depending on the kind of modal base, different worlds will be accessible. A root or circumstantial modal base refers to “what can or must happen, given circumstances of a certain kind. Circumstances of a certain kind are facts of a certain kind. Facts concerning the outside world, our bodies or our mind, for example” (Kratzer 1981: 302). Circumstances, which are “a rather slippery matter” (Kratzer 1981: 303), create possibilities, i.e. the set of worlds which are compatible with them. Kratzer (1981: 306) takes German -bar adjectives as an example of circumstantial modal base that allows all kinds of ordering sources; the specific selection of one or the other to further restrict the modal base will depend on the adjective they are attached to, as illustrated in (199), where possible worlds are ordered according to regulations or our common aims. Other possible ordering sources for circumstantial conversational backgrounds could be the Law, our hopes, or what you think is good (Kratzer 1981: 308). (199) a. Dieses Eintrittsbillet ist nicht übertragbar. (Ger) this admission ticket is not transferable ‘In view of the regulations, it is not possible to give this ticket to someone else.’ b. Dieser Vorschlag ist annehmbar. this proposal is acceptable ‘In view of our common aims, it is possible to accept this proposal’ (Kratzer 1981: 313–314) Kratzer (1981: 314) further notes that it is possible for the ordering source to be empty, as in (200), in which case, we are dealing with a case of pure circumstantial modality, where only the properties inherent in the external argument of the adjective are involved in restricting the modal operator.



The semantics of -ble   

   117

(200) Diese Tasse ist zerbrechlich. (Ger) this cup is fragile ‘In view of certain properties inherent in the cup, it is possible that it breaks.’ (Kratzer 1981: 314) Thus, I assume with Kratzer (1981) that the modal operator of -ble adjectives expresses a possibility modal relation with a circumstantial modal base of the type in view of certain properties of the external argument of the adjective. The conversational background can be further restricted by an ordering source or not. That is, the ordering source may be empty, as in a comfortable sofa or a perceptible sound, where, in principle, it is the inherent characteristiscs of the sofa and of the sound that make them comfortable and perceptible, respectively. On the other hand, in cases like an admirable attitude or a publishable manuscript, the possibility of attributing these properties to their arguments may be further restricted by general conventions on what is socially accepted or what is good or of general interest. As for the quantificational force of the modal relation, whether universal or existential, things are more complex. In possible world semantics, a possibility modal relation is generally assumed to have existential force (Kratzer 1981). Hackl (1998) develops an analysis of what he calls abilitative or ability-can within Kratzer’s model, providing the core syntactic structure or ‘basic skeleton’ for abilitative structures, which include -ble adjectives, (201a), middles, (201b), or dispositional generics, (201c). (201) a. The third math problem of the last assignment is solvable. b. This bread cuts easily. c. This car goes 20 miles an hour. (Hackl 1998: 1) Hackl (1998) points out that ability-can could be analyzed as an existential modal operator that takes a circumstantial restrictor (“RC”) and assigns a sentence like (202a) a quasi-LF structure as in (202b) that can be paraphrased as in (202c). (202) a. John can swim. b. ∃w'[w' RC w] & [John swims in w']

c. “Given John’s physical and mental properties in w there is a world w' accessible from w such that John swims in w'”. (Hackl 1998: 6)

118   

   -BLE

However, Hackl further points out that this analysis appears as problematic at first sight because, on the one hand, it lacks the universal dual that should be logically possible for a possibility modal, because a sentence like John can swim is different from John need not to swim or It is not the case that John mustn’t swim. On the other hand, the truth conditions of the existential quantification, which simply require the existence of an accessible possible world, might be too weak, in the sense that for a sentence like This elevator can lift 1500lbs universal or quasi universal force seems rather more appropriate than simple possibility, i.e. the interpretation that best suits this sentence would be that “under normal circumstances and normal operating conditions, in all cases in which the elevator has to lift 1500 lbs it will accomplish it” (Hackl 1998: 7). To avoid these problems he considers the option of having a different LF for ability-can that involves generic quantification over situations, as in (203).98 (203) a. John can swim.

b. GEN s [John in s & C(s)] ∃s' [s overlaps with s' & John swims in s']



c. “All situations that include John and certain felicity conditions are met are expandable to situations in which he swims” (Hackl 1998: 8) Hackl (1998: 8) notes that deciding between universal or existential quantification is hard, and a matter of pragmatics rather than semantics; nevertheless, he provides an analysis of abilitative can as existential quantification, pointing out that for cases with a circumstantial base, “there is a close relationship between generic quantification over situations and existential quantification over worlds”. The basic structure that Hackl (1998: 10) suggests as the skeleton of ability attributions for verbal passives is the one in (204), which corresponds to “a structure where the external argument of the main predicate is demoted, i.e. syntactically not generated in the specifier position of VoiceP (as e.g. in verbal passives, unaccusatives, etc)”.

98 Individual-level predicates are also analyzed as generalizations over situations in Chierchia (1995).

The semantics of -ble   



   119

(204) Skeleton of ability attributions

… VoiceP M

RC Voice

V

This structure contains the three key components that all ability attributions must have projected in the syntax, according to Hackl (1998: 9), listed in (205). (205) a. A modal operator (M) with existential force that relates sets of posible worlds (denoted by the restrictor Rc and the nuclear scope)

b. A circumstantial restrictor (Rc) whose content is determined by the conversational background



c. A change of state denoting predicate as a complement of the modal operator, assumed to be Cause/VoiceP.

As pointed out in Portner (2009: 202), several analyses of ability modals combine “some sort of existential quantification, corresponding to the idea that the agent chooses an action, and some sort of universal quantification, corresponding to the idea that the action guarantees a certain outcome”. Portner identifies these two operators as ModP and VoiceP in Hackl’s (1998) syntactic structure, whereas he mentions that the two functional layers are combined into a single operator in other works, e.g. Brown (1988). Thus, this would mean that whereas existential quantification seems tied to the external argument – as it is in the case of passive structures – universal quantification is related to the result of a change of state. This dual quantification of abilitatives already appears in the semantics of -ble in Di Sciullo (1997). Specifically, Di Sciullo (1997: 89) assumes that the event underlying -ble adjectives is “in the scope of the operator POSSIBLE forcing a stage-level interpretation of the verb selected by the affix “, and further, that the specifier position of the underlying verb as well as the specifier of the adjective phrase are “bound by a c-commanding generic operator GEN” that provides the arbitrary reference for the verb’s external argument. However, neither operator is represented syntactically in any of the three dimensions, the conceptual, the aspectual and the categorial, she proposes for the analysis of -ble. Note, however, that this proposal is in fact the opposite of the one for abilitatives described in

120   

   -BLE

Portner (2009), since according to Di Sciullo (1997), generic universal quantification binds the external argument, whereas the possibility operator scopes over the event. I will elaborate a bit more on the type of quantification over the agent when I compare the expression of the external argument with passives and -ble with respect to the actuality entailment in the former but not in the latter (cf. Bhatt 1999) in section 6. -Ble adjectives are individual-level adjectives and as such, they are inherently characterizing or generic predicates (Krifka et al. 1995: 7; Leonetti 1999: 873), and therefore they are among the set of intensional constructions that can create opaque contexts, like modal predicates (e.g. Leonetti 1999: 862; Pérez Saldanya 1999: 3262). On the other hand, in section 3.3.2, I have provided some empirical evidence for the presence of a syntactic possibility modal operator that interacts with the passive features of high -ble. Whether this modal operator must be understood as existentially quantifying over posible worlds or as universally quantifying over situations in -ble remains an open question and need not concern us here. The point is that the modal nature of -ble adjectives induces an intensional context. For the purposes of the analysis of -ble developed in this volume, I will assume that there is a syntactic modal component, a ModP, for which I have found empirical evidence, that may be responsible for both the generic and the possibility reading, along the lines of Menéndez-Benito’s (2013) recent analysis of dispositional sentences of the type This car goes 200 kph, which builds on Lekakou’s (2005) account of middles as inner dispositions. Menéndez-Benito (2013: 12) argues that such dispositional sentences, which “report a generalization that is true in virtue of some inherent property of the individual denoted by the subject”, contain a covert possibility modal of the kratzerian in view of circumstantial type, so that they are interpreted as This car can go 200 kph.99 Although MenéndezBenito (2013) assumes that the covert possibility operator of dispositionals comes with existential force, I will not be specifically concerned with the LF interpretation of the construction. As for the analysis of the internal syntax of high -ble, the empirical evidence reviewed in the previous sections will bring me to suggest a syntactic structure containing Hackl’s (1998) basic skeleton for abilitatives depicted in (204). The different reading obtained with evaluative or necessity low -ble adjectives will be derived from structural differences. That is, my claim is that the type of modal is one and the same kratzerian circumstantial possibility modal of the in view of

99 See also Chierchia and McConnell-Ginet (1990: 294) for the claim that “one of the main functions of generic sentences appears to be that of expressing capability or possibility”.



The semantics of -ble   

   121

type for all -ble, variation essentially results from whether the modal operator has an eventive variable in its scope or a specific tye of uncategorized root. In the next subsections I discuss the effects of the modal context created by high -ble in its external syntax.

4.4 The modal context of -ble Evidence for the presence of a modal operator, comes from two sources, the distribution of free choice items with -ble adjectives (e.g. Dayal 1998; Quer 1998, 2000, 2001; Bosque 1999b; inter alia), and the distribution of subjunctive and non-specific interpretation of NPs in restrictive relative clauses in Catalan and Spanish, (Quer 1998, 2000, 2001, 2002; Bosque 1999b; Pérez-Saldanya 1999; inter alia). These are dealt with in turn in the next subsections.

4.4.1 Licensing free choice any It is a well-known fact that free choice items are typically licensed in modal and generic environments and are incompatible with episodic predicates, as exemplified in (206). As shown in (207), -ble adjectives can license a free choice item, which is an inherently non-specific quantifier (Leonetti 1990:  151). Same facts obtain in Romance, as illustrated in (208) for Catalan. (206) a. Any student could solve that problem. b. Any owl hunts mice. c. * Any woman contributed to the fund. (Quer 2000: 251) (207) a. Any book is translatable. b. a book translatable by any professional translator (208) a. Qualsevol treball és modificable. (Cat) any work be.3sg.prs modifiable ‘Any work is modifiable.’ b. un treball modificable per qualsevol expert en la matèria a work modifiable by any expert in the matter ‘a work modifiable by any expert in the subject matter’

122   

   -BLE

4.4.2 On subjunctive and specificity In preceding sections, I have claimed that the restrictions on the interpretation of the underlying verb’s arguments in high -ble adjectives are to be related to the presence of a modal functional head. As pointed out in Leonetti (1999: 858–870), non-specific NPs are those that mention a hypothetical or possible referent that may even not exist at the moment of speech, i.e. the NP establishes some condition on possible referents without choosing any in particular. The kinds of contexts that allow this interpretation are opaque or intensional contexts; these are environments where the state of affaires described is non factual or not real, and where the reference of an indefinite NP is cancelled, i.e. they induce the nonspecific interpretation of indefinite NPs (Bosque 1999b).100 The semantic distinction between specific NPs, which presuppose the existence of a referent that meets the description, and non-specific NPs makes itself explicit in the syntax of Spanish and Catalan. A typically ambiguous sentence like (209) in Catalan, where the NP a Catalan artist can be interpreted as either specific or non-specific, when disambiguated, the use of a relative clause with indicative forces the specific interpretation in (209a), while the subjunctive results in the non-specific reading in (209b). Note also the use of any in the non-specific paraphrase. (209) La Maria es vol casar amb un artista català. (Cat) the Maria se want.3sg.prs marry.inf with a artist Catalan a. La Maria es vol casar amb un artista que the Maria se want.3sg.prs marry.inf with a artist that és català. be.3sg.prs.ind Catalan ‘Maria wants to marry a certain Catalan artist.’ b. La Maria es vol casar amb un artista que the Maria se want.3sg.prs marry.inf with a artist that sigui català. be.3sg.prs.sbjv Catalan ‘Maria wants to marry any Catalan artist.’

100 There is a vast literature on opaque contexts, which goes at least as far back as Quine (1961) or Partee (1970, 1974). Basic references on specificity are Fodor (1970), Jackendoff (1972); see also Leonetti (1990, 1999, 2012) for Spanish and Brucart (2002) for Catalan.



The semantics of -ble   

   123

Leonetti (1990:  63; 1999:  865) further points out that the relation between nonspecificity and subjunctive mood is stable in intensional contexts, whereas there is no completely strict correlation between indicative mood and specificity, because it is possible to find examples where the indicative mood does not prevent the nonspecific reading. This happens in generic contexts where indicative does not presuppose the existence of the referent denoted by the NP (Quer 2002: 2838).101 With this very basic theoretical background on specificity, let us consider the expression of arguments with high -ble adjectives. I have claimed that the contrast in (210) is due to a constraint on the specificity of the indirect object. On the one hand, note that we have a bare plural NP in (210b), which according to Leonetti (1999: 865) or Laca (1999: 899) is inherently nonspecific in Spanish (see also Krifka et al. 1995: 10), witness the fact that it can be paraphrased with a free choice item qualsevol ‘any’ in (211a). On the other hand, if we paraphrase this sentence with a relative clause, we can in principle use both the indicative and the subjunctive mood, as in (211b), even though the subjunctive is by and large the preferred mood among native speakers. However, in neither case does the NP refer to any specific children. (210) a. * La pel·lícula és mostrable als nens. (Cat) the film be.3sg.prs showable to.the children b. mostrable a nens menors de  La pel·lícula és the film be.3sg.prs showable to children younger than dotze anys. twelve years ‘The film is showable to children under twelve.’ (211) a. La pel·lícula és mostrable a qualsevol nen menor de the film be.3sg.prs showable to any child younger than dotze anys. (Cat) twelve years ‘The film is showable to any child under twelve.’

101 Leonetti (1999: 865) also mentions that verbal mood does not necessarily function as an indicator of (non)specific interpretation in generic contexts. See also Leonetti (2012: 298).

124   

   -BLE

b. una pel·lícula mostrable a nens que {són / siguin} a film showable to children that be.3sg.prs be.3sg.prs.sbjv menors de dotze anys younger than twelve years ‘a film that is showable to children under twelve.’ The same nonspecific interpretation is the only one available for external arguments. As pointed out in the NGRALE § 25.11e and § 25.10s, NPs in intensional contexts with indicative mood do not necessarily have individual reference, but they often characterize kinds or particular classes, as happens with the subjunctive, the difference lying in the fact that the subjunctive is neutral with respect to the existence of the kind or class of individuals. In other words, the possibility modal induces nonspecificity since it contributes some degree of uncertainty about the existence of an identifiable referent (see Jackendoff 1972 for the adjective possible). This is the difference found in (212b), where in neither case does a specialist refer to a person in particular, even though the indicative generally favors the specific interpretation of indefinite NPs. (212) a. un libro traducible por un especialista experto en la materia (Sp) a book translatable by a specialist expert in the matter ‘a book translatable by a specialist with expertise in the subject matter’ b. un libro traducible por un especialista que {es / a book translatable by a specialist that be.3sg.prs/ sea} experto en la materia be.3sg.prs.sbjv expert in the matter ‘a book translatable by a specialist with expertise in the subject matter’ c. un libro traducible por cualquier especialista que sea a book translatable by any specialist that be.3sg.prs.sbjv experto en la materia expert in the matter ‘a book translatable by a specialist with expertise in the subject matter’ Now, there are some examples that apparently constitute counterevidence to the specificity requirement. With respect to the indirect object, Patricia Cabredo-Hofherr (personal communication) points to the existence of examples like (213a). On the other hand, cases like (213b) exhibit exactly the same phenomenon in the expression of the external argument with -ble.



The semantics of -ble   

   125

(213) a. Una buena teoría debe ser explicable a mi hijo de a good theory must.3sg.prs be.inf explainable to my son of cinco años. (Sp) five years ‘a good theory should be explainable to my five year old son’ b. Una novela adaptable al cine por Pedro Almodóvar a novel adaptable to.the cinema by Pedro Almodóvar debe ser excéntrica. must.3sg.prs be.inf excentric ‘A novel adaptable to the cinema by Pedro Almodóvar must be excentric.’ First, in both PPs in (213), the NPs mi hijo de cinco años ‘my five-year-old son’ and Pedro Almodóvar do not have specific reference, i.e. do not denote a particular individual. Thus, although the NPs apparently refer to a particular and specific individual, the nominal is in fact recategorized so as to be interpreted as a kind to refer to a class of individuals, along the lines of (214), where the use of the subjunctive in the paraphrase reinforces this generalization. My contention is that the acceptability of this type of sentences must be due to the availability of the kind-interpretation of the NP, i.e. “the noun or complex noun must be semantically connected with a “well-established kind” to which the noun phrase can refer” (Krifka et al. 1995: 11). In the cases at hand, my five year old child is interpreted as a typical member of the class ‘five-year-old children’,102 whereas Pedro Almodóvar, who is well-known for being a very special type of director, is taken as a prototypical member for the class of very special directors. Even if Pedro Almodóvar were to be the unique member of that class, the interpretation of (213b) paraphrased in (214b) is that of a possible individual that would meet the NPs description.103

102 As pointed out in Rigau (1999), it is posible for an NP introduced by a demonstrative to receive a kind interpretation. 103 Patricia Cabredo-Hofherr (personal communication) mentions the German example Der Kontostand ist vom Bundesnachrichtendienst jederzeit abrufbar ‘The balance can be inquired anytime by the Federal Intelligence Service’, or the English The skin is damageable by the sun where the by-phrase apparently contains a specific NP, so that specificity would play no role. The existence of other possible by-phrases with -ble adjectives like by the government, by the Parliament, by the FC Barcelona, all with specific reference (in a given context) makes me think that it is possible that the restriction on specificity interacts with other semantic properties of nouns, specifically with their status as collective nouns, so that we could still say that the reading we obtain is a characterizing reading that picks out some non-specific members of the set denoted by the NP (the set of sunrays that cause the damage in the case of

126   

   -BLE

(214) a. un niño de la edad de mi hijo / un niño que tenga  a child of the age of my son a child that have.3sg.prs.sbjv la edad de mi hijo (Sp) the age of my son ‘a child the same age as my son’ b. alguien como Pedro Almodóvar / alguien que sea someone like Pedro Almodóvar someone that be.3sg.prs.sbjv como Pedro Almodóvar like Pedro Almodóvar ‘someone like Pedro Almodóvar’ These sentences are stative characterizing predicates, they express a general property of the subject, so that these NPs can only take on a kind interpretation. This meaning is triggered by the interaction between the modality and genericity of the characterizing sentence and the properties of the NP argument, which must permit a kind-referring interpretation. As I have argued above, under normal circumstances, Juan in (215) refers to a specific individual in both sentences, so that they are both judged as ungrammatical. However, it is possible to imagine a situation where these sentences are uttered in a context where say Juan’s properties are somewhat special or peculiar and people know them well, in which case, Juan could also be coerced into a kind-referring reading as in (215c).104 (215) a. * Una novela publicable por Juan debe ser excéntrica. (Sp) a novel publishable by Juan must.3sg.prs be.inf excentric

the sun). This issue requires a deeper analysis within the wider context of work on specificity as a phenomenon at the semantics-pragmatics interface (Leonetti 2012: 298) that cannot be pursued here. Necessarily so on the face of other examples like Ich bin ihm dafür dankbar ‘I am grateful to him for that’, or Das ist nicht von mir erklärbar ‘I cannot explain that (lit. That is not explainable by me)’ in German, and similar cases in Catalan like les tècniques mèdiques que li són exigibles per raó de la seva qualificació professional ‘medical techniques that can be required to him because of his profesional qualification’. As a final note, however, it may be worth pointing out that, a structure like li són exigibles ‘that can be required to him’ is generally found in contexts where the pronoun refers non-specifically to someone in her condition of member of a class (as a doctor or as a nurse, in the Catalan example). Equally so in cases like no m’és exigible ‘it cannot be required to me’, even though this is at this point just an intuition. 104 See e.g. Krifka et al. (1995) for the properties of kind-referring NPs, and the types of NPs that can obtain such interpretation, as well as the conditions under which that is posible.



The semantics of -ble   

   127

b. * Una buena teoría debe ser explicable a Juan. a good theory must.3sg.prs be.inf explainable to Juan c. un tipo como Juan a guy like Juan Thus, -ble requires its underlying verb’s arguments to be nonspecific. If they refer to some specific individual or entity, the NP in which they are contained becomes ungrammatical, unless a kind interpretation of those arguments is available in the context of utterance. The question is why this should be so, and where this restriction is to be located. I would like to argue that this effect derives from the presence of a high modal head in the structure that stativizes the structure and scopes over the eventive variable and all its arguments. In other words, the nonspecific interpretation of an argument requires the argument to be dominated by the maximal projection of a modal generic operator. Note that exactly the same stativizing effect happens when such a ModP projects as a verb can. 105 On the one hand, this is reminiscent of the interaction between passive and lexical stativity found in Spanish. Recall the contrast in passive sentences with lexically stative verbs, e.g. *María es conocida por Juan ‘Maria is known by John’ versus María es conocida por todos ‘María is known by everybody’. On the other hand, as pointed out in Pesetsky (1995: 120), -ble “semantically involves a modal that takes the external argument as one of its arguments”.

4.5 Interim summary In this section I have argued that the classification of -ble into two morphosyntactic types, a potential or high -ble and a low -ble finds further support in its semantics. Those -ble adjectives that the literature describes as expressing necessity or an evaluative judgment have been shown to belong to low -ble on the basis of standard diagnostic tests, as they show no eventive properties. I have assumed with Kratzer (1981) that -ble expresses a possibility modal relation that is restricted by a circumstantial modal base of the in view of-type, specifically, in view of certain properties inherent to the external argument of the adjective. I have further suggested that the syntactic presence of a modal component creates a modal context

105 As is well-known, modal can cannot occur (i) in the progressive (*be canning), (ii) as a complement of force/persuade (*be forced to can), (iii) in the imperative (*Can!), (iv) with the adverbs deliberately/carefully (*deliberately can), or (v) in a pseudo-cleft (*What John did was can run), which are all standard tests for stativity (Dowty 1979: 55).

128   

   -BLE

that restricts the expression and interpretation of the verb’s underlying arguments with -ble in that it stativizes the eventive structure, and can license elements that can only appear in intensional contexts, free choice items and subjunctive relative clauses. Finally, I have claimed that the interaction between the expression of the verb’s arguments and the stativity of the structure relates -ble to regular passive constructions, thus providing further support for the presence of a regular passive component within the internal configuration of -ble. In the next section, I show that these two types of -ble correspond to the two levels of word formation in the Distributed Morphology framework, above and below the first-category-assigning functional projection. Note that if on the right track, the picture that emerges is one that supports the existence of different degrees of eventivity for deverbal adjectives. On the one side, we would have eventive participles, which show the highest degree of eventivity (e.g. Bosque 1999); non-eventive participles as well as other deverbal adjectives such as -ive or low -ble would be at the other side of the scale, showing no verbal properties at all. Somewhere in between we would find potential or high -ble adjectives, which are constrained by their internal configuration. The existence of the larger subset of potential -ble adjectives, thus supports the view of two different degrees of eventivity within the set of -ble adjectives.

5 Previous approaches Due to the variety of properties described so far, different linguists have located the derivation of -ble adjectives in different modules. Among previous approaches, some authors have proposed that the -ble suffix is added in the lexicon prior to the syntax, such as Aronoff (1976), Williams (1981b), Roeper (1987), Anderson (1992) or Roeper and van Hout (1999)106 for English, ­Riehemann (1993) for German, Gràcia (1992, 1995), Gràcia et al. (2000) for Catalan, Val Álvaro (1981), de Miguel (1986), Gràcia et al. (2000), Batiukova (2006) for Spanish or Booij (1992) for Dutch. Others like Chapin (1967), Lyons (1977), Kayne (1981, 1984) or Fabb (1984) for English, or van der Putten (1997) for the corresponding Dutch affix -baar have proposed a mixed analysis: regular derivation applies in the syntax whereas lexicalized forms are created in the lexicon. Finally, there are three

106 Roeper and van Hout (1999: 185) suggest a syntactic analysis of lexical -able. That is, these authors propose, among other things, that “a VP must be projected in the lexicon” or that “-able is a lexical affix with passive properties”.



Previous approaches   

   129

purely syntax-based proposals, Nevins’ (2002), Volpe’s (2005) and McGinnis’s (2010) Distributed Morphology analyses, which account for the derivation of all -ble forms in the syntax.107 As I have already mentioned many aspects concerning the analysis of -ble adjectives in previous work, this section merely seeks to summarize the main insights and shortcomings of previous analyses with respect to the empirical data discussed so far. That is, it is not the purpose of this section to present a thorough revision of all previous work on -ble adjectives, but just point at their main contributions as well as at issues that are still to be solved in the analysis of these adjectives. Specifically, I briefly review some of the main problems and crucial insights that previous analyses have contributed to the discussion with their proposals to build -ble adjectives on a passive structure, on a middle verb and in the DM framework.

5.1 Based on passive I summarize in (216) the evidence for the parallelism between -ble adjetives and the sentential passive construction that I reviewed in the previous sections. (216) a. The external argument of the adjective (generally) corresponds to the internal argument of a (transitive) verb.

b. -Ble adjectives can only be formed from verbs that can also be passivized.



c. Transitive verbs with different uses where only one admits the passive form a -ble adjective in that passivizable use.



d. Both constructions show the same lexical constraints.



e. There are parallel interactions between passive and reflexive (at least in English).



f. They show similar constraints on the expression of a by-phrase related to eventivity.



g. Both allow subject raising with ECM structures.



h. There are parallel interactions between passive and modality.

107 I will not be concerned here with other discourse-based accounts like that of Gómez Fernández (2006). See also Polzin (2000) for the semantics of -ble adjectives in discourse.

130   

   -BLE

The relation between passive and -ble adjectives has been captured in various ways, sometimes through a semantic passive paraphrase for -ble adjectives (Aronoff 1976; Val Álvaro 1981; de Miguel 1986; Albrespit 2009; inter alia). Some works have suggested a more specific analysis of these adjectives as built on a passive structure in the lexicon (Chapin 1967; Vendler 1968; Abraham 1970; Lyons 1977; Fabb 1984; Roeper and van Hout 1999). Vendler (1968) proposes a derivational rule for -ble adjectives in English that goes through the passive; more concretely, the derivational rule for these adjectives (217b) is exactly the same as the one for past participle formation (217a) with the addition of a modality component. In a sense, Vendler analyzes -ble adjectives as being participles, differing only in the fact that the former but not the latter contain a modal feature. As any other rules of word formation, these all would be lexical rules. (217) a. Ven Ni ← Ni wh… is Ven – (by Nj)

b. Vble Ni ← Ni wh… [mod] be Ven – (by Nj)

In addition, Vendler distinguishes what he calls ‘emotive’ -ble adjectives, for which he proposes the rule in (218). Examples of these adjectives would be horrible, terrible, and abominable. (218) e(V)ble Ni ← Ni wh… [evokes, causes, etc] e (V) in Nj Similarly, Lyons (1977) acknowledges that productive -ble adjectives like justifiable in (219a) can be interpreted through a modalized passive predicative sentence as in (219b). He follows Chomsky (1972) in claiming that at least some cases, and at least for the regular sense of the adjective, could be derived from an underlying modalized passive sentential structure through a transformational rule. (219) a. His anger is justifiable. b. His anger is such that it can be justified. This has led some researchers to propose that the passive transformation (Chapin 1967; Lyons 1977) or the passive morpheme/feature is involved in the formation of -ble adjectives (Fabb 1984). At first sight, this suggestion may raise some empirical problems, though, for passives and -ble adjectives display some differences. First, as shown in (220), English resultatives are not equally licensed with -ble adjectives as they are in passives. As pointed out in section 3.3.1, this may be due to the fact that -ble already expresses a result, so that the deviance or utter ungram-



Previous approaches   

   131

maticality of (220a) is probably due to a general ban on having two result phrases in the same structure. The acceptability of the construction by some speakers would then be accounted for in terms of a kind of pragmatic repair strategy, which would allow the expression of a more specific result, in this case to ashes. (220) a. It was burned to ashes.

b. It is burnable ?/*(to ashes).108

Second, en-cliticization in Catalan in (221) shows a contrast in some native speakers’ acceptability judgments, something unexpected if -ble adjectives are to be derived from a passive construction. As I mentioned in section 3.3.1, examples like (221c) are not felt as ungrammatical, but strange, in the sense that one would not use them, even though they do not sound bad. Intuitively, it seems to me that their weirdness has to do with the stativity of the construction and the fact that -ble adjectives must express a general property, although I must acknowledge that I have no real explanation at this point. (221) a. Se’n poden confirmar dues (de notícies).  (Se-passive) (Cat) se=of.them can.3pl.prs confirm.inf two (of news) ‘Two of them can be confirmed. (of the news)’ b. En poden ser confirmades dues (de notícies). (Analytic passive) of.them can.3pl.prs be.inf confirm.part two (of news) ‘Two of them can be confirmed. (of the news)’ c. ?En són confirmables dues (de notícies).  of.them be.3pl.prs confirmable two (of news) Third, the expression of the external argument in a by-phrase is always possible in the passive construction, but not so with -ble adjectives, as shown in (222). As extensively discussed in 3.3.3 and 4.4.2, there are cases where passive does not allow the expression of a by-phrase, and they are all cases of lexically stative verbs with specific by-phrases, so that the ungrammaticality of (222b) is unproblematic and in fact expected given the presence of a modal operator that stativizes the -ble construction.

108 My native speakers’ judgments, not Fabb’s (1984), for whom (218b) would be fine.

132   

   -BLE

(222) a. This claim can be demonstrated by Kate. b. This claim is demonstrable (*by Kate). Fourth, as discussed in the previous sections, although it is possible to have verbs like deplore in a passive sentence with an external argument expressed in a byphrase, as in (223a), this verb, which belongs to the set of low-forming -ble adjectives, does not allow the expression of a by-phrase with -ble, as shown in (223b), be it generic or specific. As will be argued below, this is also unproblematic, given that low -ble adjectives do not show verbal properties and are not expected to pattern with passives. (223) a. The attack was deplored by the government.

b. *The attack is deplorable by the government / by anyone.

Fifth, in double object constructions, it is a well-known fact that in English the passive can target either object, as in (224a)–(225a). However, only the theme internal argument can become the external argument of the adjective, as shown in (224b)–(225b) (see e.g. Williams 1981b; Kayne 1981). (224) a. The money can be paid on Monday. b. payable money (225) a. The owner can be paid on Monday.

b. *payable owner

Various authors have dealt with this characteristic in different ways. So, Williams (1981b), following Wasow (1977), argues that the -ble rule is similar to the adjectival passive in that it can only externalize themes, but not goals or actors, as opposed to the verbal passive which does not exhibit this kind of thematic constancy, as illustrated in (226). Anderson’s (1992) word formation rule also includes that the theme subject of the adjective must correspond to the theme object of the verb. (226) a. Those things are promisable / perishable.

b. *Those people are runnable. (Actor)



c. *Those people are promisable. (Goal)



Previous approaches   

   133

Similarly for Spanish, according to de Miguel (1986), the subject of the adjective must be the theme of the corresponding verb, which is defined as in Anderson (1977: 367), i.e. a theme is what moves with motion verbs, the entity whose location is described with locative verbs, the patient of many transitive verbs or the entity that experiences the action expressed by the verb. Likewise, Roeper and van Hout (1999: 201) base their passive analysis of -able on a [+ Theme] passivizing feature. Specifically, “-able subcategorizes for a passive Voice phrase” with a passive [+ Theme] “in its Specifier” that percolates up to the highest available Specifier. According to these authors, -able and passive -ed are alike with respect to their passive content, the difference being that “in passive sentences, the object obligatorily raises to the subject position to satisfy [+ Theme] and case requirements”, whereas “for -able this movement is possible, but not obligatory” (Roeper and van Hout 1999: 206–7). Assuming the definition of theme in Anderson (1977) and Williams’s (1981b) proposal, the problematic examples in (226) are then not a problem. Sixth, -ble adjectives show more restrictions in the expression of indirect objects and agent-related modification. Even though they are possible, they are very constrained (see also McGinnis 2010, for whom both are considered not possible). Again, this was related to the stativity of the construction due to the presence of the modality operator. There is also a lot of variation in speakers’ judgments that would require a corpus study and experimental research. For the time being, I will assume that these constraints derive from the presence of a high ModP in the internal structure of -ble. A final potential problem has to do with the interpretation of some forms. Recall that, as it was pointed out by Flury (1964: 102), the interpretation of German active adjectives such as verwitterbar ‘weatherable’ derived from eventive intransitive verbs (Vorgangsverben) are very close to a passive reading; but this reading can certainly not come from a passive transformation or a passive component. Similarly for other intransitive verbs that, in addition to their active meaning, can also be said to have a passive meaning (e.g. Catalan transpirable ‘perspirable’). Thus, it seems that in all these cases, the passivity of the adjective comes rather from the semantic interpretation of some event or process acting upon an object, rather than from an underlying passive component in the internal structure of these -ble adjectives. After reviewing the main similarities and differences between passive and -ble, we can conclude that high -ble adjectives do contain a passive component in their internal structure, exactly the same passive component we find in passive sentences (see McGinnis 2010 for English -ble, or Anagnastopoulou and Samioti 2009, 2010 for the corresponding Greek -ble adjectives).

134   

   -BLE

5.2 Based on middle In previous sections, I have established that -ble adjectives exhibit the following properties: they express some modality, they are dispositional generics, they are non-eventive, and they are related to passive. Exactly the same properties have been argued to be displayed by middles of the type in (227a), which has led some linguists to set a semantic parallelism with (227b). (227) a. This book sells well. b. saleable book Before reviewing Gràcia (1995, and subsequent work), where this parallelism is given the form of an explicit proposal, I will briefly summarize the main properties of middles in the context of some of the many works that have examined them.

5.2.1 The middle Many authors have studied the middle construction in different languages, yet this topic appears to be far from being settled. Research on middles can essentially be divided into lexical109 (Williams 1981b; Hale and Keyser 1987, 1988; Fagan 1988, 1992; Levin and Rappaport 1988; Booij 1992; Ackema and Schoor­lemmer 1994, 1995, 2005; or Batiukova 2006)110 and syntactic analyses (Keyser and Roeper 1984; Roberts 1987; Stroik 1992, 1999; Hoekstra and Roberts 1993; ­McIntyre 2006; or Schäfer 2007), though these two alternatives do not exhaust all possibilities. Marelj (2004) proposes a mixed analysis, so that whereas middles in Dutch, English and Hebrew are created in the lexicon, in French, Italian, Polish and Serbo-Croatian the middle formation operation takes place in the syntax (LF). Other approaches have suggested a semantic analysis to middles, notably Condoravdi (1989) and more recently Lekakou (2005). Both accounts suggest that middles are in fact just an interpretation that already existing structures receive.111 Steinbach (1998, 2002)

109 This label refers to all pre-syntactic analyses, i.e. any analysis that treats middles as not being built in syntax proper. 110 Abraham and Leisiö (2006: 16) point out that a lexical analysis is generally proposed due to the “highly modal semantics of the middle construction.” 111 This is in fact a very general opinion among linguists working in Romance, as for instance Anna Bartra, Eulàlia Bonet or Joan Mascaró (personal communication) have mentioned to me, for whom there is no such thing as a middle in Romance. See Bartra (2002).



Previous approaches   

   135

defends a post-syntactic analysis for German, at the interface between syntax and semantics, with some reference to pragmatics to account for some of the properties of German middles. González Romero (2002) suggests that English and Spanish middles are essentially a pragmatic phenomenon. Based on these works, next I summarize the main properties of middles with the understanding that languages differ to various degrees with respect to some of these properties. My description essentially concentrates on those characteristics that refer to English middles.

5.2.1.1 Properties of middles The grammatical subject of middles corresponds to an underlying object or internal argument (Williams 1981b; Keyser and Roeper 1984; Hale and Keyser 1988; Condoravdi 1989; Lekakou 2005), as illustrated in (228). In some approaches, this object is taken as being responsible for bringing about the event (van Oosten 1986; Iwata 1999), though there are also proposals against this view (Condoravdi 1989). In any case, the notional object must have some inherent property or quality that is relevant for bringing the event into being (González Romero 2002). (228) a. This book sells well. b. John sold this book. Therefore, they are fine with resultatives (Hale and Keyser 1987; Hoekstra and Roberts 1993), as illustrated in (229). (229) a. This bread cuts into thin slices easily. b. This metal hammers flat easily.

(Hale and Keyser 1987: 1) (Hoekstra and Roberts 1993: 197)

Essentially all work carried out on middles has acknowledged that they have an implicit agent argument112 (Fiengo 1980; Keyser and Roeper 1984; Condoravdi 1989; Fellbaum and Zribi-Hertz 1989; Ackema and Schoorlemmer 1994; Iwata 1999; Mendikoetxea 1999b), as in passives, and unlike ergatives, that would receive an arbitrary interpretation (Lyons 1977; Fagan 1988, 1992; Fellbaum and Zribi-Hertz 1989; Lekakou 2005)113. However, contrary to the passive implicit

112 Actor for Ackema and Schoorlemmer (1994: 69). 113 González Romero (2002: 82–86) notes that extralinguistic information can give rise to various interpretations of this argument – generic, specific, collective.

136   

   -BLE

agent, the implicit agent of the middle cannot license a by-phrase, as shown in (230a). Nor is the implicit argument eligible for control, as in (230b).114 (230) a. *Bureaucrats bribe easily by managers. (Keyser and Roeper 1984: 406) b. *Bureaucrats bribe easily to keep them happy. (Keyser and Roeper 1984: 407) Agent-oriented adverbs do not seem possible in English, either (Fagan 1992: 56).115 (231) *Polyester cleans carefully However, they can license an instrumental (Condoravdi 1989: 16) and a withoutphrase (Iwata 1999).116 (232) a. This glass breaks easily with a hammer.

114 These properties apply to English. According to Condoravdi (1989: 25), the agent can be expressed in Greek or in Russian. See also Bruening (2012) for arguments based on instrumentals and external-argument-oriented commitatives that evidence the presence of an external argument in English middles. 115 Though they are fine in French, according to Hirschbühler (1988: 99), cited in Fagan (1992: 157), as illustrated in (ia) with her example. The corresponding examples are given in Catalan (ib) and in Spanish (ic), although as I mentioned in footnote 111, it is not clear that these are cases of a middle construction. (i) a. Ce livre se lit avec plaisir. (Fr) b. Aquest llibre es llegeix amb plaer.(Cat) c. Este libro se lee con placer.(Sp) this book se read.3sg.prs with pleasure ‘This book is read with pleasure.’ 116 According to Hale and Keyser (1987: 16–19), there is no agent role in the middle, i.e. it is completely absent from its LCS (Lexical Conceptual Structure). The only concrete evidence for an agent in middles comes from the possible presence of an instrumental. However, they doubt that the compatibility of an instrumental with middles can be related to the presence of an agent theta-role in the LCS of the verb. They provide as evidence examples of reflexive middles, as in (i), and conclude that the agency associated with the presence of an instrumental is due to the interpretation of the instrumental itself and its implication of a user. (i) The toughest carrots virtually slice themselves with this handy tool. As for the intuition about agentivity in middles, they suggest that it may be related either to the adverb (e.g. break) or to a means component present in the LCS of some verbs (e.g. cut). The latter suggestion is taken up in Rapoport (1999).



Previous approaches   

   137

b. This blouse washes normally without adding any special product. According to Marantz (1984), the logical subject of the middle is semantically present, but syntactically absent. Also for Condoravdi (1989), this agent is not associated with a syntactic position, but it is “an entailment of the lexical meaning of the verb.” Similarly, Schäfer (2007) appeals to the encyclopedic knowledge about the predicate. However, Stroik (1992, 1999) claims that the external argument of middles is not syntactically suppressed.117 Evidence that it is present comes from the possibility of having a subject-contained anaphora, illustrated in (233a), and the possibility of having this external role expressed in a for-phrase, as exemplified in (233b).118 Stroik (1992: 129) further points out that the impossibility of having agent-oriented adverbs and control clauses would be due to the fact that these elements require not only an agentive argument but also an eventive predicate. Since middles are statives, these elements would be excluded from middles due to their lack of eventivity (cf. Dowty 1979: 55). (233) a. Books about oneself never read poorly. b. That book read quickly for Mary. (Stroik 1992: 129, 131) As noted in Ackema and Schoorlemmer’s (1995:  179–180), on the one hand, Stroik’s assumption with respect to the expression of the external argument is too weak, because identical for-phrases occur in non-middle sentences with no implicit argument, e.g. That book is too thick for Mary;119 on the other hand, it also appears as too strong, because we would expect such a PP to be possible in every middle, which is not the case, witness examples like These books don’t sell (*for the average shopkeeper). This would relate to their being characterized as generic statements (Fagan 1988, 1992).120 As such, they are non-eventive like stative verbs, i.e. they do not

117 See also Fellbaum and Zibri-Hertz (1989). 118 This evidence has been directly criticized in Zribi-Hertz (1993), Rapoport (1999), or Lekakou (2005). 119 As Ora Matushansky (personal communication) points out, the for-phrase in middles could also be analyzed as introducing the comparison class for the adverb quickly, and not the external argument of the verb. 120 The genericity of middles is questioned in Mendikoetxea (1998), as cited in Mendikoetxea (1999b). According to her, middles do not quantify over events but over entities and, as

138   

   -BLE

describe particular events, but describe stative propositions (Keyser and Roeper 1984; Condoravdi 1989; Mendikoetxea 1999b). All middles are individual-level predicates (Ackema and Schoorlemmer 1994; Batiukova 2006), they predicate an inherent quality of the grammatical subject. According to Levin (1987), the incapacity to refer to a real occurrence of an event is a fundamental property of middles crosslinguistically, it is found in accusative languages as well as in ergative languages. As statives, they cannot occur in imperative or vocative constructions, nor can they appear in progressive structures (Dowty 1979). 121 As noted in Condoravdi (1989), in languages with the perfective-imperfective aspectual distinction, middles are marked with imperfective aspect, as in all generic sentences. Massam (1992) claims that middles contain a modal element that gives a generic meaning to the sentence; this element is equivalent to modal can. Indeed, middles typically involve some modal notion. They generally refer to the possibility, ability or likelihood of the event denoted by the predicate (Massam 1992; Fagan 1992; Iwata 1999; Steinbach 2002).122

individual predicates, they have no spatio-temporal eventive argument. Generic events, on the other hand, do quantify over events and bind situation variables. Another argument cited by this author against the genericity of middles comes from the fact that a middle is true even if the action denoted by the verb has never taken place. For her, then, middle constructions are “non-eventive propositions that express modality (ability or possibility)” (author’s translation). 121 According to Iwata (1999), although middles are non-eventive, they can be acceptable in the past tense and with progressive aspect; in those cases, they express specific, actual events. (i) a.  Grandpa went out to kill a chicken for dinner, but the chicken he selected didn’t kill easily. b. The curry digested surprisingly easily last night. Iwata (1999: 530) (ii) a. The wall is painting easily. b. The floor is waxing easily. Iwata (1999: 538) Similar examples can be found in Fagan (1992: 53), in (iii), and Van Oosten (1977: 463), in (iv). (iii) This manuscript is reading better every day. (iv) The book is selling like hot cakes. These sentences pattern with “true” middles in that they show the same adverbial effects, as shown in (v). (v) a. *The steaks you bought yesterday cut. b. The steaks you bought yesterday cut like butter. These progressive and past middles do not express modality, though, which Iwata takes as an argument for saying that modality is not an inherent property of middles (contra Massam 1992). 122 González Romero (2002: 156-189) maintains that this modal feature allows us to distinguish middles from passives, se-passives and ergatives, because the latter all lack any modal implication whatsoever.



Previous approaches   

   139

Only some transitive verbs appear in this construction in English. Roberts (1987) and Jaeggli (1986) claim that the set of verbs that may appear in the middle construction is restricted by the affectedness constraint. This claim has been further adopted in Hoekstra and Roberts (1993) or Cornips and Hulk (1999) (cf. Fagan 1992, Ackema and Schoorlemmer 1994 against this view). Finally, they require the presence of an adverb (Keyser and Roeper 1984, Condoravdi 1989, Zwart 1997, Mendikoetxea 1999b), though not all of them (Fagan 1988, 1992), as in (234)–(235). There seems to be crosslinguistic variation in this respect (Fagan 1992 for French; García Negroni 2002 for Spanish; Lekakou 2005 for Greek). For instance, according to Ackema and Schoorlemmer (1994), adverbial modification is not crucial to middles, so that a sentence like (236) would be fine given an appropriate context. (234) a. *This book reads. b. This book reads well. (235) a. This dress buttons. b. ? This dress fastens. (Fagan 1988: 201) (236) This bureaucrat bribes.

5.2.1.2 Middles and -ble Let us now compare the similarities between this middle construction and -ble adjectives. On the one hand, they share some syntactic properties, as shown in (237)–(238), that correlate with their semantic properties. First, in both constructions, a specific implicit external argument cannot be expressed, as in (237). (237) a. This text translates easily *(by John). b. This text is translatable *(by John). However, an external argument can be expressed with -ble if it is generic or less specific, as we have seen above.123 This is exemplified in (238) for Spanish.

123 Gràcia (1995: 101) observes that an adverbial easily is necessary as well, as it is with middles. She provides the following contrast:

140   

   -BLE

(238) a. un texto traducible *(por Pablo) (Sp) a text translatable *(by Pablo) b. un texto traducible por un experto en la materia a text translatable by an expert in the matter ‘a text translatable by an expert in the subject matter’ Gràcia (1995 : 102) reports the contrast in (239) for Catalan. I agree with her observation that middles are rather strange with an explicit agent, but that they become better with a generic or less specific agent.124

(i) * un llibre recomanable pel professor(Cat) a book recommendable by.the professor (ii) ? un llibre fàcilment recomanable per qualsevol que l’hagi a book easily recommendable by anyone that it=have.3sg.prs.sbjv llegit (Cat) read.part First, I find (ii) completely fine, and so do my informants. Second, although it is true that sometimes some adverbial is required, this is not always exact. There are many cases where no adverbial is present. (iii) una pel·lícula censurable per l’Església(Cat) a film reprehensible by the=Church This, in fact, could provide support for the view that middle and -ble adjectives are somehow related, for there are cases of middles that do not require the presence of an adverbial, either. On the other hand, if it turns out that the necessity for adverbial modification is somehow syntactically encoded, this could be related to the fact that there are many adjectives that tend to occur only with negation, i.e. negation would realize the same function the adverbial does. The question is then, what the function of the adverbial is. In the case of negation, Fábregas (2005: 107) suggests that the negative operator applied to a property identifies a constant value, which allows it to express a class. When negation is not present, the scale expressed by the adjective is not delimited, and so it cannot refer to a class. Applied to our -ble adjectives, we could relate the necessity of having negation (or some other device that can delimit the denotation of the adjective) to the fact that -ble adjectives are characterizing predicates (Krifka et al. 1995: 7). 124 Actually, some native speakers and I find some of the middles with an explicit generic or non-specific agent fully acceptable, given an appropriate context. For instance (239b) is perfectly fine for me and for (some of) my informants. Other examples like (i) are not accepted at all, although there is always a clear contrast with examples like (ii). (i) ?* Aquest text es llegeix fàcilment per qualsevol expert en la materia. (Cat) this text se read.3sg.prs easily by any expert in the subject (ii) * Aquest text es llegeix fàcilment per en Pau. (Cat) this text se read.3sg.prs easily by the Pau I guess that these differences are due to the relevance or possible implication of the external agent in each case. That is, whereas a briber seems absolutely necessary for an act of bribing



Previous approaches   

   141

(239) a. * Els buròcrates es corrompen fàcilment pel senyor Pere Pi. (Cat) the bureaucrats se bribe.3pl.prs easily by.the mister Pere Pi

b. ?? els buròcrates es corrompen fàcilment per qualsevol que the bureaucrats se bribe.3pl.prs easily by anyone that els ofereixi diners125 them offer.3sg.prs.sbjv money Similar contrasts can be found in Spanish, as exemplified in (240).126

(240) a. * Esta lengua se traduce bien por Pablo. (Sp) this language se translate.3sg.prs well by Pablo b. ?? Esta lengua se traduce fácilmente por un experto en this language se translate.3sg.prs easily by an expert in la materia. the subject ‘This language translates easily by an expert in the subject matter’

to be able to take place, it is less clear that the abilities of a reader are necessary for saying that a text reads well; it is more than likely that the text reads well because of its neat and clear handwriting, i.e. thanks to the inherent properties of the text, in which case a reader becomes somehow irrelevant. 125 As observed in Fellbaum and Zribi-Hertz (1989) for French, a sentence like (ia) below is ambiguous between a property reading and an eventive reading. Same facts hold for Catalan (ib) and Spanish (see Mendikoetxea 1999b). The eventive reading would correspond to the socalled pasiva refleja in Spanish or passiva perifràstica in Catalan, whereas the property reading corresponds to the interpretation of an English middle construction. (i) a. Le grec se traduit facilement. (Fr) the Greek se translate.3sg.prs easily ‘Greek translates easily.’ b. El grec es tradueix fàcilment. (Cat) the Greek se translate.3sg.prs easily ‘Greek translates easily.’ 126 According to Mendikoetxea (1999b), middle-passives imply the intervention of an agent (causer or experiencer) that cannot be specified. Given appropriate contexts, my Spanish native speakers find sentences like (240b) much less deviant than (240a), with the intended reading, that of predicating a property of the subject. See previous note. I generally do not accept by-phrases with medio-passive or passive sentences in Spanish. So, for instance, the always cited passive sentence Se firmó la paz por los embajadores ‘Peace was signed by the ambassadors’ is completely anomalous for me. However, I do find a clear contrast in (240) above.

142   

   -BLE

Second, this implicit external argument can license an instrumental adjunct, as in (241). (241) a. This glass breaks easily with a hammer.

(Condoravdi 1989: 16)

b. This glass is breakable with a hammer. Since both are stative constructions, the restrictions on the expression of agentrelated elements are in both cases tied to this property. Furthermore, as I mentioned at the outset, they are intuitively felt as being very close to each other with respect to the meaning they convey: they both express modal, passive, and generic content. Indeed, the type of modality that has been proposed in this volume for -ble adjectives based on Kratzer (1981) is the same type of root modality that has been argued for in Lekakou (2005) for middles. That is, both constructions express an inner dispositional with a circumstantial modal base of the type in view of or in virtue of. This parallelism has led Gràcia (1992, 1995) to suggest an analysis of -ble adjectives as built on a middle structure. The next section is devoted to the empirical problems posed by such a proposal given the differences between middles and -ble adjectives.

5.2.2 -ble derived from middle Gràcia (1992, and subsequent work), building on Levin and Rappaport (1988), proposes that those -ble adjectives that convey a modal passive meaning are built on a middle verb in the lexicon in Catalan. Although I agree with the basic insight relating middles and the formation of -ble adjectives, such a lexical analysis has a number of empirical problems, when applied to English. As illustrated in (242)–(248), -ble adjectives are possible with accomplishments in general, in (242), achievements (243), states (244), Subject-Experiencer psychological verbs (245), and can also be predicated of a locative, as in (246), of a selected PP, as in (247), or of a cognate object, as in (248). Since all these adjectives are both passive and modal, they should all be derived from a lexical middle verb according to Gràcia (1995), which does not seem possible, because there are no such middles (cf. Fagan 1992; Lekakou 2005). (242) a. *This food eats easily. b. eatable food



Previous approaches   

   143

(243) a. *This language acquires easily. b. an easily acquirable language (244) a. *The truth believes easily. b. a believable truth (245) a. *His behavior admires easily. b. his admirable behavior (246) a. *This place lives well. b. this liveable place (247) a. *His services dispense (with) easily. b. his dispensable services (248) a. *This music dances well. b. danceable music Furthermore, only middles license a resultative phrase, as illustrated in (249), although I have already discussed that its unavailability with -ble adjectives may obey more general constraints related to the fact that -ble incorporates a result in its internal structure. (249) a. This metal hammers flat easily.

b. ??/*This metal is hammerable flat.

Additional empirical problems for an analysis of -ble adjectives as being built on lexical middles in Catalan come from the existence of -ble adjectives that, despite not being derived from transitive verbs, convey exactly the same modal passive meaning, such as cases of denominal -ble adjectives in (250a), -ble adjectives derived from unaccusatives in (250b) or unergatives in (250c) that have a modal meaning. (250) a. ministrable ‘cabinettable’ b. durable ‘lasting’ c. fiable ‘reliable’

(Sp)

144   

   -BLE

There are also -ble adjectives that have no existing verbal base (251), and still have the same semantics and syntax as all other -ble adjectives. (251) a. assequible ‘achievable’ vs. *assequir

(Cat)

b. credible vs. to cred Finally, the middle involves some kind of implicit agentivity that is less syntactically active than in -ble adjectives,127 so that this implicit argument cannot appear in a by-phrase in English, (252a), and cannot control the PRO subject of an infinitive purpose clause (252b). However, these structures are possible with -ble adjectives as the contrasts with (253) show. (252) a. *Bureaucrats bribe easily by managers. (Keyser and Roeper 1984: 406) b. *Bureaucrats bribe easily to keep them happy. (Keyser and Roeper 1984: 407) (253) a. The grammar is learnable by the child.

(Roeper 1987 : 281)

b. Some goods are insurable to decrease the profit risks. (Roeper 1987 : 280) The differences in the behavior of middles between English and Romance are certainly due to a different status of these constructions in these languages (see footnote 111). To conclude, an analysis of -ble as being built on a middle verb does not seem feasible, even though it seems proven that both structures share an important number of syntactico-semantic properties that calls for a more detailed comparison and that should be taken into account when attempting to establish their internal configuration. Such a comparison was one of my main aims at the beginning of my research on -ble. However, I abandoned this goal in favor of an analysis of the complex constructions discussed in chapters 3 and 4, which have brought me well beyond the well-known cases of -ble adjectives. Thus, I will leave a comparison between -ble and middles for future research (but see Anagnastoupoulo and Samioti 2009, 2010 for a recent analysis of these constructions that focuses on Greek data).

127 But see Fellbaum and Zribi-Hertz (1989) for a different view.



Previous approaches   

   145

5.3 -ble in Distributed Morphology The last analyses that I will review, Nevins’s (2002), Volpe’s (2005), and McGinnis’s (2010) have been developed within Marantz’s (1999) Distributed Morphology proposal of two syntactic attachment sites, above and below a first categorizing functional head, thus suggesting that regular -ble adjectives, e.g. eatable, contain a verbalizing v head, which is absent from the more lexicalized ones, e.g. edible. Volpe (2005) does so on the basis of the morpho-phonological evidence described in Aronoff (1976 : 121–129), and applying Chomsky’s (1999) concept of phase, thus subsuming boundaries and phonological cycles under the notion of phase. More specifically, the first category-assigning functional head defines a phase (Marantz 1999), so that when the affix attaches directly to the root the structure is spelled-out semantically and phonologically. This gives rise to special semantic and phonological outputs, as in the cases in (254). (254) a. edible ‘safe to eat’ b. cómparable ‘equivalent’ c. tolerable ‘moderately good’ When the affix merges with a functional head v, it attaches to an already spelled out verb, so that neither the meaning nor the phonology of the root can be affected, as in eatable, compárable, toleratable (= capable of being V-ed). Volpe’s (2005) analysis of cómparable/compárable is as in (255)–(256). (255) Cómparable a a +abl

√compar-

(256) Compárable a a #abl

v compáre v

√compar-

146   

   -BLE

As pointed out in Nevins (2002: 2), who proposes the same structures, this distinction gives rise to contrasts such as (257). Whereas we obtain a contradiction in (257b) due to the verbal compositional meaning of compárable, the idiosyncratic and purely adjectival meaning of cómparable in (257a) is compatible with the speaker’s inability to make a comparison between the two fish. (257) a. Although the sizes of the two fish were cómparable, I couldn’t compare them. b. *Although the sizes of the two fish were compárable, I couldn’t compare them. However, since Volpe’s (2005) distinction focuses on the morpho-phonological description presented in Aronoff (1976) and the above mentioned lexical semantics of the base, it remains unclear whether this proposal correlates with a difference in the syntactic properties of these adjectives, specifically with respect to their argument structure properties, and whether and how this analysis can be made compatible with similar -ble adjectives in other languages with more complex derivational structures. More concretely, if Oltra-Massuet’s (1999, 2000) account of theme vowels is right, when Volpe’s proposed structures are applied to Catalan they cannot accommodate the theme vowel that appears in both types, the more lexicalized temible ‘fearsome’ and the regular traduïble ‘translatable’. Since this is the same structure that Nevins (2002) and McGinnis (2010) propose for root-attaching or low -ble, they would face the same morphological problem. (258) *temble instead of temible ‘fearsome’

a. Before well-formedness conditions a a



√tem-

b. After well-formedness condition on F0 and Vocabulary insertion: *temble a a √tem-

ti a

a bl

Th e



Towards an analysis of -ble   

   147

Also, the fact that all -ble adjectives receive essentially the same or similar interpretation remains obscure and unexplained. Besides, the syntactico-semantic properties of -ble adjectives we have seen so far suggest that these adjectives encode a structure that must be more complex than the languages under study morphologically show. Although McGinnis (2010) proposes a structure that includes a VoiceP for high -ble, so that -ble can be either root-selecting or Voiceselecting, her structure seems to me still incomplete given the empirical evidence suggesting the presence of aspectual as well as modal structure within the internal syntax of -ble, which I now turn to. Before that, let me add a brief note on my assumptions with respect to the formal syntactic configuration, already mentioned at the outset in chapter 1. My analysis follows proposals in Chomsky (1995), Marantz (1997) or Harley (1995, 2005) that assume that a single projection vP serves two functions, to verbalize the root and to introduce the external argument, instead of assuming the v/ Voice split structure found in recent work.128 I have done so only for simplicity reasons, since for present purposes it suffices to distinguish between a structure containing an eventive variable and a non-eventive structure for the two types of -ble.

6 Towards an analysis of -ble In the previous sections I have examined the morphology of -ble adjectives, their external syntax and the kind of meaning they convey, i.e. their syntactico-semantic properties in the four languages under investigation, while reviewing some of the main contributions in the study of these adjectives. I have provided empirical evidence that argues for the existence of two different syntactic types of -ble, potential or high -ble and low -ble, as summarized in Table 3. In this section, I provide an analysis of the internal structure of -ble adjectives that tries to accommodate all their features.

128 But see Kratzer (1996) for the split v/Voice hypothesis. See for instance Harley (2012) and references therein for a recent approach in favor of this hypothesis.

148   

   -BLE

POTENTIAL or HIGH -ble

LOW -ble

 1. always expresses possibility

 1. may express no modality, possibility or necessity  2. meaning is compositional  2. may convey idiosyncratic meaning  3. does not trigger stem allomorphy  3. may trigger stem allomorphy  4. allows derived bases  4. attaches only to underived (truncated) roots  5. [English] does not shift stress  5. [English] may shift stress  6. [English] is always spelled as -able  6. [English] may be spelled as -ible  7. [German] is always spelled as -bar  7. [German] may be spelled as -abel  8. [English] selects negative /un/  8. [English] selects negative /iN/  9. allows the expression of other arguments  9. does not allow the expression of other arguments 10. may license indirect objects 10. does not license indirect objects 11. [Romance] allows postponed adverbials 11. [Romance] does not allow postponed adverbials 12. licenses aspectual adverbials 12. does not license aspectual adverbials 13. [Romance] does not generally appear in 13. [Romance] may appear in attributive attributive position position 14. [English] allows raised subjects 14. [English] does not allow raised subjects 15. licenses by-phrases 15. does not license by-phrases 16. can control into a purpose clause 16. cannot control into a purpose clause 17. licenses (a restricted set of) agent17. does not license agent-oriented adveroriented adverbials bials 18. licenses instrumental phrases 18. does not license instrumental phrases 19. does not generally license -ly affixation 19. does generally allow -ly affixation 20. [Romance] does not generally license 20. [Romance] generally licenses -ísimo / -íssim affixation -ísimo / -íssim affixation 21. does not generally allow degree modifica- 21. generally allows degree modification by tion by very very Table 3. Two types of -ble: Main morphological, syntactic, and semantic properties

6.1 The internal syntax of -ble My analysis of the internal structure of -ble builds, on the one hand, on the basic insight in Vendler (1968), who linked the rule of -ble adjectival formation to the past participle formation by having it modalized. On the other hand, it builds on Alexiadou’s (2001) and Embick’s (2003, 2004) work on participles. As I reviewed in the introduction, Embick (2004) proposes that in addition to eventive participles, there are two types of stative participles. Resultative participial formations in (259b) differ from simple state participles (259a) in that only the former contain a verbalizing v.

Towards an analysis of -ble   



(259) a. Stative

b. Resultative

The door was open

The door was opened Asp

Asp AspS

   149

AspR ed

√ open

vP DP the door

v v

√ open

Both are stative; however, whereas Asps defines a simple state, the resultative requires an Aspr that expresses a state resulting from a grammatically represented prior event, realized by the head v. Both functional heads v and Asp include “basic information about the eventivity/stativity, along with further aspectual information about the status of the event or state.” On the other hand, only the eventive participle contains a passive little v with an agentive feature that can license agent-related elements. (260) Eventive Asp Asp

vP v [AG]

√P √ open

DP the door

If we go back to -ble, I have distinguished two main types of -ble adjectives, whose properties make them resemble the distinction between adjectival and verbal participles. Whereas high -ble shows eventive properties, low -ble is clearly stative. On the one hand, potential or high -ble adjectives have an implicit agent that can be expressed in a by-phrase and can license an instrumental, an infinitive purpose clause, and (partially constrained) agent-oriented adverbials and other verbal arguments. They also accept manner adverbials and aspectual modifiers, and, in general, do not allow -ly derivation or degree modification. Romance high -ble adjectives show additional verb-related properties, such as licensing postponed adverbials; and English allows raised subjects.

150   

   -BLE

At the same time, I have claimed that they must express a result, and assuming that only one result can be expressed in a single proposition, this prevents their co-appearance with a resultative. Thus, they express that it is possible for someone to achieve a state that results from a prior event. In a sense, then, high -ble adjectives show mixed properties of the two eventive participial structures in Embick (2003, 2004), as they are resultative and show agent-related verbal passive features. On the other hand, low -ble adjectives do not allow the expression of the external argument in a by-phrase, so that instrumentals, infinitive purpose clauses, agent-oriented adverbs or aspectual modifiers are all banned. They do all allow further derivation with -ly and degree modification, typical properties of adjectives. Thus, whereas low -ble adjectives do not show any eventive properties, high -ble adjectives show properties of mixed derivations, i.e. externally they behave as adjectives, internally they display properties associated with verbs. As pointed out in Embick (2004), this kind of mixed derivations are best “accounted for in terms of heads attaching to different structural positions in potentially clausal structure.” Based on proposals in Alexiadou (2001) and Embick (2003, 2004) with respect to participial morphology, I propose that the distinction must be understood as a distinction between a modalized adjective and a modalized resultative passive structure. Let us consider the structural differences between the two main types of -ble in more detail.

6.2 The structure for high -ble For high -ble, I propose the configuration in (261). Descriptively, high -ble adjectives express a generic property according to which it is possible for some originator to achieve a resultant state out of an eventive subcomponent.129

129 Note that this structure contains the abilitative-skeleton proposed in Hackl (1998), since we can equate the functions of the structure √-v to those of VP-VoiceP in Kratzer’s (1996) split hypothesis.

Towards an analysis of -ble   



   151

(261) Potential or high -ble, e.g. modificable, traduïble, ‘modifiable, translatable’ aP a

it creates a property

Mod ◊

creates a modalized resultant state with a circumstantial modal base

ModP AspP Rc

Aspr

creates a resultant state out of a prior event vP

v [pass]

event interpretation / implies external argument √



DP

internal argument

In this case, the root first merges with a verbalizing head v that can license other arguments as well as manner adverbials (Cinque 1999, Alexiadou 2001). As proposed in Alexiadou (2001), the head v has two types of properties, semantic (eventive interpretation and implication of an agent – or external argument, more generally) and syntactic (transitivity properties related to Case and introducing the external argument), which are dissociated in the passive. A passive v head cannot check Case, nor can it assign an external argument in its specifier, thus accounting for Burzio’s generalization. This is exactly what we find in high -ble, a passive little v head with the semantic properties of a passive v head. That means that there is no explicit external argument realized in [Spec, vP], but this passive v head ensures the presence of an external argument that can license by-phrases and other verb-related arguments, albeit in a way more constrained than in passives. This is not a stipulation, but follows from the interaction between passive and the stativity of the configuration, in the same way that such interaction bans the expression of an external argument with passive sentences of stative verbs like *La respuesta {es/fue} sabida por Juan ‘lit. The answer is/was known by John’ in Spanish. The main difference between a regular verbal passive structure and the passive structure of -ble would lie in the operator that binds the external argument. According to much previous work on passives (e.g. Chierchia 1989, 2004), passivization involves the saturation of the external thematic role by existential closure via an existential ∃ operator, which takes care of the open variable in the semantic component. This prevents the syntactic mapping of the external argument, which can only be made explicit as an adjunct by-phrase. Now, as I have already mentioned, -ble adjectives do not show the actuality entailment found in the passive construction. That means that there cannot be existential closure in

152   

   -BLE

the case of -ble adjectives, as there is in passives. Thus, a sentence like The book was translated entails the existence of an agent that carried out the translation, as represented in (262). (262) The booki was translated ti. ∃e ∃x [translate (e) & Agent (e, x) & Theme (e, the book)] However, The book is translatable expresses that it is possible for an arbitrary or generic someone to translate the book, i.e. the event need not have taken place and may never take place. In this case, the open variable of the external argument is bound by a modal operator, which induces an opacity context thus restricting the range of possible realizations of the external argument to non-specific NPs. This is in accordance with Pesetsky’s (1995: 120) suggestion that the modal present in the semantics of -ble takes the verbal external argument as one of its arguments. In my view, there cannot be arbitrary existential closure on the external argument of the type found in passives, as pointed out in Portner (2009) for the interpretation of VoiceP in Hackl (1998). Rather, as suggested in Di Sciullo (1997), there seems to be universal generic quantification over the external argument, so that The book is translatable cannot mean that there is an arbitrary someone who can translate the book, but rather that it is possible for some arbitrary someone to translate the book. The verbalizing head v also interacts with an Asp head that quantifies over event variables (Hacquard 2009) to license adverbial modification. This Asp head, which is possibly specified as a type of resultative Aspr (Embick 2004), is also linked to the licensing of aspectual modifiers (Cinque 1999). This head will turn crucial in the analysis of denominal -ble adjectives. This aspectual head is then merged with a possibility ModP that binds the verbs’s external argument and is further restricted by a circumstantial modal base of the type in view of the inherent properties of the verbal internal argument, as has been represented in Hackl (1998). The semantics of the modal will be regular throughout, as it is merged with a resultative event. Finally, this configuration merges with a head little a that creates a property. As argued for in Marantz (1999a, b, 2001), when a root is categorized, those category features are carried along through the entire derivation. In the case at hand, the root is verbalized, so that the verbal properties percolate up to little a, which is what prevents further derivation with -ly

Towards an analysis of -ble   



   153

affixation130 and severely restricts degree modification with very and -ísimo/íssim in Romance.131

6.3 The structure for low -ble As for low -ble adjectives – and evaluative adjectives such as admirable or adorable as a subset of them –, they are adjectival modalized states, as structurally represented in (263), i.e. they express a (sometimes idiosyncratically) modalized property. (263) Low -ble: e.g. admirable, credible, applicable

aP a

it creates a property

Mod ◊

modalizes a stative root

ModP AspP Rc

stativizes the root √

Asps √

DP

internal argument

In the case of low -ble, the root first merges with a stative AspP that stativizes the root, but still does not categorize it. Then merging with ModP further modalizes it. This accounts for non-compositional meanings, as well as for the difference in modality with specific roots. That is, the modality of this ModP is always the same, a possibility circumstantial root modal. When it merges with an event or the

130 It would be possible, though, after negative prefixation attachment, as in (i). (i) * (in)olvidablemente (Sp) (un)forgettably 131 As Michal Starke (personal communication) and Mercedes Tubino (personal communication) point out, there is a potential problem with the suggested structures within the DM model. The modal adjectival suffix -ble is here apparently split into different functional heads, a proposal that seems to run against basic DM tenets, whereby Vocabulary items are inserted into syntactic terminal nodes. Note that I assume that [bl] is specified for insertion into an a node, even if the features of -ble are distributed amongst different heads. See also the discussion in section 6.4.2 for the possibility of having all predicate-related functional heads merged into a single head.

154   

   -BLE

result of an event, its meaning is the expected one, it is compositionally obtained. However, when it merges with an (aspectually stative) root, its meaning is individually negotiated with the root, which may give rise to a non-potential interpretation. As discussed in § 5.3, this is possible because the root has not yet been categorized and spelled-out with low -ble, which may give rise to special semantic and phonological outputs; whereas in the case of high -ble, little v defines a phase (Marantz 1999), so that when the root merges with this category head the structure is spelled-out semantically and phonologically, and no irregularities or idiosyncrasies are possible, i.e. the first category-assigning head defines a phase and assigns an interpretation which is carried along throughout the derivation (see also Arad 2003). Thus, that most cases express what has been called an evaluative or necessity modality is due to the fact that ModP combines with still uncategorized complexes of √-Asps, so that meaning may appear as non-strictly compositional. The fact that the combination of Mod with a semantically homogeneous set of Psych-roots always results in a modality change further supports the view that the meaning of the root determines the meaning of the combination, whereas the meaning of the root has no bear on the semantics of high potential -ble, where Mod combines with an eventive variable. Note that it is possible for some high -ble adjectives to receive some additional meaning that may come from the context, from language-particular concepts or from world knowledge, as in the case of eatable or readable mentioned in footnote 3 above. This should in no way undermine a syntactic view of word formation, as it does not in the case of idioms (see Marantz 1997). The claim is that only in the case of low -ble, because of its internal structure, it is possible to obtain a modality meaning different from possibility. In the structure in (263), the functional head a, which defines properties, further merges with the modalized root and categorizes it. There is neither eventivity nor a syntactically active external argument. This means that low -ble adjectives can meet the selectional conditions for -ly affixation and degree modification, because they are (gradable) adjectives denoting properties without any eventive content. The structure in (263) would correspond to Nevin’s (2002) or Volpe’s (2005) root-affixation, which means that this is the structure I assume for forms like afable ‘affable’ or potable ‘potable’ and other idiosyncratic forms such as possible, formidable, terrible, or comfortable that exist in the different languages. In these cases, though, we would presumably have the adjective’s external argument directly merged in [Spec, aP]. This is also the structure we find with English stress shifting cómparable or allomorphic roots such as English applicable or Catalan perceptible, as well as with truncated forms like English tolerable, demonstrable or German entschuldbar ‘excusable’ diskutabel ‘debatable’. Also non-productive negative prefixation by in- would combine with low attach-

Towards an analysis of -ble   



   155

ing -ble forms only, as in imperceptible, as opposed to unperceivable that would be a case of high -ble affixation. The to my knowledge unique form assequible ‘affordable’ in Catalan or asequible in Spanish appears as puzzling, as it can be included in the set of low -ble for morphological reasons, i.e. because of the non-existence of a verb *assequir; or it may be inserted in the high -ble structure (261) on syntactic grounds i.e. due to the existence of assequible por todos ‘affordable by everybody’. Given that it exhibits properties related to an eventive component, I would rather adopt the latter, and further assume that the allomorph assequ- is selected for the corresponding verb achieve in the context of -ble.132 As Hagit Borer (personal communication) pointed out to me, this dual structural nature of -ble adjectives, i.e. that they can be inserted in these two different structures, predicts that morphologically more complex verbs, such as those that take the suffix -ize, e.g. capitalize, should always combine with a potential or high -ble. As far as I can see, the prediction is borne out, and verbs expressing nonpotential modality are morphologically simple.133

6.4 The Vocabulary item -ble crosslinguistically For English and Romance, an underspecified vocabulary item in (264), together with the requirements on the availability of a theme and an originator should provide the right results.134 (264) a. [bl] ↔ a / Mod

(Cat & Sp)

b. [əbəl] ↔ a / Mod (Eng)

132 See Bonet (1995) for a similar case, where the Catalan verb dinyar-la ‘die’ with the clitic does not exist as *dinyar. 133 A potential problem could be the adjective agradable ‘agreeable’, which would have the internal structure in (i). Examples like this could be treated as having the causative affix merge below the first-category assigning functional head, along the lines of Marantz’s (2001) analysis of destroy as [de-stroy]. Or else, could be treated as high -ble adjectives that can obtain an idiomatic interpretation, in this case, the reading ‘affable’. (i) acause – gratn – a – ble ← gratn ‘pleasure’ (Cat) ‘that pleases’ or ‘affable’ 134 There is in Spanish a fossilized -dero suffix (Lang 1990: 208) that also contributes some modal meaning. Also -dor in Catalan may have a similar fossilized modal behavior with some roots (Gràcia 1995).

156   

   -BLE

Thus, for both low and high -ble adjectives, an underspecified [bl]/ [əbəl] Vocabulary item is inserted in a little a node in the context of Mod  – possibly specified as possibility. In the case of high -ble, the conditions imposed by -ble will be met by the internal argument of the verb and the semantic properties of a passive little v present in the structure. As for low -ble with verbal roots that do not show eventive properties with -ble, the conditions imposed by -ble will be met by their internal argument and the conceptual semantics of the root, i.e. its encyclopedic meaning. Besides, it will be necessary to add a list of idiosyncratic roots that take this suffix in both languages, e.g. comfortable, formidable, etc, that is, those special roots that can select -ble but do not meet the general requirements. In addition, English will need a list that specifies the roots that are written as -ible.

6.4.1 The case of German -ble The grammar of German -ble deserves special attention. As I have shown in § 2.1 above, in addition to -bar and -wert, there is in German a third unproductive suffix -abel that is directly merged with the root and that stands in complementary distribution (see chapter 5 for a qualification on the status of German -abel). Thus, German has three different exponents that are ordered according to the Subset Principle, so that they compete for insertion in the same terminal node. Pending a thorough analysis of these German Vocabulary items135, we can simplify by saying that whereas -wert is specified for being inserted in the context of psychological roots, and -abel specializes in Latin roots, -bar is the default exponent. Assuming an informal list of phonological exponents for -ble in German as in (265), we can show that the unified analysis developed in the previous chapters supports a syntactic model like Distributed Morphology. 136 (265) a. [abel] ↔ a / Mod (with √[+ Latin])

b. [wert] ↔ a / Mod (with √[+ Psych])

135 See also Müller (2003) for a study of -bar adjectives with particle verbs in German. 136 As I say in the main text, this is an oversimplication of the facts, since as stated, we would never get diskutierbar or akzeptierbar, regular high -ble forms derived from Latin roots. As was mentioned in chapter 2, a full analysis of the German data is beyond the scope of this volumen. Here I just want to draw the reader’s attention to the general contrasting picture between German and Romance, the exact details of the morphological insertion of modal suffixes in German can thus be set aside.

Towards an analysis of -ble   





   157

c. [bar] ↔ a / Mod

In this model, which assumes Beard’s (1995) Separation Hypothesis and Late insertion, we distinguish between syntactic positions bearing only morphosyntactic features and Vocabulary items that are inserted post-syntactically if they bear a subset of the features of the target position or meet the syntactic contextual specification. This distinction makes it possible to assume the existence of a single underspecified Vocabulary item -ble in Spanish that emerges as optimal, i.e. most specific among the markers realizing a subset of the features in the syntactic morpheme – in the different syntactic contexts in which -ble appears, high and low -ble adjectives, V todo lo Vble or Nble. The late vocabulary-insertion approach makes a very simple account of the attested crosslinguistic variation. Where English, Catalan or Spanish have only one Vocabulary item -ble that appears in all contexts, German has three different pieces of vocabulary. Thus, German and Spanish or Catalan might have very similar structures for -ble formation, so that differences would reduce to the presence of theme vowels in Romance and questions of vocabulary insertion, with German having more specific markers for two of the contexts. In other words, crosslinguistic variation reduces to differences in the vocabulary of each particular language, and language-specific constraints, as one expects in the framework of the Distributed Morphology. It does not seem that such a homogeneous proposal can be derived in purely lexicalist theories of morphology for systematic reasons, i.e. lexicalist theories do not distinguish between phonological pieces and morphemes or syntactic terminal nodes.137

6.4.2 On multiple theme vowels I have just developed an analysis of -ble where the internal structure of both high -ble and low -ble contains a number of functional projections in all languages. For Romance, Oltra-Massuet’s (1999, 2000) proposal predicts that there will be a multiplicity of theme vowels, as exemplified in (266) for modificable ‘modifiable’ and traducible ‘translatable’, given that for each functional head, the wellformedness condition on functional heads will project a theme position in the morphology, i.e. at Morphological Structure. However, we find only one internal theme vowel, the one realizing the theme position adjoined to v – whether this is

137 Special thanks to Gereon Müller for helping me in the formulation of this very important point.

158   

   -BLE

phonologically realized or empty – and the external one realized as an epenthetic vowel /e/ in Spanish, schwa in Catalan, in both high and low -ble. The question is, then, how this can be made compatible with Oltra-Massuet’s well-formedness condition on functional heads. Note that even though one could assume some kind of postsyntactic phonological readjustment for -ble adjectives with underlying first conjugation verbs like modificable, such a possibility would not exist for those built on second and third conjugation verbs, where the ungrammatical *traduciable could not be applied any feasible repair strategy. (266) [[[[√ [ v Th]] [ Asp Th]] [ Mod Th]] [ a Th]] * mod ific a Ø a Ø a bl e * traduc Ø i Ø a Ø a bl e In the case of low -ble, where Volpe’s (2005) root attachment left no space for the realization of the Romance theme vowel, the structure proposed could easily accommodate the realization of the Romance theme vowel, which would be inserted in the morphology as a condition on AspP, 138 the functional head immediately following the root, and the theme position required by aP would be realized by the epenthetic vowel (see Harris 1991). However, also in this case I predict a multiplicity of theme vowels and the wrong outcome *temiable. (267) * Temiable [[[[√tem [Asp Ø iTh ]] [Mod Ø aTh ]] [a bl eTh]] As already pointed out in Oltra-Massuet and Arregi (2005), even though we find a multiplicity of theme vowels in the verbal domain, the phonological realization of theme positions in the nominal environment is more restricted, as the contrast in (268) illustrates. In nominal contexts, only the lowest or most internal theme vowel and the highest or external theme vowel seem phonologically possible, as in (268b); but then the most internal theme position is realized only if it is attached to a little v, but not to an internal little n, as in (268c). (268) a. [√ [ v Th ]] [[ T(/A/M) Th ] Agr ]] mod ific a b a s ‘you.sg modified’

(Sp)

138 This proposal is not unproblematic; apparently only v takes internal theme vowels in nonverbs (see e.g. Oltra-Massuet and Arregi 2005).



Towards an analysis of -ble   

   159



b. [√ [ v Th ]] [ n Th ] lav Ø a dor a ‘washing machine’ c. [√ [ n Th ]] [ a Th ] art ist Ø ic o ‘artistic’ To account for the multiplicity of theme positions in -ble adjectives, at this point I can only hint at two possible solutions. On the one hand, internal nominal theme positions are not phonologically realized in nominal environments due to some restriction. Only verbal heads can get their theme positions phonologically realized internally, and then if there is more than one verbal head in the nominal domain, only the lowest one can be phonologically filled. This would seem a very ad hoc solution. A perhaps less stipulative proposal would be to assume that, as in the verbal environment, aspectual and modal features (here perhaps also voice features, i.e. verbal features realized in little v) are merged together in a single functional node, where a single theme vowel is inserted in Catalan and Spanish, as illustrated in (269). Given that, leaving aside the verbalizing v head, verbal functional structure in the nominal domain is never phonologically realized, this would seem the research line to follow. (269) a. Modificable ‘modifiable’ (Cat/Sp) [[√mod [v-asp-mod ific aTh ]] [a bl eTh] ] b. Temible ‘fearsome’ [[√tem [asp-mod Ø iTh]] [a bl eTh] ] 6.4.3 Morphological evidence I have just suggested a syntactic structure containing a passive participle configuration with a modality node for -ble adjectives, so that an adjective like translatable is interpreted as ‘that can be translated’. Here I would like to provide some additional crosslinguistic morphological reflexes supporting this proposal. There are in Catalan a few idiosyncratic cases that show the connection between the verbal participle and -ble derivation. On the one hand, there are a few -ble adjectives containing the same morphological base that appears in the corresponding irregular participle, as shown in (270), which could be taken as support for the idea that -ble adjectives have participle-like content (cf. Picallo 1991). That this relation with participial morphology can be established with

160   

   -BLE

those adjectives that are supposed to be root-attaching low -ble adjectives could be taken as evidence for the presence of functional structure, specifically AspP, between the root and the categorizing head little a, even though the participial vocabulary item -t is not present in -ble adjectives in Catalan, whether these are derived from irregular roots (271a) or from regular ones (271b). (270) a. admetre – admès – admissible139 admit admitted admissible b. transmetre – transmès – transmissible transmit transmitted transmissible c. comprendre – comprès – comprensible understand understood understandable (271) a. vis-t, vis-i-ble root-part root-th-ble ‘seen, visible’ b. ball-a-t, ball-a-ble root-th-part root-th-ble ‘dance, danceable’ On the other hand, there are a couple of -ble adjectives, in (272a), where we find the theme vowel /u/ that is uniquely selected in the regular derivation of second conjugation participles, as exemplified in (272b).140 As shown in (272b), -ble adjectives derived from second conjugation verbs regularly take the third conjugation theme vowel /i/, though.

139 The vocalic change in (270a–b) is found with infinitives ending in -metre or -ndre and their derived forms in -iu, -ió, or -ble, suffixes that select the Latin participle stem. The process of n-loss in the participle comprès ‘understood’ in (270c) above or confós ‘confused’ in (ic) follows a general phonological rule for verbs ending in -ndre. (i) a. permetre, permès, permissiu, permissible ‘permit, permitted, permissive, permissible’ b. ometre, omès, omissió, omissible ‘omit, omitted, omission, omissible’ c. confondre, confós, confusió, confusible ‘confuse, confused, confusion, confusable’ 140 These are completely irregular cases, though, as the underlying verbs resoldre ‘solve’ and dissoldre ‘dissolve’ are athematic resolt ‘solved’ and dissolt ‘dissolved’. The point is that the appearance of /u/ in these irregular -ble derivatives points to the presence of some participial flavour.



Towards an analysis of -ble   

   161

(272) a. resol-u-ble, dissol-u-ble (Cat) solv-th-ble dissolv-th-ble ‘solvable’, ‘soluble’ b. sab-u-t, sab-i-ble – tem-u-t, tem-i-ble know-th-part, know-th-ble fear-th-part, fear-th-ble ‘known, knowable’, ‘feared, fearsome’ According to the NGRALE § 1.5f, Spanish -ble adjectives like revisable ‘revisable’, temible ‘fearsome’, traducible ‘translatable’ also select the participial stem, as witnessed by the fact that the theme vowel /e/ of the second conjugation underlying the verb temer ‘fear’ is neutralized to the third conjugation theme vowel /i/ in temible ‘fearsome’, even though verbs with irregular participles like romper – roto ‘break  – broken’, imprimir  – imprimido/impreso ‘print  – printed’ derive regular forms in -ble like rompible ‘breakable’ or imprimible ‘printable’. Note, however, that we cannot find the same parallelisms between the participle and the -ble form we see in (270) for Catalan. Also in Latin (see e.g. Gildersleeve and Lodge 1895, Leumann et al. 1963), the corresponding -bilis suffix adjoins to the passive perfect stem (supine). In Greek, the potential suffix -simos that parallels -ble contains the infix -m-, which is related to non-active morphology (Alexiadou 2001: 50), i.e. the one that appears in the medio-passive participle -menos (Manouilidou 2006) and in the middle voice -men/-mai (cf. Anagnastopoulou and Samioti 2009, 2010). For Nahuatl, Sullivan (1983) observes that the suffix -ni, which derives adjectives that would be analogous to -ble adjectives, is attached to a base containing a passive suffix, usually -lo/-hui, as illustrated in (273)–(274). (273) a. pahtia (Nah) cure b. pahti-lo-ni cure-pass-ni ‘curable’ (274) a. xeloa (Nah) divide b. xeli-hui-ni divide-pass-ni ‘divisible’

162   

   -BLE

I have already shown in section 4.2.1 that the corresponding adjectives in other languages contain a passive morpheme as well. For instance, in Turkish, the analogous -ble adjectives, Il(y)Abil(A)r or (I)n(y)Abil(A)r, contain the passive suffix (275b), as they do also in the corresponding nominalizations (275c). (275) a. yaz-mak (Tur) write- sub141 b. yaz-ɩl-abil-ir write-pass-mod-aor ‘writeable’ c. yaz-ɩl-abil-ir-lik write-pass-mod-aor-sufn ‘writeability’ Finally, Hungarian, which is a language that has no verbal passive, uses a participial ending -ó/-ő in the formation of the corresponding adjectives, repeated in (276). This can be easily accommodated in the proposal above, which is based on a resultative participial structure. (276) gyógy-ít-hat-ó(Hun) cure-tr-mod-part ‘curable’

7 Conclusions and remaining issues I have offered an analysis of the internal constituent structure of -ble adjectives that splits them into two main types that differ in their morphological, semantic, and syntactic properties. Support for this classification has been provided through an examination of the previous literature on -ble, but mainly through the application of well-known tests to English, German, Catalan and Spanish, which has been reinforced at times by considering parallel examples in other unrelated languages.

141 See Göksel and Kerslake (2005: 90–97) for the status of this and other subordinating or nominalizing suffixes, obligatorily present in non-finite forms.



Towards an analysis of -ble   

   163

The behavior of these two main types of -ble is associated to the presence or absence of different functional layers and their feature content, thus arguing for a syntactic approach to word formation. The claim that they are related to passive seems well-motivated, since potential -ble and passives are both eventive and, most importantly, both contain a passive v. That this verbal functional head is present in the structure of regular -ble is well-supported by the empirical evidence on argument structure, adverbial modification, raising or adjectival postponement. Also their similar behavior with respect to the interaction of passive with modality further reinforces this proposal. Indeed, I have shown that -ble does not suppress the external argument of the verb, and have provided empirical evidence showing that differences between passive and -ble, specifically restrictions on agent-related adverbial modification and the expression of verbal arguments, must be derived from general properties of the grammar of statives and the presence of modality. I have also discussed the modality of -ble as involving a type of circumstantial root modality, and have incorporated to some extent Hackl’s (1998) syntactic analysis of abilitatives. As for their relation to middles, I have only shown that -ble adjectives cannot possibly have been built on a middle verb as has been proposed in the literature, even though they share a number of syntactico-semantic properties. An account of -ble as parasitic on a middle verb makes thus a number of wrong empirical predictions, partly due to specific restrictions on middle formation, e.g. that middles require eventive verbs (Lekakou 2005). On the other hand, that both middles and -ble have an interpretation of a generic modal subject-oriented property derives from the fact that they both contain a circumstantial modal operator of the in view of certain inherent properties of the subject. The proposal advanced in this chapter has focused on examining the properties of the two main types of -ble adjectives, the requirements they impose on their possible bases, and the way their internal syntactico-semantic structure has a direct effect on their external structure. Several problems have not been dealt with and many questions arise from my analysis. For instance, I have not considered the interaction between the eventive structure of the base and the internal structure of -ble. With respect to the postulated ModP, further research is required to elucidate the exact nature of a modality head that appears in nonverbal contexts, as well as its position in the structure, and its formal relation to the genericity of -ble as individual-level adjectives. It also remains to be seen how the eventivity of the proposed structure can be made compatible with the characterization of -ble adjectives as individual-level predicates. Another open issue is to determine the best way to formalize the semantics of -ble, its LF, in relation to other structures containing the same or similar kind of modality. Equally necessary is an investigation into the pervasive and conspicuous potential-only modality meaning that we find in non-verbal contexts across languages.

164   

   -BLE

Despite these shortcomings, I will use this proposal as a working hypothesis for the analysis of two special cases: fully productive -ble forms with unaccusative and unergative verbs, and productive denominal -ble adjectives in Romance. In the next two chapters I will show that this proposal not only can accommodate these deviating though productive cases, but it correctly predicts their existence. The fact that these forms can only be found in some languages, e.g. Romance, but not others, e.g. English, will also find an answer.

3 Case Study 1: V todo lo Vble 1 Introduction The main goal of this chapter is to examine new empirical data from Spanish that can contribute to a better understanding of the internal structure and external requirements of -ble forms. These data are deviations from regular forms, i.e. those that select transitive verbal bases containing an internal argument in their argument structure (De Miguel 1986). Apart from some non-passive forms of the type durable ‘lasting’ or perdurable ‘everlasting’, which are derived from unaccusative verbs, the theme of a prototypical unaccusative verb like ocurrir ‘occur’ does not admit the derivation with -ble, as illustrated in (1b). However, it is possible to find the derived word ocurrible in certain stylistically marked constructions containing a quantifier and a neuter clitic lo,1 of the type V todo lo Vble ‘V all lo Vble’, as in (1c). (1) a. Ocurrió un accidente. (Sp) occur.3sg.pst an accident ‘An accident occurred.’ b. * un accidente ocurrible an accident occur.ble c. Ocurrió todo lo ocurrible occur.3sg.pst all lo occur.ble ‘It happened everything that could happen.’2 Similarly, even though an unergative verb like dormir ‘sleep’ does not admit -ble affixation, as shown in (2a–b), and it is deemed deviant with a cognate object

1 The status of this neuter clitic lo, whether article or pronoun, is a much debated – and still unsettled – issue in the literature on Spanish. For the analysis, I will assume with Bosque and Moreno (1990) that it is a pronoun. However, I will gloss it throughout as lo for the sake of simplicity. See also Luján (1972, 2002), Ojeda (1984), or Gutiérrez-Rexach (1999, 2009) and references therein for discussion on the particle lo. 2 English translations are provided to help the reader grasp the meaning of the construction. They are not intended to commit myself to any particular analysis. In fact, it seems that English speakers do not agree in the way these sentences should be translated. For (1c) above, some speakers prefer the translation ‘Everything happened that could happen’, others ‘There happened everything that could happen’. See the discussion below in sections 3.4 and 5 for the exact interpretation of this construction.

166   

   Case Study 1: V todo lo Vble

as the external argument of -ble (2c–d), the form dormible ‘sleep.ble’ can freely appear in constructions like (2e). (2) a. El bebe durmió. (Sp) the baby sleep.3sg.pst ‘The baby slept.’ b. ??/% un bebé dormible3 a baby sleep.ble c. dormir una siesta sleep a siesta ‘to take a siesta’ d. * una siesta dormible a siesta sleep.ble e. Luis quería dormir todo lo dormible. Luis want.3sg.pst sleep all lo sleep.ble ‘Luis wanted to sleep as much as one can sleep.’ In both cases, the semantic interpretation of the derived forms ocurrible – que puede ocurrir ‘that can happen’ and dormible – que se puede dormir ‘that can be slept’ is essentially the same as that of any deverbal -ble adjective, e.g. visible ‘that can be seen’. These data cannot be explained in a model based on the lexicon as a generative component. If words are created in the lexicon prior to insertion in the syntax, it is unclear why apparently non-existing adjectives can be created in the lexicon but cannot be used as such. Likewise, if they are not admitted as bases, they should not exist at all. Such a situation is anticipated in a syntactic theory of

3 As pointed out by Elena Benedicto (personal communication), this phrase is subject to idiolectal variation since it would be acceptable for some speakers. This is in fact expected, since the verb dormir can take an object in a causative structure (ia), which conforms to the requirements for -ble derivation. The informants I have consulted (and myself) reject the corresponding -ble form in (ib), though. This may be due to pragmatic reasons: it sounds strange of somebody to say that s/he has the property of ‘being able to be made sleep’. (i) a. He dormido al bebé. (Sp) have.1sg.prs sleep.part at.the baby ‘I have made the baby sleep.’ b. ??/% un bebé dormible a baby sleep.ble ‘a baby that can be made sleep’



Main claims and chapter overview   

   167

word formation, like DM, where the only component that can generate structures is syntax proper. These data identify a new paradox concerning the status of impossible words, such as tosible or ardible, which are impossible and at the same time they can be regularly and systematically derived, even if restricted to a very particular context. A basic question that the existence of these data raises has to do with the general architecture of grammar; specifically, how these words can be systematically generated in an appropriately constrained way. At first sight, it would seem that lexical theories of morphology undergenerate in these cases, because they predict those words that can only appear in the V todo lo Vble construction to be impossible, whereas syntactic theories of morphology may seem to overgenerate in the sense that if a word can be generated in the syntax, it is unclear why it should be constrained to appear in a particular configuration or what the restricting context should be.4

2 Main claims and chapter overview The main claim I put forward in this chapter is that syntax and semantics conspire with morphology to the effect of building such a priori aberrant forms. Such a situation is anticipated in a syntactic theory of word formation, like for instance Distributed Morphology (Halle and Marantz 1993, 1994 and related work), where the only component that can generate structure is syntax proper, which mediates between semantics and morphology. I argue that the existence of these -ble forms cannot be due to the internal or endoskeletal properties of the base, but their grammaticality must be a function of their external or exoskeletal properties in interaction with general features of quantification (Sánchez López 1999), the grammar of cognate elements (Macfarland 1995; Pereltsvaig 1999), the grammar of degrees (Kennedy 1999, 2004; Kennedy and McNally 1999, 2005), and the general grammar of achievements, unergative and unaccusative structures (Dowty 1979; de Miguel 1999), together with some general properties of the grammar of Spanish – specifically, the fact that it has a special overt morpheme lo, a variable that can range over different domains (Bosque and Moreno 1990). More specifically, I suggest an analysis of these structures in which the sequence todo lo Vble measures out the extension or duration of the event,

4 Special thanks to P. Cabredo-Hofherr for formulating these questions. See Oltra-Massuet (to appear).

168   

   Case Study 1: V todo lo Vble

either as an internal object or as a degree argument. The analysis to be defended here fully endorses Bosque and Moreno’s (1990) proposal that the denotation of the pronoun lo is contextually determined. When todo lo Vble corresponds to the object of the main verb, a referential lo is the head of a DP that provides the root with the argument required by -ble. In its second interpretation, lo materializes a degree head, which together with its complement specifies the scale along which an atelic predicate measures out the progress of the event. This degree head acts as the argument of -ble with atelic predicates. To support the analysis of todo lo Vble as a degree argument, I examine a cognate object structure with unaccusative verbs in English of the type in (3a), where the function of the cognate object parallels the one of todo lo Vble in the Spanish construction, as evidenced by (3b). (3) a.  The tree grew a century’s growth within only ten years. (Kuno and Takami 2004: 116 b. El árbol creció todo lo crecible en tan solo diez años. (Sp) the tree grow.3sg.pst all lo grow.ble in so only ten years ‘The tree grew to its highest possible degree of growth in only ten years.’ The main claim I put forward is that the verb’s event structure together with the specific type of cognate object are crucial to explain these striking data in both languages. Specifically, I argue that a century’s growth – and likewise todo lo crecible – in the example above is the instantiation of the degree argument of a gradable predicate which measures out the event, in which the cognate element essentially corresponds to the overt expression of the otherwise implicit scale with these verbs (Piñón 2000; Kratzer 2002; Rappaport Hovav 2008). Hence, I argue against previous analyses of the English construction as an example of a standard cognate object structure, be it argumental or adverbial, and suggest a joint treatment of a century’s growth and todo lo crecible. Ultimately, it is argued that the sequence todo lo Vble – as well as the analogous English cognate constituent – corresponds to the overt manifestation of the basic dimension along which the predicate is measured out, either in the form of a scale, i.e. a set of ordered degrees, or in the form of a range, i.e. a set of entities. In all cases, the cognate constituent affects the verb’s aspectual structure in that it delimits the predicate, and turns atelic predicates as well as predicates of variable telicity into telic ones. When the predicate is already telic, as with prototypical unaccusative verbs, like ocurrir ‘occur’, the function of todo lo Vble is to supply the achievement predicates with the stages that provide them with the durative component required to become an accomplishment.



Main claims and chapter overview   

   169

As for the second requirement on -ble adjectives, that of implying some originator, i.e. agent, cause, or natural force, in order to account for -ble forms derived from non-agentive unaccusatives, I adopt suggestions in Dowty (1979), Tenny and Pustejovsky (2000) or Bosque and Gutiérrez-Rexach (2009) that associate duration with agency. This will in turn hint at an explanation for the existence of adjectives like durable ’lasting‘ or perdurable ’everlasting‘, non-prototypical unaccusatives with a measure argument expressing duration. The chapter is organized as follows. I first describe in section 3 the main properties of this special construction, focusing on its effects on the predicate’s event structure, as well as on their interpretation. The next two sections concentrate on a more detailed investigation of -ble forms derived from unergatives and unaccusatives, respectively. In section 4, I also present a preliminary analysis for unergative verbs. Section 5 discusses the issue of cognate objects with unaccusative verbs, and draws a parallel between unaccusative verbs in English and Spanish. I show that an analysis of the cognate elements under study in terms of argumental versus adverbial cognate objects is not empirically adequate. In 5.3 I outline an analysis of the structure V todo lo Vble for unaccusatives, which can also be extended to unergatives, which also accounts for the parallelism between English and Spanish unaccusative cognate objects. Section 6 discusses the details on the derivation of -ble forms with unergatives and unaccusatives, which further suggests an explanation for the existence of a very small number of -ble adjectives derived from a very specific set of atelic unaccusative verbs in Spanish. Section 7 summarizes the main points and concludes with the main contributions of this chapter. There is an important caveat that I need to point out before going into the description of the main aspects of this construction with the different types of predicates. The present chapter is not intended to provide a full and detailed analysis of the sequence V todo lo Vble, for such a venture would take me too far afield. As will become clear throughout the chapter, the intricacies of this construction are numerous. Apart from our main concern here, namely the formation of independently non-existing -ble words and their internal structure, the analysis of this construction must face a number of complexities, some of them dealing with much debated – and still unsettled – issues in the linguistics literature. A first problem refers to the category and status of lo.5 Related to that is the question of what todo is quantifying over. A third question has to do with the interpretation of V todo lo Vble, which is ambiguous between the expression of quantification over objects (4) and the expression of degree quantification (5).

5 See footnote 1 above.

170   

   Case Study 1: V todo lo Vble

For the latter, as shown in (5), there are a priori three possible paraphrases available for V todo lo Vble that would correspond to three different quantificational structures, with a quantifying relative in (5a),6 a comparative structure in (5b) and a superlative in (5c).7 (4) V (todo) aquello que es V all that which be.3sg.prs

posible V (Sp) possible V

(5) a. V (todo) cuanto es V all how.much be.3sg.prs

posible V (Sp) possible V

b. V tanto como es V as.much as be.3sg.prs

posible V possible V

c. V lo máximo que es V lo maximum that be.3sg.prs

posible V possible V

These may condition and even determine the analysis of the string V todo lo Vble.8 Notice in addition that the interpretation of this construction brings us in turn to the analysis of degree quantification and scalar structure with verbs. And this further takes us to the function of todo lo Vble and the interaction between degree phrases and the determination of lexical aspect or Aktionsart with some of the predicates. Also involved in the analysis of this structure are issues pertaining to the nature of cognate objects, their status as incremental themes and therefore, their function in the determination of the telicity of the predicate. Related to this is the question of the existence of cognate objects – argumental or adverbial – with unaccusative verbs and Case assignment. Still another topic concerns the pleonastic character of the cognate construction, its source and its relation to the interpretation of the string. To make things even more complicated, it should be noted that speakers’ judgments are not uniform, so that there

6 As its name indicates, cuanto is an inherently quantificational pronoun. See Brucart (1999: 506) and Sánchez López (1999: 1053) for the properties of this pronoun. 7 Or just two, if the superlative is to be analyzed as a kind of comparative, as pointed out in Bosque and Gutiérrez-Rexach (2009: 536). 8 Due to the peculiarity of this construction as well as speaker’s idiolectal variation, it is very hard, though hopefully not impossible, to find the relevant tests that would permit an evaluation of the corresponding syntactico-semantic structures in (5), their predictions, and their formal relation to the intensive structure under study. For the analysis, I will make minimal assumptions in this respect, without committing myself to the superiority of one over the other two, and will leave a more thorough analysis for future research.



V todo lo Vble. General Properties   

   171

exists a fair amount of idiolectal variation with respect to both acceptability and interpretation.9 In this chapter I will touch on all of the aforementioned topics, insofar as this is necessary to develop a preliminary investigation of this construction that provides further support for the syntactic analysis of -ble forms suggested in the previous chapter. However, I will not elaborate on each of these – and possibly still other – properties of the construction. This means that some questions will remain open or that I will just hint at possible answers. This should in no way undermine the main empirical and theoretical contributions of this chapter; far from that, since it raises interesting questions and brings new input to longstanding debates.

3 V todo lo Vble. General Properties In the previous chapter, I have proposed that -ble imposes two basic conditions: the presence of an internal argument and the implication of an originator. In principle, both unaccusative and unergative verbs lack the relevant properties to meet these requirements, so that they should not be possible bases for -ble. Leaving aside those intransitive verbs that have a transitive counterpart – so-called ergative verbs, i.e. those entering the causative-inchoative alternation like congelar ‘freeze’ or romper ‘break’ –, unaccusative verbs, like ocurrir ‘happen’ or crecer ‘grow’, have an internal argument, but lack any implication of an originator. Conversely, unergative verbs like dormir ‘sleep’ or estornudar ‘sneeze’ do select an originator, but lack an internal argument. It is thus unexpected that these types of verbs can give rise to a -ble form.10

9 Data have been collected from electronical sources or they have been heard in real-life situations. They also come from introspection. All data have been tested against native speakers’ judgments. 10 Leaving aside cases like bailable ‘danceable’, the status of these -ble forms in terms of canonical categories may appear as unclear. These -ble forms can never appear outside of this construction, in attributive or predicative position, and as illustrated in (i)–(ii), adverbial modification does not seem possible with these -ble forms, even though it is fine with other clearly adjectival -ble forms in similar contexts, e.g. Hizo todo lo buenamente possible para ayudarla ‘S/he did everything that s/he possibly could to help her’, and also with transitive verbs in exactly the same V todo lo Vble context, e.g. Aprovecho para desearte todo lo buenamente deseable ‘I take the opportunity to wish you the best one can whish’.

172   

   Case Study 1: V todo lo Vble

However, the pleonastic construction V todo lo Vble, which is used to express intensification11 to a maximum extent or maximum quantity in colloquial Spanish, given the right context, can apparently be formed with any verb, as illustrated in (6)–(7) with unergative and unaccusative verbs, respectively, i.e. it is fully productive in that it can target any verb – though see the discussion in section 5 below. This implies that, since the unaccusative and unergative bases do not meet the requirements for -ble derivation, the conditions to meet them must come from the outside, from the contextual structure, i.e. from syntax proper. (6) a. Corrió todo lo corrible / Ladró todo lo ladrable / run.3sg.pst all lo run.ble / bark.3sg.pst all lo bark.ble / Tosió todo lo tosible. (Sp) cough.3sg.pst all lo cough.ble b. * corrible / *ladrable / *tosible run.ble  bark.ble  cough.ble (7) a.  Apareció todo lo aparecible / Brotó todo lo brotable / appear.3sg.pst all lo appear.ble / sprout.3sg.pst all lo sprout.ble / Llegó todo lo llegable. (Sp) arrive.3sg.pst all lo arrive.ble

(i) a. Ocurrió todo lo (*rápidamente / *imprevisiblemente) ocurrible. (Sp) occur.3sg.pst all lo  rapidly / unpredictably occur.ble ‘It happened everything that could happen in a fast and unpredictable way.’ b. Todo ocurrió de manera rápida e imprevisible. everything occur.3sg.pst of manner quick and unpredictable ‘Everything happened in a fast and unpredictable way.’ (ii) a. Juan durmió todo lo (*tranquilamente/ *plácidamente) dormible. (Sp) Juan sleep.3sg.pst all lo  quietly /  peacefully sleep.ble ‘Juan slept {quietly / peacefully} as much as possible.’ b. Juan durmió {tranquilamente / plácidamente}. Juan sleep.3sg.pst quietly peacefully ‘Juan slept {quietly / peacefully}.’ These forms with unergative and unaccusative verbs can hardly be manipulated. For present purposes, I will assume that they are adjectival and that the impossibility of being manipulated obeys some more general constraints still to be established. 11 Cf. Alcina and Blecua (1988: 643). This grammar mentions that the quantifier todo ‘all’ “takes an intensive value when used with abstract nouns expressing a quality, a state and psychic acts” [author’s translation], as in the groupings todo lo bastante ‘all that is enough’ or todo lo bueno ‘all that is good’.



V todo lo Vble. General Properties   

   173

b. * aparecible / *brotable / *llegable appear.ble  sprout.ble  arrive.ble A piece of evidence that it must be syntax which generates these forms comes from the fact that with transitive verbs that have root-level special forms, like admisible ‘admissible’, perceptible ‘perceptible’, risible ‘risible’, transmisible ‘transmissible’, derived from admitir ‘admit’, percibir ‘perceive’, reír ‘laugh’ and transmitir ‘transmit’, we obtain the morphologically transparent forms admitible, percibible, reíble, transmitible in this construction, whereas special (root-level) forms are clearly dispreferred. As shown in (8)–(9), exactly the opposite is what we find in all other contexts. That is, morphologically transparent forms with these verbs do not exist outside from these (and similar) constructions (but see Fábregas 2011). Interestingly, the selection of high -ble forms was fully confirmed by a small-scale experiment that I carried out among native speakers of Spanish, discussed in 3.1.1 below. (8) a. percibió todo lo ??perceptible / percibible (Sp) perceive.3sg.pst all lo   perceptible perceive.ble b. una diferencia apenas perceptible / *percibible a difference hardly perceptible  perceive.ble (9) a. rió todo lo ??risible / reíble (Sp) laugh.3sg.pst all lo  risible laugh.ble b. una propuesta risible / *reíble a proposal risible  laugh.ble

3.1 Pleonastic structure Turning to the main properties of this construction, the pleonastic nature of V todo lo Vble is illustrated in (10a). It does not seem possible to have two semantically related though different verbs, not even when they are true synonyms, as in (10b–c). (10) a. * descansó todo lo dormible (Sp) rest.3sg.pst all lo sleep.ble b. * sucedió todo lo ocurrible happen.3sg.pst all lo occur.ble

174   

   Case Study 1: V todo lo Vble

c. * (se) quemó todo lo ardible (se) burn.3sg.pst all lo flame.ble The examples in (11) show that the pleonastic structure is crucial for the grammaticality of these -ble forms, which do not exist outside of this type of structure.12 (11) a. * todo lo ladrable (Sp) all lo bark.ble b. * todo lo brotable all lo sprout.ble

3.1.1 Pleonastic structure with allomorphic verbal roots Now, the examples above in (8)–(9) on the selection of high -ble forms instead of the expected low -ble in V todo lo Vble, together with the pleonastic nature of the construction bring up an important question on root selection. Specifically, in cases of stem alternation of the verbal form, one has to determine which stem represents the root, how it can be established empirically, and how it is selected for insertion into the structure. More specifically, there are verbs in Spanish that show stem allomorphy, exemplified in (12). (12) a. tener – tuve (Sp) have.inf have.1sg.pst ‘to have – I had’ b. andar – anduve walk.inf walk.1sg.pst ‘to walk – I walked’ In the structure V todo lo Vble with these past forms, given the above-mentioned pleonastic requirement, in principle we could have either tuve todo lo tenible ‘I had everything one can have’ or tuve todo lo tuvible ‘I had everything one can have’, i.e. we must elucidate whether identity of underlying verb structure corresponds to root identity or Vocabulary item identity. The choice of one form or

12 See section 4.2 below for some exceptions to this generalization, e.g. me gusta todo lo bailable ‘I like everything that can be danced’.



V todo lo Vble. General Properties   

   175

the other is tied to factors related to the nature of root allomorphy, the process of root copying/realization in a cognate configuration – the structure I propose below for V todo lo Vble –, and Vocabulary insertion. Simplifying things a lot, we could say that the question reduces to whether we must take into account the specific phonological realization of the higher verb root V1 to determine the phonological realization of the lower adjectival root V2 in the configuration V1 todo lo V2ble. I will not specifically address this question in the analysis, as this would involve taking the status of root allomorphy into acoount, a topic in its own right. However, as I will argue below, what we have in this construction is the phonological realization of the lower copy root in a cognate structure. Note that in a Distributed Morphology model the process of Vocabulary insertion proceeds under strict locality conditions, and given that the two roots are not in a local relation, we expect that insertion of the lower copy is blind to the realization of the higher copy. Thus, since a root tuv- will be specified for insertion in a past context, it will not be able to be inserted in the context of a non-past little v c-commanded by a high possibility modal. Hence, pending a more detailed analysis of this issue, the prediction is that there should not be cases of anduvible or tuvible. As far as I can tell, the prediction is borne out. To be sure, I conducted a quick small-scale experiment among Spanish native speakers to test this prediction. I gave each speaker a first oral example with the regular transitive verb lavar ‘wash’, Juan lavó todo lo lavable and ask them to react to a series of incomplete sentences with a missing -ble form of the type Juan V-past todo lo ___ and fill the gap. Among these sentences there were examples with (a) verbs with stem allomorphy, such as anduvo ‘s/he walked’, estuvo ‘s/he was.stative’, tuvo ‘s/he had’; (b) other completely regular verbs like bailó ‘s/he danced’, confirmó ‘s/he confirmed’, robó ‘s/ he stole’; and (c) other cases with low -ble adjectives, such as admitió ‘s/he admitted’, leyó ‘s/he read’, percibió ‘s/he perceived’, rió ‘s/he laughed’. The test was carried out individually and orally in order to elicit immediate responses rather than thoughtful ones. In all cases and for all speakers the answers were systematically those expected, listed in (13). (13) a. andable, estable, tenible NOT *anduvible, estuvible, tuvible (Sp) walk.ble be.stative.ble have.ble b. bailable, confirmable, robable danceable confirmable stealable c.  admitible, leíble, percibible, reíble NOT *admisible, *legible, admit.ble read.ble perceive.be laugh.ble *perceptible, *risible

176   

   Case Study 1: V todo lo Vble

Despite the relevance of these findings for the theoretical discussion on the phonological realization of root copies in syntactic structure, I will leave the discussion of this and the other related issues mentioned above for further research.

3.2 Quantification Another crucial property of this construction is the absolutely necessary presence of a quantifier. Without a quantifier, the construction becomes ungrammatical, as exemplified in (14). (14) a. * ladró lo ladrable (Sp) bark.3sg.pst lo bark.ble b. * brotó lo brotable sprout.3sg.pst lo sprout.ble Again, verbs like bailar ‘dance’ are exceptions to this statement, since we find things like música bailable ‘danceable music’ or bailó lo bailable ‘s/he danced what could be danced’. The same -ble forms are possible with other quantifiers, as illustrated in (15). I owe these examples to Eulàlia Bonet (personal communication). (15) a. ¿Ha ocurrido algo de lo ocurrible? (Sp)  have.3sg.prs occur.part something of lo occur.ble ‘Did it happen something from what could happen?’ b. No ha ocurrido nada de lo ocurrible. not have.3sg.prs occur.part nothing of lo occur.ble ‘It did not happen anything from what could happen.’ Here I will restrict myself to discussing the construction V todo lo Vble, although the analysis should in principle be applicable to the forms in (15), with some variation due to the partitive structure the quantifiers in (15) incorporate.

3.3 Aspectual shift An additional characteristic of this construction is that it introduces an aspectual shift in the event structure of the predicate, with both atelic and telic predicates. Among intransitive atelic events, we can include activities, like dormir ‘sleep’



V todo lo Vble. General Properties   

   177

or semelfactives (Smith 1991) like estornudar ‘sneeze’.13, 14 These non-delimited predicates do not accept a frame adverbial, which focuses on the time span upon completion of the event denoted by the predicate, as shown in (16) in their intransitive use. However, such a time adverbial is fine when appearing in the construction under study in (17), i.e. they can be understood as bounded predicates in the presence of todo lo Vble.15 (16) a. Juan durmió *en una hora / durante una hora. (Sp) Juan sleep.3sg.pst  in an hour / for an hour ‘Juan slept in an hour / for an hour.’ b. Juan estornudó *en cinco minutos / durante cinco minutos.16 Juan sneeze.3sg.pst  in five minutes / for five minutes ‘Juan sneezed in five minutes / for five minutes.’ (17) a. Juan durmió todo lo dormible en una tarde / ?durante una Juan sleep.3sg.pst all lo sleep.ble in an evening /  for an tarde. (Sp) evening ‘Juan slept as much as one can sleep in an evening / for an evening.’ b.  Juan estornudó todo lo estornudable en cinco minutos / Juan sneeze.3sg.pst all lo sneeze.ble in five minutes / ?durante cinco minutos.  for five minutes ‘Juan sneezed as much as one can sneeze in five minutes / for five minutes.’

13 We could also include verbs like crecer ‘grow’ in this group, i.e. so-called degree achievements, which show variable telicity. I will deal extensively with this type of verbs in section 5. 14 Verbs like estornudar ‘sneeze’ or bostezar ‘yawn’ are punctual or instantaneous events, and therefore often considered telic. Here, I classify them as semelfactives, i.e. “atelic instantaneous events” (Smith 1991: 55) for they behave as atelic when they take todo lo Vble. Due to their special status, not all aspectual diagnoses will be applicable to them, though. 15 Although native speakers show certain degree of doubt with respect to the for-adverbial, the contrast with in-adverbials is quite sharp. That is, the crucial point is that in-frame adverbials become fine with activities when these take the cognate -ble complement. 16 The interpretation of this sentence is a repetitive one, i.e. it presents “a derived multipleevent Activity, consisting of a series of repeated Semelfactive events” (Smith 1991: 56).

178   

   Case Study 1: V todo lo Vble

Activities also have the subevent property with for-phrases (Dowty 1979). That is, Juan durmió durante una hora ‘Juan slept for an hour’ entails that at any time during that time span it was true that Juan was sleeping. There is no such entailment with accomplishments, exactly as there is no such entailment if Juan durmió todo lo dormible durante una tarde ‘Juan slept as much as it is possible to sleep in an evening’. That is, if this sentence is true, it is not the case that at any time during that evening Juan slept as much as he could sleep, because todo lo dormible can only refer to the duration of the event as a whole. This is because todo lo dormible introduces a delimiter, which is incompatible with the subevent property. An additional trait of activities is that whenever one stops doing them, it is entailed that the event has already occurred, whereas accomplishments, being delimited, must reach their final endpoint in order to occur. So, if Juan was sleeping and he interrupts this activity, as in Juan dejó de dormir ‘Juan stopped sleeping’, it follows that Juan has slept. But if Juan was sleeping todo lo dormible ‘as much as it is possible to sleep’ and stops doing it, as in Juan dejó de dormir todo lo dormible ‘Juan stopped sleeping as much as it is possible to sleep’, we cannot conclude that he has slept todo lo dormible. If we now turn to telic unaccusative predicates, like achievement verbs ocurrir ‘happen’ or llegar ‘arrive’, these can be shown to become durative in the construction under scrutiny when we apply the relevant diagnostic tests.17 Although instantaneous or punctual achievements are strange with durative time adverbials, they are not so in the presence of todo lo Vble, as exemplified in (18) with ocurrir.

17 It is not so clear that the entailments of achievement verbs change when in the construction V todo lo Vble. If (i) is true, then it can be true that all the things that could happen were happening during that hour, i.e. a plurality of events were taking place. However, if (iia) is true, one cannot infer (iib). (i) Ocurrió todo lo ocurrible en una hora. (Sp) occur.3sg.pst all lo occur.ble in an hour ‘It happened everything that could happen in an hour.’ (ii) a. Los paquetes llegaron en un minuto. (Sp) the packages arrive.3pl.pst in a minute ‘The packages arrived in a minute.’ b. Los paquetes estuvieron llegando durante un minuto. the packages be.stative.3pl.pst arrive.ger during a minute ‘The packages were arriving for a minute.’



V todo lo Vble. General Properties   

   179

(18) a. El accidente ocurrió en un instante/*durante cinco minutos.(Sp) the accident occur.3sg.pst in a moment / for five minutes ‘The accident occurred in a moment / *for five minutes.’ b. Ocurrió todo lo ocurrible en un instante / durante cinco occur.3sg.pst all lo occur.ble in a moment / for five minutos.18 minutes  ‘It happened everything that could happen in a moment / for five minutes.’ According to de Miguel (1999:  3036), punctual events do not accept adverbial modifiers that delimit the event, as in (19a). However, they license such adverbials in the presence of todo lo Vble, as in (19b), thus behaving as a durative delimited event, i.e. as an accomplishment. (19) a. * Llegaron las cartas hasta que se puso arrive.3pl.pst the letters until that se put.3sg.pst (= started) a llover.19 (Sp) to rain b. Llegó todo lo llegable hasta que se puso arrive.3sg.pst all lo arrive.ble until that se put.3sg.pst a llover. to rain ‘It arrived everything that could arrive until it started to rain.’

18 Notice that the presence of todo lo Vble does not preclude the punctual interpretation, as further illustrated with the examples in (i)–(ii). Punctual adverbial modifiers, whose appearance with durative verbs is highly restricted, are fine in the presence of todo lo Vble. (Examples adapted from de Miguel 1999: 3037). (i) Ocurrió todo lo ocurrible a las cinco en punto. (Sp) occur.3sg.pst all lo occur.ble at the five in point ‘It happened everything that could happen at five o’clock.’ (ii) En el preciso instante en que ocurrió todo lo ocurrible, se puso in the precise instant in which occur.3sg.pst all lo occur.ble, se put.3sg.pst a llover. (Sp) to rain ‘At the precise moment when it happened everything that could happen, it started to rain.’ 19 The examples are adapted from de Miguel (1999: 3036).

180   

   Case Study 1: V todo lo Vble

In both cases, the appearance of the string todo lo Vble with these verbs has the same effect as when we have a plural or mass noun object, i.e. NPs with a cumulative denotation, so that repetition confers the predicate with duration. That is, the difference in aspectuality is directly tied to the plural reference of the quantified element in subject position, which triggers an iterative and durative interpretation of the event (de Miguel 1999: 3033). (20) a.  Los accidentes ocurrieron en un momento / durante cinco the accidents occur.3sg.pst in a moment / for five minutos.(Sp) minutes ‘The accidents occurred in a moment / during five minutes.’ b. Llegó gente hasta que se puso a llover. arrive.3sg.pst people until that se put.3sg.pst to rain ‘People arrived until it started to rain.’ Hence, whereas activities and semelfactives become delimited in the presence of todo lo Vble, achievements become durative when they appear in the construction V todo lo Vble, i.e. there is an aspectual reclassification. Smith (1991:  28) defines accomplishments as a “process of successive stages and a natural final point” (see also Vendler 1967; Dowty 1979; Pustejovsky 1995; de Miguel 1999; and much related work on aspectual classes). So, we can say that the presence of todo lo Vble causes an aspectual type shift: it turns activities, semelfactives and achievements into accomplishments, i.e. it provides successive stages to those that have an endpoint, and a final endpoint to those that are already durative, i.e. that contain stages, in their event structure. That this is so can be further illustrated with the almost test (Dowty 1979: 58). Accomplishments, which have been said to involve two subevents, a process and a final endpoint, are ambiguous with the adverb casi ‘almost’, because it can refer to either subevent, so that either the event did never take place (the adverb has scope over the initial subevent) or the event was interrupted before reaching the final state (the adverb scopes over the final subevent). Thus, whereas sentences like Juan casi durmió ‘Juan almost slept’, Juan casi estornudó ‘Juan almost sneezed’, Casi ocurrió un accidente ‘It almost occurred an accident’, and Juan casi llegó ‘Juan almost arrived’ are all unambiguous and express that the event did not take place, the same verbs become ambiguous when followed by todo lo Vble.



V todo lo Vble. General Properties   

   181

(21) a. Juan casi durmió todo lo dormible. (Sp) Juan almost sleep.3sg.pst all lo sleep.ble 1. Juan almost slept, but he didn’t. 2. Juan slept almost as much as possible. b. Juan casi estornudó todo lo estornudable. Juan almost sneeze.3sg.pst all lo sneeze.ble 1. Juan almost sneezed, but he didn’t. 2. Juan sneezed almost as much as possible. (22) a. Casi ocurrió todo lo ocurrible. (Sp) almost occur.3sg.pst all lo occur.ble 1. Everything that could happen almost happened, but it didn’t. 2. It happened almost everything that could happen. b. Casi llegó todo lo llegable. almost arrive.3sg.pst all lo arrive.ble 1. Everything that could arrive almost arrived, but it didn’t. 2. It arrived almost everything that could arrive. A crucial task of this chapter is to develop a preliminary analysis that accounts for this aspectual shift, i.e. the source of this type-shifting, since it is crucially related to the interpretation of this construction as well as to the derivation of the elsewhere illicit -ble forms.

3.4 The interpretation of V todo lo Vble The interpretation of this construction varies depending on the verb, as illustrated in (23)–(24).20 There are two possible readings: todo lo Vble may denote the set of all objects that can be Ved / can V in reading 1; or else, it expresses the

20 One may think that it is likely that the intensifying interpretation in (23b).2 is due to a semantic implication rather than to a different reading. Part of the discussion below will be devoted to showing that the two readings are not the result of an implication, but must be derived from two different syntactic configurations. Furthermore, it will be shown that each interpretation corresponds to a different type of lo (Bosque and Moreno 1990. See also Bartra and Villalba 2006; Villalba and Bartra 2009).

182   

   Case Study 1: V todo lo Vble

maximum possible extent, measure, or degree21 to which the event named by the verb takes place in the second interpretation. The former has a referential denotation, whereas the latter does not. As shown in the examples below, both unergatives and unaccusatives may or may not be ambiguous. (23) a. Durmió todo lo dormible. (Sp) sleep.3sg.pst all lo sleep.ble 1. * S/he slept everything that could be slept: siesta, drunkenness, etc. 2. S/he slept as much as possible / up to the highest possible extent.22 b. Bailó todo lo bailable. dance.3sg.pst all lo dance.ble 1. S/he danced everything that could be danced: salsa, rock, etc. 2. S/he danced as much as possible / up to the highest possible extent. (24) a. Ocurrió todo lo ocurrible. (Sp) occur.3sg.pst all lo occur.ble 1. It happened everything that could happen: accident, disaster, etc. 2. * It happened as much as possible / up to the highest possible extent b. Floreció todo lo florecible. bloom.3sg.pst all lo bloom.ble 1. It bloomed everything that could bloom: plants, trees, the garden, etc. 2. It bloomed as much as possible / up to the highest possible extent As evidence for the existence of two independent readings, consider the examples in (25)–(28). In (25), we have coordination of todo lo llorable with the adverbial quantifier poco. If we want to question the string todo lo llorable in this sentence, we can only use the adverbial interrogative quantifier cuánto ‘how much’, as shown in (26a); the use of the objective interrogative qué ‘what’ is completely unacceptable, as illustrated in (26b).

21 Throughout, I use degree, extent, or measure as general cover terms that are to be understood as referring to quantity, intensity, degree, measure, or extent. I am not claiming that they are the same, only that possible differences among them will be irrelevant for my proposal. 22 I have used here two possible English translations, one using the comparative of equality as much as, and the other using a superlative, to avoid misunderstandings. Both are possible, and none should be understood as committing myself to any particular analysis, as I mentioned in footnote 8.



V todo lo Vble. General Properties   

   183

(25)  Lloró todo lo llorable por la muerte de su padre, pero poco cry.3sg.pst all lo cry.ble for the death of her father but little por la de su madre. (Sp) for the of her mother ‘She cried as much as one can cry for her father’s death, but very little for her mother’s.’ (26) a. ¿Cuánto lloró por la muerte de su abuelo? – Todo lo  how.much cry.3sg.pst for the death of her father – all lo  llorable. (Sp)  cry.ble ‘How much did she cry for her father’s death? – As much as one can cry.’ b. *¿Qué lloró por la muerte de su padre? – *Todo lo llorable.  what cry.3sg.pst for the death of her father –  all lo cry.ble In (27)–(28) we find exactly the opposite behavior of the string todo lo llorable. In (27a), notice that todo lo llorable in the list reading need not correlate with the highest/longest possible extent reading found in the degree interpretation. In fact, this second example could not be coordinated with an adverbial quantifier, but it can be coordinated with an object, as in (27b). As a consequence, we cannot use in this case the interrogative cuánto ‘how much’ in (28a), but qué ‘what’ in (28b) is fine. (27) a.  En ese instante lloró todo lo llorable: la muerte de su in that instant cry.3sg.pst all lo cry.ble the death of her padre, la ausencia de su marido, la pérdida de su trabajo, father the absence of her husband the loss of her job y otras desdichas. Pero fue eso, sólo un instante. (Sp) and other misfortunes but be.3sg.pst that only an instant ‘At that moment she cried for everything one can cry: her father’s death, her husband’s absence, the loss of her job, and other misfortunes. But it was that, just a moment.’ b.  En ese instante lloró todo lo llorable, menos la ausencia de in that instant cry.3sg.pst all lo cry.ble less the absence of su marido. her husband ‘At that moment she cried for everything one can cry, except for her husband’s absence.’

184   

   Case Study 1: V todo lo Vble

(28) a. * ¿Cuánto lloró en ese instante? – Todo lo llorable. (Sp)  how.much cry.3sg.pst in that moment – all lo cry.ble b. ¿Qué lloró en ese instante? – Todo lo llorable.  what cry.3sg.pst in that moment – all lo cry.ble  ‘What did she cry for at that moment? – Everything one can cry for.’ Thus, an additional task is to determine whether such variation is due to the grammar or it is just an arbitrary fact. Below I show that, as with degree quantifiers like mucho ‘a lot’, variation depends on the semantic properties of the predicate (cf. Bosque and Masullo 1998:  18; Sánchez López 1999:  1094; Brucart and Rigau 2002: 1579; Solà i Pujols 2002: 2918), specifically, on its event structure. I propose that the first interpretation, the referential reading of todo lo Vble, arises from a quantifier phrase configuration where todo quantifies over a predicative structure – a small clause – whose subject DP is headed by lo, which corresponds to the external argument of a -ble form that is the predicate of the small clause (represented as FP, which stands for functional projection, possibly a PredP), which is in the direct object position of the matrix clause (not shown in (29)).

QP

(29)

Q' Q todo

SC DP

FP

lo

…ble…

In the second reading, the degree reading, todo lo Vble is a quantized degree phrase, where lo materializes a degree head that takes a -ble adjective as its complement, thus denoting a scale, i.e. the entire set of ordered possible degrees. In this case, todo closes the scale by assigning the maximal value to the event’s extension, resulting in the entire interval or extent, i.e. the entire set of degrees. This degree interpretation is licensed only with atelic predicates (cf. Rigau 2002: 2090).23

23 The interpretations of mucho and todo lo Vble, though very similar, do not fully coincide. According to Sánchez López (1999: 1094), a sentence with a gradable predicate like Juan

V todo lo Vble. General Properties   



(30)

   185

DegP Deg'

QP todo

Deg

aP

lo

…ble…

Interestingly, when considering both the aspectual shift and the different interpretations available for this structure, we will be able to account for additional contrasts in the realm of transitive verbs of the type exemplified in (31)–(34). (31) a. * Juan alcanzó la cima todo lo alcanzable. (Sp) Juan reach.3sg.pst the top all lo reach.ble b. * Juan la alcanzó todo lo alcanzable. Juan it.acc.f reach.3sg.pst all lo reach.ble (32) a. * Juan comió la cena todo lo comible. (Sp) Juan eat.3sg.pst the dinner all lo eat.ble

grita mucho ‘Juan shouts a lot’ has three possible interpretations: iterative, i.e. often, which would correspond to an aspectual modifier, available with any verb; one that measures the extension of the shouting event, which corresponds to a quantified indefinite direct object, acceptable only with transitive verbs; and a degree interpretation, possible only with gradable predicates. A sentence like Juan grita todo lo gritable has only two readings, which correspond to the second and the third, the list and the degree readings to be precise. Further differences between the two quantifier expressions have to do with the kind of degree they express. Whereas mucho expresses an indefinite degree above normal, todo lo Vble denotes the maximum degree on the relevant scale, so that a sentence such as (ii) contrasts with (i) in that the former is fine only with a hyperbolic interpretation. (i) Juan gritó mucho, pero María gritó más. (Sp) Juan shout.3sg.pst a.lot but María shout.3sg.pst more 1. ‘Juan shouted very often, but María shouted more often.’ 2. ‘Juan shouted many things, but María shouted more.’ 3. ‘Juan shouted a lot, but María shouted more.’ (ii) Juan gritó todo lo gritable, pero María gritó más. (Sp) Juan shout.3sg.pst all lo shout.ble but María shout.3sg.pst more 1. ‘Juan shouted as many things as one can shout, but María shouted more.’ 2. ‘Juan shouted as much as one can shout, but María shouted more.’

186   

   Case Study 1: V todo lo Vble

b. * Juan la comió todo lo comible. Juan it.acc.f eat.3sg.pst all lo eat.ble (33) a. ? Juan acercó la lámpara todo lo acercable. (Sp) Juan bring.closer.3sg.pst the lamp all lo bring.closer.ble ‘Juan brought the lamp as much closer as one can.’ b. Juan la acercó todo lo acercable. Juan it.acc.f bring.closer.3sg.pst all lo reach.ble ‘Juan brought it as much closer as one can.’ (34) a. ? Juan pateó la puerta todo lo pateable. (Sp) Juan kick.3sg.pst the door all lo kick.ble ‘Juan kicked the door as much as one can.’ b. Juan la pateó todo lo pateable. Juan it.acc.f kick.3sg.pst all lo kick.ble ‘Juan kicked it as much as one can.’ In (31) we have a transitive punctual achievement. Todo lo Vble is ungrammatical when it co-appears with a direct object. Exactly the same facts obtain with accomplishment predicates like the one in (32).24 These two sentences contrast with example (33), a transitive activity predicate, and example (34), a transitive semelfactive, where todo lo Vble can be found with a direct object, resulting in a degree interpretation. In all cases, they can have todo lo Vble in the position of direct object: Juan alcanzó todo lo alcanzable ‘Juan reached everything that could be reached’, Juan comió todo lo comible ‘Juan ate everything that could be eaten’, Juan acercó todo lo acercable ‘Juan brought closer everything that could be brought closer’, and Juan pateó todo lo pateable ‘Juan kicked everything that could be

24 According to Kovacci (1999: 733), quantity adverbials like mucho, poco, bastante must appear in a position adjacent to the verb, as shown in (i). (i) ? Tú has madrugado hoy mucho. (Sp) You have.2sg.prs got.up.early.part today a.lot ‘You have got up very early today.’ The slight deviance of examples (33a)–(34a) suggest that todo lo Vble is also subject to this contiguity requirement, and that is why I also give the examples with the cliticized direct object, where todo lo Vble occupies the immediate post-verbal position. Below I will argue that the degree phrase todo lo Vble functions as a telicity marker, and as such it must occupy the same position as an incremental theme. If right, this could perhaps explain all these grammaticality judgments with mucho ‘a lot’ as well.



Unergative verbs   

   187

kicked’. In all these sentences, the interpretation is that of a list, i.e. quantification over objects. At first sight, one could think that the contrast between (31)–(32) and (33)–(34) is due to the (a)telicity of the predicates: only atelic predicates can license a degree phrase; the first two are telic and therefore do not allow this type of modifier. However, the ungrammaticality of (35) shows that such an explanation is not enough to account for all cases. We know that the presence of a plural or a mass noun in object position makes the predicate atelic. Thus (35), despite being atelic, is still ungrammatical and contrasts with the slightly deviant (33a)–(34a). (35) * Juan comió patatas / queso todo lo comible. Juan eat.3sg.pst potatoes / cheese all lo eat.ble Below I will argue that these cases are ruled out by a general requirement on having just one scale in a particular sentence (Tenny 1994) that measures out the event. Thus, in (35), patatas ‘potatoes’ or queso ‘cheese’ represent the external scale (Rappaport Hovav 2008) along which the eating event progresses, despite not being delimited. In contrast, la lámpara in (33) does not have this function, so that a (path) scale todo lo acercable becomes felicitous. Note that the presence of another path scale would render the sentence ungrammatical as well. (36) * Juan la acercó todo lo acercable a la mesa. (Sp) Juan it.acc.f bring.closer.3sg.pst all lo reach.ble to the table Next, I examine the specific behavior of unergative verbs in this construction, pointing to the properties they show, as well as the problems they pose. I also flesh out a preliminary analysis for the list reading of unergatives based on a cognate object configuration, and show how extending it to the other interpretation cannot resolve all their complexities.

4 Unergative verbs In the case of unergatives, that they can appear in the construction V todo lo Vble is certainly less surprising than in the case of unaccusatives, or not surprising at all, because we know that unergatives are possible bases for -ble derivation when

188   

   Case Study 1: V todo lo Vble

they select a cognate or hyponymic object [henceforth CO and HO], as exemplified in (37).25 (37) a. música bailable (Sp) music danceable ‘danceable music’ b. un material trabajable a material workable ‘a workable material’ c. una experiencia vivible a experience liveable ‘a liveable experience’ Next, I first explore the possibility of a CO analysis for todo lo Vble with unergative verbs, and show that such an analysis cannot account for the variation they display.

4.1 todo lo Vble as a cognate object Indeed, unergative verbs have been analyzed as being transitive in their lexical conceptual structure within Hale and Keyser’s (1993, and much subsequent work) framework. In this model, denominal verbs like laugh or sneeze would have the lexical structure in (38), and would be derived through conflation (a copy process, subsumed under Merge in their 2002 work) under a strict head-complement relation (see the next two footnotes). (38)

V V

N laugh sneeze

25 See e.g. Haugen (2009) on hyponymous objects within Distributed Morphology and the Copy Theory of Chomsky (1995).



Unergative verbs   

   189

For true COs (which Hale and Keyser distinguish from HO), these authors suggest that “under certain conditions, copies are realized overtly in both positions, that is, target and source” (Hale and Keyser 2002: 76),26 specifically when the lower copy, the source, appears in its extended projection, dominated by some functional category, as in (39) (their example (60) on page 76).27 (39)

V V sleep

D D the

N N

P

sleep

of the just

For Marantz (2005), the CO is realized in the same position as incremental theme objects, which are built on an activity verb and are interpreted as a change of state event in that they measure out the event. Marantz’s (2005: 2) structure is repeated in (40) with a cognate object.

26 In previous versions, Hale and Keyser have analyzed COs and HOs as “the result of ‘reinsertion’ into the trace position created by conflation”, a solution explicitly rejected in their (2002) book on theoretical grounds and also “because conflation is not head movement and hence does not leave a trace in any conventional sense” (Hale and Keyser 2002: 88). 27 See Hale and Keyser (2002) for technicalities on the process of conflation. In fact, Hale and Keyser (2002: 98) eliminate conflation from their theory of argument structure altogether, in that conflation is identified with Merge, as a relation of binding between (the semantic features of) a head and (those of) its complement. A denominal verb like sleep would be analyzed as an uncategorized root that becomes verbal or denominal depending on the context, e.g. it is a verb if it heads a verbal projection and establishes inflectional relations like tense and mood. As a verb, it “imposes a selectional claim on its object, licensing the empty N” or it establishes a hyponymous relation with an overt complement. The first case is illustrated in (i); here the verb takes an empty category N as a complement which is linked by selection with the verb.

190   

   Case Study 1: V todo lo Vble

(40) v v

a dance

Change of state event (incremental theme)

√dance Relation of causation

Activity verb Thus, in both analyses, the CO appears as sister of a verbal head, which is the position of the internal argument that measures out the event.28 Hence, we may expect this to be exactly the same position for todo lo Vble in the sentences in (41), since, as we saw in the previous section, the function of todo lo Vble with unergatives is that of making the predicate telic. Thus, if we assume an analysis that merges the CO in the same position as any other direct internal argument, as also proposed in Massam (1990) or Macfarland (1995) for COs, these cases should be relatively straightforward.29, 30 (41) a. En la fiesta bailamos todo lo bailable. (Sp) in the party dance.1pl.pst all lo dance.ble b. Mi padre trabajó todo lo trabajable. my father work.3sg.pst all lo work.ble

V

(i) V{sleep}

N{sleep}

sleep 28 See also Harley (2005) for evidence that supports the view that the boundedness of the conflated/incorporated root of denominal verbs in Hale and Keyser’s framework determines the telicity of the event denoted by the verbal predicate, i.e. they behave as incremental themes. In other words, she argues for the position of sister-to-v as the one that “licenses the event-object homomorphism” in the sense of Krifka (1989, 1998). 29 ‘Straightforward’ because we can pattern them with COs. ‘Relatively’ because the analysis of COs is far from being straightforward. 30 The interpretation of the sentences in (41) is given in (45) for expository reasons.



Unergative verbs   

   191

c. Los niños de la guerra vivieron todo lo vivible. the children of the war live.3pl.pst all lo live.ble Additional support for this analysis comes from the fact that COs have been argued to express the result of an action that delimits the predicate (Macfarland 1995), which seems to be the same function conveyed by todo lo Vble in these examples.31 There are some problems, though, with this hypothesis – apart from other complications stemming from the intricacies of COs. First, verbs like dormir or toser do not allow their objects to become the external argument of a -ble adjective, despite their being able to appear with a CO, as shown in (42)–(43).32 Thus, we would expect that no -ble form can be derived in Spanish for these verbs. Still, their -ble forms do exist, as we have seen in (6) and (23) above, even though these are restricted to special constructions of the type under study. Therefore, any analysis that treats those special -ble forms as derived from a structure with an internal argument cannot account for these contrasts. (42) a. dormir una siesta / dormir una borrachera (Sp) sleep.inf a siesta / sleep.inf a drunkenness ‘to take a siesta / to sleep it off’ b. * una siesta dormible / *una borrachera dormible a siesta sleep.ble /  a drunkenness sleep.ble

31 Although this seems true with respect to the delimiting function, the expression of a result or created object is not true with all verbs. For instance, note that with the verb trabajar in (41) above this is clearly not the case. The meaning of this sentence is that my father worked everything that could be worked, e.g. wood, leather, gold, etc, all of these already existing materials. Similarly perhaps with the verbs live and even dance. However, in these cases, the verbs allow two readings, one of them being that of an effected object or result, the one that appears in the CO construction. Thus, if one dances a tango, it is possible to say that the tango already existed and that it also came into existence. It is this last reading the one that is required in the CO. See Macfarland (1995), particularly section 1.3.5. 32 As pointed out in Hale and Keyser (2002: 70f), we must distinguish between cognate objects (i.e. root-identical to the verbal base) and hyponymous argument constructions. There seem to be very few cognate objects in Spanish (see Real Puigdollers (2008) for the contrast between English and Romance COs), but I have included one in the text above with the verb toser ‘cough’, to show that they seem to behave in the same way. Notice in passing that the requirement on the pleonastic nature of the construction V todo lo Vble may be taken as an additional argument for the distinction between true cognate and hyponymic objects.

192   

   Case Study 1: V todo lo Vble

(43) a. toser mucosidad / toser sangre / toser una tos seca (Sp) cough.inf mucus cough.inf blood  cough.inf a cough dry ‘to cough up phlegm / to cough up blood / to cough a dry cough’ b. * mucosidad tosible / *sangre tosible / *una tos seca tosible mucus cough.ble   blood cough.ble   a cough dry cough.ble On the other hand, there are some unergative verbs that never take a CO, such as estornudar ‘sneeze’ or viajar ‘travel.’33 In this case, we would expect them to reject -ble affixation altogether. The contrast in (44) shows that the analysis of these forms cannot be reduced to a simple cognate object structure. (44) a. Creo que con este resfriado he estornudado todo think.1sg.prs that with this cold have.1sg.prs sneeze.part all lo estornudable. (Sp) lo sneeze.ble ‘I think that I have sneezed as much as one can sneeze with this cold.’ b. Ha viajado todo lo viajable y aún tiene have.3sg.prs travel.part all lo travel.ble and still have.3sg.prs ganas de emprender otro viaje. wishes of embark.inf other trip ‘She has travelled as much as she could and is still looking forward to embark on another trip.’ c. * estornudable / * viajable sneeze.ble travel.ble Interestingly, such uneven behavior among unergative verbs correlates with a difference in interpretation. The forms bailable / trabajable / vivible ‘dance-

33 We can find examples like (i), where the verb takes a direct object. Note, however, that this is not a typical cognate object, since COs are completely banned with these verbs, as shown in (ii), but it is an expression that manifests degree. In §5.3, I will argue that cases like those in (44) above are possible precisely because the cognate object is in fact a degree cognate construction. (i) He viajado todo lo que he querido. (Sp) have.1sg.prs travel.part all lo that have.1sg.prs want.part ‘I have travelled as much as I wanted’ (ii) * He viajado un viaje rápido. (Sp) have.1sg.prs travel.part a travel quick ‘I have travelled a quick travel’



Unergative verbs   

   193

able / workable / liveable’ in (41) above are ambiguous. As shown in (45), they can be interpreted as quantifying over objects, so that we obtain the list reading in interpretation (1), or as having scope over the event, in which case we get a degree – measure or extent – interpretation, as in reading (2). (45) a. En la fiesta bailamos todo lo bailable. (Sp) in the party dance.1pl.pst all lo dance.ble 1. We danced everything that could be danced: salsa, rock, swing, etc. 2. We danced as much as possible / up to the highest possible extent b. Mi padre trabajó todo lo trabajable. my father work.3sg.pst all lo work.ble 1. He worked everything that could be worked: wood, marble, stone, etc. 2. He worked as much as possible / up to the highest possible extent c. Los niños de la guerra vivieron todo lo vivible. the children of the war live.3pl.pst all lo live.ble 1. They lived everything that could be lived: a war, famine, exile, etc. 2. They lived as much as possible / up to the highest possible extent It turns out that the referential reading is absent with verbs like dormir/toser or estornudar/viajar, i.e. those that (almost) never take a CO, as illustrated in (46).34 Note that if the degree reading had to be analyzed as an implication of the list reading (e.g. if one dances many dances, the inference is that one has danced a lot), these cases could not be explained, because there is no list reading from which to derive such an implication.

34 The verb toser seems special in that it does not allow the form tosible with cognate objects, as shown in (43) above, but it may (marginally) allow a list reading, as illustrated in (i). (i) (?) Ha tosido todo lo tosible: sangre, moco, flema. have.3sg.prs coughed all lo cough.ble blood mucus phlegm ‘She has coughed up everything that one can cough up: blood, mucus, phlegm.’ Joan Mascaró (personal communication) suggests that the unavailability of a referential reading with the verb dormir could be due to pragmatic reasons, namely the fact that one can only sleep siestas and borracheras. Also Jaume Mateu (personal communication) observes that the oddness of the list reading with dormir is unexpected and does not seem grammar-based. I essentially agree with them, i.e. that the grammar alone cannot account for the ungrammaticality of these forms. Below I will suggest that both interpretations are available with verbs like dormir, and will hint at a possible explanation for the absence of the list reading with some verbs related to the nature of the quantifier todo ‘all’ and its interaction with the pragmatics of these verbs.

194   

   Case Study 1: V todo lo Vble

(46) a. Durmió todo lo dormible. (Sp) sleep.3sg.pst all lo sleep.ble 1. * S/he slept everything that could be slept: siesta, drunkenness. 2. S/he slept as much as possible / up to the highest possible extent b. He estornudado todo lo estornudable. have.1sg.prs sneeze.part all lo sneeze.ble 1. * I have sneezed everything that could be sneezed: ??? 2. I have sneezed as much as possible / up to the highest possible extent A possible line of analysis could be to suppose that the contrast between verbs like bailar and verbs like dormir may be due to a difference in the kind of object they admit. Whereas the former type admits argumental COs, i.e. objects that have been generated in an internal argument position and give rise to the list reading, these are not licensed in the case of dormir. The latter would only accept a so-called adverbial CO (Pereltsvaig 1999; Rodríguez Ramalle 2003; or Nakajima 2006), also possible with bailar-type verbs, which results in the degree or measure interpretation. According to Rodríguez Ramalle (2003), adverbial cognate objects can predicate a property of an action, process or state, like Spanish adverbs ending in -mente (that correspond to -ly adverbs in English). This seems to be the function of todo lo dormible in (46a). We can apply some tests that show the different status of the sequence todo lo Vble with different verbs. First, the constituent V todo lo Vble can be used as an answer to the questions what? or how much? with verbs like bailar. However, with verbs like dormir, we can only ask and answer the question with how much? The examples in (47)–(48) illustrate this contrast, which argues for the non-argumental status of todo lo dormible. (47) a. ¿Qué bailó / trabajó / vivió? – Todo lo bailable / trabajable /  what dance / work / live.3sg.pst? – all lo danceable / workable / vivible. (Sp) liveable ‘What did she dance/work in/live? – Everything that can be danced/ worked in /lived.’ b.  ¿Cuánto bailó / trabajó / vivió? – Todo lo bailable / how.much dance / work / live.3sg.pst? – all lo dance.ble / trabajable / vivible. work.ble / live.ble ‘How much did she dance/work/live? – As much as one can dance/ work/live.’



Unergative verbs   

   195

(48) a. * ¿Qué durmió / estornudó / viajó? – *Todo lo dormible /  what sleep / sneeze / travel.3sg.pst? –  all lo sleep.ble / estornudable / viajable. (Sp) sneeze.ble / travel.ble b.  ¿Cuánto durmió / estornudó / viajó? – Todo lo dormible /  how.much sleep / sneeze / travel.3sg.pst? – all lo sleep.ble / estornudable / viajable. sneeze.ble / travel.ble ‘How much did she sleep/sneeze/travel? – As much as one can sleep/ sneeze/travel.’ With verbs like dormir, the DP todo lo Vble can alternate with degree or quantificational adverbs like mucho ‘a lot’, as shown in (49), which could be taken as additional evidence for its non-argumental status.35 (49) Durmió todo lo dormible / mucho. (Sp) sleep.3sg.pst all lo sleep.ble / a.lot ‘She slept as much as one can sleep / a lot.’ Despite the fact that these two tests argue for the adverbial status of todo lo Vble in one of its readings, we obtain the opposite result when applying another wellestablished test for argumenthood in Spanish. According to de Miguel’s (1992: 71) well-formedness condition on reduced participial clauses, in Spanish only verbs with an internal argument, i.e. “that is projected in the structural object position” (author’s translation), can appear in such structures, as illustrated in (50) with her examples: the transitive verb vender ‘sell’, the unaccusative morir ‘die’ and the unergative nadar ‘swim’.

35 This last statement needs some caution, for the status of mucho as either object or adjunct in sentences like (i) is still an unsettled issue. See Bosque and Masullo (1998) and the discussion below. Indeed, mucho is as ambiguous as todo lo Vble. (i) Juan come mucho. Juan eat.3sg.prs much ‘Juan eats a lot.’ a. ‘in great quantities’ (adverbial interpretation) b. ‘a lot of food’ (direct object interpretation) (Bosque and Masullo 1998: 27) Still, the fact that mucho in (49), like todo lo dormible, can only have an adverbial interpretation would speak for its status as non-argumental.

196   

   Case Study 1: V todo lo Vble

(50) a. Vendida la casa, abandonamos para siempre el valle. (Sp) sell.part the house, leave.1pl.pst for always the valley ‘We left the valley forever, once the house was sold.’ b. Muerto el perro, se acabó la rabia. die.part the dog se finish.3sg.pst the rabies ‘The best way to solve a problem is to attack the root of it.’36 c. * Nadado Juan, se sintió mejor. swim.part Juan se feel.3sg.pst better (de Miguel 1992: 71) As shown in (51), a verb like dormir can appear in this construction when together with a CO or the sequence todo lo dormible. Even more puzzling, verbs that never take a CO are possible in this construction when followed by todo lo Vble, as exemplified in (52). Thus, we get contradictory results with respect to the argumenthood of todo lo Vble.37 (51) a. Una vez dormida la siesta, se puso a trabajar. (Sp) a time sleep.part the siesta se put.3sg.pst to work.inf ‘Once she had slept the siesta, she started to work.’ b. Una vez dormido todo lo dormible, se puso a trabajar. a time sleep.part all lo sleep.ble se put.3sg.pst to work.inf ‘Once she had slept as much as possible, she started to work.’ c. *Una vez dormido Juan, empezó a soñar. (Juan = agent)   a time sleep.part Juan start.3sg.pst to dream.inf

36 This is a Spanish saying that literally means something like ‘once the dog [affected by rabies] is dead, the rabies is finished.’ 37 As I mentioned at the very beginning, the structure todo lo Vble is a special construction used in colloquial Spanish under special circumstances of strong intensification, i.e. it is not an everyday expression. This makes that the manipulation of the sentences containing this structure for testing purposes may sound often forced and rather bizarre. In those cases, either they are acceptable under the right intensifying intonation, or they show some sharp contrast that I consider enough to support the discussion.



Unergative verbs   

   197

(52) a. Una vez viajado todo lo viajable, me instalaré en a time travel.part all lo travel.ble me.refl settle.1sg.fut in la costa. (Sp) the coast ‘Once I have travelled as much as one can travel, I will move to the coast.’ b.  Una vez estornudado todo lo estornudable, me a time sneeze.part all lo sneeze.ble me.dat salió sangre por la nariz. come.out.3sg.pst blood through the nose ‘Once I sneezed as much as one can sneeze, my nose bleeded.’ c. *Una vez viajado Juan, se instaló en la costa.   a time travel.part Juan se settle.3sg.fut in the coast Thus, since the question test in (47)–(48) is a syntactico-semantic test, the alternation test in (49) is rather semantic in nature, and the last test applied is properly syntactic, we can say that todo lo Vble in its degree interpretation displays a mixed behavior, it acts semantically as an adverbial, but syntactically as an internal object. It does not seem that an analysis of these examples as adverbial COs could satisfactorily account for such a conflicting behavior.38 Additional problems arise from the comparison between todo lo Vble and properties of adverbial COs in other languages. According to Pereltsvaig (1999: 547), this type of COs can only have a manner interpretation – exactly as the adverbial COs in Spanish reported in Rodríguez Ramalle (2003: 328) –, do not need to be strictly root-identical with the verb, and do not delimit the event denoted by the main predicate. However, our Spanish data not only must be cognate with the verb, but they do not predicate the manner of the event, and they affect the telicity of the predicate.39

38 According to de Swart (2007), COs in English behave semantically like manner adverbials, but syntactically as direct arguments of the verb, a position that seeks to reconcile the longstanding debate about the adverbial – because of its semantic contribution – versus argumental – based on syntactic grounds – function of COs (e.g. Jones 1988 for the former view; Massam 1990 or Macfarland 1995 for the latter). There is a crucial difference between the data my statement above refers to and de Swart’s. Whereas there is no doubt that the semantic expression of manner modifies the event, the status of a measure or degree constituent is not so clear. See Real Puigdollers (2006) for a review of the different stances. 39 An additional problem has to do with the form of adverbial COs in Spanish and other languages. These always contain a cognate noun plus an obligatory adjective. According to Rodríguez Ramalle (2003), the noun in (ib) is interpreted as an event nominalization (as opposed to the result nominalization found with argumental COs in (ia)); the adjective must

198   

   Case Study 1: V todo lo Vble

There are still some remaining problems with an analysis of these data as adverbial COs. It does not account for the fact that dormir-type verbs can license a CO but have only the adverbial reading with todo lo Vble. A possible solution would be to assume that cases like dormir la siesta or toser sangre are not cases of true CO, probably a right assumption, but examples of HO.40 More problematic would then be the existence of toser una tos seca ‘cough a dry cough’, for the object is root-identical to the main verb, and therefore more difficult to be analyzed as a non-true CO, in fact impossible or stipulative. Thus, on the basis of the above description, it seems that we must distinguish between three different types of unergative verbs exemplified by bailar, estornudar and dormir,41 because they show a divergent behavior with respect to the derivation of regular -ble adjectives, the cognate object selection, and their interpretation in the presence of todo lo Vble. The former, bailar-type verbs, can form regular -ble adjectives, can take COs and HOs, and are ambiguous in the construction V todo lo Vble, i.e. they can have the list and the degree readings. The latter, the dormir-type verbs, can have COs and/or HOs but they can only have the degree reading with todo lo Vble, and they cannot produce regular -ble adjectives. Finally, estornudar-type verbs never take COs, cannot give rise to regular -ble adjectives and can only get a degree interpretation in this construction. These properties are summarized in the table in (53).

denote some verbal property, as the contrast in (ii) shows. Examples are from Rodríguez Ramalle (2003: 327–328). None of these characteristics is found in V todo lo Vble. (i) a. Juan lee una lectura divertida. (Sp) Juan read.3sg.prs a reading enjoyable ‘Juan is reading an enjoyable reading.’ b. Juan lee con una lectura rápida. Juan read.3sg.prs with a reading quick ‘Juan reads quickly.’ (ii) a. * Juan vive una vida probable. (Sp) Juan live.3sg.prs a life probable b. Juan vive una vida tranquila. (= Juan vive tranquilamente) Juan live.3sg.prs a life quiet (= Juan live.3sg.prs quietly) ‘Juan lives a quiet life. (= Juan lives quietly)’ 40 Interestingly, la siesta cannot delimit an event of sleeping, as COs/HOs generally do, witness the contrast Juan durmió la siesta durante una hora / *en una hora ‘Juan took a siesta for an hour/*in an hour’. 41 Notice that if the special behavior of verbs such as toser and dormir is due to pragmatic reasons, then we are left with only two types of unergative verbs (see footnote 34). In fact, it might be that all unergative verbs pattern the same, if it turns out that verbs like estornudar ‘sneeze’ do not have a list reading because of pragmatic reasons, as well, i.e. there is nothing that one can sneeze in Spanish. I will come back to this issue below.



Unergative verbs   

   199

(53) Properties of unergative verbs bailar trabajar

dormir toser

estornudar viajar

(a) CO / HO





*

(b) Degree reading







(c) List reading



*(/?)

*

(d) Derived -ble



*

*

Among these, the most remarkable property is undoubtedly that apparently nonexisting words like *dormible or *estornudable become possible, i.e. that despite line (53d), we can create Durmió todo lo dormible and Estornudó todo lo estor­ nudable. Their grammaticality cannot be tied to the internal properties of their verbal bases, but it must be a function of the external syntactic context and the specific nature of the DP in which they appear, a fact that lexicalist frameworks cannot predict, but that it is anticipated in syntactic theories of word formation. Also important is to find the source of variation among unergative verbs. Notice that the hypothesis I have advanced, i.e. to treat todo lo Vble as a cognate object, cannot derive these differences from any more general property. What is more, assuming an analysis of all unergative verbs as being derived from a transitive structure à la Hale and Keyser, the above stated hypothesis would rather have to stipulate the properties in (53). Finally, the fact that the adverbial interpretation conveys a meaning of measure or degree, whereas adverbial COs are in general restricted to express manner appears as rather mysterious under such a hypothesis. Next, I work out a preliminary analysis of the list interpretation with unergative verbs. The analysis of the degree interpretation will be deferred to until after the analysis of unaccusative verbs, which will provide the clues to a unified analysis of all cases.

4.2 lo as D The analysis of V todo lo Vble in its list reading seems rather uncomplicated. Essentially, in the syntax, the root merges with a little v and, as suggested in Hale and Keyser (2002), the CO is realized in the position of complement of the verbal head, when this appears in an extended projection. That is, in a CO configuration, the root is realized in two different places. The fact that the CO is realized in the source position would account for the fact that they must necessarily be identical.

200   

   Case Study 1: V todo lo Vble

And since this is the position of incremental themes, this complement will be able to measure out the progress of the event, which means that it turns the verbal activity into an accomplishment. (54)

v v √i

QP v

… √i

For the internal analysis of todo lo Vble, I assume with Bosque and Moreno (1990), that the particle lo is a pronoun that acts as a head;42 specifically, an individuative lo is a variable that takes non-human entities as its range. In the case at hand, lo is a variable that ranges over a set of entities restricted by the -ble predicate. It is further quantized by todo, so that a sentence like Juan bailó todo lo bailable ‘Juan danced everything that could be danced’ could be represented informally as in (55), i.e. we have a quantification over the set of all possible (non-human) entities that can be danced. (55) ∀x [[x is an entity] [x can be danced → Juan danced x]] I propose that this lo materializes the head of a DP that has been generated as an internal argument in the specifier position of a verbalizing head underlying regular -ble adjectives, thus satisfying the sufix’s requirement on having an internal argument. As shown in the structure in (56), it will later raise up to a small clause structure where it establishes a predicative relation with a modalized predicate,43 i.e. where lo is the subject of a predicate. FP refers to a functional projection that turns the adjectival structure into a predicate.

42 See footnote 1 above. Here I assume that lo is a cliticized form of a strong neuter pronoun ello. See also Leonetti (1999) or Luján (2002), inter alia. 43 See Luján (2002), based on Kayne (1994), for an analysis of lo + adjective as involving a small clause structure.



Unergative verbs   

   201

SC

(56) DP lo

FP aP

-ble configuration

ModP AspectP vP DP D’

v' v



D t

In principle, this structure would apparently suffice for unergatives, because they are agentive, i.e. all the requirements of -ble suffixation are met, the implication of an originator and an internal argument. Indeed, forms such as the ones in (57) are perfectly fine outside this construction. (57) a. bailable, trabajable, vivible (Sp) danceable workable liveable b. lo bailable, lo trabajable, lo vivible lo danceable lo workable lo liveable However, (57b) cannot express a multiplicity of objects, i.e. a list reading, as shown in (58). As pointed out in Bosque and Gutiérrez-Rexach (2009:  485), “a singular pronoun, like lo, can only obtain reference to multiple individuals if it is c-commanded by a quantifier” (author’s translation).

202   

   Case Study 1: V todo lo Vble

(58) un estilo que oscila entre lo experimental y lo a style that fluctuate.3sg.prs between lo experimental and lo bailable (Sp) danceable ‘a style that fluctuates between that which is experimental and that which is danceable’ In the case of dormir, which can take cognate objects but disallows -ble and cannot have the list reading, there are two possibilities. It could be that expressions like dormir la siesta ‘take a siesta’, dormir la borrachera ‘sleep off a hangover’, or dormir el sueño de los justos ‘sleep the sleep of the just’ are fixed, idiomatic expressions, so that dormir would rather be like estornudar in that it cannot take cognate objects freely. The examples in (59) show that such a stipulative solution cannot account for all cases, so that no insight is gained. (59) a.  He dormido {la borrachera/*/? una borrachera/*/? have.1sg.prs sleep.part  the drunkenness a drunkenness varias borracheras esta semana}. (Sp) various drunkenness.pl this week ‘I have slept off the hangover.’ b.  He dormido la siesta/ una siesta/ varias siestas esta have.1sg.prs sleep.part the siesta a siesta various siestas this semana/ una buena siestecilla. week a good little.siesta ‘I have taken a (good little) siesta (many times this week).’ c.  Han dormido el sueño de los justos44 / *un sueño / have.3pl.prs sleep.part the dream of the fair.people / a dream / *varios sueños.   various dreams ‘They have slept the sleep of the just.’

44 As Mascaró (personal communication) points out to me, it seems that the verb dormir ‘sleep’ admits sueño ‘dream’ as an object as long as this is somehow modified. So, we find dormir with the complements in (i) (examples are from CREA). (i) el sueño eterno, el sueño profundo de la juventud, el sueño del olvido, un the dream eternal the dream deep of the youth the dream of.the oblivion a sueño de paz, … (Sp) dream of peace ‘the eternal dream, the deep dream of the youth, the dream into oblivion, a dream of peace’



Unergative verbs   

   203

Another more promising option has to do with the number of entities that the expression todo lo dormible refers to. In principle for a verb like dormir one can only sleep siestas45 and borracheras,46 and such a reduced number of possible entities is what precludes the use of the quantifier todo.47 There is in Spanish, as in English, a specific universal quantifier that is restricted to two entities, namely ambos ‘both’, which would block the use of todo for a quantification of a number fewer than three. That is, the quantifier todo needs a minimum number of entities to quantify over, possibly ‘three or more’; when there are fewer than that, quantification over objects with todo is not felicitous. Similarly for the verb toser ‘cough’: depending on whether the set of coughable entities (blood, phlegm, mucus) is considered to be two or more, native speakers may (marginally) admit a list reading or not. This suggests that the availability of the list reading with intransitive verbs is a pragmatic rather than grammatical issue. That is, a list interpretation would only exist when there are enough elements to build a list, when the inventory is large enough to restrict the universal quantifier todo. Hence, the general impossibility of a referential reading with dormir ‘sleep’ or toser ‘cough’ cannot be linked to the grammar of -ble, nor can it be linked to general properties of unergative verbs; it is a function of the quantifying properties of todo together with the individual pragmatico-selectional properties of each particular verb.

45 Siestas would presumably include also the various expressions with sueño as they would stay in a relation of hyponymy. 46 Depending on the variety, native speakers may use dormir la mona, dormir la zorra, dormir el vino, but they are all colloquial expressions for dormir la borrachera ‘sleep it off’, so that they would not count as different objects. 47 Joan Mascaró (personal communication) provided me with the crucial example that brought me to my suggestion. Suppose that Gulliver reaches Marmotland, where people sleep four hours at night (la oscurada ‘the dark.ada’); during the daytime, they always sleep one or, exceptionally, two naps out of six different types that have different names: the nap at ten lasts five quarters and it is done on a chair (la sillada ‘the chair.ada’); the one at eleven is done by standing for ten minutes (la depieada ‘the on.foot.ada (standing.ada)’); the one at noon lasts for one hour and takes place on the ground in the garden (la jardinada ‘the garden.ada’); the one at two lasts two hours and people must snore (la ronquetada ‘the snore.ada’); the one at three lasts for three hours and it is done in the doghouse (la perrada ‘the dog.ada’); and the one at five which lasts three quarters and it is done on the sofa after a shower (la sofatada ‘the sofa.ada’). It is most likely that people take the same kind of nap during the week, perhaps two or three different ones. Rarely, though not impossibly, one can sleep the six different ones during the same week. In that case one can say Esta semana Gulliver ha dormido todo lo dormible: nochadas, perradas, jardinadas,… “This week Gulliver has slept all the sleep.ble: nochadas, perradas, jardinadas, …”.

204   

   Case Study 1: V todo lo Vble

In any event, the structure in (56) is not enough for dormir/toser or estornu­ dar/viajar and results in the ungrammatical forms in (60), thus contrasting with (57) above. (60) a. * estornudable / *viajable, *lo estornudable / *lo viajable (Sp) sneeze.ble  travel.ble  lo sneeze.ble  lo travel.ble b. * dormible / *tosible, *lo dormible / *lo tosible sleep.ble  cough.ble  lo sleep.ble  lo cough.ble The explanation for the ungrammaticality in (60a) can be immediately related to the absence of CO/HO with these verbs. There is nothing in Spanish that can be sneezed or travelled, and therefore it is not possible to predicate the property of being able to be sneezed or travelled. As for (60b), I must say that I have not found any other verb that can have some type of CO or HO, but does not allow a -ble form in some context or other.48 It seems to me that there are in fact just two types of unergative verbs, those that can take a number of CO/HO and those that do not. The former, as expected, can derive -ble adjectives in a natural way, and they behave as such under normal circumstances. The latter do not give rise to any -ble adjectives, unless a cognate structure is provided. To create a -ble adjective we need an internal argument – apart from an originator, already implied with unergatives; and to have a cognate object, this must be able to delimit the event expressed by the activity in that it expresses a result or a resultant state. All requirements are met in the structure V todo lo Vble: an internal argument materialized by lo, a cognate structure, and todo that delimits the set of possible entities. That is, in the list reading, todo lo Vble materializes a set of objects which is used to measure out the event, exactly as an incremental theme. Notice that this is in accordance with Hale and Keyser’s suggestion that unergative verbs contain

48 I have not solved the problem posed by the contrast between *moco tosible ‘mucus cough. ble’ and música bailable ‘music dance.ble’. If they were both cases of CO/HOs, they should have the same structure and behave in the same way. The fact that they do not, together with the contrast *toser una tos ‘cough a cough’ versus bailar un baile ‘dance a dance’ suggest that they are in fact instances of different kinds of complements. Indeed, additional evidence for their different status may come from the fact that verbs like bailar ‘dance’ can form regular -ble adjectives that can appear in non-pleonastic contexts, such as (i), as was mentioned in footnote 12 above; and from the fact that they do not require the presence of a quantifier either, as we just saw in (57). (i) Me gusta todo lo bailable. (Sp) me.dat like.3sg.prs all lo danceable ‘I like everything that can be danced.’



Unaccusative verbs   

   205

an unbounded noun (root) that denotes the result of the event in their lexical relational structure.49 As for the degree interpretation, I have already mentioned that todo lo dormible behaves as a degree or quantificational adverbial similar to mucho ‘a lot’. According to Sánchez López (1999: 1094), one of the functions of mucho as a verb phrase or verb intensifier is to measure the extension of the action expressed by the predicate, “in which case it would be equivalent to an indefinite quantified direct object, admitted only by transitive verbs” [author’s translation]. Since the function of todo lo Vble with verbs like dormir is that of being an adverbial intensifier, this DP could be licensed by the AspP present in the structure of -ble, a head responsible for licensing aspectual modifiers. We could then further assume that, since its function is the same as the function of a cognate object, which turns the lexical aspect of the verbal base from an activity into an accomplishment (Rosen 1999), as shown in (61), so that it measures out the event like an internal argument, the DP would take over the function of the object in that it can become the external argument of the -ble adjective in the specific pleonastic construction. (61) a. * Juan durmió en una tarde. (Activity) (Sp) Juan sleep.3sg.pst in one afternoon b. Juan durmió todo lo dormible en una tarde. (Accomplishment) Juan sleep.3sg.pst all lo sleep.ble in one afternoon ‘Juan slept as much as one can sleep in an afternoon.’ Such an analysis could account for a number of issues. On the one hand, it would explain the different interpretation of dormible ‘sleep.ble’, which lacks the list reading of bailable ‘dance.ble’, as well as the contrast between the existence of bailable and the ungrammaticality of the independent item dormible. On the other hand, it would also explain the fact that lo dormible only exists in the construction under analysis, where it appears as a complement of the quantifier todo that gives the structure a quantificational reading, which allows it to measure out

49 In fact, as mentioned in Bosque and Masullo (1998: 43), some of the nouns underlying unergative verbs can also be bounded, as shown in (i)–(ii) with their examples. (i) a. Se hace camino al andar. (Unbounded) (Sp) se make.3sg.prs way at.the walk Lit. ‘One makes way as one goes along.’ b. Hay dos caminos posibles a seguir. (Bounded) there.have.3sg.prs two ways possible to follow ‘There are two possible ways to follow.’

206   

   Case Study 1: V todo lo Vble

the extension of the predicate, which is a function typical of the internal argument. In the case of bailable, which exists independently of this construction, it becomes ambiguous in the presence of the quantifier todo, i.e. it obtains a second reading without losing the one it already had. The main problems with such a proposal are that it neither accounts for its obligatory cognate nature, nor can it be argued to follow from some more general property. In addition, it is unclear how an adverbial intensifier should take over the function of the incremental theme, or how it could be related to the same interpretation available with some, but not all, unaccusative verbs. Thus, I will not investigate this option any further. The next section explores the main properties of this construction with unaccusative verbs. A thorough description of the facts, together with an interesting correlation of this construction with some special English data, will give rise to a possible solution for the degree interpretation not yet accounted for and, more importantly, will provide a unified analysis for all verbs.

5 Unaccusative verbs With respect to unaccusative verbs, things are much more complex. Recall that it was precisely the fact that there are no -ble adjectives regularly derived from unaccusative bases, like those in (62), that brought me to the proposal that there has to be a second requirement on -ble forms. That is, in this case there is a set of entities that are capable of undergoing the change of state denoted by the verb, which would meet the internal argument requirement for -ble derivations. The problem is that these forms lack the implication of an originator. Whereas in the case of unergatives, these can often enter into the CO alternation, the predicates in (62) do not enter into any alternation that could provide the sentence with an external agent, causer, or natural force, as illustrated in (63). (62) a. * brotable, *crecible, *florecible, *llegable, *ocurrible (Sp) sprout.ble  grow.ble  bloom.ble  arrive.ble  occur.ble b. * lo brotable, *lo crecible, *lo florecible, *lo llegable, lo sprout.ble  lo grow.ble  lo bloom.ble  lo arrive.ble * lo ocurrible lo occur.ble



Unaccusative verbs   

   207

(63) a. * {Juan / El agua} ha crecido el árbol.50 (Sp)  Juan / the water have.3sg.prs grow.part the tree b. * La mala suerte ha ocurrido el accidente. the bad luck have.3sg.prs occur.part the accident Surprisingly, all these forms, based on prototypical unaccusative verbs, are fine in the sentences in (64), when they appear in the construction V todo lo Vble.51 (64) a. En esa familia ocurrió todo lo ocurrible. (Sp) in that family occur.3sg.past all lo occur.ble ‘In that family it happened everything that could happen.’ b. Esta semana llegó todo lo llegable. this week arrive.3sg.pst all lo arrive.ble ‘This week it arrived everything that could arrive.’ c. En el jardín floreció todo lo florecible. in the garden bloom.3sg.pst all lo bloom.ble ‘In the garden it bloomed everything that could bloom.’ d. Después de las intensas lluvias, en el jardín creció todo lo after of the intense rains in the garden grow.3sg.pst all lo crecible. grow.ble ‘After the intense rain, in the garden it grew everything that could grow.’ As we saw in (24) above, repeated in (65) with additional data, the structure V todo lo Vble distinguishes two subtypes of unaccusative verbs, depending on whether they express a list reading, as in (1), or they can also state the extent or degree achieved, as in (2). This difference correlates with a well-known subclassification of unaccusative verbs. Wherea (65a–b) are cases of prototypical telic unaccusative verbs, (65c–d) are examples of so-called degree achievements52 (Dowty 1979),

50 The transitive English grow of John grows roses in the garden corresponds to a different lexical item in Spanish, cultivar as in Juan cultiva rosas en el jardín. 51 In (64c–d) I provide the most salient interpretation for V todo lo Vble in these contexts, but see (65). 52 A misnomer, as pointed out in Hay et al (1999). I will keep using this label, though, as it has been customary in the literature on the topic (with a few exceptions, e.g. Levin and Rappaport’s (1995) atelic verbs of change of state or Bertinetto and Squartini’s (1995) gradual completion verbs). See also Rothstein (2008) on degree achievements.

208   

   Case Study 1: V todo lo Vble

i.e. verbs with variable telicity. Note that the list reading of unaccusative verbs derives from a structure where todo lo Vble is the underlying object that corresponds to the surface subject, which presumably appears in post-verbal position. In the degree reading, there is a non-explicit subject (translated in English as something), so that todo lo Vble instantiates a degree or measure. (65) a. Ocurrió todo lo ocurrible. (Sp) occur.3sg.pst all lo occur.ble 1. It happened everything that could happen: the phone rang, someone knocked at the door, the mobile rang, the child fell down, my porcelain vase got broken, etc. 2. * It happened as much as possible / up to the highest possible extent b. Llegó todo lo llegable. arrive.3sg.pst all lo arrive.ble 1. It arrived everything that could arrive: letters, bills, parcels, etc. 2. * It arrived as much as possible / up to the highest possible extent c. Floreció todo lo florecible.53 bloom.3sg.pst all lo bloom.ble 1. It bloomed everything that could bloom: plants, trees, bushes, etc. 2. Something bloomed as much as possible / up to the highest possible extent d. Creció todo lo crecible. grow.3sg.pst all lo grow.ble 1. It grew everything that could grow: plants, trees, bushes, weeds, etc. 2. Something grew as much as possible / up to the highest possible extent The variable behavior of degree achievements with respect to telicity, illustrated in (66), has been mentioned in the literature at least since Perlmutter (1978).54 (66) a. The soup cooled in / for two minutes.

53 De Miguel and Fernández Lagunilla (2000), Demonte (2006), or Pérez Jiménez and Moreno Quibén (2003) classify the verb florecer ‘bloom’ as an achievement. That it is a degree achievement will be shown later on in the main text. 54 See also Tenny (1994), Hay et al. (1999) or Kennedy and Levin (2008) for English, and Bosque and Masullo (1998) or Fábregas (2003) for Spanish, to mention just a few.



Unaccusative verbs   

   209

b. El Euribor ha bajado en / durante dos meses. (Sp) the Euribor have.3sg.prs drop.part in / during two months ‘The Euribor has dropped in / for two months.’ Also their variable status as unaccusatives or as unergatives has been and still is a matter of debate (e.g. Levin and Rappaport Hovav 1995; Bosque and Masullo 1998; and more recently Kratzer 2002; Fábregas 2003; or Pérez Jiménez and Moreno Quibén 2005). The examples in (65c–d) show that degree achievements pattern with unergative verbs when in the structure V todo lo Vble. That is, they can express a list reading, and they can also license a degree phrase, as all other atelic verbs. There is an additional set of unaccusative verbs for which a -ble form in the construction under study is acceptable only in the degree reading. These are unaccusative verbs that express an internally caused change of state which necessarily select a personal theme, like palidecer ‘turn pale’ or enrojecer ‘blush’ in (67). Their ungrammaticality in the list reading stems from the incompatibility between the conceptual semantics of the verb and the type of lo that appears in this structure, Bosque and Moreno’s (1990) individuative lo, which requires a set of non-human entities in its domain. That is, the absence of a list reading with these verbs is due to independent properties of the grammar of lo, which means that these verbs do not constitute a separate class within unaccusatives. (67) a. Al ver el cadáver, Juan palideció todo lo at.the see.inf the corpse Juan turn.pale.3sg.pst all lo palidecible. (Sp) turn.pale.ble ‘Juan turned pale up to the highest possible extent when he saw the corpse.’ b. * Al ver el cadáver, palideció todo lo palidecible. at.the see.inf the corpse turn.pale.3sg.pst all lo turn.pale.ble ‘Everything capable of turning pale turned pale when seeing the corpse.’ (68) a. El ridículo fue tan grande que Juan enrojeció todo lo the ridicule be.3sg.pst so big that Juan blush.3sg.pst all lo enrojecible. (Sp) blush.ble ‘The situation was so absurd that Juan blushed up to the highest possible extent.’

210   

   Case Study 1: V todo lo Vble

b. * El ridículo fue tan grande que enrojeció todo lo the ridicule be.3sg.pst so big that blush.3sg.pst all lo enrojecible. blush.ble ‘Everything capable of blushing blushed due to the ridiculous situation.’ It is likely that the same type of incompatibility is responsible for the ungrammaticality of -ble forms with inherently reflexive verbs in (69), because these verbs also select personal complements, i.e. only personal entities can kneel down or blush. (69) a. * Todo lo arrodillable se arrodilló. (Sp) all lo kneel.down.ble se kneel.down3sg.pst b. * Todo lo sonrojable se sonrojó. all lo blush.ble se blush.3sg.pst As for the degree reading with these verbs, it can be admitted as long as the verb is atelic, and therefore it can license a degree phrase, as evidenced by the contrast in (70). Thus, we may talk about degrees of blushing, but not about degrees of kneeling-down.55

55 Notice that (70a) is possible for native speakers who accept (i), i.e. those that acknowledge the possibility of having different degrees of a kneeling event. Or more generally, as Mascaró (personal communication) suggests, it may just depend on the conceptualization of the verb on each occasion, as in (ii). (i) Juan se arrodilló un poco para alcanzar la caja. (Sp) Juan se kneel.down3sg.pst a bit to reach the box ‘Juan kneeled down a little bit in order to be able to reach the box.’ (ii)  Juan empezó a arrodillarse y no alcanzaba la caja, Juan start.3sg.pst to kneel.down and not reach.3sg.ipfv.pst the box, siguió arrodillándose … y tampoco. Cuando, finalmente, se continue.3sg.pst kneeling.down and not.either when finally se había arrodillado todo lo arrodillable … (Sp) have3sg.ipfv.pst kneel.down.part all lo kneel.down.ble … ‘Juan started kneeling down but he didn’t reach the box; kept on kneeling down, but didn’t either. When he finally had kneeled down as much as possible …’ As for (70b), there seems to be a contrast between sonrojable and ruborizable, both meaning ‘blush’, in that the latter is marginally accepted in this construction. Both facts may be related to the internal structure of these verbs. On the one hand, whereas arrodillarse and ruborizarse are denominal, sonrojarse is deadjectival, which would explain the gradability of the latter, but not of the former. The existence of another ‘blush’ verb enrojecer



Unaccusative verbs   

   211

(70) a. * Juan se arrodilló todo lo arrodillable. (Sp) Juan se kneel.down.3sg.pst all lo kneel.down.ble b. Juan se sonrojó todo lo sonrojable. Juan se blush.3sg.pst all lo blush.ble ‘Juan blushed as much as one can blush.’ Weather verbs have also been analyzed as unaccusatives in Spanish (Calzado Roldán 2000) and French (Ruwet 1989). Like the other unaccusatives, they can also appear in the intensive construction, although they seem to admit only the degree reading.56, 57

with an adjectival base, which also accepts de degree reading, as we have seen in (68) above, would favor this suggestion. On the other hand, the marginal gradability of arrodillarse may be due to a difference in the conceptual semantics of the verb: whether it is understood as ponerse (*más) de rodillas ‘get down (*more) on her knees’ or as doblar (más) las rodillas ‘bend (more) the knees’. Since the construction is colloquial, Mascaró (personal communication) points out that additional aspects may have an influence on the acceptability of the construction, such as the frequency of use of the verb itself (iii), or the degree of formality (iv), i.e. the register. (iii)  Durante su enfermedad, Juan adelgazó todo lo adelgazable / ?? during his illness Juan lose.weight.3sg.pst all lo lose.weight.ble  enflaqueció todo lo enflaquecible. (Sp) get.thin.3sg.pst all lo get.thin.ble ‘During his illness, Juan lost as much weight as one can lose / got as thin as one can get.’ (iv) a. Mira tú, puestos a considerar, ya he considerado todo look you, put.part.m.pl to consider already have.1sg.prs consider.part all lo considerable, o sea que …. (Sp) lo consider.ble or be.3sg.prs.sbjv that ‘Look, if we have to consider pros and cons, I have already taken into account everything that can be taken into account, so …’ b.  La estructura del SD, una vez *considerado todo lo considerable / the structure of.the DP one time  consider.part all lo consider.ble  consideradas absolutamente todas las posibilidades, … consider.part.f.pl absolutely all the possibilities ‘DP structure, once all possibilities have been considered, …’ 56 The literature shows disagreement on the status of weather verbs as either unergatives or unaccusatives. For instance, Bosque and Masullo (1998: 43) include them among unergative verbs. 57 Interestingly, Dixon (2005) reports the examples of COs in English with weather verbs in (i)– (ii). I think that these cases should also be treated as cases expressing the degree or intensity of the thunder or rain, parallel to the examples above. That is, (ii) cannot be interpreted as *It rained (absolutely) tremendously, as Dixon himself notes, but its meaning rather corresponds to It rained a lot.

212   

   Case Study 1: V todo lo Vble

(71) Ayer llovió todo lo llovible. (Sp) yesterday rain.3sg.pst all lo rain.ble 1. ?? / * Yesterday it rained everything that could rain: freezing rain, acid rain, sleet, etc. 2. Yesterday it rained as much as possible / up to the highest possible extent. (72)

Esta semana ha nevado todo lo nevable. (Sp) this week have.3sg.prs snow.part all lo snow.ble 1. * This week it has snowed everything that could snow: ??? 2. This week it has snowed as much as possible / up to the highest possible extent.

With respect to V todo lo Vble, they seem to pattern with unergative verbs of the estornudar-type, i.e. they cannot generally take COs, and therefore they do not accept a list reading. This is in fact expected, given our knowledge of the world, e.g. only snow can snow. An exception would be non-weather verb llover, in which case it can have the list reading with todo lo Vble, as shown in (73).58 (73)  Me llovió todo lo llovible: problemas, desgracias, regalos, me.dat rain.3sg.pst all lo rain.ble problems misfortunes presents ofertas, … (Sp) offers  ‘Everything possible came over me: problems, misfortunes, presents, offers …’ We can conclude that all unaccusative verbs, whether telic or atelic, admit a cognate element in object position resulting in the expected list reading. The impossibility of this interpretation with some verbs is due either to the intervention of some other independent grammatical requirement (e.g. palidecer ‘turn pale’) or to conceptual semantics (e.g. snow ‘nevar’).59 In addition, atelic unac-

(i) It thundered the most ear-splitting cracks of thunder that I’ve ever heard. (ii) It rained an absolutely tremendous storm while we were on holiday. 58 Note that for all these cases, palidecer ‘turn pale’, sonrojarse ‘blush’, nevar ‘snow’, since only the degree reading is available, an implicational analysis is completely banned, thus providing further evidence against an analysis of the degree interpretation as an inference from the list reading. 59 It may be that this second option is due to some grammatical reason or that it is language specific, if it is related to the fact that Spanish has so few COs in contrast to languages like



Unaccusative verbs   

   213

cusative verbs admit the sequence todo lo Vble in the presence of their internal argument, giving rise to a degree reading. Hence, assuming as I have assumed, that there has to be an implication of an originator involved in the derivation of -ble forms, a first obvious question is where such an implication comes from in the case of unaccusative bases. A second question has to do with the source of variation in the interpretation of some verbs (florecer ‘bloom’, crecer ‘grow’) but not others (ocurrir ‘happen’, llegar ‘arrive’), which is unrelated to the conceptual semantics of lo (palidecer ‘turn pale’). And still a third more general question has to do with the status of the NP todo lo Vble with unaccusatives. For unergative verbs, I have suggested that it may be analyzed as an argumental cognate object in the list reading or, not without problems, as an adverbial cognate object in its degree interpretation. I will start in 5.1 by investigating this last question, namely the possibility of having COs with unaccusative verbs. This will in turn answer the second question on the various interpretations. The first question will be postponed until section 6.

5.1 Cognate objects with unaccusative verbs In section 4, I have provided an analysis of todo lo Vble as the realization of an argumental CO in the list reading. Since this interpretation is also available with most unaccusative verbs, I would like to consider the possibility of having COs with unaccusative verbs. As far as I can tell, leaving aside exceptional cases like die a terrible death, such a possibility has only been discussed in the literature for English. It has been generally assumed that the cognate object construction serves as a diagnosis for the unergative-unaccusative distinction (e.g. Keyser and Roeper 1984; Massam 1990; Levin and Rappaport Hovav 1995; Macfarland 1995): only unergative verbs can appear in this construction, as the contrast in (74)–(75) illustrates. This different behavior has generally been related to the fact that the position of internal argument is already occupied in the case of unaccusative verbs. An additional position for the CO is banned by Case theory: whereas unergatives

English. That is, it strikes me that in English one can sneeze a big sneeze, but one cannot estornudar un gran estornudo in Spanish. On the other hand, the fact that all cases of sneeze a sneeze appear with adjectives modifying the degree or intensity of the sneezing (a big sneeze or the most tremendous sneeze I have ever heard, the latter reported in Dixon (2005)) is very suspicious and, in my view, hints at a parallelism with Spanish estornudar todo lo estornudable.

214   

   Case Study 1: V todo lo Vble

have the potential to assign accusative Case to a complement, unaccusatives do not. (74) a. Malinda smiled her most enigmatic smile. (Levin and Rappaport Hovav 1995: 40) b. Last night the candidate slept a restless sleep. (Macfarland 1995: 19) (75) a. *The actress fainted a feigned faint. (Levin and Rappaport Hovav 1995: 40) b. *She arrived a glamorous arrival. (Levin and Rappaport Hovav 1995: 148) Next, I briefly review two works that report the existence of COs with unaccusative verbs in English, which challenges this well-established generalization. As I will show, a proper understanding of these data will be relevant for the analysis of the construction V todo lo Vble, as discussed in 5.1.2.

5.1.1 Unaccusative COs in English Kuno and Takami (2004) provide examples of cognate object constructions with typical unaccusative verbs in English, repeated in (76), that challenge the validity of the CO as an unaccusativity diagnostic, and therefore the well-established generalization about the occurrence of cognate objects – what these authors call the Unergative Restriction on the Cognate Object Construction. According to Kuno and Takami, all the verbs in (76) are unaccusative because they express non-volitional events that denote a change of state or location of their theme subject. (76) a. The tree grew a century’s growth within only ten years. b. /? The gale blew its hardest blow yet in the next hour.60 c. The stock market dropped its largest drop in three years today. d. The stock market slid a surprising 2 % slide today.

60 These are Kuno and Takami’s (2004) judgments.





Unaccusative verbs   

   215

e.  Stanley watched as the ball bounced a funny little bounce right into the shortstop’s glove.

f. The apples fell just a short fall to the lower deck, and so were not too badly bruised. (Kuno and Takami 2004: 116) In essence, these authors use these data to argue against the Unergative Restriction on the Cognate Object Construction, and suggest a Functional Constraint on the Cognate Object Construction whose first part is repeated in (77).61 (77) The Functional Constraint on the Cognate Object Construction A. In the cognate object construction, (i) the intransitive verb must represent an activity or event involving a temporal process, and (ii) the object NP must represent a specific state or event that belongs to the set of possible states or events resulting from the activity or event. N.B. ‘die a … death’ is an exceptional construction that has historically a different derivational process from the ordinary cognate object construction, and its cognate object represents the manner, rather than the result, of the event described by the verb. That is, such a constraint states that only activities or processes, i.e. durative events, can take a CO, which accounts for the ungrammaticality of e.g. *She arrived a glamorous arrival in (75); besides, the CO must be a possible resultant state of that activity or process, which explains the contrast between (78a) and (78c). According to Kuno and Takami (2004: 124), the ungrammaticality of the CO in (78a) is “undoubtedly due partly to the fact that its cognate object describes the manner, rather than the resultant event/state, of the falling of the apples.” (78) a. * The apples fell a smooth fall. (Levin and Rappaport Hovav 1995: 148) b. ??The apples fell a short fall.62

61 The second part of their constraint essentially claims that, as a marked structure, the use of COs “must be justifiable” (Kuno and Takami 2004: 130). 62 The native speakers I have consulted judge sentence (78b) as fine as (78c).

216   

   Case Study 1: V todo lo Vble

c.  The apples fell just a short fall to the lower deck, and so were not too badly bruised.

On the other hand, these authors explain the marginality of (78b), which contrasts to (79), in terms of some conventional facts, as stated in (80). (79) a. The stock market dropped its largest drop in three years today. b. The stock market slid a surprising 2 % slide today. (80) “when the stock market drops (or rises), we conventionally pay attention to the resulting event with respect to how big the drop (or increase) has been, while when apples fall, we do not conventionally pay attention to how long or how short the fall has been” (Kuno and Takami 2004: 125). They continue to say that (78c) is acceptable “because it explains contextually why the speaker has chosen to mention the event resulting from the apples falling” (Kuno and Takami 2004: 125). I see different problems with this proposal. First, note that, as stated, this constraint does not seem to follow from any more general property of the grammar, nor does it establish any relationship between its two subparts (i)–(ii), i.e. being durative and expressing result. Second, it is unclear to me that the examples in (78) contain two different instances of a cognate object, manner and result. Note that the same kind of contrast could be applied to the sentences in (81), but they are both fine. (81) a. John smiled a beautiful smile. b. John smiled a short smile. It seems to me rather stipulative to say that (78a), but not (81a) describes just the manner of the event, although they both can be said to express the parallel meanings in (82). (82) a. The apples fell smoothly. b. John smiled beautifully. To me, all sentences in (78) contain a resultant event fall, the difference being that (78a) expresses the manner in which this resultant event is achieved, which corresponds to the interpretation of a standard cognate object construction, and this



Unaccusative verbs   

   217

is not possible with unaccusative verbs because of Case theory; whereas (78b–c) express the degree or extent achieved by the resultant event of falling. That is, in these two examples the cognate phrase materializes the degree argument of these predicates, and therefore they are not ruled out by Case theory. The examples in (81) would get similar interpretations, so that their different behavior cannot be based on the interpretation of the cognate object as manner or result, but must be due to the fact that they are different types of cognate configurations. Kuno and Takami exclude achievement verbs, occur or arrive, because they do not involve a temporal process, and their cognate objects cannot expres a result. However, they do not distinguish the fine unaccusative cognate object structures from regular unergative cognate objects. This is problematic, because it cannot explain the differences between them. For instance, Kuno and Takami argue with Macfarland (1995), and against Jones (1988), that passive sentences are possible with cognate objects, as exemplified in (83). According to my informants, the cognate element appearing with unaccusative verbs can never be passivized, as shown in (84).63 (83) a. Pictures were taken, laughs were laughed, food was eaten. (Kuno and Takami 2004: 128) b. It is a smile that could be smiled by the whole country. (N. Woodsworth, Financial Times, IX. Cited in Macfarland 1995: 112) (84) a. *It is a century’s growth that was grown by the tree. b. *Large drops were dropped by the stock market. I will provide additional differences between COs and unaccusative cognate phrases in the discussion of Nakajima’s (2006) work. For now, we can already conclude that, perhaps even more importantly, a deeper analysis of these facts is necessary before giving up the reliability of the CO construction as an unaccusativity diagnostic (Levin and Rappaport Hovav 1995: 40), as Kuno and Takami have proposed.64

63 But see the discussion in section 6, where Spanish data show that a bound degree or measure phrase selected by unergative verbs can become the external argument of a sepassive sentence. 64 This reliability has been challenged, to my knowledge, only by the verb die in different languages. Macfarland (1995) devotes a whole chapter to show that it is in fact unergative, and therefore no longer a problem. Note that any analysis of COs must face the analysis of this verb as being exceptional in one sense or other.

218   

   Case Study 1: V todo lo Vble

Nakajima (2006) undertakes the task of developing a preliminary structural analysis of these data, and suggests that they must be regarded as adverbial COs (e.g. Pereltsvaig 1999 or Rodríguez Ramalle 2003), as opposed to argumental COs. The latter would appear in an object position – already filled, and therefore banned, in the case of unaccusatives –, whereas the former would be inserted as adjuncts – a position still available for unaccusatives. As for the different behavior of unaccusative verbs, this author accounts for the contrast between (76) and (85) as a difference in Case properties: verbs like grow would have the potential to assign accusative Case to an adverbial cognate object, which verbs like occur lack. (85) a. *The coup d’état occurred a sudden occurrence in the Middle East. b. *The train arrived a punctual arrival.

c. *The submarine emerged an abrupt emergence.



d. *The sun appeared a bright appearance this morning.

e. * He existed a dangerous existence. (Nakajima 2006: 681) Nakajima (2006) provides some evidence based on the corresponding nominalizations, in (86), where we find the preposition of “which is the realization of accusative Case in nominalization”. (86) The tree’s growth of a century’s growth in only ten years surprised me. (Nakajima 2006: 681) For this author, also the fact that in languages like German, COs appear with accusative Case as in (87) would favor his analysis. (87)

Maria ist einen weiten Sprung gesprungen. Maria is a wide jump jumped ‘Maria jumped a long jump.’ (Nakajima 2006: 681)

Nakajima leaves open the question of how accusative case would be assigned to these adverbial objects, although two possibilities are suggested. Either accusative case is assigned to adverbial COs in some maximal functional projection, along the lines of Pereltsvaig’s (1999) analysis of Russian and Hebrew adverbial COs; or some sort of ‘pseudo’ accusative case is assigned by the verb under linear adjacency.



Unaccusative verbs   

   219

There are some theoretical and empirical shortcomings with Nakajima’s analysis. First and foremost, such an analysis runs against serious difficulties related to the well-known Burzio’s generalization, stated in (88). (88) Burzio’s Generalization All and only the verbs that can assign a θ-role to the subject can assign ­accusative Case to an object. (Burzio 1986: 178) Although there is currently extensive agreement that this generalization must be stated in some other terms (see e.g. Reuland 2000; Alexiadou et al. 2004), its exact formulation does not affect the theoretical problem raised by having unaccusative verbs assign accusative case. On the one hand, Nakajima does not explicitly clarify the reasons for having a subset of English unaccusative verbs assign some kind of (pseudo-)accusative Case, whereas the rest of unaccusatives cannot. Furthermore, it is also unclear why Case should be available only for the cognate object, but not for the theme object selected by the verb.65 More­ over, analyzing these facts as stemming from some particular – and mysterious – Case property of an arbitrary subset of unaccusative verbs, without taking into account their semantics, simply loses another well-known – though probably not yet well-understood – and pervasive generalization: the use of accusative Case for measure, degree or extent phrases in different languages,66 from Latin and Greek (e.g. so-called accusative of extent or accusative of duration) to German (e.g. accusative is used for time, duration, and measure) or Finnish (e.g. accusative is associated with duration, frequency, and measure phrases).67, 68 As pointed out in Kratzer (2002: 38), who attributes it to Wechsler and Lee (1996), “it cannot

65 A possibility, not mentioned in Nakajima, would have been to reclassify the set of atelic unaccusative verbs as unergatives, as has already been proposed in the literature (e.g. Levin and Rappaport Hovav 1995; Pérez Jiménez and Moreno Quibén 2003, 2005; Pérez Jiménez 2006). Still, this would not explain the differences in the interpretation of the cognate object, i.e. manner versus degree. 66 Perhaps for this reason, some authors have included this type of complements, with a semantic value of (temporal or spatial) extent, within the group of cognate objects (e.g. Cano Aguilar 1981). 67 See Maling (1993), who also includes COs in the set of “adverbials” that get assigned accusative Case. 68 As is well-known, accusative Case is a marker of telicity in Finnish (Kiparsky 1998). See also Tenny (1994) or Svenonius (2002) for non-object constituents with accusative Case that affect the aspectual properties of events. As I will show, English unaccusative verbs with a CO become telic.

220   

   Case Study 1: V todo lo Vble

be an accident, that many languages use accusative case for measure and degree phrases, but not for other kinds of adjuncts” (see also Pereltsvaig 2000 on accusative adverbials). As for the empirical problems, it has been shown that standard COs can be questioned by the interrogative what or how (e.g. Jones 1988; Massam 1990), as exemplified in (89)–(91). Indeed, for Nakajima (2006:  681), COs of unergative verbs are all ambiguous, so that all COs can have a result reading or a modification reading expressing manner, duration, frequency, or extent. This ambiguity arises from the two functions of COs: “that of a measuring-out argument, which changes the eventuality of the verbs”, and that of a “modifier adjunct, which just elaborates the meanings of the verbs and does not alter their eventuality.” (89) a. The resting girl dreamed a secret dream. (Nakajima 2006: 680) b. What / *How did the resting girl dream? A secret dream / *Secretly. (90) a. Harry lived an uneventful life. (Jones 1988: 89) b. What / How did he live? An uneventful life / Uneventfully. (91) a. John died a gruesome death. (Jones 1988: 89)

b. *What / How did he die? *A gruesome death / Gruesomely.

However, unaccusative COs can only be an answer to the very specific how much, as shown in (92).69 (92) a. The tree grew a century’s growth within only ten years. (Kuno and Takami 2004: 116) b. *What did the tree grow?

c. *How did the tree grow?

d. How much did the tree grow? A century’s growth.

69 See also Nakajima (2006: 677), who uses the interrogatives what kind of and how much/ how far to argue that “all these facts indicate that the cognate objects in (4) [my (76) above] are not argumental, but adverbial.”



Unaccusative verbs   

   221

Although adverbial COs can express different types of modification, it cannot be a mere coincidence that COs with unaccusatives only and always express one type of adverbial modification, namely measure, degree or extent. As I have already mentioned, according to Pereltsvaig (1999: 547), adverbial COs can only have a manner interpretation. Interestingly, the kind of entailments exhibited by manner adverbials (Levinson 2010) are semantically distinct from the ones we find with unaccusative cognate objects, as illustrated in (93). (93) a. The dog ran quickly. →The running event was quick. (Levinson 2010) b. The tree grew a century’s growth. 1. ⇏ The growing event was a century’s growth 2. → The extension of the growing event was a century’s growth Nakajima (2006: 679) already mentions that the unnaccusative CO in (92) is an “extent DP” that refers to the “resultant extent of the action” expressed by the verb, but this information has no value beyond the mere descriptive use in this work, i.e. his proposal is completely unrelated to the meaning conveyed by these COs. In fact, Nakajima concludes that “unaccusative verbs may take cognate objects with adverbial meanings (manner, extent, etc)” (Nakajima 2006: 683), i.e. he does not restrict the semantic type of unaccusative COs. Examples like (92c) already show that unaccusative COs cannot express manner. According to Macfarland (1995) or Kuno and Takami (2004), the interpretation of an argumental CO is a “created or result object”. If we include unaccusatives in this statement, as Kuno and Takami have suggested, we cannot account for the contrast in (94). Whereas (94a) corresponds to an unergative CO construction, the unaccusative CO in (94b) is ungrammatical, because, the argument goes, it does not express a resultant state. (94) a. John smiled a beautiful smile. b. *The tree grew a beautiful growth. I argue that this contrast cannot be due to the expression, or lack thereof, of a resultant state or event. Whenever there is an entity growing, there is some change of state and some growth of that entity as a result. And this result can be expressed in the form of a measure argument, as in (95a), or as in (76a), repeated in (95b) for convenience. (95) a. The tree grew three inches.

222   

   Case Study 1: V todo lo Vble

b. The tree grew a century’s growth within only ten years. This can be related to another difference between unergative COs and unaccusatives that has to do with restrictions on the type of cognate noun. It has often been pointed out that, in standard CO structures, the noun appearing as object must be cognate with the verb, which explains the contrast in (96).70 (96) a. He smiled a sardonic smile. b. * He smiled a stupid grin. (Wanner 2000: 89) However, Pereltsvaig (1999:  547) indicates that adverbial COs do not need to be strictly root-identical with the verb. That seems true of English unaccusative COs, as well, which would argue for their adverbial status. As pointed out by Nakajima (2006), English unaccusative verbs allow a non-cognate object, as shown in the (b) examples (97)–(98). Crucially, though, this object denotes some kind of measure or extent, as further evidenced by the approximate paraphrases in the (c) examples, which runs against Pereltsvaig’s (1999) semantic restriction on adverbial COs.71 (97) a. The tree grew a century’s growth within only ten years. b. The tree grew a century’s expansion. c. The tree grew by a century’s expansion. (Nakajima 2006: 676) (98) a. The apples fell just a short fall to the lower deck, […]. (Kuno and Takami 2004: 116)

70 This restriction excludes Massam’s (1990) transitivizing objects in (i), which are different from pure cognate objects (they would correspond to Hale and Keyser’s HO). See Massam (1990) for this distinction. (i) a. Bernadette danced the Irish jig. b. Tosca sang an aria. Notice that Macfarland (1995: 90) rejects the existence of the restriction exemplified in (96) above on the face of examples like the following. (ii) “Let’s wipe our brows and smile a graduation grin,” said Ms. Ator of Reisterstown. (T.W. Waldron, The Baltimore Sun, 16B) 71 Recall that although Nakajima mentions that these adverbial cognates are extent phrases, this fact is completely dodged and not exploited to account for the data.



Unaccusative verbs   

   223

b. The apples fell the length of my arm. c. The apples fell {by/to} the length of my arm. (Nakajima 2006: 676) A final significant aspect of these unaccusative COs in English, not explicitly mentioned in Kuno and Takami or Nakajima, is that, despite the fact that all verbs appearing in this construction show variable behavior with respect to telicity, all of them become telic in this construction, as proven through the application of the frame versus durative adverbial test (99). Interestingly enough, Pereltsvaig (1999: 547) states that adverbial COs do not delimit the event denoted by the main predicate. And recall that Nakajima (2006: 681) maintains that the CO “does not alter” the eventuality of the predicate in its adverbial reading. The contrast in (99) shows that the presence of the cognate phrase determines the telicity of the predicate, which clearly contradicts Nakajima’s analysis. (99) a. The stock market slid in one day / for one day. b. The stock market slid a surprising 2 % slide in one day / *for one day. All these facts suggest that the type of CO is not only relevant, but decisive for the acceptability of such unaccusative cognates. Therefore, any analysis of these data that does not take it into account will be doomed to fail. Before moving towards the analysis of unaccusative -ble forms, I will present and discuss an as yet unnoticed correlation between the English construction with unaccusative verbs just described and the structure V todo lo Vble, which will shed light on the nature of V todo lo Vble. Both constructions are rather unusual in both languages,72 and they occur with the same type of unaccusative verbs conveying the same kind of meaning. This connection should reinforce the proposal I will present that the CO with unaccusative verbs is the materialization of a scale.

72 There are languages where the CO can appear with a wide range of predicates. For instance, Pereltsvaig (1999) reports that Hebrew COs can appear with unergatives, transitives, passives, ditransitives, unaccusatives, psych-verbs, adjectival predicates, stative verbs, and individuallevel predicates. My point above is not that having COs with unaccusatives in general is unusual, but that it is so for languages like English – or Spanish – where COs are supposed to be restricted to unergative verbs.

224   

   Case Study 1: V todo lo Vble

5.1.2 Unaccusative COs in English and Spanish: A crosslinguistic parallelism I have described the properties of unaccusative COs with English verbs. Let us now see the relevance of these data for the analysis of todo lo Vble with Spanish unaccusative verbs. On the one hand, the English acceptable cognate objects contain some measure, which often expresses an extreme degree, which corresponds exactly to one of the interpretations of the construction V todo lo Vble, the degree reading to be precise. That is, the interpretation of the unaccusative English verbs drop or grow corresponds to a measure or degree reading, which is the additional interpretation that the corresponding Spanish unaccusatives bajar ‘drop’ or crecer ‘grow’ accept. On the other hand, the English data display the same type of contrast that we have seen with the Spanish data. Namely, in English the appearance of a CO with unaccusative verbs distinguishes between occur/arrive-type verbs and grow/drop-type verbs, which is essentially the same distinction established by the possible interpretations of the construction V todo lo Vble between ocurrir/ llegar-type ‘occur/arrive’ verbs and crecer/florecer-type ‘grow, bloom’ verbs. ­Furthermore, in both languages, the cognate phrase makes the predicate telic, or better said, it makes the telicity of the predicate explicit. Notice that the examples in (100)–(101) are essentially symmetric translations. (100) a. The tree grew a century’s growth within only ten years. b. El árbol creció todo lo crecible en tan solo diez años. (Sp) the tree grow.3sg.pst all lo grow.ble in so only ten years ‘The tree grew to its highest possible degree of growth in only ten years.’ (101) a. The stock market dropped its largest drop in three years today. b. La bolsa ha bajado todo lo bajable / the stock.market have.3sg.prs drop.part all lo drop.ble ha caído todo lo caíble en tres años hoy. (Sp) have.3sg.prs fall.part all lo fall.ble in three years today ‘The stock market dropped down to the lowest possible degree in three years today.’ This kind of modification is not possible with the other subset of unaccusative verbs, as illustrated in (102), because they are punctual or instantaneous events. Exactly the same facts obtain in Spanish, as shown by the parallel examples in (103).



Unaccusative verbs   

   225

(102) a. Something {grew / fell} {a little bit / to some extent / a lot}. b. *Something {occurred / arrived} {a little bit / to some extent / a lot}. (103) a. Algo {creció / cayó} {un poco / hasta cierto something  grow.3sg.pst fall.3sg.pst  a bit until certain punto / mucho}. (Sp) point much ‘Something {grew/fell} {a Little bit / to a certain extent / a lot}.’ b. *Algo {ocurrió / llegó} {un poco / hasta cierto  something  occur.3sg.pst arrive.3sg.pst  a bit until certain  punto / mucho}.  point much This parallelism requires a joint explanation, which must be related to the event structure of these verbs,73 i.e. degree achievements, and, crucially, also to the type of object they take in these particular constructions, namely a measure, degree or extent phrase. In both languages, the same set of atelic unaccusative verbs, i.e. degree achievements, admit a CO that expresses some degree or measure. That is, the cognate object must be analyzed as a degree phrase (DegP). That the expression of a degree is crucial in the English examples is further illustrated by the contrasts in (104)–(105). That it is equally necessary in Spanish is exemplified in (106)–(107). (104) a. The stock market slid a surprising 2 % slide today. b. *The stock market slid a surprising slide today. (105) a. The stock market dropped its largest drop in three years today.

b. *The stock market dropped a bad drop today.

73 De Swart (2007: 36) does acknowledge the key role of the verb’s aspectual structure for the analysis of unaccusative verbs with cognate objects in his Aspectual Restriction on the Cognate Object Construction, stated in (i). (i) Only atelic verbs, i.e., verbs without an inherent endpoint, can occur in the cognate object construction. Still, such a restriction, apart from being stipulative, cannot explain all the facts, since the type of cognate element also plays a role, whether a degree or measure or a regular direct object.

226   

   Case Study 1: V todo lo Vble

(106) a. El árbol creció todo lo crecible. (Sp) the tree grow.3sg.pst all lo grow.ble ‘The tree grew as much as it could grow.’ b. * El árbol creció lo crecible. the tree grow.3sg.pst lo grow.ble (107) a. El Euribor bajó todo lo bajable. (Sp) the Euribor drop.3sg.pst all lo drop.ble ‘The Euribor dropped down as much as it could / to its lowest possible degree.’ b. * El Euribor bajó lo bajable. the Euribor drop.3sg.pst lo drop.ble If what we have is a degree cognate phrase, three immediate questions arise: first, the nature of this DegP with verbal predicates; second, where this DegP is realized, whether as an argument or as an adverbial adjunct; and third, how we can account for its cognate nature. An additional important question is how this analysis extends to the degree interpretation of unergative verbs. Other key questions have to do with the internal structure of the DegP or accusative Case assignment. In the next section, I will first review some of the main points that have been introduced in the literature on gradable predicates and degree achievements in particular, which will be relevant for my analysis of these data. In addition, I will provide extra support for a treatment of todo lo Vble (and English unaccusative cognate objects) as a DegP, before discussing the above stated questions in section 5.3 where I will present the analysis of these forms.

5.2 On degrees74 It is generally assumed that the possibility for a predicate to license degree quantification depends on the (inherent) semantics of the quantified element (Sánchez

74 On the one hand, this section is not intended as a review of the literature on degrees and gradable predicates, which goes back to the seminal work by Bolinger (1972) or Cresswell (1977). On the other hand, most of the works dealing with gradable predicates have been carried out within semantics. In this brief section, I will abstract away from the technicalities of their semantic formalizations, and will concentrate on the main contributions of recent works to the understanding of the general key notions underlying the concept of gradability.



Unaccusative verbs   

   227

López 1999: 1094). It is probably for this reason that most works assume that gradability is somehow lexically encoded. For instance, Piñón (2000) suggests that degree achievements “have a scale of change built into their meaning”. According to Rappaport Hovav (2008: 17), “verbs which denote events of scalar change are those which lexically specify a scale” where a scale and a scalar change are as defined in (108). (108) a. “A scale is an ordered set of values for a particular attribute.”

b. “A scalar change is one which involves an ordered set of changes in a particular direction of the values of a single attribute and so can be characterized as movement in a particular direction along the scale.” (Rappaport Hovav 2008: 17) Also Kennedy and Levin (2008: 167) assume that “the variable aspectual properties of DAs [degree achievements] derive from the semantic properties of the adjectival part of their (decomposed) lexical meanings”, and further that “gradable adjectives in English directly lexicalize measure functions”, which are “functions that associate objects with ordered values on a scale, or degrees.” Thus, the most usual semantic analysis of gradable predicates is some version of (109), implemented as (110). (109) “They are analyzed as expressions that map their arguments onto abstract representations of measurement, or scales. Scales have three crucial parameters, the values of which must be specified in the lexical entry of particular gradable predicates: a set of degrees, which represent measurement values; a dimension, which indicates the property being measured (cost, temperature, speed, volume, height, etc.); and an ordering relation on the set of degrees, which distinguishes between predicates that describe increasing properties (like tall) and those that describe decreasing properties (like short). (Kennedy 2004: 3) (110) “Gradable predicates have (at least) two arguments: an individual and a degree. Gradable predicates further contain as part of their meanings a measure function and a partial ordering relation such that the value of the measure function applied to the individual argument returns a degree on the relevant scale that is at least as great as the value of the degree argument.” (Kennedy 2004: 4)

228   

   Case Study 1: V todo lo Vble

Rappaport Hovav (2008: 18) further notes that “while all dynamic verbs are potentially associated with a scale, (at least in English) with some verbs this is a lexical property and with others this is not.” Among verbs denoting events of scalar change, this author distinguishes three different types, depending on the type of scale they specify. Verbs of directed motion lexically specify a path scale, which can be a multi-point scale, e.g. ascend, descend, or a two-point scale, e.g. enter, exit, come. Change of state verbs lexically specify what she calls a property scale, which can also be multi-point or two-point, e.g. lengthen, open, widen. Finally, incremental theme verbs like read, eat, or build can specify an extent scale. In this case, the scale is not lexically specified, but introduced by the theme object.75 Interestingly, a crucial property of the degree reading of todo lo Vble is that it is only available with the first two types of verbs. With respect to the function of measure phrases, Kratzer (2002: 37) points out that “if a verb is modified by a measure or degree phrase, it is that very phrase that provides the measure for the success of the event.” Also Kennedy and Levin (2008) suggest that an explicit extent phrase with degree achievements of the type (111a) plays the role of an incremental theme in that it determines the telicity of the predicate, as demonstrated (111b). (111) a. The soup cooled in / for ten minutes. b. The soup cooled (by) 5 degrees in / *for two minutes. As I mentioned at the outset, almost all works dealing with degree quantification have been developed within semantics. Among syntactic approaches to this type of quantification I will briefly review the inspiring work by Bosque and Masullo (1998). Working on Spanish, these authors suggest that degree achievements, and gradable predicates in general contain “a gradable component in their sub-lexical structure. That is, their gradability is compositionally determined by lexical structure and not specified by lexical features” (Bosque and Masullo 1998: 12). This proposal is implemented within Hale and Keyser’s framework in the area of verbs. Specifically, they develop the idea that gradability derives from the presence of a hidden argument in the lexical relational structure of the predi-

75 Rappaport Hovav’s scalar predicates essentially correspond to Tenny’s (1994) measuring out verbs: her change of state verbs, route verbs, and incremental theme verbs; and to Kennedy and Levin’s (2008) verbs of variable telicity, which are their degree achievements, (a)telic motion verbs, and incremental theme verbs, respectively. For Kennedy and Levin (2008: 156), they share a function “that measures the degree to which an object changes relative to some scalar dimension over the course of an event.”



Unaccusative verbs   

   229

cate, which is bound by a degree modifier. This means that only verbs containing degree quantification in their sublexical structure will be able to license a quantifier adjunct. To illustrate their analysis, for a predicate like sangrar mucho ‘bleed a.lot’, they suggest the structure in (112), where the complement of an unergative verb is a QP instead of a NP; the Q0 head licenses the quantifier adjunct in that it provides a variable to be bound by the quantifier adjunct. The measure phrase is understood as a predicate of the inner noun, which would be interchangeable with a definite measure phrase like dos litros ‘two liters’ in sangrar dos litros (de sangre) ‘bleed two liters (of blood)’. VP

(112) VP

QP

Vj

QP

sangrar

Q'

muchoi

Q

NP

ei,j

N' Nj

For degree achievements, Bosque and Masullo (1998: 40) suggest that they “incorporate a comparative head in their corresponding LRSs [Lexical Relational Structures].” The structure they propose for a degree achievement like adelgazar ‘slim down’ would be something like (113). In this case, licensing of the QP in adjunct position would be instantiated via spec/head agreement between the maximal projection in the specifier of DegP and the degree head; this way they can still maintain their proposal that a Q head licenses the adjunct.

230   

   Case Study 1: V todo lo Vble

VP

(113) VP spec

QP V'

Vi adelgazar

muchoj DegP Deg'

QPj Deg

AP

t'i

A' A ti

This interesting proposal allows their authors to account for a number of issues within the area of quantification in Spanish, which I will not review here. It seems to me that most of this proposal could be used to account for my Spanish data. However, although it is very appealing and I will make use of some of the insights contained in this piece of research, I will not fully adopt their proposal. At this point it is unclear to me how their analysis could account for the two interpretations, as well as the lack thereof, of todo lo Vble.76 After this brief revision of basic concepts and proposals, we can conclude that, in one way or other, it has been assumed that gradable predicates express or contain some abstract (or hidden) representation of measurement or scale. The main idea I want to put forward is that the cognate constituent in the above discussed structures materializes this scale, i.e. it makes the scale explicit in the form of a degree phrase. More specifically, we can say that the cognate word represents an abstract dimension along which a set of (possible) ordered measure-

76 See also Fábregas (2003) for some theoretically-based criticism to Bosque and Masullo’s proposal. Roughly, he disapproves of the fact that a property that uniquely identifies the category of adjectives, gradability, is attributed to the verbal category. But see Baker (2003: 213), or Cabredo Hofherr (2010: 4) for empirical arguments against the treatment of gradability as the defining property of adjectives.



Unaccusative verbs   

   231

ments of change, i.e. degrees, is built giving rise to a scale, which must be further bound by a quantifier that assigns some value to an argument on that scale. For instance, in the sentences The tree grew a century’s growth or El árbol creció todo lo crecible, there is some entity, the tree, that undergoes some degree of change, made explicit by a century’s and todo, on a scale, a scale of growth or a scale of possible degrees of growing, respectively, in the property expressed by the predicate, rendering the sentence telic. The cognate element is allowed to appear as cognate, i.e. root-identical with the event in the position of a CO because, like COs, degrees denote a resultant extent (Macfarland 1995). Hence, the function of this extent phrase, i.e. the cognate object, is similar to Kennedy and Levin’s (2008) proposal that an explicit extent phrase with degree achievements of the type in (111) plays the role of an incremental theme in that it determines the telicity of the predicate. This is what we find in (114). (114) a. El Euribor ha bajado en / durante dos meses. (Sp) the Euribor have.3sg.prs drop.part in / during two months ‘The Euribor has dropped in / for two monts.’ b. El Euribor ha bajado 2 puntos en /*durante dos meses. the Euribor have.3sg.prs drop.part 2 points in / during two months ‘The Euribor has dropped 2 points in /*for two monts.’ c. El Euribor ha bajado todo lo bajable en / *durante the Euribor have.3sg.prs drop.part all lo drop.ble in / during dos meses. two months ‘The Euribor has dropped down to its lowest possible degree in / *for two months.’ They are not licensed with achievements, because they cannot materialize a scale. Moreover, they already contain a built-in end-result in their event structure, i.e. they are telic (see also Nakajima 2006).77 If the function of a DegP is to delimit the predicate, i.e. making it telic, this possibility is necessarily banned for achievements. In the next section I discuss the details of this proposal that further explain the issues that I brought up at the end of the previous section. The idea is that the DegP of degree achievements materializes a scale in argument position and that its cognate nature derives from these two facts.

77 See also Tenny (1994: 39–40), who argues that COs only occur with verbs that do not have an affected or measuring argument in their basic sense, thus excluding cases like *Meg broke her break.

232   

   Case Study 1: V todo lo Vble

5.3 lo as Deg For the analysis of todo lo Vble in the degree interpretation, I propose that in this case lo materializes a degree, measure, or quantity, not entities as with the individuative lo. In this case, lo is an instance of what Bosque and Moreno (1990) call a quantitative lo, a head pronoun, a variable that denotes a set of quantities, as in No duermo lo suficiente ‘I don’t sleep lo enough’. According to Bosque and Moreno (1990: 34–35), this type of lo can only be modified by a very restricted set of epistemic adjectives like suficiente ‘enough’, necesario ‘necessary’, justo ‘enough’, imprescindible ‘essential’, so that a sentence like Duerme lo necesario ‘S/he sleeps lo necessary’ means duerme lo que es necesario dormir ‘s/he sleeps lo that is necessary sleep’, or more specifically, it can be paraphrased as duerme la cantidad (de tiempo) que es necesario dormir ‘s/he sleeps the quantity (of time) that is necessary sleep’, according to these authors. There are three aspects of this lo that I want to emphasize in relation to todo lo Vble in its degree reading. Consider the meaning of durmió todo lo dormible, which is durmió toda la cantidad (de tiempo) que se puede dormir ‘s/he slept all the quantity (of time) that one can sleep’. First, both expressions denote a quantity. Second, the adjective following lo is a modalized adjective, so that it could be included in the restricted set of modal adjectives selected by the quantitative lo, even though they do not express epistemic modality, but circumstantial. And third, although in the case of duerme lo necesario ‘s/he sleeps lo necessary’, we omit the second instance of the verb, which expresses the scale on which a necessary quantity of time is measured, this scale is contained in the -ble adjective. Thus, as shown in (115)–(116), the two expressions can be said to be exact parallels with respect to their denotations. The only difference is in the characteristic I have presented in the third place, the one that relates to the cognate nature of V todo lo Vble. And, of course, the other difference lies in the explicit quantification todo we find in one construction but not in the other. (115) a. Juan durmió lo necesario. (Sp) Juan sleep.3sg.pst lo necessary ‘Juan slept as much as is necessary.’ b. Juan durmió la cantidad de tiempo que es Juan sleep.3sg.pst the quantity of time that be.3sg.prs necesario dormir. necessary sleep ‘Juan slept the quantity of time that is necessary to sleep.’



Unaccusative verbs   

   233

(116) a. Juan durmió todo lo dormible. (Sp) Juan sleep.3sg.pst all lo sleep.ble ‘Juan slept as much as one can sleep.’ b. Juan durmió toda la cantidad de tiempo que es Juan sleep.3sg.pst all the quantity of time that be.3sg.prs posible dormir. possible sleep ‘Juan slept all the quantity of time that is possible to sleep.’ According to Kennedy (2004:  3), gradable predicates are generally analyzed as “expressions that map their arguments onto abstract representations of measurements, or scales.” What I want to propose, then, is that the cognate element materializes the abstract scale of the event that is used to evaluate its progress (Tenny 1994), so that the event can establish a relation of measurement with its argument, by assigning some value to it on that scale (Kennedy 1999; Kennedy and McNally 1999; Kennedy and McNally 2005; and related work). This is possible when the root is unbounded and thus corresponds to an open or non-finite scale, on which the quantifier todo can set a terminus; this excludes instantaneous events, i.e. achievements, which contain a bounded root. Abstracting away from all (semantic and other) details, the (naïve) syntactic tree in (117) visually represents the kind of relation that is established by a gradable event. (117)

predication argument

property event

degree on a scale

Consider degree achievements first. In a sentence like El árbol creció todo lo crecible ‘The tree grew all lo grow.ble’, we locate the tree’s growth on an open growing scale formed by possible degrees of growing and we assign to it the maximum value denoted by the quantifier todo, resulting in the whole extent.78

78 Another possibility would be to analyze it as a comparative structure, where crecible ‘grow. ble’ would correspond to the standard of comparison (akin to the last part of the English translation ‘as much as it can grow’, against which the degree expressed by todo is compared. The clitic lo would equally materialize a quantity variable. That is, the sequence would be analyzed as an ordering of equality as in as much as comparative structures, which would

234   

   Case Study 1: V todo lo Vble

Thus, lo is a variable ranging over quantities, degrees, extents or some measure, which is restricted by the modalized -ble adjective. The quantifier todo provides a quantized value for the possible degree of change, and closes the scale, thus supplying the scale with an (upper) bound that makes the whole predicate telic. In other words, the quantifier identifies a terminus for the event described by the verb, which leads to a telic interpretation of the phrase.79 This terminal point is fixed non-contextually on the scale, as shown by the contradiction in (118a), which contrasts with the fine context-dependent interpretation of (118b). That is, since the telicity of (118a) is made explicit through the value assigned by the QP, it cannot be cancelled; the sentence in (118b) contains an empty telicity-marking position (a covert degree position in Kratzer (2002), that allows the predicate to give rise to a delimited interpretation), which results in an atelic predicate (Heim and Kennedy 2002). (118) a. # Ha crecido todo lo crecible, pero aún puede have.3sg.prs grow.part all lo grow.ble but still can.3sg.prs crecer más. (Sp) grow.inf more ‘# It has grown as much as it could grow, but it can still grow more.’ b. Ha crecido, pero no del todo. have.3sg.prs grow.part but not of.the all ‘It has grown, but not completely.’ Thus, in (118a), todo lo Vble specifies a (maximal) degree on a scale, therefore it delimits the event, and no other reading is possible. In (118b), there is no explicit degree, so that the sentence is atelic. However, since we have an internally caused change of state verb (a root merged with a little v), it is possible to assign a standard or default endpoint value on the unbounded root, i.e. an open scale, resulting in the telic reading we can obtain in e.g. La planta ha crecido en una semana

relate the actual degree that an object possesses some property to the set of all possible degrees that property can be possessed. 79 Additional support may come from the existence of cases like ocurrió lo inocurrible ‘it occurred lo in.occur.ble’ where negative quantification closes the scale (cf. Fábregas 2005), thus delimiting the predicate. The impossibility of sentences like *ocurrió lo ocurrible, may be then also related to contrasts like un chico infatigable/??fatigable ‘a boy indefatigable / tire. ble’, un comentario inolvidable/??olvidable ‘a comment unforgettable / forgettable’ (cf. Varela 1990). If we can establish such a relationship, a more general important question arises, namely why the presence of negative quantification makes the phrases fine or better in both cases.



Unaccusative verbs   

   235

‘The plant has grown in a week’. A similar behavior can be observed with verbs of directed motion like subir ‘raise’, bajar ‘go.down’, which can be said to have an unbounded root that would specify a path scale. Since quantification happens on an explicit basic scale denoted by the event, we obtain a pleonastic structure. Simplifying things a lot, the meaning of El árbol creció todo lo crecible ‘The tree grew up to the maximum extent that it could grow’ would be something like (119).80 That is, the tree grew up to some extent which is the maximal extension that the tree can grow,81 where todo lo Vble has the function of an incremental theme and delimits the event. (119)  ∃d ∈ Dd [[d= max {the tree can grow to the extent d'} and the tree grew to the extent d] The internal structure of todo lo Vble in the degree reading is the one depicted in (120). (120)

DegP Deg'

QP todo

Deg

aP

lo

…ble…

80 The simplified formula in (113) is only intended to provide the basic meaning, and it is therefore devoid of the complications of a semantic formalization with a maximalization over degrees. The details of such formalization are irrelevant for present purposes. Here, I am merely interested in illustrating the contrast between quantification over degrees in (119) and quantification over entities in (55) above, repeated in (i). See next footnote for the different function of the quantifier todo ‘all’. (i) ∀x [[x is an entity] [x can be danced → Juan danced x]] 81 As Louise McNally (personal communication) points out, the quantifier todo is not a universal quantifier in this case, but it has a maximalizing effect. In El árbol creció todo lo crecible ‘The tree grew up to the maximum extent that it could grow’, todo maximalizes over the set of possible degrees of growing. That is, the function of todo is that of a maximality operator that takes the whole set of degrees in its domain and assigns a single value, the maximal degree. As Louise McNally, and also Richard Larson (personal communication), further point out, this is the same interpretation that we find with so-called amount relatives (Carlson 1977b; see also Heim 1987; Grosu and Landman 1998; McNally 2008) of the type in (i). (i) I took with me the three books that/Ø there were ____ on the table. (Grosu and Landman 1998: 128)

236   

   Case Study 1: V todo lo Vble

Note that this structure corresponds to the extended projection of the adjective (e.g. Kennedy 1999). Recall that for the analysis of todo lo Vble with unergative verbs, I have suggested, following proposals in Hale and Keyser (2002), that the CO can be realized in the lower position when it appears in its extended projection. For adjectives, these authors rule out this possibility on the basis of the unacceptability of COs with adjectives in English. I would like to propose that this is precisely what we find in Spanish, i.e. this possibility does exist in Spanish, so that the structure for El árbol creció todo lo crecible ‘the tree grew all lo grow.ble’ is the one represented in (121). In chapter 5 I will suggest an initial course of action to deepen our understanding of the difference between the two languages with respect to adjectival COs. v

(121)

v

DP el árbol

v creció

DegP Deg'

QP todo

Deg

aP

lo

…ble…

Given this proposal for degree achievements, I want to consider whether this analysis can be extended to the degree reading of unergative verbs, and see how this proposal can be applied to the analysis of the English examples with unaccusative verbs that also express a degree, i.e. Kuno and Takami’s examples. Let us first consider the analysis of unergative verbs in their degree interpretation, since treating them as cases of adverbial COs was shown to be very problematic. Except for a few non-gradable verbs like pernoctar ‘to spend the night’ (Bosque and Masullo 1998: 44),82 unergative verbs, like sleep or travel are gradable, so that they license a degree or measure phrase with poco ‘little’ or mucho

82 As noted in Bosque and Masullo (1998: 44), the reduced number of non-gradable unergative predicates all contain a bounded root denoting a period of time, e.g. veranear ‘to summer’, invernar ‘to winter’. It seems that when the base refers to long periods of time like verano ‘summer’ a non-iterative reading of (i) becomes more plausible and even fine for some speakers.



Unaccusative verbs   

   237

‘a lot’, e.g. Juan duerme poco ‘Juan sleeps little’, Luis viaja mucho ‘Luis travels a lot’. Assuming as I have assumed that a degree interpretation is possible when we have an unbounded root, the analysis should be straightforwardly applied to the relevant unergative verbs. Indeed, whereas Juan durmió todo lo dormible is interpreted as ‘Juan slept as much as it was possible to sleep’, Juan pernoctó todo lo pernoctable cannot mean ‘Juan spent the night as much as it was possible to spend the night, it can only have, if any, an iterative reading ‘Juan spent many nights’. Now, with this analysis, in principle we expect that all verbs that have an unbounded root show variable telicity, as it was the case with degree achievements. Since we have contrasts like El árbol ha crecido en/durante un año ‘the tree has grown in/during a year’ versus Juan ha dormido *en/durante dos días ‘Juan has slept *in/during two days’, it is obvious that there has to be something else that prevents some, though not all, atelic verbs from presenting such variability. As I have advanced above, I think that the dissimilarity comes from a difference in the kind of semantic root, and in the Aktionsart, which is represented in their syntactic configurations, perhaps also through different types of little v heads. For instance, verbs like crecer ‘grow’ merge with a vbecome, take an internal argument and express an internally caused change of state, whereas verbs like dormir ‘sleep’, which express activities, merge with a vdo and have an external argument (see e.g. Harley 1999; Harley and Noyer 2000; Folli and Harley 2004, 2007; McGinnis 2000; Lin 2004; and related work). Thus, a degree interpretation is available for all verbs that contain an unbounded root, which correspond to atelic verbs. Now, I have already mentioned (see footnote 49) that Bosque and Masullo (1998) note that some of the nominals internal to the lexical relational structure of unergative verbs are ambiguous with respect to boundedness. We could entertain the possibility that when the root is unbounded, we get the degree reading; in the list reading, we have a bounded root. Thus, in the list-reading configuration, repeated in (122), lo materializes a variable that stands in an abstract relation with the bounded root. For clarity, I represent this relation with an index. The interpretation of a sentence with a bound root like √bail ‘√dance’ in Bailó todo lo bailable ‘s/he danced all lo dance.ble’ would be something like Bailó todos los bailes que se podían bailar ‘s/ he danced all the dances that could be danced’. Since there is a multiplicity of entities that can be categorized as bailes ‘dances’, quantification over the variable

(i) De pequeño, Juan veraneó todo lo veraneable. (Sp) of little Juan summer.3sg.pst all lo summer.ble ‘As a child, Juan summered many summers (= many times) // Juan summered for long times.’

238   

   Case Study 1: V todo lo Vble

gives us the expected list reading. When the root is unambiguously unbounded, the list reading is simply unavailable, and we can only get the degree interpretation, as in Durmió todo lo dormible ‘s/he slept all lo sleep.ble’. When the root can be bounded or unbounded, but the lo variable cannot range over a minimum number of entities related to the conceptual representation of the bounded root, the list reading is not available. This is what happens in Viajó todo lo viajable ‘s/ he travelled all lo travel.ble’.83 When the root is unambiguously bounded, as in the case of pernoctar ‘spend the night’, we can only get an iterative reading, as with semelfactives (see footnote 83). That is, Juan pernoctó todo lo pernoctable en el hotel ‘Juan spent the night all lo spend.the.night.ble in the hotel’, can only mean Juan pernoctó todas las noches que era posible pernoctar en el hotel ‘Juan stayed overnight all the possible nights’. (122)

vP QP

v

Q' Q

SC DPi lo

FP aP

-ble configuration ModP AspectP vP v

√boundi

83 In this case, the interpretation of the list reading in the proposed structure would be something like Viajó todos los viajes que se pueden viajar ‘s/he travelled all the travels that one can travel’. Note that this looks like the iterative reading that we obtain with semelfactives like Estornudó todo lo estornudable ‘s/he sneezed all lo sneeze.ble’, which would be interpreted as Estornudó todos los estornudos que se pueden estornudar ‘s/he sneezed all the sneezes that one can sneeze’. Since the noun estornudo ‘sneeze’ is a bounded root, it would actually not be licensed in the degree reading configuration, although its interpretation very much resembles it.



Unaccusative verbs   

   239

Notice that this way we can unify the analysis of all verbs. Their different interpretations are the result of the interplay of different factors. These include the type of root they contain, which correlates with differences in Aktionsart, and with a crucial difference in the type of scale/set-of entities along which they can measure out the event; their extended projections, which correlate with different types of lo that range over different domains, entities or quantities. Let us now turn to the English unaccusative examples. I would like to suggest that we have exactly the same structure I have proposed for the degree reading with a crucial difference. As I have already mentioned, according to Hale and Keyser (2002), English does not allow a CO where the underlying base is an adjective (or adjectival root), so that for those cases, they propose that English makes use of an incorporation process, instead of Merge. Thus, in English, only nominal COs are licensed, and that is what we find in examples like The tree grew a century’s growth. Whereas the cognate noun instantiates the basic scale represented by the verb along which the progress of the event will be measured, a century’s provides the quantification necessary to bind a non-overt degree variable. In a sentence like The stock market dropped its largest drop, the degree variable would presumably be realized by the superlative morpheme. Whereas we talk about possible degrees of growing in the case of El árbol creció todo lo crecible ‘the tree grew all lo grow.ble’, here we speak of degrees of growth. And, since degree achievements derive from gradable adjectival bases (Hay et al. 1999, or Fábregas 2003), they can license a DegP, a possibility not available for telic achievements.84 That this proposal is on the right track can be further supported by the contrasting examples in (123)–(124). De Swart (2007: 39) reports that examples like (123a ) are not possible with some atelic unaccusative verbs, because the existence of a transitive counterpart would block them, as in (123b). However, the unaivailability of a CO reading of (123a) contrasts with the perfectly fine example in (124). As I have argued, the puzzle can be immediately solved, as soon as we acknowledge that we are dealing with a degree argument in (124) and a manner CO in (123a). Unaccusative verbs do not license COs, but unaccusative degree achievements do allow for degree arguments to be made explicit, and this is unrelated to the existence of transitive uses of a verb.85

84 A detailed analysis of the English unaccusative construction is necessary to determine, among other things, the realization of an underlying adjective in the verb grow (= become bigger) as a noun in grow a century’s growth. 85 There are indeed some atelic unaccusatives that seem to disallow a CO, like cool. As suggested in Kuno and Takami (2004) and de Swart (2007), this is due to their not having a cognate noun that can express a result.

240   

   Case Study 1: V todo lo Vble

(123) a. #He descended a fast descent. b. He descended the hill /the steps of the temple/the ladder. (124)  That group descended the largest descent in one week. It descended twenty positions. This was achieved by their song “It’s Gonna Be Me” way back in 2000. An important consequence of this analysis of unaccusative cognate objects as degree arguments, if on the right track, is that we can still maintain the CO as a diagnostic test to distinguish between unergatives and unaccusatives. The position where this DegP is realized accounts for various issues. First, it explains the obligatory cognate nature of the adjective in this construction. Second, being in the position of immediate sister of v equates it with an incremental theme, and explains the fact that it does determine the telicity of the sentence. It also explains that no other scale can be introduced in the predication, e.g. *El árbol creció dos centímetros todo lo crecible ‘the tree grew two centimeters all lo grow. ble’. This predicts that the presence of an incremental theme in object position that explicitly measures out the event by providing a measuring scale and a temporal bound (Tenny 1994) should prevent the apearance of the Degree or Extent Phrase. The prediction is borne out, as illustrated by the contrast in (125).86 Whereas la cena ‘the supper’ is an incremental theme, la silla ‘the chair’ is clearly not. (125) a. Gregorio la acercó todo lo acercable, la Gregorio it.acc.f bring.closer.3sg.pst all lo bring.closer.ble the silla. (Sp) chair b. * Paula la comió todo lo comible, la cena. Paula it.acc.f eat.3sg.pst all lo eat.ble the supper Thus, when the only scale is the one expressed by the event itself, an internal argument is compatible with the expression of a measure, as witness the contrast in (126)–(127). The verb acercar ‘bring.closer’ is derived from a gradable root cerc(a) ‘close’, so that a DegP is compatible with the object la lámpara ‘the lamp’. By contrast, for the verb to eat, the direct object introduces an incremental theme, even if unbounded as in (127), since the progress of the event is expressed by the

86 As already mentioned in footnote 24, I use here the clitic construction to avoid the immediate adjacency requirement of degree quantifiers like mucho ‘a lot’ in Spanish (Kovacci 1999: 733f).



Unaccusative verbs   

   241

consumption of the potatoes or the cheese.87 In Rappaport Hovav’s (2008) terms, a verb like acercar lexicalizes an internal property scale, whereas eat has a direct object introducing an external extent scale. This is why the former, but not the latter, can license an explicit measurement scale. (126) a. ? Juan acercó la lámpara todo lo acercable. (Sp) Juan bring.closer.3sg.pst the lamp all lo bring.closer.ble ‘Juan brought the lamp as much closer as one can.’ b. Juan la acercó todo lo acercable. (Sp) Juan it.acc.f bring.closer.3sg.pst all lo bring.closer.ble ‘Juan brought it as much closer as one can.’ (127) * Juan comió patatas / queso todo lo comible. (Sp) Juan eat.3sg.pst potatoes / cheese all lo eat.ble Another prediction of this analysis has to do with transitive verbs that show variable behavior with respect to telicity (Kratzer 2002: 13). We expect them to be able to license a todo lo Vble complement expressing degree that turns the predicate telic, so that its behavior parallels that of intransitive degree achievements. The prediction is borne out, as illustrated in (128)–(129).88 (128) a. Luis limpió el apartamento en / durante una hora. (Sp) Luis clean.3sg.pst the appartment in / during an hour ‘Luis cleaned the appartment in / for an hour.’ b. Luis lo limpió todo lo limpiable en / *durante una hora. Luis it clean.3sg.pst all lo clean.ble in / during an hour ‘Luis cleaned it as much as one could clean it in / *for an hour.’

87 I should note that todo lo Vble in the sentences in (126) sounds rather strange. However, there is a clear and robust contrast between them, which I take as the basis for my argument. 88 The examples in (128)–(129) appear as rather bizarre, I claim due to pragmatic reasons. The key point is that they sharply contrast with examples like (i), which parallels contrasts of the type in (ii). (i) * Luis bebió la limonada todo lo bebible en cinco minutos. (Sp) Luis drink.3sg.pst the lemonade all lo drink.ble in five minutes’ (ii) a. * Luis bebió mucho/ bastante la limonada. (Accomplishment) (Sp) Luis drink.3sg.pst a.lot / quite.a.lot the lemonade b. Luis ordeñó mucho/ bastante la vaca. (Activity) Luis milk.3sg.pst a.lot / quite.a.lot the cow

242   

   Case Study 1: V todo lo Vble

(129) a. Luis ordeñó la vaca en / durante cinco minutos. (Sp) Luis milk.3sg.pst the cow in / during five minutes ‘Luis milked the cow in / for five minutes.’ b. Luis la ordeñó todo lo ordeñable en / *durante cinco minutos. Luis it milk.3sg.pst all lo milk.ble in / during five minutes ‘Luis milked it as much as one can milk it in / *for five minutes.’ This difference can be represented schematically by the basic structures depicted in (130)–(131). The key difference lies in the divergent position of the internal argument, which appears as sister of the root with acercar ‘bring.closer’, and therefore it is non-measuring, but as sister to little v with comer ‘eat’ which corresponds to the incremental theme position. Thus in (130) the gradable verb acercar has not projected its degree argument (signaled by degP) and is ambiguous between a telic and an atelic interpretation. v

(130)

√P

vdo √

acercar

DP

(degP)

la lámpara

‘to bring.closer the lamp’

(cf. Acercar mucho la lámpara ‘to bring.closer a.lot the lamp’)

(131)

v v comer

QP/NP v

(muchas) patatas

‘to eat (many) potatoes’

The analysis also predicts that, since the cognate object and the cognate degree argument are realized in the same position, they are precluded from appearing in the same sentence. The examples in (132)–(134) show that this is a right prediction. In the presence of an internal object that measures out the event, i.e. that introduces an external scale, the expression of some measure or quantity must necessarily apply to that object, i.e. it must necessarily assign a value on that



Unaccusative verbs   

   243

scale and no other; it can never quantify directly over the event, as the expression of a basic scale.89 (132) a. * Luis bailó un baile mucho. (Sp) Luis dance.3sg.pst a dance much b. Luis bailó muchos bailes. Luis dance.3sg.pst many dances ‘Luis danced many dances.’ (133) a. * Luis tosió una tos seca bastante. (Sp) Luis cough.3sg.pst a cough dry plenty b. Luis tosió bastante tos seca. Luis cough.3sg.pst plenty cough dry ‘Luis coughed plenty of dry cough.‘(= He coughed dry quite a lot) (134) a. * Luis soñó un sueño mucho. (Sp) Luis dream.3sg.pst a dream much b. Luis soñó muchos sueños.90 Luis dream.3sg.pst many dreams ‘Luis dreamt many dreams.’

89 In all cases, the ungrammatical sentence becomes grammatical if the quantifier appears preceding the object, but with a different reading. Thus, in a sentence like Luis bailó mucho un baile ‘Luis danced a.lot a dance’, the reading we obtain is that of iteration, not the list reading that results from a quantification over objects. In De joven Luis bailó mucho el tango ‘When young, Luis danced the tango a lot’ it has either an iterative interpretation, or the degree reading. In the latter, el tango is not an incremental theme (it corresponds to ‘dance tango’ as an activity and not ‘dance a tango’ as an accomplishment), as shown by the impossibility of coappearance with a frame adverbial, in (i). (i) De joven Luis bailó el tango *en una tarde / durante una tarde. (Sp) of young Luis dance.3sg.pst the tango  in an afternoon  during an afternoon ‘When young, Luis danced tango *in an afternoon / for an afternoon ‘ 90 Most speakers, including myself, would certainly prefer the light verb construction in (i), rather tan the cognate object construction in (134b). Similarly for (ii) with respect to (133b). (i) Luis tuvo muchos sueños. (Sp) Luis have.3sg.pst many dreams ‘Luis had many dreams.’ (ii) Luis tuvo bastante tos seca. (Sp) Luis have.3sg.pst plenty cough dry ‘Luis had quite a lot of dry cough.’

244   

   Case Study 1: V todo lo Vble

Additional support for the analysis of COs with unaccusative verbs as materializing a degree scale may come from an investigation into COs in other languages. For instance, de Swart (2007: 16–17) reports an example with the Arabic verb understand that takes a second accusatively marked CO that “expresses the degree to which the person in question has understanding” of the relevant subject expressed by the regular internal argument of the verb. This example would parallel the examples in (125a) or (126b), where the degree cognate object expresses the degree that the agent in question brings the chair or the lamp close. (135) Yafhamu l-‘adab-a fahm-an ‘amîq-an (Ar) he.understand the-literature-acc understanding-acc deep-acc ‘He has a profound understanding of literature’ (Lit. He understands the literature a profound understanding) (de Swart 2007: 16, citing Lazard 2003: 8) As for the list reading available with unaccusative verbs of the type Llegó todo lo llegable ‘arrived all lo arrive.ble’ or Creció todo lo crecible ‘grew all lo grow.ble’, they must have the structure in (136). For instance, for crecer, the unbounded root merges with a verbal head and takes an internal argument, where todo lo crecible is realized. At this point, it is unclear to me the process that leads to this configuration, as well as its exact formalization. Thus, todo lo crecible has now the same function as el árbol ‘the tree’ in El árbol creció todo lo crecible ‘the tree grew all lo grow.ble’.91 In (136), the fact that the main verbal root must be root-identical with the -ble form has been symbolized through a common index for expository reasons. The same configuration would derive a sentence with a telic unaccusative verb like Ocurrió todo lo ocurrible ‘it occurred all lo occur.ble’ or Llegó todo lo llegable ‘it arrived all lo arrive.ble’.

91 A sentence like the one in (i) is possible, which clearly shows that the two kinds of todo lo Vble can co-occur. (i) Ha llovido tanto que en el campo todo lo crecible ha have.3sg.prs rain.part so.much that in the field all lo grow.ble have.3sg.prs crecido mucho. (Sp) grow.part much ‘It has rained so much that in the countryside everything that can grow has grown as much as it could grow.’

Unergative and unaccusative -ble   



   245

v

(136)

v

QP Q' Q todo

v

√i

SC DP lo

FP …√ible…

5.4 Case assignment With respect to the Case properties of these CO structures with unaccusatives, a key point in Nakajima’s (2006) analysis of unaccusative COs in English, given the analysis of todo lo Vble / a century’s growth as incremental themes, we can assume that Case is assigned to the cognate constituent by virtue of its being what Wechsler and Lee (1996: 630) call a situation delimiter, i.e. “an extensive measure function which temporally quantifies the event or state depicted by the clause”. These authors propose to extend the domain of direct case assignment so as to include such situation delimiters for languages like Korean or Finnish. I have identified unaccusative cognate objects as degree phrases that delimit the situation (events or states) expressed by the predicate and can be said to satisfy the transitivity property (e.g. if x is equal or greater in degree of growth than y, and y is equal or greater in degree of growth than z, then x is equal or greater in degree of growth than z) that defines measure functions as well as the condition of additivity (the degree of growth of x+y will be the sum of the degree of growth of x plus the degree of growth of y) that classifies measure functions as extensive, so that they can be included in the set of situation delimiters.92

92 See Maling (1993), Maling et al. (2001), or Kiparsky (2001) for the possibility of analyzing accusative Case with measure, duration or frequency adverbial phrases as an instance of structural Case in languages like Finnish or Korean. See Lee (1999) for an LFG analysis of adverbial Case marking in Korean. See also Svenonius (2001, 2002) or McFadden (2004), for semantic or inherent Case assignment.

246   

   Case Study 1: V todo lo Vble

For todo lo Vble with unergatives in the list reading, accusative Case is assigned as to any internal object of a transitive verb.

6 Unergative and unaccusative -ble I have proposed an analysis of the structure todo lo Vble as either a DegP that expresses a quantification over degrees which in turn quantifies over the main event, which is possible with unbounded (atelic) roots; or as a QP that denotes a quantification over entities. Now the question is how the -ble form that appears in this sequence meets the requirements we have established for -ble derivation. This section is devoted to clarify how the analysis developed so far for this construction can solve the main problem we have started this chapter with, namely the occurrence in V todo lo Vble of the independently non-existing forms *dormible ‘sleep.ble’ or *ocurrible ‘occur.ble’. For unergatives like bailar ‘dance’, I have already shown that the corresponding -ble forms not only do not pose a problem, but they are in fact expected. That is, the CO/HO is selected as the external argument of -ble. I showed that the case of dormible ‘sleep.ble’ is more complex, because it cannot take a cognate object as its external argument. Analyzing the string todo lo Vble in the degree reading as containing a degree variable provides us with the answer. Namely, the external argument of -ble with gradable predicates is the degree (measure or quantity) that these predicates select, and that is explicitly realized by the clitic pronoun lo. This degree variable must be bound by a quantifier, which explains the ungrammaticality of *lo dormible.93 This solution is also applicable to the degree reading found with degree achievements, where the external argument of the -ble form is not the internal argument of the main predicate but the degree that lo materializes. In the case of telic unaccusative verbs, the requirement is met by the set of entities that the variable lo instantiates. That means that the requirement of -ble on having an internal theme, also includes incremental themes, which could subsume the notion of location as in pista esquiable ‘skiable piste’ or riu navegable ‘navigable river’, if we understand it as a kind of incremental theme (in fact a path along which the event can be measured out. See chapter 5).

93 This proposal seems to predict that other telic quantified expressions could become the external argument of a -ble adjective, e.g. *creció dos centímetros crecibles ‘it grew two growable centimeters’. The impossibility of such constructions must be related to the gramar of the variable lo, and the fact that it is possibly the only variable that can instantiate such a degree head.



Unergative and unaccusative -ble   

   247

Support for having a degree or quantity as the external argument of a passive construction like -ble may come from the existence of se-passive constructions with unergative verbs, like the ones in (137), where we find agreement between the verb and the measure phrase, showing that a measure phrase can become the external argument of a se-passive construction.94 Also, the attested examples in (138) support the view that these verbs have a degree argument that can be made explicit in the form of a measure or quantity phrase. It functions as an incremental argument in that it makes the predicate telic, exactly as todo lo Vble.95 As shown in these examples, it behaves as an internal argument with respect to Case. (137) a. Se puede {dormir/ viajar} {mucho o poco / una cantidad se can.3sg.prs  sleep.inf travel.inf  much or little a quantity de tiempo}. (Sp) of time ‘One can {sleep / travel} {a lot or not much / a certain amount of time}’ b. No se pueden dormir ocho horas seguidas cuando se not se can.3sg.prs sleep.inf eight hours consecutive when se navega solo. sail.3sg.prs alone ‘One cannot sleep eight consecutive hours when one sails alone.’ (138) a. ¡Cuatro horas, no las viajo ni loca! (Sp) four hours not them.acc.f travel.1sg.prs not.even mad ‘Four hours, I’d never travel them!’

94 Note that this is imposible with the periphrastic passive construction, as shown in (i). (i) * Mucho o poco / una cantidad de tiempo puede ser viajada / dormida. (Sp) much or little a quantity of time can.3sg.prs be.inf sleep.part travel.part 95 But see Bosque (1998: 57), who claims that native speakers either do not accept sentences such as (i), which are very similar to (138); or when they do, as in (ii), the interpretation is no los alcanza ‘does not reach them’. (i) * Esta ranura mide medio centímetro y aquella también lo this slot measure.3sg.prs half centimeter and that too it.acc mide. (Sp) measure.3sg.prs (ii) Su hijo mide los dos metros. her son measure.3sg.prs the two meters ‘Her son is two meters tall.’

248   

   Case Study 1: V todo lo Vble

b. ¡Veinte veces, Juan no las estornuda seguidas! twenty times Juan not them.acc.f sneeze.3sg.prs consecutive ‘Twenty times, Juan doesn’t sneeze them one after the other!’ c. ¡Cinco puntos, la bolsa no los subirá {en/ five points the stock.market not them.acc.m raise.3sg.fut  in *durante} una semana!  during a week ‘Five points, the stock market won’t raise them {in/*for} a week! d. ¡Dos centímetros, tu hijo no los ha two centimeters your son not them.acc.m have.3sg.prs podido crecer {en/*durante} un mes! can.part grow.inf  in during a month ‘Two centimeters, your son can’t have grown them {in/*for} a month!’ More challenging is the source of the implication of an originator for unaccusative verbs. The idea I would like to suggest is that the cognate string that measures out the extension of the predicate provides the structure with a durative component. And it is this durative component that supplies the structure with the implication of an originator. This is perhaps best seen in the case of telic unaccusative verbs. When discussing the general properties of this construction, I proved that the main function of the sequence todo lo Vble seems to be an aspectual reclassification of predicates into accomplishments, by providing a telos and/or the stages that bring durativity to the predicate. For telic predicates this became clear in the light of contrasts like *Llegaron las cartas hasta que se puso a llover ‘arrived the letters until it started to rain’ versus Llegó todo lo llegable hasta que se puso a llover ‘arrived all lo arrive.ble until it started to rain’. Different authors have already associated duration with agentivity, among them Dowty (1979), Tenny (1987, 1994), Piñón (1997) or Bosque and Gutiérrez-­Rexach (2009), in the sense that, intuitively, intentional activities take time, whereas instantaneous events lack the necessary temporal extension to be intentional.96, 97

96 Interestingly, Levin (2007: 10) mentions that “the name accomplishment suggests agency, though Mourelatos (1978) prefers development”. If such a suggestion could be formalized, we could state the relationship between agency and duration on the one hand, and between agency and the formation of -ble in a more formal way. 97 As Joan Mascaró (personal communication) points out, we can utter the sentence Ocurrió todo lo ocurrible ‘occurred all lo occur.ble’ in a situation in which all happens at once, instantaneously. True, although my claim is that the sequence contributes some degree of linguistic duration as evidenced by the results of the tests.



Unergative and unaccusative -ble   

   249

A piece of evidence that duration is what saves the structure comes from the fact that, from the whole set of unaccusative verbs, only a reduced group of nonprototypical atelic verbs can become bases for regular -ble affixation in Spanish, as shown in (139). (139) durable, perdurable (Sp) lasting everlasting It has been argued that these verbs select an additional measure argument by different authors, among them Bosque (1989), Sánchez López (1999), Real Puigdollers (2006). This measure or degree argument would be parallel to the underlying degree argument of atelic verbs that todo lo Vble can materialize. Thus, in (140), both examples have an implicit measure, of duration, like mucho ‘long’ or largo tiempo ‘long time’ in (140a); or specifying some extent or degree in (140b). (140) a. El recuerdo perduró. (Sp) the memory last.3sg.pst ‘The memory lasted long.’ b. El jardín floreció. the garden bloom.3sg.pst ‘The garden bloomed.’ My claim is that it is precisely this measure argument, which can be made explicit or can be left unspecified but implied, as shown in (141), that contributes the structure with the implication of an originator. If this proposal is on the right track, then, these forms are in fact expected.98 (141) a. Esta charla dura veinte minutos. (Sp) this talk last.3sg.prs twenty minutes ‘This talk lasts twenty minutes.’ b. Esta charla dura (mucho). this talk last.3sg.prs (much) ‘This talk lasts (a lot).’

98 It is interesting to note that Maling et al. (2001: 325) seem to tie, at least descriptively, the assignment of accusative Case on degree and frequency adverbials to the presence of an external argument with eventive (non-stative) verbs, thus relating agentivity-eventivity to the occurrence of a measure adverbial.

250   

   Case Study 1: V todo lo Vble

This group of verbs belongs to the set of verbs expressing Continuation of a pre-existing state in Sorace’s (2000, 2004) classification of unaccusative verbs,99 which is based on their choice of auxiliary, be or have, in languages like Italian, German or Dutch. According to Sorace (2000: 868), these verbs are sensitive to the agentivity of the subject in Italian, which weakens their preference for auxiliary be. Also for anticausative verbs, the presence of an external cause in their lexical representation also affects their choice of auxiliary, have or be. There is no longer auxiliary selection in modern Spanish, all verbs select auxiliary have. However, there was such a choice in Old Spanish. Interestingly, verbs like durar ‘last’ always selected auxiliary have (Benzing 1931). If this suggestion is on the right track, the fact that prototypical unaccusatives are non-durative and telic would explain their impossibility of becoming regular -ble bases.100

7 Conclusions The most important contribution of this chapter is certainly empirical. I have presented a set of new empirical data, which should a priori be impossible: -ble forms regularly derived from intransitive verbs. Among the questions posed at the beginning of this paper three were related to the crucial paradox that the formation of such impossible words creates in the system: (i) how these -ble forms can be generated in an appropriately constrained way; (ii) why they should be constrained to appear in a specific construction; and (iii) what the restricting context should be. The basic answer to the three questions is that any form is systematically restricted to appear when all the conditions for insertion of its component parts are met. In the case at hand, for crecible ‘growable’, ocurrible ‘occurrable’, or tosible ‘coughable’ this happens only in certain contexts where the interaction of various general and language-particular grammatical subsystems provide the structure with the elements required for -ble insertion. In other words, it is the grammatical system of the language that constrains the generation of these forms. This, which is possibly the most important contribution of this chapter

99 I wish to thank Jaume Mateu for bringing this work to my attention. 100 The impossibility of a -ble adjective with other unaccusative verbs that also select a measure argument, like costar ‘cost’ *costable ‘cost.ble’ reinforces the suggestion that it is not just having some measure or degree, but that the relevant measure unit is in the dimension of time, specifically duration.

Conclusions   

   251

and perhaps of this volume, may have far-reaching consequences for the nature of morphology and the architecture of grammar in general. Syntactic approaches to morphology such as DM do not overgenerate, but they set the right scenario for the interaction of the various grammatical components. And, since there is no generative lexicon in a syntactic model like DM, we do not expect such regularly derived forms to be stored and function anywhere else and in the same way. The lexicalist model cannot predict this pattern of data, since the grammaticality of these -ble formations cannot only be due to the internal or endoskeletal properties of the base, but must be a function of the external or exoskeletal properties of the base together with the specific nature of the constituent todo lo Vble, thus providing support for syntactic theories of word formation. This proposal has further allowed me to unify the analysis of all verbs in all their readings, so that their different interpretations are the result of the interplay of different factors. These include the type of root they contain, which correlates with differences in Aktionsart, and with a crucial difference in the type of scale/ set-of entities along which they can measure out the event; their extended projections, which correlate with different types of lo that range over different domains, entities or quantities. I have also identified a kind of cognate object construction in English and Spanish that does not seem to correspond to any of the two standard cognate object constructions, the argumental and the adverbial, as these have been described in the literature. The parallelism between the two languages reinforces my proposal that the cognate phrase with unaccusative verbs is the materialization of a degree argument. More importantly, my proposal, which shows that unaccusative cognate objects pattern with incremental themes and other measure phrases with respect to their syntax, semantics and case properties, allows us to maintain the cognate object structure as a diagnosis to distinguish between unergatives and unaccusatives. And it is also compatible with the properties of adverbial cognate objects described in Pereltsvaig (1999), as well as with the properties of standard cognate object constructions. The analysis of V todo lo Vble just proposed supports the grammar of -ble presented in this volume, although there are still aspects of this complex structure that certainly need to be scrutinized more carefully or have not been dealt with yet, and new questions arise from my analysis. Among them, the exact way in which duration contributes to the originator interpretation, the formal instantiation for the materialization of a cognate dimension, be it a scale or a range of entities, or the detailed analysis of the internal structure of the degree phrase with English unaccusative verbs, where degree quantification can come in the form of a superlative or in the form of a quantity-denoting lexical expression.

252   

   Case Study 1: V todo lo Vble

There is still a key question that remains to be raised, namely, why this structure is possible in Spanish but it is not in all the other Romance languages, although they do have a productive process of –ble formation. I will discuss this question in chapter 5, where I will point to further correlations and possible extensions of this proposal.

4 Case study 2: N-ble 1 Introduction The main goal of this chapter is to provide an analysis of the internal structure of  denominal -ble forms in English, Catalan and Spanish that allows for a unified analysis of all -ble forms. These data are deviations from regular forms, i.e. those that select verbal bases. Although the existence of denominal -ble has been acknowledged in the descriptive literature on Spanish word formation (e.g. Rainer 1999), the theoretical implications for their being possible bases for -ble in a (relatively) productive way have not been explored yet. Leaving aside lexicalized forms such as saludable ‘healthy’, both Catalan and Spanish -ble can select nouns denoting posts, positions or functions, as in (2). These are interpreted as ‘that can become N’, i.e. they express a change of state. In English, however, although there are some -ble forms derived from nominal bases like saleable, -ble cannot productively attach to nouns. (1) a. témer / temible (Cat) fear.inf fearsome ‘to fear / fearsome’ b. resistir  / resistible resist.inf resistible ‘to resist / resistible’ c. assequible achievable (2) alcaldable, bisbable, ministrable, papable, presidenciable, rectorable  (Cat) mayor.ble bishop.ble minister.ble Pope.ble presidency.ble rector.ble ‘that can {become / be made / be appointed} N’ or ‘candidate for N’ These data raise a number of questions. First, if -ble selects transitive verbal bases, it is unexpected that the deviating forms in (2) are created in the absence of the corresponding verbs in (3).

254   

   Case study 2: N-ble

(3) * alcaldar, *bisbar, *ministrar, *papar, *presidenciar, *rectorar1  (Cat) mayor.inf  bishop.inf  minister.inf  Pope.inf  presidency.inf  rector.inf A second question concerns the exact characterization of this specific set of nouns which makes them appropriate as -ble bases. Third, it is necessary to determine where and how the relevant properties of these nouns are encoded to meet the requirements of -ble, i.e. having a theme argument and the implication of an originator, in addition to being eventive. A fourth problem relates to the source of crosslinguistic variation, i.e. why these forms are possible in Romance but not in English. Different authors have proposed that there are two different affixes -ble, e.g. Aronoff (1976) or Anderson (1992) for English, Rainer (1999) for Spanish. However, as observed in Val Álvaro (1981), the suffix behaves in exactly the same way with both nouns and verbs, both semantically and syntactically. The interpretation of denominal forms is essentially the same as that of any deverbal -ble adjective, “they both imply some process of change towards some state or quality.” Syntactically, Val Álvaro (1981:  197) advances the proposal that denominal adjectives “acquire modal content by virtue of the effect of the suffix on a predicative component” [author’s translation],2 thus suggesting that an adjective like alcaldable would be analyzed as the semantic head of an attributive sentential structure, as in (4), where the predicative component would be implicit as opposed to deverbal -ble adjectives where such component would be explicit. (4) ser alcalde + ble → alcaldable (Sp) be.inf mayor + ble → mayor.ble (Val Álvaro 1981: 197fn) As I understand it, such a proposal implies that the suffix -ble attaches to a predicative structure. Although this seems intuitively adequate, it would be rather difficult to capture within a lexicalist approach to word-formation, where affixes attach to either roots or stems, but not to syntactic predicative structures. Neither

1 According to the DIEC, Catalan has the verbs ministrar ‘supply’ and papar ‘swallow without chewing; see’. These are not commonly used, and more importantly, they are unrelated to the adjectives ministrable and papable, so that the latter cannot be said to derive from the former. 2 Note that this citation is essentially all that is said about the analysis of these denominal -ble forms in Val Álvaro (1981: 197).

Introduction   

   255

can affixes spell out syntactic structures, without recourse to the existence of lexical or pre-syntactic configurations.

1.1 Main claims and chapter overview In this chapter, I show that a syntactic model like Distributed Morphology is especially well-equipped to deal with such cases, since the only component that can generate structures is syntax proper. I propose that -ble can only be added to those nominal bases that can be temporally located and delimited (Lecarme 1999, 2008), i.e. those that accept deictic and event-temporal modifiers that can manipulate some temporal interval of the noun (Demonte 2005). The temporalaspectual features of the noun are realized in a predicative component that corresponds to the little v-Asp layer present in the internal structure of regular deverbal -ble formations, thus arguing for a uniform analysis of all -ble forms. Variation is restricted to the vocabulary (or lexicon) in the sense that it mainly arises from differences in the syntactic behavior of the base nouns and the semantic status of the lexical class of post-denoting nouns across languages. Crucially, my proposal need not be stipulated. It just follows from the general requirements on -ble affixation I have established so far, i.e. from the context of insertion of -ble, as well as from the grammatical status and conceptual nature of the lexical class of social roles3 in the different languages. Specifically, one of the main insights of this chapter is that I identify the subset of post-denoting nouns as unique among the set of role-denoting nouns in that they can be said to contain an originator in their internal (semantic) structure, and can be shown to behave differently with respect to their aspectual properties in the languages under study. The chapter is organized as follows. I first briefly review the treatment of denominal -ble forms in the literature. Then, I describe the main characteristics of denominal -ble adjectives as well as the main features of their bases in English and Catalan/Spanish. In section 3, I develop the analysis of productive denominal -ble adjectives and point at a possible explanation for their non-productive nature in English. My working hypothesis is that productive denominal -ble adjectives share their internal structure with regular -ble adjectives.

3 See for instance Masolo et al. (2004) for a mainly philosophical discussion of the ontological nature of (social) roles.

256   

   Case study 2: N-ble

2 Denominal -ble adjectives I first briefly review some of the previous analyses of denominal -ble adjectives in English and in Romance and the problems they pose. Then, concentrating on the productive cases in Catalan and Spanish, I describe their main properties, showing that only a particular subset of nouns can become a legitimate base for Nble derivation in these languages. To this end, I examine the main characteristics of these nominal bases, and briefly review some of the analyses that have been proposed in the literature to account for their special behavior.

2.1 Previous approaches In this section I review the very few works that have been concerned with denominal -ble adjectives for completeness.

2.1.1 English Nble In the analyses of English -ble adjectives, denominal forms have in general been either analyzed on a par with bound roots such as viable or potable, as in Fabb (1984); they have been presented as resulting from the application of a different morphological rule (Aronoff 1976); or they have been analyzed more or less on a par with deverbal adjectives despite being considered as lexical idiosyncrasies (Di Sciullo 1997). As I already pointed out in chapter 2, the existence of English denominal -ble adjectives is problematic for Aronoff’s (1976: 48) Unitary Base Hypothesis to begin with, which states that “the syntacticosemantic specification of the base, though it may be more or less complex, is always unique. A WFR [Word Formation Rule] will never operate on either this or that.” As a consequence, Aronoff is forced to say that there are two different -ble adjectives, and hence two different word formation rules. Aronoff provides two arguments for this division. His first argument concerns the process of nominalization of -ble forms. According to Aronoff (1976: 48), N-based items would take the suffix -ness, whereas V-based adjectives would form nominalizations in -ity/ness.

Denominal -ble adjectives   

   257

(5) a. fashionableness, sizableness, but *fashionability, *sizability4 b. acceptability/acceptableness, movability/moveableness However, this argument is challenged by the counterexamples provided in Anderson (1992: 192) or Adams (2001: 40n). (6) fissionability, pleasurability, reputability, roadability, saleability Furthermore, according to Marchand (1969) denominal adjectives, such as the ones in (7), have all passive meaning, which is the same type of meaning that we find with deverbal forms. (7)  carriageable, clubbable, magazinable, marketable, palatable, razorable, ­saleable For instance, clubbable is defined as “fit to be a member of a social club”, palatable means “acceptable or agreeable to the palate or taste”, and saleable is said of something “salable; subject to or suitable for sale”.5 In all cases, there is a passive sense in that the object that is attributed the property expressed by the -ble adjective is somehow affected, influenced or acted upon. So, we are left with just a “slightly less palpable evidence” in Aronoff’s own words (Aronoff 1976: 48) for the view that there are two different -ble adjectives in English. Aronoff’s second piece of evidence is related to the interpretation of these forms: N-based -ble means “characterized by N”, whereas V-based -ble means “capable of being Ved” in Aronoff’s terms. To postulate the existence of two different affixes based solely on an alleged difference in meaning seems a rather undesirable solution. Furthermore, according to Adams (2001: 35), most N-based cases can either be related to verbs, as shown in (8),6 or to events, as in (9).

4 According to Adams (2001: 33), a word like fashionability is equally fine, though it has a meaning different from fashionableness. Note that the meanings that I quote from Adams would apparently support Aronoff’s division of -ble into two different affixes, since only (i) expresses modality and has a passive meaning (see the discussion in the text). Such difference would be accounted for within DM by having root-attachment in (ii), but not in (i). (i) fashionability ‘quality of being able to be fashioned’ (ii) fashionableness ‘quality of being in fashion’ 5 Definitions are from the Webster’s Encyclopedic Unabridged Dictionary of the English Language. 6 Adams exemplifies (8) above with actionable as meaning “subject to a legal action”.

258   

   Case study 2: N-ble

(8) a. impressionable, objectionable, saleable, trailerable b. to impress, to object, to sell, to trail (9)  cabinetable, fissionable, overcoatable, pelletable, roadable, sudsable, ­tenurable This means that we would have just a handful of denominal adjectives that cannot be related to verbs or events, such as the ones listed in (10). (10) companionable, knowledgeable, peaceable, pleasurable Di Sciullo’s (1997: 90) proposal is claimed to account for Nble lexical idiosyncracies such as companionable, reputable or knowledgeable. In her proposal these nouns “have stage-level properties”, and “designate THINGS” that have transitory properties and not permanent ones” which make them possible bases for -ble derivation as opposed to forms such as *sunable, *airable, *earthable. On the one hand, if I understand it correctly, these cases would fit her conceptual dimension, but it is unclear to me how they would fair with the categorial and aspectual dimensions in her multidimensional structural account, since she does not spellout the details of the analysis for Nble. On the other hand, her proposal predicts that forms such as *warable, *accidentable or perhaps even *mayorable should exist, if we understand that war, accident and mayor have the same properties as the bases underlying those possible Nble forms in English. That is, Di Sciullo’s analysis cannot account for the lack of productivity of -ble with nominal bases in English. Note that the Catalan and Spanish productive cases take posts as bases, which are nouns denoting transitory or temporal properties, but are not possible in English.

2.1.1.1 Analyzing English Nble I agree with Di Sciullo’s (1997) suggestion that there is a single suffix -ble. Unless we want to multiply the number of suffixes in a way that does not help us understand how they are related to the rest of -ble derivations, it is certainly preferable to have just one suffix for all sets of adjectives, whether derived from verbs, nouns or roots. Indeed, in a DM approach we expect to have a single Vocabulary item and derive all variation either from divergences in their layered structure or in the level of attachment of some functional head(s). Additional differences may arise from distinct semantic properties of the base, i.e. differences may be due to the conceptual meaning of the base.

Denominal -ble adjectives   

   259

At first sight, the simplest DM analysis of the adjectives in (10) would seem to be an appeal to their being cases of low -ble adjectives, i.e. they would be formed by attaching the -ble suffix at the level of the root, i.e. below the first categorygiving head, which would correspond to + affixation in SPE or Level 1 affixation in Lexical Phonology,7 and have their encyclopedic meaning listed in the Encyclopedia.8 Such an analysis would work for reputable or peaceable; it would be problematic, though, for cases like knowledgeable, pleasurable, companionable, because they are constructed on the basis of derived stems, e.g. √know-ledge-able or √please-ure-able.9 Thus, the presence of these affixes points to the presence of nominal functional structure, thus banning the option of root-attachment. There are two possible ways to go within the DM framework. On the one hand, one could assume a root phrase internal analysis of these forms along the lines of Marant’z (2001) analysis of destroy as a manner root √stroy and a particle de below the first category-giving head v, which suggests the possibility of having complex structure in the root domain and therefore opens up the possibility of having categoryassigning functional heads merged with complex root-structures (see Acquaviva 2009 for discussion on roots). On the other hand, we could assume with Di Sciullo that these nouns bear the conceptual kind of content that can meet the requirements on -ble adjectives, i.e. expressing an event with an internal argument and the syntactic or semantic implication of an external argument, both related to the underlying presence of a verbal component, know and please. In this second option, we would be dealing with cases of a high -ble attachment which would have to be treated like idioms, in the sense that there would be a special meaning assigned to a phrasal structure as happens with idioms such as kick the bucket, a complex phrasal construction that must be listed in the Encyclopedia where it is assigned a special meaning (Marantz 1997, 1999a, b, 2001). Note that either option, though available in the framework, is stipulative and requires a number

7 Notice that Aronoff’s division between a +ble suffix and a #ble suffix first proposed in SPE, which has essentially remained in all subsequent work in morphology did not include N-ble formations, since this was a different suffix. See also Siegel’s (1979) Class I and Class II affixes, Selkirk’s (1982) Root-type affixes versus Word-type affixes, Fabb’s (1984) syntactic versus lexical affix -ble, or the most recent Volpe’s (2005) above and below v, among others. 8 Cf. Lyons (1977: 534), who lists all non-deverbal forms as “simple lexemes in the lexicon,” i.e. “all the other adjectives with stems in -able which do not satisfy the formula Vtr + able → Az.” 9 The analysis of companionable is unclear to me. On the one hand, it could be treated as a simple case of root attachment, √companion-able, if we are to look at its lexicalized meaning and its etymology (from Old French compaignon). On the other hand, one may want to reanalyze it as √company-ion-able, in which case, it would be as problematic as the cases discussed in the main text above.

260   

   Case study 2: N-ble

of assumptions on the face of unproductive idiosyncratic cases that need special and ad hoc treatments. Taking this into account, the absolute lack of productivity for Nble in English and the presence of unproductive suffixation, -ure, -ledge, the best solution seems to be the first, where the domain of the root in interaction with the Encyclopedia is taken as the repository for idiosyncrasies. Nothing else needs to be added as for the unproductivity of these cases. An analysis of the second type would have to face the impossibility of creating new idiomatic denominal -ble forms. Finally, consider again the cases in (8)–(9). At least from a synchronic point of view, they can be formally related to some ‘event’, so that they could be analyzed as regular cases of -ble adjectives. Note that the Webster’s Encyclopedic Unabridged Dictionary of the English Language contains the transitive verbs in (11), which could in principle be taken as bases for -ble derivation. (11) to fission, to overcoat, to pellet, to suds, to tenure Observe that all cases in (9) above contain some kind of modality as well, as can be inferred from their corresponding paraphrases: sudsable is interpreted as ‘that can be suds out’ and pelletable ‘that can be turned into pellets’. However, as far as I can tell, none of these -ble adjectives passes any of the tests we have seen in chapter 2 to establish the presence of verbal functional structure in -ble, which means that they must also be treated as cases of low -ble. There are two examples, cabinetable and roadable, which, at least from a synchronic point of view, cannot be directly tied to a base verb – though Adams (2001: 35) claims that they are related to events. If we consider their definitions, they both can be said to contain some passive and modal meaning. So, cabinetable means “fit to belong to the Cabinet” and roadable is interpreted as “capable of travelling on roads as well as in the air”. In fact, cabinetable is very similar to the Catalan/Spanish examples under study, in that it can be interpreted as ‘that can become a member of the Cabinet’, exactly as ministrable ‘minister.ble’, whose interpretation is ‘that can become a minister’. Despite this resemblance to the productive Romance cases, I think that these must also be treated as cases of low -ble affixation with idiosyncratic bases in English. To conclude, English Nble adjectives must be treated as cases of low -ble affixation. From a synchronic point of view, their existence is either due to the conceptual eventive meaning of the nominal base, as suggested in Di Sciullo (1997), as in knowledgeable, or it is due to their being reanalyzed as deverbal -ble adjectives, e.g. pelletable. I will have nothing else to say about the analysis of existing English Nble. Let us now move on to the more productive cases of Nble in Romance.

Denominal -ble adjectives   

   261

2.1.2 Catalan and Spanish Nble In the analyses of Spanish -ble adjectives, denominal forms have in general been either completely ignored, as in de Miguel (1986) or they have just been mentioned and left unanalyzed (Gràcia et al. 2000: 392). Two works stray away from this general tendency. According to Rainer (1999: 4633), there is in Spanish a denominal suffix -able “that should not be confused with deverbal -ble”. It is found with two different collections of words. There is a heterogeneous array of words that correspond to lexicalized forms, as in (12a), and a second set of “discreetly productive” -able adjectives, where the base refers to a cargo ‘post’, which can be paraphrased as “worth of being N”, in (12b). (12) a. confortable, entrañable, favorable, razonable, saludable, sociable (Sp) comfortable, intimate, favorable, reasonable, healthy, sociable b. alcaldable, ministrable, papable, rectorable mayor.ble, minister.ble, Pope.ble, rector.ble ‘that can {become / be made / be appointed} N’ or ‘candidate for N’ There are two main problems with this description. First, it overlooks the fact that all instances of -ble are preceded by a theme vowel. All cases in (12b) contain the vowel a that corresponds to the default theme vowel.10 Second, I completely disagree with the semantics attributed to these forms by Rainer. The words in (12b) express the possibility of becoming the post denoted by the base noun, i.e. the same type of modality and the same kind or similar verbal predication that we find with regular -ble adjectives. There is no component of worthiness or any other evaluative feature in the meaning of these words. Indeed, there are too many (shameful) examples showing that a candidate for mayor or president can be totally unworthy of being or becoming mayor or president, and he or she would still be referred to as alcaldable or presidenciable in Catalan or Spanish. A more promising analysis of -ble adjectives is Val Álvaro’s (1981). This author rejects Aronoff’s distinction (and therefore, also Rainer’s (1999) proposal) as not empirically adequate for Spanish, and points out that the suffix behaves exactly in the same way with both nouns and verbs in Spanish. In both cases, it expresses a

10 In fact, all denominal -ble adjectives I can think of, whether lexicalized or regularly derived from a post, have the default theme vowel a, an unsurprising fact if we assume Oltra-Massuet’s (1999, 2000) analysis of theme vowels. See the discussion on theme vowels in chapter 2.

262   

   Case study 2: N-ble

“virtual process of change towards some state or quality” (Val Álvaro 1981: 197n). For this author, both types of bases get modal content through the effect of the suffix on “a predicative component”. Whereas this component is explicit in the verbal base, in the case of nominal bases, he suggests the possibility that the noun heads an attributive structure, as we saw in (4) above, according to which, alcaldable ‘mayor.ble’ would be derived through adjunction of the affix -ble to the predicate ser alcalde ‘be mayor’. Despite the accuracy of the description, it seems to me that it is not possible to capture Val Álvaro’s insight within a lexicalist approach to word-formation, as I mentioned at the outset. In a lexicalist model, affixes attach to either roots or stems, but not to syntactic structures, and in this case, we seem to need a morphological rule that attaches a suffix to a sentential structure. In this chapter, I show that Val Álvaro’s intuition is best captured within a syntactic framework such as Distributed Morphology, where the only generative component is syntax proper. As for Catalan, there is to my knowledge no analysis of denominal -ble forms. Gràcia (1995: 99) just mentions the possibility of assuming a possible but not real intermediate verbal stage (*)alcaldar ‘to mayor’ for these forms, without providing any further details (cf. Gràcia et al. 2000: 197). Note that such an assumption is problematic due to the non-existence of these verbal forms, and it cannot be assessed without an explicit analysis.

2.2 General properties In this section I first describe the Catalan and Spanish denominal -ble data. Then I discuss the productivity of this construction and assess it against the absence of productive Nble in English. The last past of this section deals with the special properties of possible bases for denominal -ble.

2.2.1 Romance Nble The first thing we have to note is that -ble can only attach to a very restricted set of nominal bases, whose denotation corresponds to a post/position/office (Catalan càrrec, Spanish cargo). (13)  alcaldable, bisbable, ministrable, papable, presidenciable, rectorable  (Cat) mayor.ble, bishop.ble, minister.ble, Pope.ble, presidency.ble, rector.ble ‘that can {become / be made / be appointed} N’ or ‘candidate for N’

Denominal -ble adjectives   

   263

(14) alcaldable, obispable, ministrable, papable, presidenciable, rectorable  (Sp) mayor.ble, bishop.ble, minister.ble, Pope.ble, presidency.ble, rector.ble ‘that can {become / be made / be appointed} N’ or ‘candidate for N’ Posts, the set of nouns that allows -ble, are generally classed together with nouns denoting professions, which are in turn generally classified together with nouns referring to religions, nationalities, titles, and other occupations or roles in society, due to their special syntactic behavior. However, apart from the subset of ‘posts’, no other lexical subset can give rise to the formation of denominal -ble forms, as shown in (15) for Catalan.11 (15) a. * advocatable, *lampistable, *metjable, *politicable (Professions) (Cat) lawyer.ble,  plumber.ble,  doctor.ble,  politician.ble b. * budistable, *catolicable, *musulmanable(Religions) Buddhist.ble,  Catholic.ble,  Muslim.ble c. * alemanyable, *francesable, *portuguesable, *xinable(Nationalities) German.ble,  French.ble,  Portuguese.ble,  Chinese.ble d. * comtable, *ducable, *principable, *reiable(Titles) count.ble,  duke.ble,  prince.ble,  king.ble e. * empresariable, *capable, *socialistable, *veïnable, *marable  (Other) businessman.ble,  boss.ble,  socialist.ble,  neighbor.ble,  mother.ble Exactly the same facts obtain in Spanish, as shown in (16) for the various subtypes of social roles.

11 Since many of these words contain some agentive suffix before -ble, one could think that there may be some kind of morphological constraint at stake, as well. The ungrammaticality of previously underived forms such as Cat.*metjable/ Sp.*medicable ‘doctor-ble’ or Cat.*veïnable/ Sp.*vecinable ‘neighbor-ble’ shows that if such a constraint existed, it would not affect the point.

264   

   Case study 2: N-ble

(16) a. * abogadable, *fontanerable, *medicable,12 *politicable  (Professions) (Sp) lawyer.ble,  plumber.ble,  doctor.ble,  politician.ble b. * budistable, *catolicable, *musulmanable(Religions) Buddhist.ble,  Catholic.ble,  Muslim.ble c. * alemanable, *francesable, *portuguesable, *chinable(Nationalities) German.ble,  French.ble,  Portuguese.ble,  Chinese.ble d. * condable, *ducable, *principable, *reyable(Titles) count.ble,  duke.ble,  prince.ble,  king.ble e. * empresariable, *jefable, *socialistable, *vecinable, *madrable (Other) businessman.ble,  boss.ble,  socialist.ble,  neighbor.ble,  mother.ble It is certainly remarkable that only posts accept -ble affixation. I will defer until section 3 the elucidation of the special feature of post-nouns that singles them out from among the well-established set of social-role-nouns; a key feature that makes them appropriate as bases for -ble derivation. The Spanish examples in (17) could appear, at first sight, as potential problems for the statement that only post-denoting nouns admit -ble affixation, because they derive from gerencia ‘manager’s office’, ministerio ‘ministry’ and presidencia ‘presidency’ which are generally taken to denote locations.13 (17) gerenciable, ministeriable, presidenciable(Sp) manager/management.ble, ministry.ble, presidency.ble ‘that can {become / be made / be appointed} {manager, minister, president} However, upon closer examination, we can conclude that these examples pattern with all the others in that they also express a post that has temporally contingent properties. According to the DRAE, as well as their specific use, these words can

12 The Spanish word medicable is possible with a regular meaning ‘that can be administered medication’ when derived from the verb medicar ‘to administer medication to’. The star in the text refers to the meaning derived from the noun médico ‘doctor’, i.e. medicable cannot be interpreted as ‘that can become a doctor’. 13 Similar examples are found in Catalan, although perhaps ministeriable ‘ministery.ble’ sounds less familiar to my informants and myself; ministrable ‘minister.ble’ would be used instead. On the other hand, examples like gerenciable ‘manager/management.ble’ appear as fully acceptable to native speakers working in the world of business.

Denominal -ble adjectives   

   265

denote, among other things, a post as well as the place where this post is held,14 so we just need to assume that -ble is only compatible with the post interpretation of these bases. As I will argue below, only the meaning associated with a post can give rise to a legitimate -ble form, since, only posts can supply the structure with the kind of information necessary to meet the requirement on the availability of an originator for the derivation of -ble. Note that a specification of these nominal bases in terms of being either [human] or [animate] would not be enough, as the ungrammatical forms in (18) show. (18) a. * abellable, *crisalidable, *granotable, *papallonable (Animate N) (Cat) bee.ble,  chrysalis.ble,  frog.ble,  butterfly.ble b. * adultable, *donable, *homable adult.ble,  woman.ble,  man.ble

(Human N)

Thus, although we can think of the possibility of different kinds of larvae, tadpoles or chysalises to become bees, chysalises, frogs or butterflies, as exemplified in (19a); or of children, girls and boys to become adults, women and men, because they all can be said to express a transition to a resultant state – which is the regular meaning found with -ble adjectives, as shown in (19c) –, none of these is admitted to be expressed with a -ble adjective, as illustrated in (19b). (19) a. Una larva pot esdevenir abella en vint-i-un dies.(Cat) a larva can.3sg.prs become.inf bee in twenty-one days ‘A larva can become a bee in twenty one days.’ b. * una larva abellable en vint-i-un dies a larva bee.ble in twenty-one days c. un programa modificable en tres dies a program modifiable in three days ‘a program modifiable in three days’

14 Apart from denoting a post and a location, these nouns show multiple meanings. The noun gerencia ‘management’ additionally means (i) period of time that a person occupies the management, (ii) management. Other meanings of ministerio ‘ministry’ include (i) state government, (ii) the term of office of a minister, (iii) the body of ministers of state, (iv) each administrative governmental department that the state government is divided into. In the case of presidencia ‘presidency’, it also means (i) the action of being president of, (ii) the term of office of the president, and (iii) person or group of people that are presidents of something (all meanings are from DRAE, with author’s translations).

266   

   Case study 2: N-ble

Not only must these nouns denote a post, but they are also subject to an additional requirement on having unique reference, i.e. they must be uniquely identifiable. So, denoting a post is a necessary, though not sufficient condition, for accepting -ble. The contrasts in (20) should suffice to illustrate such a requirement for Catalan. For instance, while there is only one bishop per diocese, each diocese may have many cardinals. That is, uniqueness is evaluated relative to the relevant social institution.15 That is, even though there are many ministers in a given government, each of them has their own attributions in the individual institution and can therefore be uniquely identified. The same with directors or managers, whose uniqueness requirement would be evaluated relative to the company or institution where they are assigned unique and individual properties. (20) a. alcaldable, bisbable, directorable, ministrable, papable, rectorable (Cat) mayor.ble, bishop.ble, director.ble, minister.ble, Pope.ble, rector.ble ‘that can {become / be made / be appointed} N’ or ‘candidate for N’ b. * cardenalable, ?regidorable,16 *vicariable, *secretariable cardinal.ble,  city.councilor.ble,  vicar.ble,  secretary.ble Again, the same facts apply in Spanish, as illustrated in (21). (21) a. alcaldable, obispable,17 directorable, ministrable, papable, rectorable (Sp) mayor.ble, bishop.ble, director.ble, minister.ble, Pope.ble, rector.ble ‘that can {become / be made /be appointed} N’ or ‘candidate for N’ b. * cardenalable, ?concejalable, *vicariable, *secretariable cardinal.ble,  city.councilor.ble,  vicar.ble,   secretary.ble

15 The contrasts between ministrable ‘minister.ble’ and ?regidorable ‘city.councilor.ble’ in Catalan, or the corresponding forms ministrable and ?concejalable in Spanish, may just be due to a question of popularity of use. I have found a couple of examples of regidorable and concejalable on a Google search, which are deemed as acceptable by the native speakers consulted once they are supplied with an appropriate context that favors a uniqueness interpretation. 16 Native speakers disagree with respect to the acceptability of this word. For me, it is perfectly fine, even if perhaps not as usual as the ones in (20a), as it is among politicians and in the world of press. See the discussion on productivity in § 2.2.2 below. 17 There is a verb obispar meaning ‘obtain a bishopric, be appointed for it’ (DRAE, author’s translation). This verb is not used, so that its existence can only be diachronically relevant.

Denominal -ble adjectives   

   267

This uniqueness requirement seems to reduce the number of possible bases considerably, which may be taken to question the relevance of these data. A note on the productivity of this process of Nble derivation is thus advisable, and is dealt with in the next subsection.

2.2.2 The productivity of Nble in Romance Even though it is true that these examples are not so numerous, there is a clear regular and systematic pattern, even if very constrained by the reduced number of possible bases. Indeed, Carme Picallo (personal communication) points out that one could virtually make a list of the nouns that can be the target of -ble affixation, i.e. this process is grammatically not very productive with nouns in her opinion, so that expected forms such as *vicerrectorable or *deganable would not be possible. It is true that the forms that are most commonly used are very few, but I disagree about the grammatical productivity of -ble affixation with nouns on the basis of examples such as those in (22), attested by Google search and filtered out by native speaker’s positive judgments. (22) a. Sebastià Serrano, deganable de la llista continuïsta…(W) (Cat) Sebastià Serrano, dean.ble of the list continuous.ist ‘Sebastià Serrano, candidate for dean on the list of the people that maintains the same ideas as those in power’ b.  me recuerda (…) a cierto vicerrectorable…(W) (Sp) me remind.3sg.prs (…) to certain vice-rector-ble ‘(He) reminds me (…) of a certain candidate for vice-dean’ I take it as part of native speaker’s grammatical competence that, given the right type of noun, i.e. those that meet all grammatical prerequisites, a -ble form is possible with a nominal base, which means that -ble affixation is fully productive with nouns. Some of the forms can be said to belong to some particular slang, which should not minimize its productive power, though. For instance, as I mentioned in footnote 13 above, while gerenciable ‘management-ble’ may sound rather strange at first to non-business people like me or the reader, it is perfectly fine and used in the world of business. I assume that a word like arquebisbable ‘archbishop-ble’ will equally be fine in the world where bishops can be candidates to an archbishopric. Indeed, less familiar forms are deemed to be right as soon as these are placed in an appropriate context.

268   

   Case study 2: N-ble

To see the true productive status of these forms, consider the additional example in (23), provided to me by Joan Mascaró (personal communication). (23) X is presidenciable a França, per tant coprincipable a Andorra. x is presidency.ble at France, by so co-prince.ble at Andorra ‘X is a candidate for president in France, therefore a candidate for co-prince in Andorra.’ So, even though a prince and a co-prince are non-elected titles, Andorra’s coprinces are designated like posts, and therefore, they immediately become possible bases for -ble derivation. Similarly, Mascaró further adds that although kings and monarchs do not allow -ble affixation, because they do not denote appointed posts, we could create the form monarcable ‘monarch.ble’ referred to a monarch in the old Germanic world, where these titles were not inherited but elected. As I will argue below, such examples clearly support my proposal that -ble insertion requires the implication of an originator that can be related to encyclopedic knowledge. Most importantly, though, they contribute clear empirical support for the productive status of this process, if we understand that a process is productive when all possible bases that it targets can indeed become bases, be they numerous or reduced in number, and that they result from the application of the rules of the grammatical system (e.g. Bauer 2001, 2005 on productivity; see also Baayen 1992, and much related work for ways of quantifying productivity in morphology). In the case at hand, this is clearly so. Also crucial is the fact that these -ble forms exhibit the same morphology and semantics as all regular deverbal -ble adjectives. Hence, in my opinion, any analysis of -ble adjectives that does not take them into account will necessarily be flawed.

2.2.3 Productive Nble: Romance versus English As opposed to Romance, English denominal -ble is never productive, not even with nouns denoting posts or positions, as shown in (24). (24) *deanable, *mayorable, *popeable, *presidentiable If we assume that -ble formation targets essentially the same kind of roots for regular high -ble derivation in English and Romance, it is puzzling that they do not behave the same in this respect. The question is whether this divergence must be included in the specification of the context of insertion of the Vocabulary item -ble, i.e. in the morphological rule, say as a constraint on its domain of applica-

Denominal -ble adjectives   

   269

tion, or on its selectional properties. Or else, the difference lies in some particular feature of these nominal bases, which cannot license -ble insertion in English, so that variation would not be due to a specific difference in -ble formation. More generally, the question is whether this is due to some language particular constraint (or ability in the case of Romance), or whether it can be linked to some more universal property found in the languages of the world. My claim is that posts can meet the two main requirements necessary for -ble insertion. On the one hand, they are predicated of a theme (which corresponds to the person who is assigned the post denoted by the N-base). On the other hand, I will argue that they are the only nominal bases that can be said to imply an originator. Therefore, nothing prevents the application of the rule. The line of research I will pursue is that the different behavior of denominal -ble in English and Romance would derive from a well-studied more general feature of the lexical class posts are a subset of. Specifically, the difference would lie in the fact that in Romance, posts, like all other nouns in the lexical class of professions, seem to have adjectival properties as has been suggested in Bosque (1989, 1996), Stowell (1991), and more recently in Alexiadou (2005b), Beyssade and Dobrovie-Sorin (2005), or Depréz (2005), whereas in English uniquely identified posts like president are analyzed as concealed definites, as in Munn and Schmitt (2005). Interestingly enough, if this proposal is on the right track, it relates the morphological derivation of -ble to the syntactico-semantic status of their bases which become either predicates or definite descriptions. That is, I will suggest that in Romance, the set of profession-referring bases have some adjectival property that I identify with the presence of an aspectual functional projection in their syntax, which makes them compatible with the internal structure of -ble suggested in chapter 2. In view of the crucial role that the lexical class of the base noun plays in determining its suitability for -ble derivation, I will now examine the main properties of these special nominal bases. At the same time, I will briefly review some of the different proposals available in the recent linguistic literature that try to account for their distinctive behavior.

2.2.4 Properties of the N-base Probably one of the most relevant properties of nouns denoting professions, occupations and social roles in general is that they can appear as bare singular nouns in predicate position in a number of languages, among them Catalan and

270   

   Case study 2: N-ble

Spanish, languages where singular count nouns cannot head a bare NP in predicate position, as illustrated in (25)–(26).18 (25) a. * És cadira / dona / gos / ocell.(Cat)  be.3sg.prs chair woman dog bird b. És bisbe  / manyà  / paleta  / terrissaire / vídua. be.3sg.prs bishop locksmith bricklayer potter widow ‘He is a bishop / a locksmith / a bricklayer / a potter / a widow.’ (26) a. * Es silla / mujer  / perro / pájaro.(Sp) be.3sg.prs chair woman dog bird b. Es obispo / cerrajero / albañil  / alfarero / viuda. be.3sg.prs bishop locksmith bricklayer potter widow ‘He is a bishop / a locksmith / a bricklayer / a potter / a widow.’ This special syntactic behavior has been studied in a number of recent semantic papers (Roy 2004; Alexiadou 2005b; Matushansky and Spector 2005; Munn and Schmitt 2005; Winter 2005; de Swart et al. 2005, 2007).19 As pointed out in Roy (2004), the English sentence in (27) is ambiguous in that it can be analyzed as a predicational sentence or as an identificational sentence, and therefore it can be an answer to the two questions in (27). (27) John is a teacher. a. What does John do?  (Predicational) b. Who is John? (Identificational)

18 See also Espinal (2004) and Espinal and Dobrovie-Sorin (2006) for special cases such as (i)–(ii), where a bare count noun appears bare with a light verb and in complement position. (i) tenir gana  (Cat) have.inf hunger ‘to be hungry’ (Espinal 2004: 19) (ii) Lleva sombrero.(Sp) wear.3sg.prs hat ‘She wears a hat.’ (Espinal and Dobrovie-Sorin 2006: 5) 19 See also Fernández Leborans (1999) or Laca (1999) for general properties of nouns that can head a bare NP.

Denominal -ble adjectives   

   271

However, in languages like French, and also Catalan and Spanish, the corresponding sentence can only be identificational, as shown in (29b) for Catalan. To answer the questions What is John? or What does John do? these languages require a bare singular noun as a predicate, as illustrated in (29a) for Catalan. (29) a. Què és en Joan? / De què fa en Joan? – En Joan what be.3sg.prs the John of what do.3sg.prs the John the John és metge. (Predicational) (Cat) be.3sg.prs doctor ‘What is Joan? / What does he do? – He is a doctor.’ b. Qui és en Joan? – En Joan és un metge / who be.3sg.prs the John the John be.3sg.prs a doctor és el metge del poble.  (Identificational) be.3sg.prs the doctor of.the town ‘Who is Joan? – Joan is a doctor / is the town’s doctor.’ At first sight, one could think that being able to appear as head of a bare NP or not might be the key property to account for the contrast between Catalan alcaldable and English *mayorable. However, as I have already mentioned, although all occupations, professions and alike can occur as bare NP heads, only posts are appropriate as bases for -ble derivation. Perhaps even more importantly, as was already noted in Stowell (1991), despite the fact that bare singular nouns are generally not possible in English, not even with the lexical class denoting professions, and alike, as shown in (30a), it is possible in English to find a very restricted subset of that lexical class appearing as bare singular nouns in predicate position, what this author calls the president-class nouns, as exemplified in (30b–c). (30) a. *John is doctor. b. The people elected Nixon (??the) president. 

(Stowell 1991: 49)

c. Obama is president of the United States. The president-class of nouns does not fully correspond to the subset of nouns that admits -ble derivation in Catalan/Spanish, for it includes “elected, inherited, or appointed office, such as king, president, or treasurer” (Stowell 1991: 49), i.e. this is a larger subset than the class of post-nouns that can become -ble bases in Romance (e.g. inherited office such as king cannot be the base for -ble derivation). However, since English president-class nouns and Romance post-nouns are both subclasses of the same lexical class of social roles, and in fact, the latter is

272   

   Case study 2: N-ble

a subset of the former, I will briefly review some of the proposals that have been suggested in the literature, to the extent that they will help us understand the special characteristics of posts. The large, though partial, overlap between the two subsets may also offer some insight into different conceptualization patterns in these languages, which may be helpful in the analysis of the forms under study. According to Stowell (1991), the crucial feature of president-nouns is that they “can be used prenominally as titles” as shown in (31), and he takes titles as a special type of adjectives, so that the NPs that they head are reanalyzed as adjectival predicates when they refer to the title (they are kinds otherwise). (31) a. King George b. President Roosevelt Similarly, Alexiadou (2005b :  817) suggests that nouns like English president, which denote titles, do not have independent reference, i.e. they “do not exist independently from the individual that carries them”, which would explain the impossibility of having them bare in argument position, a position where they cannot establish the appropriate relation. That is, such a relation is only possible in predicative structures, as the contrast in (32) shows. (32) a. *We met President yesterday. b. We met President Bush yesterday. Note that this feature, i.e. denoting a title, cannot be the relevant property of the subset of nouns that allow -ble derivation. First, titles are not possible as bases for -ble derivation, as the exclusion of nouns like king, prince, or viscount in (33) make evident. (33) a. * princepable, *reiable, *vescomtable (Cat) prince.ble   king.ble  viscount.ble b. *principable, *reyable, *vizcondable (Sp) prince.ble   king.ble  viscount.ble Second, in Catalan and Spanish, titles – or English president-type nouns more generally – cannot appear bare in prenominal position, as shown in (34)–(35).

Denominal -ble adjectives   

   273

(34) a. El Rei Jaume I va escriure la primera crònica the King James first aux.3sg write.inf the first chronicle medieval catalana.(Cat) medieval Catalan ‘King Jaume I wrote the first Catalan medieval chronicle.’ b. * Rei Jaume I va escriure la primera crònica medieval king James first aux.3sg write.inf the first chronicle medieval catalana. Catalan (35) a. El Presidente Zapatero no cumple sus promesas.  the president Zapatero not keep.3sg.prs his promises ‘President Zapatero does not keep his promises.’

(Sp)

b. * Presidente Zapatero no cumple sus promesas. president Zapatero not keep.3sg.prs his promises On the other hand, the identification of this type of nouns as a special type of adjectival predicates in Stowell’s proposal seems more promissing, as this has been a recurrent idea in the literature on Romance languages. According to Munn and Schmitt (2005), Pollock (1983) already attributes the special properties of these nouns to their adjectival status, and the fact that professions, like adjectives, do not have inherent gender but agree in gender with their subjects. However, Kupferman (1991) provides different tests, applied to French, showing that bare nouns pattern differently from adjectives, thus arguing against the classification of president-nouns as adjectives. A similar, though crucially different line of research is presented in Alexiadou (2005b), who argues that it is the property that these nouns have in common with adjectives what allows them to function as predicates. Neither adjectives nor president-type nouns are inherently specified for number and gender, but these features are determined by the individual they are predicated of. Thus, the special behavior of president-type nouns is not due to their classification as adjectivals, but to their having some adjectival property, which would overcome the problems posed in Kupferman for analyses that equate these nouns to adjectives. Indeed, in Spanish, according to Fernández Leborans (1999:  2371), the set of nouns that denote socio-cultural roles and functions, which includes posts and positions, have an “adjectival performance”, in the sense that they behave as qualifying adjectives, because they attribute to the subject the social category denoted by the bare noun as a property. Similarly, Ramos (2002: 1961) points out

274   

   Case study 2: N-ble

that the ascription of a social characteristic, such as a profession or a post, can be perceived as a simple property, which shows that the limits between the category noun and the category adjective are rather vague with this lexical class. Bosque (1996:  59f)20 suggests that the members of this grammatical class have in common their “inherently predicative nature”. They show a unique behavior among nouns in Spanish, in that they allow se-passives (pasiva refleja) with definite personal subjects, although se-passives do not license definite personal subjects. Thus, whereas the sentence in (36) allows a se-passive interpretation in which the post-verbal subject does not refer in (36b), this interpretation is not available in (37), where the only possible reading is a reciprocal one. (36) Cuando se nombraron los nuevos embajadores.(Sp) when se appoint.3pl.pst the new ambassadors

a. Referential: some ambassadors have been appointed (to another post)



b. Predicative: some people have been appointed to the post of ambassador

(37) Cuando se saludaron los nuevos embajadores.  when se greet.3pl.pst the new ambassadors

a. Referential: the new ambassadors greeted each other



b. Predicative: Ø

(Sp)

According to Bosque (1996: 60), this means that these nouns have a special treatment in the grammar, as further evidenced with his examples in (38), where the definite phrase in (38a) is again ambiguous. It has a referential interpretation where it denotes an individual and a predicative reading that corresponds to “representative of the class ambassador”. This second reading is the only one available for (38b). (38) a. Han nombrado al embajador.(Sp) have.3pl.prs appoint.part at.the ambassador 1. ‘They have appointed the ambassador (to another post).’ 2. ‘They have appointed somebody to the post of ambassador.’

20 See also Bosque (1989: 109).

Denominal -ble adjectives   

   275

b. Han nombrado embajador. have.3pl.pst appoint.part ambassador ‘They have appointed somebody to the post of ambassador.’ More recently, Beyssade and Dobrovie-Sorin (2005) also attribute the capacity of a noun to appear as a singular bare nominal to its status as a non-sortal noun (i.e. non-object-denoting nouns). Non-sortal nouns include the set of professions, titles, posts, etc, and are analyzed on a par with adjectives, which attribute a property to an individual, as opposed to sortal nouns, which identify an individual as a member of a set. Also Depréz (2005) attributes their unusual behavior in all types of languages to their adjectival status. We can conclude, then, that the most important property of president-type nouns, and therefore of its subclass post-nouns, is their being able to appear in bare NPs, and that this property has been attributed, in one way or other, to their adjectival or predicative nature. As pointed out in Lecarme (1999), it has been a widely held assumption that, semantically, all instances of bare NP are predicates (Williams 1981b; Higginbotham 1985), and therefore, they are time sensitive (Enç 1981; Musan 1995). Indeed, it has also been acknowledged that these nouns denote transitory, non-permanent, properties (e.g. Kupferman 1991). Thus, another basic feature of these nouns is that they are stage-level predicates (Carlson 1977a; Kratzer 1988/1995). According to Kratzer (1988/1995), stage-level predicates have an extra (Davidsonian) argument position for events or spatio-temporal locations. Indeed, nouns denoting posts can be temporally located and aspectually delimited with deictic and event-temporal modifiers, as it is characteristic of episodic predicates (Higginbotham 1985, 1987; Demonte 2005). This is illustrated in (39) for English and in (40) for Catalan. (39) a. the {former / then} president of the United States b. the {future/current} mayor of Barcelona (40) a. l’ {aleshores / ara / ex-} president dels Estats Units(Cat) the=  then now ex- president of.the United States b. el futur alcalde de Barcelona the future mayor of Barcelona Note, in passing, that being eventive or not is irrelevant for -ble derivation, as shown in (41) with Catalan eventive nouns. That is, the eventive property alone is not enough for being an appropriate base for -ble, although it is a necessary condition, as I will show below.

276   

   Case study 2: N-ble

(41) a. * ampollable, *artable, *homable, *gossable   bottle.ble,   art.ble,  man.ble,   dog.ble

(Non-eventive N) (Cat) 

b. * festable, *guerrable, *partidable, *tempestable  party.ble,  war.ble,  match.ble,  storm.ble

(Eventive N)

Munn and Schmitt (2005), building on Kupferman (1991), who argues that the class of nouns that can appear bare must denote socio-cultural sets and are therefore non-inherent and non-permanent,21 take the eventivity of profession nouns as their fundamental property to account for their being able to appear bare in post-copular positions in languages like Brazilian Portuguese (and also other Romance languages like Catalan and Spanish). For them, profession nouns are stage-level predicates; as such, they have an event argument that must be bound by Tense, instead of the referential argument bound by the functional head D that we find in regular nouns. This is an option, though, because professions can also behave as regular nouns with a referential argument. Since in English only nouns like president can appear bare, they propose a uniqueness restriction, which is similar to the uniqueness of proper names in that none of them require a determiner in English. They analyze president-type nouns as concealed definites, like proper names. To distinguish president-nouns from proper names, they suggest that the former are inherently specified for number in the lexicon. I represent this specification as a feature [Num] in (42) for convenience, since they do not provide the structure for these cases. This inherent marking allows them to raise to a head NumP, so that the NumP need not be spelled-out as the indefinite determiner. That makes them pattern with indefinite phrases such as a teacher in John is a teacher, which are said to be NumPs. As bare NumP, president-type nouns can thus appear in predicative position. For completeness, I include in (43) their proposal for profession nouns in French, which can appear bare in predicate position as in Catalan and Spanish. These are analyzed as NP predicates that merge with a DP in the lexical domain, i.e. the NP domain, and this DP subject is θ-marked by the head noun.22

21 As exemplified in Munn and Schmitt (2005: 846), whereas plumber or teacher denote contingent and temporal properties, so that one and the same individual can have both at the same time, this is not possible with inherent and natural classes, such as bulldog and fox terrier. 22 See Munn and Schmitt (2005: 835) for details on different types of predication (θ-marking, predication and identification), depending on the level at which the relation between a DP argument and the (extended projection of a) noun is established (within the lexical NP domain, at NumP, or at the DP level respectively).

Denominal -ble adjectives   

NumP (42)  DP John this man

(Predication)

NumP Num a

NP teacher president[Num]

NP (43)  DP

   277

(θ-marking)

NP

cet homme médicin this man (is a) doctor In a similar vein, Winter (2005) analyzes president-like nouns as “the singleton set that consists of the (unique) president”, i.e. “a predicate with a uniqueness requirement on its extension.” Their lexical structure in (44a) corresponds partially to the lexical structure he proposes for proper names in (44b), i.e. as bare nouns with an empty Spec, NP that imposes a uniqueness requirement. Proper names require additional lexical structure to account for their semantics, i.e. that they denote an entity. (44) a. [ Øthe [N' president ] ]

b. [ d' Øcf [np Øthe Einstein ] ]

president-type nouns proper names23

Although such a uniqueness requirement seems to be at stake in the case of -ble derivation for posts, I do not think that these analyses can be straightforwardly applied to Catalan, since, in this language, proper names do bear a determiner, whereas professions and posts can appear bare in predicative position, as illustrated in (45).

23 In Winter (2005), NPs are predicates denoting sets of entities, whereas a DP is a generalized quantifier, i.e. a set of sets of entities. Thus, proper names need an additional functional layer in order to end up denoting an entity. This is achieved through the application of an empty choice function variable in D, as with indefinites, which takes a set X as its argument and returns an element of X as its value.

278   

   Case study 2: N-ble

(45) a. En Magí és professor.(Cat) the Magí be.3sg.prs professor ‘Magí is a professor.’ b. La Laia és alcaldessa. the Laia be.3sg.prs mayor.f ‘Laia is a mayor.’ (46) a. El professor és en Magí.(Cat) the professor be.3sg.prs the Magí ‘The professor is Magí.’ b. L’alcaldessa és la Laia. the=mayor.f be.3sg.prs the Laia ‘The mayor is Laia.’ Furthermore, in Munn and Schmitt’s analysis, president-type nouns pattern structurally with non-profession nouns, whereas for my analysis of -ble, I need some property that distinguishes posts as a subset from professions, which seem to behave differently from all other nouns, proper names included. According to Partee and Borschev (2003:  67f), an additional property of English nouns like mayor is that they are relational nouns that take arguments. (47) *The mayor of Boston has more power than the one of Baltimore. Interestingly, whereas in Romance it is possible for some inalienable relational nouns like Catalan mare ‘mother’ to appear as a bare NP in the absence of its complement, i.e. it can be used in absolute terms, this is not possible for the corresponding English nouns, which in general can never appear bare.

Denominal -ble adjectives   

   279

(48) a. L’Andrea és mare. / L’Andrea és mare de dos the=Andrea be.3sg.prs mother the=Andrea be.3sg.prs mother of two nens.24(Cat) boys ‘Andrea is a mother. / Andrea is a mother of two boys.’

b. *Andrea is mother. / */?Andrea is mother of two boys.

(49) a. No t’amoïnis, en Joan és amic. / En Joan not you.refl=worry.2sg.prs the Joan be.3sg.prs friend the Joan és amic de la família.(Cat) be.3sg.prs friend of the family ‘Don’t worry, Joan is a friend. / Joan is a family friend.’

b. *Don’t worry, John is friend. / *John is friend of the family.

It turns out that there is a similar contrast with posts in English but in the opposite direction. Consider the examples in (50). Although sentences such as John is president or John is mayor are fine, they become ungrammatical in the presence of the prefix ex- or the adjective former which are supposed to target the eventivity of the noun. However, the sentences become fine in the presence of the complement, as illustrated in (50c). Notice that no such interaction seems to be relevant for Catalan or Spanish, as exemplified in (51) for Catalan. (50) a. *John is {ex-president / ex-mayor / ex-director}.25

b. *John is former {president / mayor / director}.



c.  John is {ex-president of the US / ex-mayor of Barcelona / ex-director of the company}.

24 However, this cannot be generalized to all relational nouns, not even to all kinship terms, as shown in (i). (i) a. Sóc tieta.(Cat) be.1sg.prs auntie ‘I am an auntie.’ b. * Sóc germana. be.1sg.prs sister ‘I am a sister.’ 25 Some speakers report these sentences as deviant, rather than outright bad. The crucial point, however, is that they are in any event worse than the ones with the complement in (50)c. I have got the same judgments with former.

280   

   Case study 2: N-ble

(51) a. En Joan és ex-president / ex-alcalde / ex-director.(Cat) the Joan be.3sg.prs ex-president ex-mayor ex-director ‘Joan is an {ex-president / ex-mayor / ex-director}.’ b.  En Joan és {ex-president dels Estats Units / ex-alcalde the Joan be.3sg.prs {ex-president of.the States United / ex-mayor de Barcelona / ex-director de la companyia}. of Barcelona / ex-director of the company} ‘Joan is {ex-president of the United States / ex-mayor of Barcelona / exdirector of the company}.’ To account for this contrast, we may follow two lines of inquiry. On the one hand, there could be some interplay between the eventivity of the base nouns and its relational status in English but not in Romance. Unless we could show that such interaction exists somewhere else in the language – and preferably also in other languages –, this suggestion should be abandoned. A more plausible idea is that such different behavior is due to a different status of president-type nouns in English, but not in other languages, i.e. that we are dealing with a lexical difference. Specifically, I would like to argue that this contrast is related to Munn and Schmitt’s (2005) or Winter’s (2005) uniqueness requirement of president-nouns in English, i.e. English, unlike Catalan or Spanish, disallows bare predicates with nouns that do not denote singleton sets. Indeed, according to Payne and Huddleston (2002: 409), the interpretation of bare NPs that denote a social role is “invariably definite” in English and therefore bare role NPs are “invariably replaceable by their counterparts with determiner the.” Accordingly, Payne and Huddleston’s (2002:  409) example (52a) could be replaced by (52b) without a difference in meaning, because in both cases, “we are concerned with the office of treasurer in some particular organization” [my emphasis]. That is, in both cases the treasurer is uniquely identifiable; it does not have a predicational interpretation. (52) a. Henry became treasurer. b. Henry became the treasurer. (Payne and Huddleston 2002: 409) Crucially, the corresponding sentences in Catalan or Spanish are not mutually replaceable, because they do not have the same meaning. Whereas the (a) examples in (53)–(54) have a predicational interpretation, i.e. the individual named Enric/Enrique is said to be a member of the set of treasurers, the (b) examples identify an individual as the unique treasurer of some contextually identified organization, i.e. as the unique member of the singleton set.

Denominal -ble adjectives   

   281

(53) a. L’Enric va esdevenir tresorer. (Cat) the=Enric aux.3sg become.inf treasurer ‘Enric became a treasurer’ b. L’Enric va esdevenir el tresorer. the=Enric aux.3sg become.inf the treasurer ‘Enric became the treasurer’ (54) a. Enrique se convirtió en tesorero. (Sp) Enrique se become.3sg.pst in treasurer ‘Enrique became a treasurer’ b. Enrique se convirtió en el tesorero. Enrique se become.3sg.pst in the treasurer ‘Enrique became the treasurer’ In section 3 I will present additional empirical evidence which provides support to the view that the different behavior of post-nouns in English and Romance with respect to -ble, but also to ex- or former, can be reduced to a difference in the internal constituent structure, i.e. in the lexical status, of this type of social roles. Let me summarize the main properties of the possible nominal bases for -ble derivation in (55). (55) a. They denote a post;

b. they must be a uniquely identifiable post;



c. they are episodic, stage-level predicates;



d. as stage-level predicates, they license deictic and event-temporal modifiers;



e. they can appear as bare singular predicates in languages where bare singular nouns cannot appear bare;



f. they seem to have some adjectival property and possibly lack inherent gender features;



g. they are relational nouns that can be used in absolute terms in Romance.

I have also presented three important differences between English and languages like Catalan or Spanish with respect to post-nouns, which will be crucial for the analysis of N-ble forms in Romance, and to account for their absence in English. Specifically, English president-type nouns can function bare as prenominal titles,

282   

   Case study 2: N-ble

whereas Catalan and Spanish require the presence of the determiner; English president-nouns cannot appear bare with event-temporal modifiers in the absence of the complement, while this is possible in Romance; bare NPs denoting social roles are invariably identificational in English but predicational in the Romance languages under study. There is of course a fourth difference, which is the object of inquiry of this chapter, namely, that post-nouns allow -ble derivation in Romance, whereas neither president-nouns nor post-nouns admit -ble in English. In the next section I will show that all these properties are interrelated and are accounted for if we assume a basic difference between English and Romance: English president-nouns must be analyzed as concealed definites, as proposed in Munn and Schmitt (2005), whereas the corresponding Catalan and Spanish nouns, as well as all other nouns denoting social roles in these languages, have some adjectival, i.e. predicational, property, as suggested for instance in Alexiadou (2005b).

3 Towards an analysis of Nble Let me go over the main questions I posed at the beginning of this chapter. First, why it is possible to derive alcaldable if there is no verb *alcaldar and more importantly, where and how these forms meet the two basic requirements on -ble derivation. The second question relates to the crosslinguistic variation, i.e. why these forms are possible in Romance but not in English, and where the source of this variation is to be found. Additional questions relate to the specificity of these nominal bases that make them stand out among the larger class of nouns identified as capacities in de Swart, Winter, and Zwarts (2007), among other things, in being appropriate bases for -ble derivation. The idea I want to develop is that there is only one morpheme -ble. Lexicalized forms like English denominal peaceable are cases of root affixation in Distributed Morphology, where we expect the kind of special or idiosyncratic meaning that we indeed find, i.e. they would be analyzed on a par with edible or tolerable.26 All other cases belong to the regularly derived -ble adjectives, i.e. they are cases of derivation above first-category-assigning functional head in Distributed Mor-

26 The analysis of lexicalized cases in Catalan or Spanish, such as Catalan saludable ‘healthy’, are more complex, since in those cases we also find the default theme vowel following the root, which signals the presence of some additional functional head (Oltra-Massuet 1999 and related work). See chapter 2 for -ble affixation below the first category-assigning functional head.



Towards an analysis of Nble   

   283

phology. That is, my claim is that denominal adjectives like alcaldable share their internal structure with all deverbal -ble adjectives in (56). (56)

ModP Mod

AspP Aspr

vP DP

v v



Let us examine this proposal in more detail. I will essentially concentrate the discussion in providing evidence for the presence of eventivity and an aspectual layer, i.e. I take the presence of a modal head for granted. Note that it is the possibility modal content of -ble in alcaldable ‘mayor.ble’ that can give rise to a reading that expresses a contradiction in (57) when we try to cancel the possibility of becoming a mayor expressed in alcaldable. That is, we predicate both that Joan can be candidate for mayor and that he cannot.27 (57) # En Joan és alcaldable, però no pot ser  the Joan be.3sg.prs mayor.ble but not can.3sg.prs be.inf  candidat a alcalde.(Cat)  candidate to mayor  ‘Joan is candidate for mayor but he cannot be candidate for mayor’ For denominal adjectives in Romance, we have seen that these nominal bases are stage-level predicates. According to Kratzer (1988/1995) and subsequent work, this means that these nouns have a semantic event argument. Also, Higginbotham (1985, 1987) points out that temporal modification such as former is direct evidence for an event position in nominals. In the Distributed Morphology framework, being eventive in principle amounts to having a verbal functional head little v in the syntactic structure, there is no other type of eventivity,

27 I have not marked the sentence as ungrammatical, because it could in principle have an interpretation where Joan can be appointed as a mayor because he has all necessary properties to become one, but for some reason, say personal, political or other reasons, he cannot be appointed as candidate for mayor.

284   

   Case study 2: N-ble

semantic or other.28 There are at least two potential problems with postulating the presence of such eventive functional head in the internal structure of these nominal bases. First, I have found a crucial contrast in (50)–(51), repeated as (58)–(59). (58) a. *John is {ex-president / ex-mayor / ex-director}.

b. *John is former {president / mayor / director}.



c.  John is {ex-president of the US / ex-mayor of Barcelona / ex-director of the company}.

(59) a. En Joan és ex-president / ex-alcalde / ex-director.(Cat) the Joan be.3sg.prs ex-president ex-mayor ex-director ‘Joan is an {ex-president / ex-mayor / ex-director}.’ b.  En Joan és {ex-president dels Estats Units / ex-alcalde the Joan be.3sg.prs {ex-president of-the States United / ex-mayor de Barcelona / ex-director de la companyia}. of Barcelona / ex-director of the company} ‘Joan is {ex-president of the United States / ex-mayor of Barcelona / ­ex-director of the company}.’ In principle, if post-nouns in both English and Romance are stage-level predicates, i.e. they are equally eventive, we cannot explain this contrast. For a noun like presidency, both languages would have the basic structure in (60). (60)

n n

v v

√presid

Below, I will tie the difference between English and Romance to the divergent behavior of the underlying base nouns, though. Whereas in Catalan and Spanish the root is semantically compatible with a predicative or an identificational interpretation, the noun in English has an additional uniqueness requirement that prevents it from appearing in a non-unique context, or better said, the cor-

28 Thanks to Jaume Mateu for reminding me of this point.



Towards an analysis of Nble   

   285

responding English nouns can only get the identificational meaning.29 Thus, the Catalan examples in (59a) say that the individual Joan is no longer a member of the set of presidents/mayors/directors whereas the sentence in (59b) expresses that the individual named Joan no longer holds the post described by the NP. Only the latter reading is available in English. A more problematic issue for the cases under study is that if this eventivity is syntactically represented in the form of a little v layer in Romance, something else must be said to explain the lack of the corresponding verbs, the main problem I started with. That is, in principle, nothing prevents the eventive head from licensing more verbal functional structure, so that it ends up becoming a verb. A proposal that makes use of some special little v head to prevent this result appears rather stipulative and would not account for the systematic rejection of such verbs by native speakers in a principled way. I have carried out a small-scale experiment among a few native speakers who are non-linguists, asking them to provide a meaning for sentences such as (61a–b). In essentially all cases, my informants were unable to produce a single interpretation for them, as opposed to the equally non-existing sentence in (61c), for which they all coincided in assigning the same interpretation, the one we find in the existing English sentence. (NB. There is no verb to shelve in Catalan). (61) a. #En Joan ha alcaldat.(Cat)  the Joan have.3sg.prs mayor.part b. #Han alcaldat en Joan.  have.3pl.prs mayor.part the Joan c. En Joan ha emprestatjat els llibres. the Joan have.3sg.prs shelve.part the books ‘Joan has shelved the books’

29 As Ora Matushansky (personal communication) points out, it is possible to have sentences like There are three presidents in the room which appear as counterevidence to my claim. As I will argue below, even in cases like these the native speakers I have consulted do not seem to get the predicative interpretation by which we would be talking of three individuals that have as a characteristic that they are members of the set of presidents. The default interpretation always assigns them the identifying property of being presidents of a specific country. Also a quick Google search shows that quantified instances of the noun president always refer to some US President. Although the question on the reference of these nouns calls for a largerscale experimental study with native speakers, see the discussion below, especially the one on example (95).

286   

   Case study 2: N-ble

Perhaps, one could entertain the possibility of having the root merge directly with Asp in Catalan and Spanish, so that there would be no v present in the internal structure of these nouns. Such a suggestion would overcome the problem of having a little v layer in a configuration that cannot license further verbal structure, i.e. that cannot form verbs. The presence of Asp in Romance but not in English, due to the adjectival or predicative nature of the relevant nouns in the former languages, would also account for crucial contrasts in the number and type of adverbial modification these languages allow. This would be in accordance with Cinque’s (1999) and Alexiadou’s (2001) proposals that this type of adverbials are licensed by an AspP. See the examples and discussion that immediately follows. However, as pointed out to me by Jaume Mateu and Gemma Rigau (personal communication), the eventivity of post-nouns clearly differs from the eventivity of nouns like destruction, as illustrated in the contrast in (62) for Catalan. (62) a. la destrucció de la ciutat pels invasors(Cat) the destruction of the city by.the invaders ‘the destruction of the city by the invaders’ b. *el candidat alcaldable de la ciutat pels votants  the candidate mayor.ble of the city by.the voters How this difference in the degree of eventivity is to be represented in the DM model, where the only possible eventivity is in the form of a v layer, would be a matter of research. Also, the theoretical problem of having an aspectual head merge directly with the root would involve a further significant theoretical weakness: in the absence of an eventive head, we would find a semantic change of state – alcaldable as become mayor – represented as just a result, instead of being the result of a prior event. A possible solution, suggested to me by Antonio Fábregas (personal communication) is to assume that the verbs *alcaldar ‘to.mayor’ or *presidenciar ‘to.president’, which would be stative verbs, are lexicalized in Catalan and Spanish as be N. Although this suggestion needs to be worked out, together with a detailed investigation on the phonological realization process of these syntactic structures, note that in a syntactic framework like DM there is no generative lexicon that stores the interpretation of the structure, which allows for the same structure to be phonologically realized in different ways. Despite this shortcoming, I will continue to assume that the eventivity of these roots is realized in a little v head. In Romance, apart from licensing the temporal location of the event, these nouns can also license a number of aspectual delimiters. I assume with Cinque (1999) and Alexiadou (2001) that aspectual modifiers like deictic and temporal



Towards an analysis of Nble   

   287

adverbials are linked to an AspP (Cinque 1999). As shown in the (a) sentences in (63)–(65), aspectual delimiters are not licensed as modifiers of posts-nouns in English, but they are fine in Catalan/Spanish, as the corresponding translations in the (b–c) examples show. (63) a. *The until then {president /mayor / bishop} pleaded not guilty. b. El fins aleshores {president / alcalde / bisbe} es va the until then  president mayor bishop se aux.3sg declarar no culpable.(Cat) plead.inf not guilty c.

El hasta entonces {presidente / alcalde / obispo} se declaró the until then president mayor bishop se plead.3sg.pst no culpable.(Sp) not guilty

(64) a. *The liberal party remained in the hands of the now {president / mayor}. b. El partit liberal va quedar en mans de l’ara the party liberal aux.3sg remain.inf in hands of the.now {president / alcalde}.(Cat) president mayor c. El partido liberal quedó en manos del ahora the party liberal remain.3sg.pst in hands of.the now {presidente / alcalde}. (Sp)  president mayor (65) a. ?*The still president has announced a new rise of taxes. b. L’encara president ha anunciat una nova pujada de the=still president have.3sg.prs announce.part a new rise of taxes.(Cat) taxes c.

El aún presidente ha anunciado una nueva subida the still president have.3sg.prs announce.part a new rise de impuestos.(Sp) of taxes

As further shown in (66), in Catalan these post-nouns admit a great variety of aspectual modification. Similar facts obtain in Spanish, as shown in (67).

288   

   Case study 2: N-ble

(66) El {tantes vegades / fins fa poc / per ara / per molts the  so.many times / until does little / for now / for many anys / fins ara} alcalde(Cat) years / until now mayor  ‘The {so many times/until recently/for now/for many years/until now} mayor’ (67)  El {tantas veces / hasta hace poco / por ahora / por muchos años / the  so.many times / until does little / for now / for many years / hasta ahora} alcalde(Sp) until now mayor  ‘The {so many times/until recently/for now/for many years/until now} mayor’ According to my English informants, all these modifiers are anomalous, at least in the absence of the nominal complement.30 (68) *The {until now / until then / until recently / so many times} mayor31 In addition, Bosque (1989: 143) also reports examples such as (69a)–(70a) as parallel to (69b)–(70b) in Spanish, where post-nouns are premodified by a deictic adverb. According to this author, adverbs are acceptable in the NP “when the substantive designates predicates – generally person predicates – whose meaning is specifically linked to a temporal state, as it is the case with occupations, posts, activities and other attributions that involve some chronological boundary (cf. therefore, *la actualmente carretera de Barcelona ‘the currently road of Barcelona’)” [author’s translation]. Additional examples are given in (71). (69) a. El actualmente Primer Ministro del Japón(Sp) the currently prime minister of.the Japan

30 See the discussion below for the relevance of the nominal complement. 31 Andrea Rackowski (personal communication) reports the fine phrases in (i)–(ii). Note that they differ crucially in that they no longer contain adverbial modification, but adjectival, which is otherwise expected. That is, English admits only temporal modification as long as this is adjectival, whereas in Catalan and Spanish, it is possible to have adverbial modification with this type of nouns (cf. examples (69)–(72) in the main text). (i) the long-time mayor of Barcelona (ii) the several-time mayor of Barcelona



Towards an analysis of Nble   

   289

b. El actual Primer Ministro del Japón the current prime minister of.the Japan (70) a. El antiguamente Primer Ministro(Sp) the formerly prime minister b. El antiguo Primer Ministro the former prime minister (71) a. El recientemente alcalde de Barcelona32(Sp) the recently mayor of Barcelona b. El hoy alcalde de Madrid the today mayor of Madrid

(Bosque 1989: 143)

Not all my informants accept the corresponding examples in Catalan, in (72).33 (72) a. {/?/*} L’actualment Primer Ministre del Japó34(Cat) the=currently prime minister of.the Japan b. {/?/*} L’antigament Primer Ministre the=formerly prime minister c. {?/*} El recentment alcalde de Barcelona the recently mayor of Barcelona d. alcalde de Barcelona  L’avui the=today mayor of Barcelona As for English, the equivalent examples in (73) are ungrammatical to all speakers I have consulted.

32 Although Bosque (1989) does not provide an explicit example with reciente/recientemente he does mention that exactly the same effect is obtained with these deictic forms. 33 Joan Mascaró (personal communication) suggests that the contrast between Catalan and Spanish may be due to differences in register; specifically the fact that the construction is rather learned in Spanish; thus, whereas Spanish speakers easily accept learned constructions, Catalan speakers are reluctant to do so. Indeed, for bilingual speakers most of these sentences seem better in Spanish than in Catalan. 34 It would be interesting to see whether the differences in judgments among Catalan native speakers correlate with some more general variation in eventive or aspectual properties.

290   

   Case study 2: N-ble

(73) a. *The formerly Prime Minister

b. *The currently Prime Minister of Japan



c. *The recently mayor of Barcelona



d. *The today mayor of Barcelona

Additional evidence that these nouns must have this kind of functional structure associated with the spatio-temporal location of the event comes from examples such as (74) for Catalan, where last year and in Barcelona locate the post in time and space. Same facts obtain for Spanish. (74) a.  El director de la companyia l’any passat a Barcelona the director of the company the=year last in Barcelona va ser elegit president.(Cat) aux.3sg be.inf elect.part president ‘Last year’s director of the company in Barcelona was elected president.’ b. {/?/*}L’alcalde del PSC l’any passat a Barcelona va  the=mayor of.the PSC the.year last in Barcelona aux.3sg ser elegit president del partit.35 be.inf elect.part president of.the party ‘Last year’s mayor of Barcelona for the socialist party was elected president of the party.’ Interestingly, all these adverbial modifiers are possible when we have a nominal -ble form, as well. (75) a.  el {fins ara / durant molts anys / tantes vegades} alcaldable  the until now during many years so.many times mayor.ble d’esquerres  (Cat) of=left-wing ‘the {until now / for many years / so many times} left-wing candidate for mayor’

35 I should note that these examples are rather difficult to parse, and some speakers reject them altogether, while others consider them only marginal or even fine. As Jaume Mateu (personal communication) points out, they seem to be acceptable only in the journalistic style.



Towards an analysis of Nble   

   291

b. … el hasta ahora alcaldable renunciará a entrar en la lista the until now mayor.ble give.up.3sg.fut to enter.inf in the list electoral (El País, 11/02/03) (Sp) electoral ‘the until now candidate for mayor will give up entering the electoral list.’



c. {/?}els alcaldables fa dos anys pel PSC the mayor.ble do.3sg.prs two years for.the PSC s’han retirat tots de la política.36(Cat) se.have.3pl.prs retire.part all from the politics ‘The two years ago candidates for mayor for the socialist party have all retired from politics.’ Note that these adverbial modifiers are not restricted to appear with posts. As noted in the NGRALE § 13.8i, and exemplified in (76) for Spanish, they are possible with nouns denoting titles, posts or any other similar temporary situations, where the adverbial refers to the temporal interval that the title or post is held, or the temporal point at which this is obtained. (76) a. Perdió en 1972 con el posteriormente campeón(Sp) lose.3sg.pst in 1972 with the subsequently champion ‘He lost with the later champion’ b. Solo queda uno, el inicialmente catedrático de Derecho only remain.3sg.prs one the initially professor of law Político political ‘There remains only one, the one who was initially professor of Political Law.’ c. La actualmente señora Kane gritó: “¡Es una mentira! the currently Mrs Kane shout.3sg.pst  be.3sg.prs a lie ‘The current Mrs Kane shouted: That’s a lie!’ (NGRALE § 13.8i)

36 The contrast in judgments between (74b) and (75c) could be taken as an argument for the presence of v in the case of alcaldable ‘mayor.ble’, even if for alcalde ‘mayor’ this presence is much more controversial (Jaume Mateu, personal communication, for whom (74b) would be ungrammatical). As Jaume Mateu further points out, this would reinforce my analysis of alcaldable as containing an eventive functional layer.

292   

   Case study 2: N-ble

We can link the contrast between Romance and English to a difference in the status of these nominal bases. As I already mentioned, profession-type nouns – as well as all other social roles – have been argued to have adjectival properties in Romance (e.g. Bosque 1989). According to Alexiadou (2001) and Embick (2003, 2004a), adjectives merge with an AspP, which in Embick’s terms is a functional head that “defines a state” that results from a prior event if present. Notice that if we add modality to such an AspP, we obtain the interpretation of a regular -ble adjective, i.e. the possibility of a state resulting from a previous event. As was already suggested in Val Álvaro, denominal -ble adjectives such as alcaldable have a meaning of “some process of change towards some state or quality”. As with unaccusative crecer ‘grow’ in the previous chapter, we can assume a type of vbecome (or perhaps vbe) that merges with a (result) AspP in this case. Thus, on the one hand, these nominal bases are compatible with an AspP in Romance, and its presence is further evidenced by the possibility of having aspectual modification. When these roots combine with AspP, they become predicational and are therefore compatible with -ble. If they merge with nP, they express a referential reading.37 The structure for an adjective like Catalan presidenciable is given in (77) for illustration, after all morphological operations and Vocabulary insertion has taken place. Thus, the key point is whether the highlighted structure corresponds to a predicative head, as in the diagram, or it is an nP, in which case, it could not license ModP and we would obtain a referential noun presidència ‘presidency’.38, 39

37 See also Fábregas (2005: 119), who follows Bosque (1989: 109) in suggesting that roots denoting professions and nationalities have the property of being compatible with either a little n, denoting “the professional group”, or a little a, in which case they denote the function attributed to the professional group. It is this last reading the one we find with -ble, which corresponds to the AspP in the structure (77) above. 38 As is well-known, there is a phonological alternation nt-nci with this set of adjectives/ nouns. Note that this alternation can be accounted for in phonological terms. That -i- is not the locative suffix appearing in minister-i-able ‘ministry-ble’ is well-supported by the fact that the locative suffix selects the class marker or theme vowel /o/ in Spanish, /Ø/ in Catalan, but not /a/ as in presidencia ‘presidency’. On the other hand, the suffix -i in minister-i-able ‘ministryble’ would be realized in the same position as -nci- in the structure (77), as would the suffix -or of other derived stems, such as director or rector. 39 See chapter 2 for the multiplicity of theme vowels in the nominal domain. There I suggested that it is possible that verb-related functional heads are merged together in the nominal domain. Note, however, that this could not be applied to (77). This could possibly be related to the specification of Asp with post-denoting nouns.



   293

Towards an analysis of Nble   

aP

(77)

a

ModP AspP

Mod

v

Asp v

√ presid

v Ø

Asp Th e nt

Mod

Th

Ø

Ø

a

Th

bl

e

Th a

The impossibility of aspectual modification in English would derive from the absence of AspP in their internal structure, which is related to the status of these nouns as non-predicative or non-adjectival in English. This proposal guarantees that the reading of presidenciable ‘presidency.ble’ or alcaldable ‘mayor.ble’ is something like ‘that can become a member of the set of elements described by the predicate president or mayor’. This raises the question on the specification of the vocabulary item nt-nci – and any other nominal suffix in Romance for that matter, which can be inserted both in a little n, where it obtains a referential reading (c.f. Baker 2003, Fábregas 2005) but also in Asp, in which case it will end up obtaining a predicate reading. Additional support for the idea that the difference between English and Romance lies in the presence versus absence of an aspectual head may come from a difference in the semantics of relational nouns in both languages. As Anna Bartra (personal communication) points out, the Catalan examples in (78), that we saw in (48) above, are more often used when the (grand)child, niece or nephew are just born. Despite being relational nouns they can express this recently acquired property, which involves a change of state, a possibility that is not available in English. The same facts obtain in Spanish, as shown in (78c). (78) a. Sóc mare / tieta / àvia.(Cat) be.1sg.prs mother auntie grandmother ‘I am {a mother / an aunt / a grandmother}.’

b. *I’m mother / auntie / grandmother.

294   

   Case study 2: N-ble

c. Soy {madre / tía / abuela}.(Sp) be.1sg.prs  mother aunt grandmother ‘I am {a mother / an aunt / a grandmother}.’ The English facts in (78b) are certainly not surprising, given the overall distribution of the indefinite determiner with predicative nouns, as even professions require its presence, e.g. I am *(a) lawyer. There is extensive literature on this issue, see Roy (2004), Matushansky and Spector (2005), de Swart et al (2007), inter alia. However, the crucial point is that when the presence of some aspectual element makes the change of state explicit in English, such as become or the adverb already, which would involve the presence of an AspP in the structure, ungrammaticality turns into marginality only, as illustrated in (79). (79) a. *Andrea is mother of a boy.

b. ?Andrea {became / is already / has just become} mother of a boy.

As for the requirements imposed by -ble, that of having a theme and implying an originator, my claim is that these are met through the specification of these nouns as posts. Specifically, when they merge with AspP, they obtain a reading related to the social role they denote, i.e. the property reading that is attributed to the class, as suggested in Bosque (1989: 109), be it an occupation, a profession, a nationality, a post, etc. When they denote a post or office, this is necessarily attributed to some person, so that this person corresponds to the theme argument. On the other hand, posts are by definition appointed, designated, or elected by somebody; my claim is that this encyclopedic information can be taken as the implication of an originator. As discussed in chapter 2, the originator required by -ble need not be syntactically present, but it can be merely conceptually implied, i.e. this implication can come from the encyclopedic meaning contributed by the base. That a temporal transition or change of state is a necessary but not sufficient condition is further illustrated by the ungrammaticality of (80). Although we can have in Catalan or Spanish relational kinship terms appearing as bare NPs in predicate position as in (78) above, where they can get a reading of a change of state, and could be said to denote some kind of occupation, which is attributed to a theme argument, they cannot become bases for -ble derivation. This is expected, since the second basic prerequisite of -ble insertion cannot be met, the implication of an originator.40

40 Carme Picallo (personal communication) mentions *catedraticable ‘professor.ble’ (highest title awarded to a professor) as a potential problem. It should be possible, because, she



Towards an analysis of Nble   

   295

(80) * marable, *tietable / *tiable, *aviable(Cat) mother.ble,  auntie.ble  aunt.ble,  grandmother.ble That the specification as a post does not come with the root receives support from the fact that there are cases where the root can give rise to a post despite not denoting a post. Recall the discussion on the example in (23), repeated in (81a). Although the nouns príncep ‘prince’ and co-príncep ‘co-prince’ are titles and are not generally possible as -ble bases, they become posts in Andorra, where these functions are designated, a context that makes them become suitable for -ble derivation. Similarly in (81b), the acceptance of this -ble form will depend on whether we are dealing with a designated monarch or an inherited title, the former would merge with AspP whereas the latter would merge with nP. (81) a. X is presidenciable a França, per tant coprincipable a Andorra.(Cat) x is presidency.ble at France, by so co-prince.ble at Andorra ‘X is a candidate for president in France, therefore a candidate for coprince in Andorra.’ b. monarcable monarch.ble Thus, the difference in their internal configuration could account for the contrast in (81), but it is not enough to deal with the contrast in (82). The interpretation of (82a) is ambiguous. It can mean that the person who holds the position has lasted two years (and is now dead), or, more likely, that he has held the post of mayor for two years only. In (82b), the latter interpretation is only possible if we add something like en el càrrec ‘in the post’/ en el lloc de treball ‘in the workplace’/ a l’empresa ‘in the company’. Otherwise, the sentence can only mean that he is dead, and cannot refer to his quitting his job or his profession.

continues, it could be said to imply an originator, an examination committee. I think that the reason for its ungrammaticality is related to the fact that it is not a post, i.e. catedràtic ‘professor’ does not refer to a temporal post designated or elected, but it is a permanent title awarded by some institution, exactly as any other title, such as llicenciat ‘bachelor’, where we cannot have llicenciable meaning ‘that can become a bachelor’. The fact that we can have doctorable would possibly be related to the fact that this derives from the verb doctorarse which, despite being intransitive, contains the underlying noun doctor, so that it would be interpreted as ‘to become a doctor’ for the purposes of -ble derivation.

296   

   Case study 2: N-ble

(82) a. L’alcalde només ha durat dos anys.(Cat) the=mayor only have.3sg.prs last.part two years ‘The mayor has lasted only two years.’ b. # El lampista només ha durat dos anys. the plumber only have.3sg.prs last.part two years Since both alcalde ‘mayor’ and lampista ‘plumber’ belong to the class of social roles, they can appear as bare NPs in predicative position. This means that they both can license an AspP or an nP. I assume that the difference that gives rise to the contrast in (82) comes from encyclopedic knowledge, probably related to the, also encyclopedic, implication of an originator.41 In English, my informants judge the corresponding sentences as deviant, and weird. Interestingly, they mention that one could coerce the meaning the position of mayor for a sentence like (83a),42 whereas (83b) is outright bad. (83) a. ?The mayor has lasted two years.

b. *The plumber has lasted two years.

This would mean that English does not make a distinction between posts and non-posts, as seems to be the case. Indeed, as I already mentioned, presidenttype nouns in English do not correspond to posts, because they also include titles. As pointed out already in Stowell (1991) and recently developed in Munn and Schmitt (2005), Alexiadou (2005b), and Winter (2005), president-type nouns are defined as singletons. So, the property that they have in common is that they are

41 Note that (i) would be equally ambiguous in the old Germanic world, i.e. either the monarch died after two years in the throne, or he lasted for a period of two years and was dismissed from his post. (i) El monarca només va durar dos anys.(Cat) the monarch only aux.3sg last.inf two years ‘The monarch lasted only two years.’ This would support the view that the specification of these nouns as posts must be linked to the Encyclopedia. Thanks to Anna Bartra and Carme Picallo for comments on this issue. 42 That is, the same kind of coercion we could apply for the second reading in (82b), which shows that also in English mayors and plumbers have different properties. In this case, (83) would illustrate the contrast between Stowell’s (1991) president-type nouns and all other social roles.



Towards an analysis of Nble   

   297

unique, and that is why they include kings, and princes, and treasurers, none of which can be the base of a -ble form.43 Now, things are not so easy, though, because English native speakers report sentences like (84) as fine. So, the presence of the complement seems to be relevant for the licensing of aspectual modification in English, even if temporal -ly adverbs are equally impossible when the complement is made explicit. (84) a. The until then {president of Catalonia / mayor of Barcelona / bishop of Girona} pleaded not guilty.

b. *The currently president of Catalonia

The question is why there is such a difference between Romance and English. It is striking that the presence of the complement can save the sentences that were ungrammatical in (63)–(65), repeated in (85)–(87) for convenience.44 (85) a. *The until then {president /mayor / bishop} pleaded not guilty.

b.  The until then {president of Catalonia / mayor of Barcelona / bishop of Girona} pleaded not guilty.

(86) a. *The liberal party remained in the hands of the now {president / mayor}.

b.  The liberal party remained in the hands of the now {president of the US / mayor of Barcelona}.

(87) a. ?*The still president has announced a new rise of taxes.

b. ?The still president of Spain has announced a new rise of taxes.

Interestingly, there is no such contrast in Catalan or Spanish, for they are all equally fine, as exemplified in (88)–(90) for Catalan.

43 Notice that the Catalan sentence El rei ha durat dos anys ‘the king has lasted two years’, when rei ‘king’ refers to the inherited title, can only mean that he is dead, as with lampista ‘plumber’. Otherwise, one must add en el tro ‘in the throne’. See footnote 41. 44 Perhaps, this could bear some relation to the fact that these nouns are relational nouns and, according to Barker (to appear), English makes a syntactic distinction between alienable and inalienable relational nouns. As shown in (i), “only relational nouns are able to take a postnominal genitive of-phrase”. (i) a. the brother of Mary b. *the cloud of Mary

298   

   Case study 2: N-ble

(88) a. El fins aleshores {president / alcalde / bisbe} es va the until then  president mayor bishop se aux.3sg declarar no culpable.(Cat) declare.inf not guilty ‘The until then {president/mayor/bishop} pleaded not guilty.’ b.  El fins aleshores {president de Catalunya / alcalde de Barcelona / the until then  president of Catalonia mayor of Barcelona bisbe de Girona} es va declarar no culpable. bishop of Girona se aux.3sg declare.inf not guilty ‘The until then {president of Catalonia/ mayor of Barcelona /bishop of Girona} pleaded not guilty.’ (89) a. El partit liberal va quedar en mans de l’ara the party liberal aux.3sg remain.inf in hands of the=now {president / alcalde}. (Cat)  president mayor ‘The liberal party remained in the hands of the current {president /  mayor}.’ b.  El partit liberal va quedar en mans de l’ara the party liberal aux.3sg remain.inf in hands of the=now {president dels Estats Units / alcalde de Barcelona}.  president of.the States United mayor of Barcelona ‘The liberal party remained in the hands of the current {president of the United States / mayor of Barcelona}.’ (90) a. L’encara president ha anunciat una nova pujada de the=still president have.3sg.prs announce.part a new rise of taxes.(Cat) taxes ‘The still president has announced a new rise of taxes.’ b.  L’encara president de l’Estat ha anunciat una the=still president of the=state have.3sg.prs announce.part a nova pujada de taxes. new rise of taxes ‘The still president of the State has announced a new rise of takes.’ At first sight, it seems that in English, the post is understood as a relation and this relation can be delimited by some aspectual modifier.



Towards an analysis of Nble   

   299

An additional piece of evidence that the eventivity of these nouns seems to interact in some way with its relational nature is given in (91). Again, no such contrast exists in Catalan or Spanish, as shown in (92) with Catalan examples. (91) a. *John is {ex-president / ex-mayor / ex-director}.

b. *John is former {president / mayor / director}.



c.  John is {ex-president of the US / ex-mayor of Barcelona / ex-director of the company}.

(92) a. En Joan {és ex-president / ex-alcalde / ex-director}.(Cat) the Joan  be.3sg.prs ex-president ex-mayor ex-director ‘Joan is an {ex-president / ex-mayor / ex-director}’ b.  En Joan és {ex-president dels Estats Units / ex-alcalde the Joan be.3sg.prs {ex-president of.the States United ex-mayor de Barcelona / ex-director de la companyia}. of Barcelona ex-director of the company} ‘Joan is {ex-president of the US / ex-mayor of Barcelona / ex-director of the company}.’ All these contrasts are accounted for once we assume that there is a crucial difference between English and Romance with respect to these nouns. That is, these contrasts arise from the fact that English president-type nouns can only identify the unique element of a singleton set, and therefore it is possible to express that that element is no longer the unique member of that set. No such restriction is active in Catalan or Spanish. These languages allow non-singleton NP predicates to appear without an indefinite article. Thus, English president-type nouns must be analyzed as Munn and Schmitt’s (2005) concealed definites, which denote the unique member of a set, like proper nouns. There is further empirical evidence that this is the right interpretation for these examples.45 In English, if we ask the question in (93a), we can get the answer in (93b), but only if there is sufficient context for it to be clear which city we are talking about to interpret the last clause. The context can come from the

45 Thanks to Eulàlia Bonet and Joan Mascaró for discussing this issue with me and suggesting me the first example.

300   

   Case study 2: N-ble

city the question is asked in, or the city under discussion where all those people live. Without context, the sentence is not good; one would ask mayor of what?46 (93) a. What do the members of your family do for a living?

b.  My father is a doctor, my mother is a lawyer, my sister is a teacher, and her husband is mayor.

Likewise, the sentence in (94b) is fine as long as context provides what country my father is president of.  This is good if we are in the country that the father is president of, or if we are talking about the country that all those family members live in.  If it said a president, one could interpret it as that he holds the position of president in some company or business, but that would involve some kind of coercion. (94) a. What do the members of your family do for a living?

b.  My father is president, my mother is a lawyer, my sister is a teacher, and her husband is a doctor.

Similarly, if I wanted to describe some picture of three people my addressee has never seen before, and I uttered the sentence in (95), it would be fine as long as they are mayor, bishop, and president of the same area, i.e. the president should preside over the country that includes the city the mayor is from and the domain of the bishop.  (95)  The man on the right is mayor, the one on the left is bishop, and the woman is president. Elissa Flagg (personal communication) further points out that if we are in Boston and looking at the mayor of Boston and the president of Mexico, the sentence in (96) would be bad. If we are in Boston and looking at the mayor of Boston and the president of the US, that would be good.  If we are in Toronto and looking at

46 Thanks also to Elissa Flagg and Andrea Rackowski for judgments. Elissa Flagg (personal communication) further specifies that “if were telling you about my family who live in a far away country that I think you know nothing about, I could say My father is a doctor, my mother is a lawyer, my sister is a teacher, and her husband is a mayor but to me that reads like I’m assuming it’s no use to name the city he’s mayor of because you wouldn’t have heard of it anyway. I think in that case omitting the indefinite becomes bad.”



Towards an analysis of Nble   

   301

the mayor of Toronto and the concurrent president of the US, that could be fine because there is no president of Canada, and Canadians are always aware of who the US president is.  (96) The man on the right is mayor and the woman is president. All these examples and judgments clearly argue for a uniqueness restriction in English. Crucially, the corresponding examples in Catalan or Spanish are not constrained in this way, and posts can be interpreted as denoting any member of the set of elements that fit the description, as illustrated in (97)–(98) for Catalan. (97) a. De què fan els membres de la teva família?(Cat) of what do.3pl.prs the members of the your family ‘What do the members of your family do for a living?’ b.  El meu pare és metge, la meva mare és the my father be.3sg.prs doctor, the my mother be.3sg.prs advocada, la meva germana és mestra, i el seu lawyer, the my sister be.3sg.prs teacher, and the her marit és alcalde. husband be.3sg.prs mayor ‘My father is a doctor, my mother is a lawyer, my sister is a teacher, and her husband is a mayor.’ c. El meu pare és president, la meva mare és the my father be.3sg.prs president the my mother be.3sg.prs advocada, la meva germana és mestra, i el seu lawyer the my sister be.3sg.prs teacher and the her marit és metge. husband be.3sg.prs doctor ‘My father is a president, my mother is a lawyer, my sister is a teacher, and her husband is a doctor.’ (98)  L’home de la dreta és alcalde, el de l’esquerra és the=man of the right be.3sg.prs mayor the of the=left be.3sg.prs bisbe, i la dona és presidenta. (CAT) bishop and the woman be.3sg.prs president ‘The man on the right is a mayor, the one on the left is a bishop, and the woman is a president.’

302   

   Case study 2: N-ble

3.1 Predictions of the analysis There is an interesting prediction of this proposal, pointed out to me by Carme Picallo (personal communication). Assuming that in English president-type nouns are concealed definites, they are predicted to cause a definiteness effect in the relevant contexts, whereas the predicative status of the corresponding nouns in Romance would not create such effect. The prediction is borne out in English and Spanish, as illustrated in (99)–(100), respectively. (99) a. *Is there Pope / president / minister? b. *Is there Peter / the Pope / the president? (100) a. (Ya) hay Papa / presidente / ministro?(Sp) (already) there.have.3sg.prs Pope president minister ‘Do we (already) have a {Pope / president / minister}?’ b. *Hay Pedro / el Papa / el president / el ministro?  there.have.3sg.prs Pedro the Pope the president the minister There is no definiteness effect in this type of construction in Catalan,47 as illustrated in (101). (101) a. (Ja) hi ha Papa?(Cat) (already) clloc have.3sg.prs Pope ‘Do we (already) have a Pope?’ b. Hi ha en Pere / el Papa / el president? clloc have.3sg.prs the Pere / the Pope / the president ‘There is Pere / the Pope / the president.’ However, we can find such an effect in Alguerese, the Catalan variety spoken in l’Alguer (Sardinia).48 As predicted, in the absence of the definite article, the question in (102a) is fine, whereas the presence of the definite article in (102b) forces the change of hi ha ‘there has’ to hi és ‘there is’. (102c)–(102d) show that the opposite constructions are impossible, i.e. the construction with haver ‘have’ is incom-

47 But see Rigau (1994), Ramos (1998) and Brucart (2002) for a definiteness effect with pronouns in Catalan existential constructions. 48 Thanks to Joan Mascaró for this remark, and to his informant, Luca Scala for the data.

Conclusions   

   303

patible with the presence of the definite determiner; the absence of a determiner or the presence of an indefinite determiner cannot license the construction with ser ‘be’. (102) a. Ja hi ha Papa?(Alg) already clloc have.3sg.prs Pope ‘Do we already have a Pope?’ b. Ja hi és lo Papa? already clloc be.3sg.prs the Pope ‘Is the Pope already there?’ c. *Hi ha lo Papa  clloc have.3sg.prs the Pope d. *Hi és (un) Papa  clloc be.3sg.prs a Pope There is a second more general prediction of this proposal that I discuss in chapter 5. If on the right track, a distinction based on the predicative nature of the base predicts that other languages where posts can be predicates as well as referring expressions, and where posts are not restricted to express the singleton set, should be able to allow posts as bases for -ble derivation, if they also have a similar -ble rule.

4 Conclusions There are many different factors that interact in the building of denominal -ble adjectives, which should be scrutinized more carefully. The proposal I have outlined is a preliminary one, but it addresses all the data that ultimately needs to be accounted for, and provides a first answer to the questions I posed at the outset. I asked why these adjectives are possible at all given the generally held assumption that -ble attaches to verbs. My answer is that this description is not adequate, to start with. The suffix -ble does not attach to verbs, it needs an eventive (verbalized) root that expresses a transition and that can meet the two requirements imposed by -ble. Posts can meet all these requirements in that they are syntactically eventive, and their specification as ‘posts’ provides the other two requirements: as posts they must be predicated of some human being that provides the external argument of the adjective; and since posts are designated, appointed or elected, they imply the existence of some originator. This also

304   

   Case study 2: N-ble

answers the question of why this set of nouns in particular can become appropriate bases for -ble and how they are characterized. Posts are the only nouns that meet all -ble requirements. That is all that needs to be said. Another question was related to the non-existence of the corresponding verb, that I attribute to either the type of little v present in the structure, or, most probably, to the fact that the structure is phonologically interpreted as a copulative predicate be N. Finally, crosslinguistic variation is due to a difference in the characterization of posts in the different languages. Posts in English are grouped with other concealed definites, whereas in Catalan/Spanish they are rather syntactic predicates. Of course, there are problems that have not been dealt with and new questions arise from this proposal. Among them, the reason why denominal -ble adjectives are more often than not used as nouns or the interaction between the eventivity of these nominals and their classification as relational nouns in English but apparently not so, or at least less so, in Catalan or Spanish. Also, the exact formalization of Romance post-predicates versus English concealed definites in the building of -ble adjectives within the Distributed Morphology framework. Other issues relate to the exact characterization of nouns that admit adverbs in Spanish and Catalan, such as la ahora viuda ‘the now widow’. Further research should include a study to establish which English nouns admit similar uses, such as the now widow Baker, and how this difference is manifested more generally in the contrast between adjectives and nouns in Romance and English. Summing up, this proposal turns an apparently exceptional morphological phenomenon into a process that is predictable from the interaction of the syntactic-semantic properties of the base and the general properties of -ble. To conclude, this chapter shows that the analysis of even a small subset of empirical data can be relevant to and may shed light on the interaction between different components of the grammar.

5 Conclusions and directions for further research I have suggested an analysis of -ble that selects a theme, understood in a broad sense, as its external argument. That is, it can be a theme as in admirable; but also an incremental theme, either in the form of an explicit degree or measure,1 or in the form of a regular DP; or it can be a theme that can be understood as the state resulting from a change, as in alcaldable ‘than can {become / be made / appointed} N’, too. In addition, it requires an originator, which can be syntactically realized (to some extent) or just semantically or encyclopedically implied. The discussion in the previous chapters has demonstrated that -ble adjectives provide strong empirical support for a view of morphology that is the result of the operations of the computational syntax. The two case studies investigated in chapters three and four have proven that the syntactico-semantic and morphological properties of the -ble forms appearing in the constructions V todo lo Vble and Nble result from the interaction of the syntactico-semantic properties of already familiar phenomena with the grammar of -ble suggested in chapter two. In other words, the conclusion is that whenever its requirements are met, a -ble form can be created. Before summarizing the main contributions of this work and concluding in § 3, let me point to some additional theoretical observations and parallelisms worth exploring in future research on V todo lo Vble in § 1 and Nble in § 2.

1 Summary and further research on V todo lo Vble In the case of V todo lo Vble, it has been shown that the prerequisites on -ble are met by the structural properties of the grammar of cognate elements, and the grammar of degrees in interaction with (universal) quantification and the general grammar of achievements, unergative and unaccusative structures, together with some general facts of the grammar of Spanish – concretely, its having a special overt morpheme with the properties of an individuative and a quantitative lo.

1 Here, I include apparent cases of a locative argument, such as (i), where the external argument of the adjective is an incremental theme in that it shows the same kind of eventobject homomorphism. Example (i) was heard on TV3 – the Catalan public television, December 2009. (i)  10 quilòmetres esquiables de pistes (Cat) 10 kilometres skiable.pl of ski-pistes ‘10 skiable kilometres of ski-pistes’

306   

   Conclusions and directions for further research

That is, I have shown that general features of the syntax and semantics of Spanish make available the appropriate kind of configuration that can give rise to a -ble form, and Spanish has decided to make use of this possibility.

1.1 Crosslinguistic variation It would be interesting to investigate whether the same cluster of properties has been made available in other languages. Note that it is precisely in this cluster of features where we should find an answer to the striking fact that V todo lo Vble is possible in Spanish, but it is not in all the other Romance languages, as far as I can tell. It seems to me that the answer could lie in the fact that only Spanish has a specific Vocabulary item lo to express both the individuation of an abstraction (individuative lo), as well as a degree (quantitative lo), unique properties of this particle which may support its relevance to answer this question. Neither Catalan, or French, or Italian, or Portuguese have any element that corresponds to Spanish neuter lo with such functions. According to Anna Bartra (personal communication), it is possible that various other properties can be derived from this fact. As for the parallelism between English and Spanish in the cognate object construction, and the question why it is possible to have an adjectival cognate object in Spanish where English can only have a cognate noun, the answer must also be related to the existence of lo in Spanish. Note that, as in English, adjectives are not possible as cognate objects either, as shown in (1). (1) * enfriar frío, *ensuciar sucio (Sp) to.cool cold  to.dirty dirty Thus, these deadjectival verbs have been shown to convey the meaning become more A, that is they contain a DegP which is the extended projection of the adjective (see e.g. Fábregas 2003, 2005). My contention is that in the absence of an overt realization of the degree head, the adjective must merge with this degree head before raising to merge with the verbalizing head, so that it gets a default comparative interpretation. Exactly the same happens in Spanish, except when the degree head is overtly realized by the Vocabulary item lo, in which case, the lower adjectival root can be realized if all other conditions are met. Further research is certainly necessary, also to explain the very striking fact that in such a configuration English resolves to surface its lower adjectival root as a noun.



Summary and further research on V todo lo Vble    

   307

1.2 Cognation across languages There are still other questions related to the grammar of cognate objects that have not been addressed. Specifically, although languages like Hebrew or Standard Arabic make extensive use of COs and can form a cognate object with essentially any verb (e.g. Horrocks and Stavrou 2010) to express manner adverbial modification, as in (2), Spanish does not have a productive process of forming N-based cognate objects (e.g. Real Puigdollers 2008). Apart from a few cases like (3a), we cannot say things like John sneezed a big sneeze or John smiled a malicious smile in Spanish, as shown in (3b–c). Spanish prefers these objects to be introduced as adjuncts, like (4a), or in a light verb construction, as in (4b). (2) a. Hu nafal nefila kaša (Heb) he fell falling hard (Mittwoch 1998: 314) b. dhahaba dhaha:ba:n sari:’a:n (Ar) he.went a.going quick (Horrocks and Stavrou 2010: 289) (3) a. reír la risa de un niño (Sp) laugh.inf the laugh of a child (Mendikoetxea 1999a: 1610) b. * Juan estornudó un gran estornudo Juan sneeze.3sg.pst a big sneeze c. * Juan sonrió una sonrisa malévola Juan smile.3sg.pst a smile malicious (4) a. Juan sonrió con una sonrisa malévola. Juan smile.3sg.pst with a smile malicious ‘Juan smiled a malicious smile’ (Real Puigdollers 2008) b. Juan hizo un gran estornudo. Juan male.3sg.pst a big sneeze ‘Juan sneezed a big sneeze’ This may appear as problematic for the analysis developed in chapter 3, which is based on a cognate structure. Interestingly, though, Spanish does have a productive process of adjective/participle-based cognate phrases, exemplified in (5) (e.g.

308   

   Conclusions and directions for further research

Bosque 1990; Demonte 1991, 1999; Demonte and Masullo 1999; Armstrong 2011). That means that a kind of cognate configuration is available in the grammar of Spanish, but apparently, only with adjectival bases. This is certainly striking, and even more when we consider it in comparison to English, where cognate adjectives are not possible, as noted in Hale and Keyser (2002). Crucially, in all cases in Spanish, the adjectival cognate phrase corresponds to a measure phrase, i.e. a measure that instantiates the degree achieved in carrying out the action expressed by the verb. This parallelism calls for a detailed contrasting study that has not been pursued in this volume. As a starting point, note that, as in the case of V todo lo Vble, the examples in (5) contain bounded roots, i.e. atelic verbs that become telic in the presence of the cognate measure phrase. (5) a. Ordeñó la vaca bien ordeñada. (Sp) milk.3sg.pst the cow well milk.part ‘S/he milked the cow well milked (=completely)’ b. Lavó la camisa bien lavadita. wash.3sg.pst the shirt well wash.part.dim ‘S/he washed the shirt well washed’ (Demonte 1991) c. limpiarlo bien limpio clean.inf.it well clean ‘to clean it well cleaned’ (Bosque 1990) Interestingly, this cognate adjectival construction is also found in Catalan, which would in a sense reinforce the idea that Catalan does not make use of V todo lo Vble due to the lack of a variable with the properties and functions of Spanish lo.2 (6) a. Aquests pebrots, s’han de trinxar ben trinxats. (Cat) these peppers se=have.3pl.prs of carve.inf well carve.part.pl ‘These peppers must be carved well carved.’

2 Note that there is a particle lo in some varieties of Catalan, and there is a widespread use of lo in certain contexts in colloquial Catalan. However, all speakers consulted reject the parallel Catalan construction *V tot lo Vble. In the case of bilingual speakers, those that accept V todo lo Vble in Spanish radically reject the corresponding construction in Catalan.



Summary and further research on V todo lo Vble    

   309

b. T’hauries de maquillar ben maquillada, com you.refl=have.2sg.cond of make.up.inf well make.up.part.f like una artista. an artist ‘You should make yourself up well made up, like an artist’ (Rigau 2002)

1.3 Degree cognates and resultatives as incremental themes Let me show here the directions to incorporate these degree cognate phrases into the larger set of incremental themes as these have been analyzed in recent work. The proposal I have suggested in chapter 3 equating the explicit expression of a measure phrase with an incremental theme finds further support in recent work developed by Wyngaerd (2001), Wechsler (2005) or Beavers’ (2008, 2011, 2012a, 2012b, in press) on the analysis of resultative phrases, which raises further issues. Although they differ in scope and technical implementation, the main idea behind all these proposals coincides in analyzing the resultative construction as involving an abstract path or scale that measures out the event and must, thus, be homomorphic and coextensive with the event. As with the cognate constructions described in this chapter, adding a resultative secondary predicate turns a sentence telic, as illustrated in (7). (7) a. John hammered the metal {for an hour / *in an hour} b. John hammered the metal flat {*for an hour / in an hour} Wechsler (2005 : 259) The similar analysis of resultatives and unaccusative degree cognate objects as scales along which the event is measured out, puts forward the question about the possibility of analyzing unaccusative cognates as degree-resultatives that require explicit quantification to map the argument onto a specific value-result on the scale denoted by the cognate element. If such a parallelism was established, it would raise additional issues with respect to the existence of resultatives with unaccusative predicates in English, or the absolute impossibility of complex resultatives of the English type with unergative verbs in Spanish, which could

310   

   Conclusions and directions for further research

have remarkable consequences for the grammar of resultative constructions.3 In other words, since both constructions have a similar interpretation and function, if it turns out that the construction presented in chapter 3 can be equated to a resultative predicate, a possibility suggested to me by Richard Larson, we would have discovered the existence of a resultative construction with unergative verbs, a possibility that has been assumed not to exist in the literature on resultatives. The investigation of these facts is already in process.4 There is still a final issue that I would like to mention in this connection. The analysis I have suggested assumes that the degree cognate object materializes a degree argument of gradable verbs, i.e. degree achievements discussed in this paper, but also activities with an unbounded root, since these allow a degree reading of todo lo Vble, in the position of sister to v – recall that this is the position for measuring out the event (Tenny 1994), which has also allowed me to explain the obligatory pleonastic nature of the cognate construction. This brings up the question of the status of measure phrases (e.g. Morzycki 2004, Real Puigdollers 2006; van Riemsdijk 2012), as well as the syntactic relationship between the two internal arguments of degree achievements like el árbol and todo lo crecible in the sentence El árbol creció todo lo crecible ‘The tree grew to its highest possible extent’. On the one hand, according to Tenny’s (1994) Measuring-Out constraint “direct internal arguments are the only overt arguments which can measure out the event”, which implies that todo lo crecible must be a direct internal argument, as I have assumed, even though el árbol is Dowty’s (1979) incremental theme, Wechsler’s (2005) affected argument or Beavers’ (2008, 2011) force-recipient that changes homomorphically with the event as this proceeds, which would mean that we have in fact two arguments that measure out the event, against Tenny’s well-established Measuring-Out constraint. On the other hand, in a very recent manuscript, Marantz (2012 : 15) raises the question about whether “empirically, a root may or may not modify the event and the end state of a change of state VP”

3 See e.g. Mateu (2002) for simple resultatives of the type in (i), which are possible only with very specific verbs. See also Mateu (2010). (i) caer enfermo, volver loco fall.inf ill turn.inf crazy ‘fall ill, turn crazy’ 4 As pointed out to me by Noah Constant, there are in English sentences like He slept everything he could sleep that almost literally correspond to Spanish Durmió todo lo dormible, where we equally find a kind of cognate construction containing modality and obligatory quantification in an object that receives the interpretation of a measure phrase that has the function of an incremental theme, i.e. everything he could sleep could be analyzed along with degree cognates and resultatives.



Summary and further research on Nble   

   311

given that “verbal roots may modify either the event introduced by little v or the state associated with the complement of v” in his model of argument structure, and asks “whether, theoretically, the architecture of grammar prevents this”. If I understand it correctly, it seems to me that this is what seems to occur in examples like El árbol creció todo lo crecible ‘The tree grew up to its highest possible extent’, where the same complex root grow that modifies the verbal event also modifies the final state appearing in complement position. Whether this empirical fact is theoretically welcome or not will have to await further research, though.

1.4 todo lo Vble as an amount relative? There is still another research line that one could pursue with respect to V todo lo Vble. As I already mentioned in the introduction to chapter 3, this construction seems to involve, among other issues, some relation to a relative quantifier cuanto ‘that much’. Indeed, as pointed out to me by Louise McNally (personal communication) and Richard Larson (personal communication), the interpretation of the construction V todo lo Vble resembles that of so-called amount relatives (Carlson 1977b, Grosu and Landman 1998, McNally 2008), illustrated in (7), which refers to the quantity of books that there is on the table, not to the individual books, so that the effect of the quantifier is a maximalization over degrees, similar to El árbol creció todo lo crecible ‘The tree grew to its highest possible degree of growth’. (8) The books there are on the table.

2 Summary and further research on Nble Turning to the other case study, in the investigation of Nble, I have shown that the conditions imposed by the grammar of -ble are fulfilled by the structural properties of the grammar of posts in connivance with the encyclopedic content of this type of nouns. The analysis I have developed in chapter 4 for denominal -ble adjectives makes an interesting prediction for languages that have a similar Vocabulary item [bl] with similar syntactico-semantic properties. In these languages, if nouns denoting posts can appear as heads of a bare NP, i.e. have predicative properties, they should license forms similar to Catalan or Spanish alcaldable ‘mayor.ble’.

312   

   Conclusions and directions for further research

2.1 Post-denoting -ble across languages When we consider other Romance languages like French, Italian or Portuguese, the prediction relating the grammar of post-denoting nouns and -ble adjectives seems to be borne out. Thus, as discussed in work by Roy (2004) for French, or Munn and Schmitt (2005) for Brazilian Portuguese, it turns out that these languages behave like Catalan and Spanish with respect to the predicative status of posts and other social roles, as illustrated in (9) for French. (9) a. Que fait Paul dans la vie? – Paul est dentiste. (Predicational) (Fr) what does Paul in the life Paul is dentist ‘What does Paul do for a living? – Paul is a dentist’ b. Qui est Paul? – Paul est un dentiste. who is Paul Paul is a dentist ‘Who is Paul? – Paul is a dentist.’ (Roy 2004 : 32)

(Identificational)

As expected, similar denominal adjectives are possible in French (10), Italian (11) and Portuguese (12), even if to different degrees of productivity.5 (10) ministrable, présidentiable, sénateurable (Fr) minister.ble presidency.ble senator.ble ‘that can {become / be made /appointed} N’ or ‘candidate for N’ (11) ministrabile, papabile6 (It) minister.ble Pope.ble ‘that can {become / be made /appointed} N’ or ‘candidate for N’ (12) prefeitável, ministeriável, presidenciável (Por) mayor.ble minister.ble presidency.ble ‘that can {become / be made /appointed} N’ or ‘candidate for N’ The novel fact is that I have found a few examples of denominal -ble adjectives in languages like German, in (13), and also in Dutch, as in (14).

5 Thanks to Cosima Vergari (Italian) and Marion Florencio (French) for help with data. The Portuguese examples have been checked against the Portal da Língua Portuguesa from the Ministério da Ciéncia, Tecnologia e Ensino Superior. 6 According to Val Álvaro (1981 : 191n), the Spanish papable would be a loan from Italian papabile, and the rest of forms would seem to have been derived by analogy.



Summary and further research on Nble   

   313

(13) ministerabel, rektorabel, bürgermeisterabel(Ger) minister.ble dean.ble mayor.ble ‘that can {become / be made /appointed} N’ or ‘candidate for N’ (14) ministerabel, burgemeesterabel(Dut) minister.ble mayor.ble ‘that can {become / be made /appointed} N’ or ‘candidate for N’ As exemplified in (15) for German, and in (16) for Dutch, their meaning is exactly the same as the meaning of the corresponding forms in Romance. (15) a. Der Kandidat muss durch den Senat als rektorabel angesehen the candidate must through the senate as dean.ble look.part werden. (W) (Ger) become ‘The candidate must be considered as ‘deanable’ by the senate’ b.  Genauso offen wie die Frage, wer bei der FDP den eigentlich just.as open like the question who at the FDP then actually ministerabel sei. minister.ble be.3sg.kjv1 ‘Just as open as the question of who should actually be appointed candidate to mayor in the FDP’

c. Es gehe darum, ob die Kandidaten grundsätzlich it be.3sg.kjv1 about.that whether the candidates in.principle direktorabel seien. director.ble be.3pl.kjv1 ‘It is about whether the candidates could in principle be appointed director.’ (16) Burgemeesterabel was ze toch al niet …  mayor.ble be.3sg.pst she but although not ‘Although she was not a candidate for mayor …’

(W) (Dut)

Note that for instance in German, nouns like Bürgermeister ‘mayor’, like posts in Spanish or Catalan, are episodic, i.e. eventive, and can therefore be temporally and aspectually delimited with deictic and temporal event modifiers (Higginbotham 1985), as shown in (17).

314   

   Conclusions and directions for further research

(17) a.  Die Idee dazu hatte der noch Bürgermeister dem the.f idea for.it have.3sg.pst the.m still mayor the.dat Rat vor einigen Monaten selbst vorgetragen. (W) (Ger) council before some.dat months self report.part ‘The person who was still mayor had reported the idea for that to the Council some months ago’ b.  Der früher Bürgermeister Walser hat das Haus in einem the.m former mayor Walser have.3sg.prs the.n house in a.dat schwäbischen Gedicht beschrieben. Swabian poem described ‘Former mayor Walser has described the House in a Swabian poem.’ The generalization is that these languages must behave rather like Romance languages with respect to post-nouns, and not like English. Indeed, as noted in de Swart et al. (2005, 2007), the prediction is borne out since both Dutch and German can take bare NPs headed by nouns denoting professions, nationalities and alike, as exemplified in (18)–(19).7 (18) Er ist Arzt. he be.3sg.prs doctor

(Ger)

(19) Jan is leraar. Jan be.3sg.prs teacher (de Swart et al. 2005 : 447)

(Dut)

Thus, it seems that there may be a parametric difference based on the type of N, whether a predicate or a concealed definite, posts are specified for in the different languages. The different degrees of productivity of these forms are merely the consequence of a language-particular decision on making use or not of mechanisms made available by the grammar.

7 As Artemis Alexiadou (personal communication) pointed out to me, Greek is a potential counterexample for this claim. Although it behaves like Catalan or Spanish in all relevant respects, it has no denominal -ble forms. This counterevidence to my proposal will have to await further research.



Summary and further research on Nble   

   315

2.2 The case of Italian Nble With respect to the type of base that can be selected by -ble across languages, the exceptionality of Italian deserves a special consideration. As was mentioned in footnote 46 in chapter 2, Italian shows additional contrasts, in that it allows the suffix -bile to attach to vehicles to form adjectives that modify roads and alike, with the meaning that the road can be driven by that type of vehicle, e.g. strada camionabile ‘road for heavy vehicles’ or strada rotabile ‘carriageway’. Additional examples are repeated in (20). (20) camionabile, carrozzabile, ciclabile, rotabile (It) truck.ble carriage.ble cycle.ble rail.ble Also, with nouns denoting a place, -bile can be used to predicate of a noun that the object denoted by the noun can be placed there, as in libro tascabile ‘pocket book’. None of these forms are possible in Catalan or Spanish. The analysis of Catalan and Spanish Nble developed in the previous chapter cannot be extended to account for these cases, since a form like camionabile, if anything – because it is non-eventive, it would be interpreted as ‘that can become a truck’. The strategy for an inclusive analysis of all Nble forms in the Romance languages could go through a uniform treatment of Nble forms as meaning ‘designated or appointed for N’, so that we could have someone appointed for mayor in Romance alcaldable, and road designated (or designed) for trucks in Italian camionabile. This would possibly imply the introduction of an additional hidden or implied zero root with the meaning of appoint/designate.8

2.3 Intralinguistic variation: German -bar/-wert/-abel There is a final point that I would like to mention with respect to the grammar of German -ble. Note that the German Nble forms in (15) above show neither -bar nor -wert, but the very specific suffix -abel. We saw in chapter 2 that -abel is generally described as a non-productive suffix traditionally reserved for bases of Latin origin, so that in fact we would have expected the default -bar in German new Nble formations. Interestingly, this a priori unproductive affix seems to have found a revival in that it has taken over the selection of nominal post-denoting bases that are not necessarily of Latin origin like bürgermeisterabel ‘mayor.ble’.

8 Thanks to Ora Matushansky for suggesting this possible line of inquiry.

316   

   Conclusions and directions for further research

(21)  Handfest, nicht unsympathisch, aber etwas desinformiert und keinesfalls bürgermeisterabel. Auch er wollte sich ganz offensichtlich den Traum erfüllen, einmal Bürgermeisterkandidat gewesen zu sein. (Ger) ‘Reliable, not unpleasant, but a bit uninformed and under no circumstances candidate for mayor. Obviously he wanted to fulfill the dream of being once candidate for mayor.’ In the absence of a detailed analysis of the grammar of -bar/-wert/-abel, I informally include this possibility in my oversimplified list of German Vocabulary items in (22), where -abel now specializes for insertion with Latin roots and nominal post-denoting bases, -bar still being the default exponent. (22) a. [abel] ↔ a / Mod (with √[+Latin] & √Post)

b. [wert] ↔ a / Mod (with √[+Psych])



c. [bar] ↔ a / Mod

As I argued for in chapter 2, the existence of three different pieces of vocabulary in the grammar of German, as opposed to a single piece in English or Romance, not only poses no additional problem for the unified analysis of -ble, but in fact provides strong empirical support for the syntactic Distributed Morphology model assumed in this volume in that both intralinguistic as well as crosslinguistic variation are easily and elegantly reduced to a question of vocabulary insertion in a late insertion approach that distinguishes phonological exponents from morphosyntactic and semantic features.

3 Concluding remarks One of the best studied and most intriguing areas of inquiry in morphology has been the productive suffix -ble. However, although its morpho-phonology has generated an immense interest, and its internal structure and external syntax has received some attention in the literature, its semantic properties and their interaction with its morpho-syntax have rather been disregarded. In this respect, I have filled the gap to a certain extent by describing various ways in which the semantics of -ble interacts with its external syntax, e.g. to restrict the expression of event-related arguments in high -ble, as well as with their morphology, e.g. to give rise to different modal interpretations.



Concluding remarks   

   317

More importantly, this volume makes a new empirical and descriptive contribution with theoretical consequences for morphological theory. Despite the fact that the suffix -ble has been examined to different degrees by many linguists and it appears prominently in grammars and other works on morphology none of the previous accounts has ever mentioned the attested -ble formations productively derived from unaccusative and unergative verbs in constructions of the form V todo lo Vble in Spanish, and have thus overlooked the important theoretical contribution to the nature of morphology that these involve. As I have argued, these forms identify a new paradox, a completely new theoretical problem concerning the status of words that can only appear in a particular construction. Their existence raises a number of theoretical questions, e.g. how the generation of these words can be properly constrained, or what the restricting context is. I have shown that only a syntactic approach is able to accommodate the complexities of the elements involved in a principled way. As for denominal -ble adjectives, their importance and relevance for linguistic analysis has been completely ignored in essentially all previous work. The discussion on Nble, despite important remaining puzzles, has also revealed how syntax and semantics interact in the morphological derivation of these forms. I have shown that a decompositional analysis of -ble is well-supported by empirical data, and can accommodate simple as well as morphologically more complex languages. Such syntactic account has allowed me to develop a unified approach to the four types of -ble  – potential or high -ble, low -ble, V todo lo Vble, and Nble – that provides support for a Distributed Morphology model where one distinguishes between syntactic positions containing morpho-syntactic features, and Vocabulary items that are inserted post-syntactically. It is important to emphasize that the four types of -ble that are found in Spanish have never been analyzed in depth from a unified perspective, nor has the analysis of -ble been approached from a comparative perspective in an attempt to parameterize the different types of -ble adjectives that are available in this variety of languages. Such crosslinguistic investigation sheds some light into the much debated question on why certain types of word formation operations are attested in some languages but are absent in others, a kind of variation that seems tied to the vocabulary in the case of V todo lo Vble, a welcome result. Also in the case of German -bar/-wert/-abel, which contrasts with the single -ble in Romance or English, the source of variation must be located in the vocabulary. To conclude, the work reported here contributes new empirical data to the study of grammatical interfaces and introduces a strong empirical argument for syntactic approaches to word formation. It presents two clear arguments showing that derivational operations are the result of syntactic computation and post-syntactic vocabulary insertion. This study also illustrates some of the benefits to be

318   

   Conclusions and directions for further research

gained from investigating into an apparently very small phenomenon, such as the grammar of -ble. Also, the examination of apparently marginal phenomena, such as V todo lo Vble or Nble constructions in Romance, and how they contrast with the grammar of English or German not only provides striking data and challenging theoretical observations, but I hope that it contributes to the understanding of the nature of morphology and its crucial place at the interface with syntax and semantics.

References Abraham, Werner. 1970. Passiv und Verbableitung auf E. -able, Dt. -bar. Folia Linguistica 4: 38–52. Abraham, Werner and Larisa Leisiö. 2006. Passivization and Typology. Form and Function. Amsterdam, Philadelphia: John Benjamins. Ackema, Peter and Maaike Schoorlemmer. 1994. The middle construction and the syntaxsemantics interface. Lingua 93: 59–90. Ackema, Peter and Maaike Schoorlemmer. 1995. Middles and nonmovement. Linguistic Inquiry 26: 173–197. Ackema, Peter and Maaike Schoorlemmer. 2005. Middles. In: Martin Everaert and Henk van Riemsdijk (eds.) The Blackwell Companion to Syntax, 131–203. Malden, MA and Oxford: Blackwell. Acquaviva, Paolo. 2009. Roots and lexicality in Distributed Morphology. In: Alexandra Galani, Daniel Redinger and Norman Yeo (eds.) York Papers in Linguistics 2. York-Essex Morphology Meeting 2, 1–21. York: Department of Language and Linguistic Science, University of York. Adams, Valerie. 2001. Complex Words in English. Edinburgh, Harlow, Essex: Pearson Education Limited. Albrespit, Jean. 2009. Is the trust-ee trust-able? On English suffixes with a passive meaning. Groninger Arbeiten zur Germanistischen Linguistik. Alcina Franch, Juan and José Manuel Blecua. 1988. Gramática española. Barcelona: Ariel. Alexiadou, Artemis. 2001. Functional Structure in Nominals: Nominalization and Ergativity. Amsterdam, Philadelphia: John Benjamins. Alexiadou, Artemis. 2005a. Gerund types, the present participle and patterns of derivation. In: Claudia Marienborn and Angelika Wöllstein-Leisten (eds.) Event Arguments in Syntax, Semantics and Discourse, 139–152. Alexiadou, Artemis. 2005b. Possessors and (in)definiteness. Lingua 115: 787–819. Alexiadou, Artemis and Elena Anagnostopoulou. 2008. Structuring participles. In: Charles B. Chang and Hannah J. Haynie (eds.) Proceedings of the 26th West Coast Conference on Formal Linguistics, 33–41. Somerville, MA: Cascadilla Proceedings Project. Alexiadou, Artemis, Elena Anagnostopoulou and Martin Everaert (eds.). 2004. The Unaccusativity Puzzle. Explorations of the Syntax-Lexicon Interface. Oxford: Oxford University Press. Alexiadou, Artemis and Florian Schäffer. 2006. Instrument subjects are Agents or Causers. In: Donald Baumer, David Montero, and Michael Scanlon (eds.) Proceedings of the 25th West Coast Conference on Formal Linguistics, 40–48. Somerville, MA: Cascadilla Proceedings Project. Altmann and Kemmeling. 2000. Wortbildung fürs Examen. Wiesbaden: Westdeutscher Verlag. Anagnostopoulou, Elena. 2003. Participles and voice. In: Artemis Alexiadou, Monika Rathert, and Arnim von Stechow (eds.) Perfect Explorations, 1–36. Berlin: Mouton de Gruyter. Anagnostopoulou, Elena and Yota Samioti. 2009. Domains for idioms. Hand-out of a talk, Roots Workshop, Universität Stuttgart, June 10–12th. Anagnostopoulou, Elena and Yota Samioti. 2010. Locality constraints on the interpretation of words. Hand-out of a talk, On Linguistic Interfaces, University of Ulster, December 2–4th. Anderson, Stephen R. 1977. Comments on the paper by Wasow. In: P. Culicover, T. Wasow and A. Akmajian (eds.) Formal Syntax. New York: Academic Press.

320   

   References

Anderson, Stephen R. 1992. A-Morphous Morphology. Cambridge: Cambridge University Press. Arad, Maya. 1998. VP-structure and the syntax-lexicon interface. Ph.D. Dissertation, University College London. Arad, Maya. 2003. Locality constraints on the interpretation of roots: The case of Hebrew denominal verbs. Natural Language and Linguistic Theory 21: 737–778. Armstrong, Grant. 2011. On the adjectival component of change of state verbs in Spanish. University of Wisconsin-Madison, Ms. [Retrieved from author’s webpage in December 2011] Aronoff, Mark. 1976. Word Formation in Generative Grammar. Cambridge, MA, London, England: The MIT Press. Aronoff, Mark. 1994. Morphology by Itself. Stems and Inflectional Classes. Cambridge: The MIT Press. Arregi, Karlos. 2000. How the Spanish verb works. Handout of a talk given at LSRL 30, at the University of Florida, Gainesville. [Available at http://home.uchicago.edu/~karlos/] Asatiani, Rusudan. 2001. Conceptual structure of reflexive and middle. Ms, Georgian Academy of Sciences, Oriental Institute. Azkarate, Miren and Lluïsa Gràcia. 1995. Agentivity and modality in deverbal adjectives in Basque and in Catalan. Rivista di Grammatica Generativa 20: 3–32. Baayen, Harald. 1992. Quantitative aspects of morphological productivity. In: Geert Booij and Jan van Marle (eds.) Yearbook of Morphology 1991, 109–149. Dordrecht: Kluwer. Baker, Mark. 1985. The Mirror Principle and morphosyntactic explanation. Linguistic Inquiry 1: 373–415. Baker, Mark. 1988. Incorporation: A Theory of Grammatical Function Change. Chicago & London: University of Chicago Press. Baker, Mark. 2003. Lexical Categories: Verbs, Nouns, and Adjectives. Cambridge: Cambridge University Press. Bartra Kaufmann, Anna. 2002. La passiva i les construccions que s’hi relacionen. In: Joan Solà, Maria Rosa Lloret, Joan Mascaró, and Manuel Pérez Saldanya (eds.) Gramàtica del català contemporani, Vol. 2: 2111–2179. Barcelona: Empúries. Bartra Kaufmann, Anna and Xavier Villalba. 2006. Agreement and predicate inversion in Spanish DP. In: Jenny Doetjes and Paz González. (eds.) Romance Languages and Linguistic Theory 2004, 23–41. Amsterdam/Philadelphia: John Benjamins. Batiukova, Olga. 2006. Las oraciones medias como proyección de estructuras subléxicas. In: Milka Villayandre Llamazares (ed.) Actas del XXXV Simposio Internacional de la Sociedad Española de Lingüística. León: Universidad de León. [Available at http:/www3.unileon.es/ dp/dfh/SEL/actas.htm] Bauer, Laurie. 1983. English Word-Formation. Cambridge: Cambridge University Press. Bauer, Laurie. 2001. Morphological Productivity. Cambridge: Cambridge University Press. Bauer, Laurie. 2005. Productivity: Theories. In: Pavol Štekauer and Rochelle Lieber (eds.) Handbook of Word-Formation, 315–334. The Netherlands: Springer. Beard, Robert. 1995. Lexeme-Morpheme Base Morphology. Albany , New York: SUNY Press. Beavers, John. 2008. Scalar complexity and the structure of events. In: Johannes Dölling, Tatjana Heyde-Zybatow, and Martin Schäfer (eds.) Event Structures in Linguistic Form and Interpretation, 245–265. Berlin: Mouton de Gruyter. Beavers, John. 2011. On affectedness. Natural Language and Linguistic Theory. 29: 335–370, doi 10.1007/s11049-011-9124-6.

References   

   321

Beavers, John. 2012a. Lexical aspect and multiple incremental themes. In: Violeta Demonte and Louise McNally (eds.) Telicity and Change of State in Natural Language: Implications for Event Structure. Oxford: Oxford University Press. Beavers, John. 2012b. Resultative constructions. In: Robert I. Binnick (ed.) The Oxford Handbook on Tense and Aspect. Oxford: Oxford University Press. Beavers, John. in press. Aspectual classes and scales of change. Linguistics. [Updated draft 5th December 2011, retrieved from the author’s webpage in January 2012] Belletti, Adriana and Luigi Rizzi. 1988. Psych verbs and theta theory. Natural Language and Linguistic Theory 6.3: 291–352. Bennis, Hans and Pim Wehrmann. 1990. On the categorial status of present participles. In: Reineke Bok-Bennema and Peter Coopmans (eds.) Linguistics in the Netherlands 1990, 1–12. Dordrecht: Foris. Benzing, Joseph. 1931. Zur Geschichte von ser als Hilfszeitwort bei den in­transitiven Verben im Spanischen. Zeitschrift für romanische Philologie 51: 385–460. Bermúdez-Otero, Ricardo. 2006. Morphological structure and phonological domains in Spanish denominal derivation. In: Fernando Martínez-Gil and Sonia Colina (eds.) OptimalityTheoretic Studies in Spanish Phonology, 278–311. Amsterdam: John Benjamins. Bertinetto, Pier Marco and Mario Squartini. 1995. An attempt at defining the class of ‘gradual completion verbs’. In: Pier Marco Bertinetto, Valentina Bianchi, James Higginbotham, and Mario Squartini (eds.) Temporal Reference, Aspect and Actionality. Vol 1: Semantic and Syntactic Perspectives, 11–26. Torino: Rosenberg and Sellier. Beyssade, Claire and Carmen Dobrovie-Sorin. 2005. A syntax-based analysis of predication. In: Effi Georgala and Jonathan Howell (eds.) Proceedings of Semantics and Linguistic Theory 15, 44–61. Cornell University, Ithaca, NY: CLC Publications. Bhatt, Rajesh. 1999. Covert modality in non-finite contexts. Ph.D. Dissertation, University of Pennsylvania, Institute for Research in Cognitive Science Technical Report No. IRCS-00-01. Bhatt, Rajesh and Roumyana Pancheva. 2005. Implicit arguments. In: Martin Everaert and Henk van Riemsdijk (eds.) The Blackwell Companion to Syntax, Vol. II: 554–584. Malden, MA and Oxford: Blackwell. Bolinger, Dwight. 1972. Degree Words. The Hague, The Netherlands: Mouton. Bonet, Eulàlia. 1995. The where and how of clitic order. Revue Québécoise de linguistique 24.1: 61–81. Bonet, Eulàlia and Daniel Harbour. 2012. Contextual allomorphy. In: Jochen Trommer (ed.) The Morphology and Phonology of Exponence – The State of the Art, 195–235. Oxford: Oxford University Press. Booij, Geert E. 1992. Morphology, semantics and argument structure. In: Iggy M. Roca (ed.) Thematic Structure: Its Role in Grammar, 47–64. Berlin, New York: Foris. Borer, Hagit. 1990. V + ing: It walks like an adjective, it talks like an adjective. Linguistic Inquiry 21: 95–102. Borer, Hagit. 1994. On the projection of arguments. In: Elena Benedicto and Jeffrey Runner (eds.) Functional Projections, 19–48. GLSA Amherst: UMass. Borer, Hagit. 1998. Deriving passives without theta roles. In: Steven G. Lapointe and Diane K. Brentari (eds.) Morphology and Its Relation to Syntax, 60–99. Stanford: CSLI Publications. Borer, Hagit. 2001. Exo-skeletal vs. endo-skeletal explanations: Syntactic projections and the lexicon. In: Maria Polinsky and John Moore (eds.) Explanation in Linguistic Theory, 31–67. Stanford: CSLI.

322   

   References

Borer, Hagit. 2004. The grammar machine. In: Artemis Alexiadou, Elena Anagnostopoulou and Martin Everaert (eds.) The Unaccusativity Puzzle. Explorations of the Syntax-Lexicon Interface. Oxford: Oxford University Press. Borer, Hagit. 2005. The Normal Course of Events. Oxford: Oxford University Press. Borer, Hagit. 2008. The syntax of words. Lecture Series held at the 10th Seoul International Conference on Generative Grammar, Dongguk University, Seoul, July 2008. Borgonovo, Claudia. 1999. Participios activos. Nueva Revista de Filología Hispánica 47: 281–303. Bosque, Ignacio. 1989. Las categorías gramaticales. Madrid: Síntesis. Bosque, Ignacio. 1990. Sobre el aspecto en los adjetivos y en los participios. In: Ignacio Bosque (ed.), Tiempo y aspecto en español, 177–214. Madrid: Cátedra. Bosque, Ignacio. 1996. El sustantivo sin determinación: la ausencia de determinante en la lengua española. Madrid: Visor. Bosque, Ignacio. 1998. Sobre los complementos de medida. In: Nicole Delbecque and Christian de Paepe (eds.) Estudios en honor del Profesor Josse de Kock, 57–73. Leuven: Leuven University Press. Bosque, Ignacio. 1999a. El sintagma adjetival. Modificadores y complementos del adjetivo. Adjetivo y participio. In: Ignacio Bosque and Violeta Demonte (eds.) Gramática descriptiva de la lengua española, Vol. 1: 217–310. Madrid: Espasa. Bosque, Ignacio. 1999b. Sobre la gramática de los contextos modales: Entornos modales y expresiones inespecíficas en español. In: José Antonio Samper Padilla and Magnolia Troya Déniz (eds.) Actas del XI Congreso Internacional de la Asociación de Lingüística y Filología de la América Latina, Vol.1: 43–58. Universidad Las Palmas de Gran Canaria: Servicio de Publicaciones. Bosque, Ignacio and Javier Gutiérrez-Rexach. 2009. Fundamentos de sintaxis formal. Madrid: Akal. Bosque, Ignacio and Pascual José Masullo. 1998. On verbal quantification in Spanish. In: Olga Fullana and Francesc Roca (eds.) Studies on the Syntax of Central Romance Languages. Proceedings of the III Workshop on the Syntax of Central Romance Languages 1996, 9–63. Girona: Publicacions de la Universitat de Girona. Bosque, Ignacio and Juan Carlos Moreno. 1990. Las construcciones con lo y la denotación del neutro. Lingüística 2: 5–50. Brown, Mark. 1988. On the logic of ability. Journal of Philosophical Logic 17.1: 1–26. Brucart, Josep M. 1999. La estructura del sintagma nominal: Las oraciones de relativo. In: Ignacio Bosque and Violeta Demonte eds. Gramática descriptiva de la lengua española, Vol. 1: 395–522. Madrid: Espasa. Brucart, Josep M. 2002. Els determinants. In: Joan Solà, Maria Rosa Lloret, Joan Mascaró, and Manuel Pérez Saldanya (eds.) Gramàtica del català contemporani, Vol. 2: 1435–1516. Barcelona: Empúries. Brucart, Josep M. and Gemma Rigau. 2002. La quantificació. In: Joan Solà, Maria Rosa Lloret, Joan Mascaró, and Manuel Pérez Saldanya (eds.) Gramàtica del català contemporani, Vol. 2: 1517–1589. Barcelona: Empúries. Bruening, Benjamin. 2012. By-phrases in passives and nominals. Syntax 16: 1–41. Burzio, Luigi. 1986. Italian Syntax. Dordrecht: Reidel. Cabredo Hofherr, Patricia. 2010. Adjectives: An introduction. In: Patricia Cabredo Hofherr, Ora Matushansky (eds.) Adjectives. Formal Analyses In Syntax And Semantics, 1–26. Amsterdam: John Benjamins.

References   

   323

Calzado Roldán, Araceli. 2000. La impersonalidad de los verbos meteorológicos: una explicación pragmático-discursiva. Dicenda 18: 85–108. Camacho, José. 2012. Ser and estar: The individual/stage-level distinction and aspectual predication. In: José Ignacio Hualde, Antxón Olarrea, and Erin O’Rourke (eds.) The Handbook of Hispanic Linguistics, 453–476. Oxford: Blackwell. Cano Aguilar, Rafael. 1981. Estructuras sintácticas transitivas en el español actual. Madrid: Gredos. Cano Cambronero, Mª Ángeles. in progress. Las derivaciones agentivas en -nte y -dor: estructura argumental y complejidad estructural en una morfología sintáctica. Ph.D. Dissertation, Universidad Autónoma de Madrid. Carnap, Rudolf. 1947. Meaning and Necessity. Chicago: University of Chicago Press. Carlson, Greg N. 1977a. A unified analysis of the English bare plural. Linguistics and Philosophy 1: 413–457. Carlson, Greg N. 1977b. Amount relatives. Language 53: 520–542. Chapin, Paul. 1967. On the syntax of word-formation in English. Ph.D. Dissertation, Massachusetts Institute of Technology. Chierchia, Gennaro. 1989. A semantics for unaccusatives and its syntactic consequences. Ms., Cornell University, Ithaca, New York. Chierchia, Gennaro. 1995. Individual-level predicates as inherent generics. In: Carlson, Gregory N. and Francis Jeffry Pelletier (eds.) The Generic Book, 176–223. Chicago and London: The University of Chicago Press. Chierchia, Gennaro. 2004. A semantics for unaccusatives and its syntactic consequences. In: Artemis Alexiadou, Elena Anagnostopoulou, and Martin Everaert (eds.) The Unaccusativity Puzzle: Explorations of the Syntax-Lexicon Interface, 22–59. Oxford: Oxford University Press. Chierchia, Gennaro and Sally McConnell-Ginet. 1990. Meaning and Grammar. An Introduction to Semantics. Cambridge, MA, London, England: The MIT Press, second edition. Chomsky, Noam. 1972. Language and Mind. New York: Harcourt. Chomsky, Noam. 1986. Barriers. Cambridge, MA: The MIT Press. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: The MIT Press. Chomsky, Noam. 1999. Derivation by Phase. MIT Occasional Papers in Linguistics, Department of Linguistics and Philosophy, MIT, Cambridge, MA. Chomsky, Noam and Morris Halle. 1968. The Sound Pattern of English. New York: Harper & Row. Cinque, Guglielmo. 1990. Ergative adjectives and the lexicalist hypothesis. Natural Language and Linguistic Theory, 8: 1–39. Cinque, Guglielmo. 1999. Adverbs and Functional Heads: A Cross-Linguistic Perspective. New York, Oxford: OUP. Cinque, Guglielmo. to appear. A note on mood, modality, tense and aspect affixes in Turkish. In: Eser Erguvani-Taylan (ed.) Verb in Turkish: the core element of clause structure. Amsterdam: Benjamins. Condoravdi, Cleo. 1989. The middle voice: Where semantics and morphology meet. MITWPL, 11, 16–30. Cornips, Leonie and Aafke Hulk. 1999. Affected objects in Herleen Dutch and Romance. Languages in Contrast 1: 191–210. CREA. Corpus de Referencia del Español Actual. [Available at http://corpus.rae.es/creanet.html] Cresswell, M.J. 1976. The semantics of degree. In: Barbara Partee (ed.) Montague Grammar, 261–292. San Diego, CA: Academic Press.

324   

   References

Dayal, Veneeta. 1998. Any as inherently modal. Linguistics and Philosophy 21: 433–476. Demonte, Violeta. 1991. Temporal and aspectual constraints on predicative APs. In: Héctor Campos and Fernando Martínez-Gil (eds.) Current Studies in Spanish Linguistics, 165–200. Washington, DC: Georgetown University Press. Demonte, Violeta. 1999. El adjetivo: Clases y usos. La posición del adjetivo en el sintagma nominal. In: Ignacio Bosque and Violeta Demonte (eds.) Gramática descriptiva de la lengua española, Vol. 1: 129–215. Madrid: Espasa. Demonte, Violeta. 2005. Meaning-form correlations and the order of adjectives in Spanish. In: Chris Kennedy y Louise McNally (eds.) The Semantics of Adjectives and Adverbs, 71–100. Oxford: Oxford University Press. Demonte, Violeta and Pascual José Masullo. 1999. La predicación: Los complementos predicativos. In: Ignacio Bosque and Violeta Demonte (eds.) Gramática descriptiva de la lengua española, Vol. 2: 2461–2523. Madrid: Espasa. Depréz, Viviane. 2005. Morphological number, semantic number and bare nouns. Lingua 115: 857–883. Di Sciullo, Anna Maria. 1997. Selection and derivational affixes. In: Wolfgang U. Dressler, Martin Prinzhorn, and John R. Rennison (eds.) Advances in Morphology, 79–95. Berlin, New York: Mouton de Gruyter. Di Sciullo, Anna Maria and Edwin Williams. 1987. On the Definition of Word. Cambridge, MA: The MIT Press. Dixon, R.M.W. 2005. A Semantic Approach to English Grammar. Oxford: Oxford University Press. Dixon, R.M.W. and Alexandra Y. Aikhenvald (eds.). 2004. Adjective Classes. A Cross-linguistic Typology. Oxford: Oxford University Press. Dowty, David R. 1979. Word Meaning and Montague Grammar. Dordrecht: Reidel. Dryer, Matthew S. 1985. The role of thematic relations in adjectival passives. Linguistic Inquiry 16.2: 320–326. Dubinsky, Stanley and Silvester Ron Simango. 1996. Passive and stative in Chichewa: Evidence for modular distinctions in grammar. Language 72: 749–781. Duo de Brottier, Ofelia. 2000. La formación de adjetivos en -able/-ible. In: Gerd Wotjak (ed.) En torno al sustantivo y adjetivo en español actual, 359–368. Frankfurt am Main: Vervuert/ Madrid: Iberoamericana. Embick, David. 2000. Features, syntax, and categories in the Latin perfect. Linguistic Inquiry 31: 185–230. Embick, David. 2001. Participial structures and their morphological realization. Talk presented at International Workshop on Participles, Tübingen, 17.–18. 04. 01. Embick, David. 2003. Locality, listedness, and morphological identity. Studia Linguistica 57: 143–169. Embick, David. 2004. On the structure of resultative participles in English. Linguistic Inquiry 35: 355–392. Embick, David and Morris Halle. 2000. Aspects of the Latin conjugation. Ms, MIT, Cambridge, Mass. Embick, David and Morris Halle. 2005. On the status of stems in morphological theory. In: Twan Geerts, Ivo van Ginneken, Haike Jacobs (eds.) Romance Languages and Linguistic Theory, Selected Proceedings from Going Romance 2003, Amsterdam, 37–72. Amsterdam: John Benjamins. Embick, David and Alec Marantz. 2008. Architecture and blocking. Linguistic Inquiry 39 : 1–53. Embick, David. 2010. Localism versus Globalism in Morphology and Phonology. Cambridge, MA: MIT Press.

References   

   325

Enç, Mürvet. 1981. Tense without scope: An analysis of nouns as indexicals. Ph.D. Dissertation, The University of Wisconsin-Madison. Enç, Mürvet. 1991. The semantics of specificity. Linguistic Inquiry 22.1 : 1–25. Érsek, Iván. 1977. Praktisches Lehrbuch Ungarisch. Berlin, München, Wien, Zürich, New York: Langenscheidt. Espinal, Ma. Teresa. 2004. Lexicalization of light verb structures and the semantics of nouns. Catalan Journal of Linguistics 3­: 15–43. Espinal, Ma. Teresa and Carmen Dobrovie-Sorin. 2006. Tipología semántica de los nombres escuetos. El caso particular de los nombres escuetos singulares contables. Research report CGT-06-12, UAB. Fabb, Nigel. 1984. Syntactic affixation. Ph.D. Dissertation, Massachusetts Institute of Technology. Fabb, Nigel. 1988. English suffixation is constrained only by selectional restrictions. Natural Language and Linguistic Theory 6 : 527–539. Fábregas, Antonio. 2003. Los verbos de realización gradual. Revista Española de Lingüística, 322: 475–506. Fábregas, Antonio. 2005. La definición de la categoría gramatical en una morfología orientada sintácticamente: Nombres y adjetivos. Ph.D. Dissertation, Universidad Autónoma de Madrid. Fábregas, Antonio. 2011. Propiedades sintácticas y alomorfía: Alternancias con el sufijo -ble. In: José Pazó, Irene Gil, and Mª Ángeles Cano (eds.) Teoría morfológica y morfología del español, 59–85. Madrid: UAM Ediciones. Fagan, Sarah M. B. 1988. The English middle. Linguistic Inquiry 19 : 181–203. Fagan, Sarah M. B. 1992. The Syntax and Semantics of Middle Constructions. Cambridge/New York: CUP. Fellbaum, Christiane and Anne Zribi-Hertz. 1989. The Middle Construction in French and English: A Comparative Study of its Syntax and Semantics. Bloomington, Indiana: Indiana University Linguistics Club Publications. Fernández Leborans, Ma. Jesús. 1999. La predicación. Las oraciones copulativas. In: Ignacio Bosque and Violeta Demonte (eds.) Gramática descriptiva de la lengua española, Vol. 2: 2357–2460. Madrid: Espasa. Fintel, Kai von. 2005. Modality and language. In: Donald M. Borchert (ed.) Encyclopedia of Philosophy. Second Edition. Fleischer, Wolfgang and Barz, Irmhild. 1995. Wortbildung der deutschen Gegenwartssprache. Tübingen: Niemeyer. Flury, Robert. 1964. Struktur- und Bedeutungsgeschichte des Adjektiv-Suffixes -bar. Winterthur: Verlag P. Keller. Fodor, Janet Dean. 1970. The Linguistic Description of Opaque Contexts. Bloomington, Indiana: Indiana University Linguistic Club Publications. Folli, Raffaella and Harley, Heidi. 2004. Flavors of v: Consuming results in Italian and English. In: Roumyana Slabakova and Paula Kempchinsky (eds.) Aspectual Inquiries, 95–120. Dordrecht: Kluwer. Folli, Raffaella and Heidi Harley. 2007. Causation, obligation, and argument structure: On the nature of little v. Linguistic Inquiry 38: 2, 197–238. Fu, Jingqi, Thomas Roeper and Hagit Borer. 2001. The VP within process nominals: Evidence from adverbs and the VP anaphor do-so. Natural Language and Linguistic Theory 19: 549–582.

326   

   References

García Negroni, M. Marta. 2002. La construcción media con SE. In: Cristina Sánchez López (ed.) Las construcciones con SE, 275–308. Madrid: Visor Libros. Gavarró, Anna and Brenda Laca. 2002. Les perífrasis temporals, aspectuals i modals. In: Joan Solà, Maria Rosa Lloret, Joan Mascaró, and Manuel Pérez Saldanya (eds.) Gramàtica del català contemporani, Vol. 3: 2663–2726. Barcelona: Empúries. Gildersleeve, Basil, and Gonzalez Lodge. 1895. Latin grammar. London: Macmillan. Givón, T. 1970. Notes on the semantic structure of English adjectives. Language 46, 4, 816–837. Göksel, Aslɩ and Kerslake Celia. 2005. Turkish. A Comprehensive Grammar. London and New York: Routledge. Gómez Fernández, Araceli. 2006. Estudio morfosemántico de los adjetivos deverbales en -ble en francés y español desde la Teoría Sentido-Texto. Salamanca: Ediciones Universidad de Salamanca. González Romero, Luisa. 2002. Estudios de gramática contrastiva: las construcciones medias en inglés y español. Huelva: Universidad de Huelva. Gràcia, Lluïsa. 1992. -Ble adjectives and middle constructions: A problem for inheritance. Catalan Working Papers in Linguistics, 163–182. Gràcia, Lluïsa. 1995. Morfologia lèxica. L’herència de l’estructura argumental. València: Servei de Publicacions de la Universitat de València. Gràcia, Lluïsa, M. Teresa Cabré, Soledad Varela, Miren Azkarate, Mercè Lorente, Elisenda Bernal. 2000. Configuración morfológica y estructura argumental: Léxico y diccionario. Gipuzkoa: Servicio Editorial Universidad del País Vasco. Grosu, Alexander and Fred Landman. 1998. Strange relatives of the third kind. Natural Language Semantics 6: 125–170. Gruber, Jeffrey. 1965. Studies in lexical relations. Ph.D. Dissertation, Massachusetts Institute of Technology. Guerzoni, Elena. 2000. Stress and morphology in the Italian verb system. Ms, MIT. Gumiel-Molina, Silvia and Isabel Pérez-Jiménez. 2012. Aspectual composition in structures: Adjectival scalarity and verbal aspect in copular constructions. Borealis. An International Journal of Hispanic Linguistics 1.1: 33–62. Gutiérrez-Rexach, Javier. 1999. The structure and interpretation of Spanish degree neuter constructions. Lingua 109 : 35–63. Gutiérrez-Rexach, Javier. 2009. Correlativization and degree quantification in Spanish. In: Pascual José Masullo, Erin O’Rourke and Chia-Hui Huang (eds) Romance Linguistics 2007: Selected Papers from the 37th Linguistic Symposium on Romance Languages, 121–141. Amsterdam: John Benjamins. György, Dorogman. 1992. Diccionario Español-Húngaro. Magyar-Spanyol Kéziszótár. Budapest: Akadémiai Kiadó. Hackl, Martin. 1998. On the semantics of ‘ability attributions’. Ms, MIT. Hacquard, Valentine 2009. On the interaction of aspect and modal auxiliaries. Linguistics and Philosophy 32: 279–315. Hale, Ken and Samuel Jay Keyser. 1987. A view from the middle. Lexicon Project Working Papers 10. Cambridge, MA: MITWPL. Hale, Ken and Samuel Jay Keyser. 1988. Explaining and constraining the English middle. In: Carol Tenny (ed.) Studies in Generative Approaches to Aspect. Lexicon Project Working Papers 24: 41–58. Cambridge, MA: MIT.

References   

   327

Hale, Ken and Samuel Jay Keyser. 1993. On argument structure and the lexical expression of syntactic relations. In: Ken Hale and Samuel Jay Keyser (eds.) The View from Building 20: Essays in Honor of Sylvain Bromberger, 53–110. Cambridge, MA: The MIT Press. Hale, Ken and Samuel Jay Keyser. 2002. Prolegomenon to a Theory of Argument Structure. Cambridge, MA: The MIT Press. Hall, Fitzedward. 1877. On English Adjectives in -Able, with Special Reference to Reliable. London: Trübner and Co., Ludgate Hill. Halle, Morris. 1997. Distributed Morphology: Impoverishment and fission. MIT Working Papers in Linguistics 30, 425–449. Cambridge, MA: MITWPL. Halle, Morris and Alec Marantz. 1993. Distributed Morphology and the pieces of inflection. In: Ken Hale and Samuel Jay Keyser (eds.) The View from Building 20: Essays in Honor of Sylvain Bromberger, 11–176. Cambridge, MA: The MIT Press. Halle, Morris and Alec Marantz. 1994. Some key features of Distributed Morphology. MIT Working Papers in Linguistics 21: 275–288. Harley, Heidi. 1995. Subjects, events and licensing. MIT, Ph.D. Dissertation. Harley, Heidi. 1999. Denominal verbs and Aktionsart. In: Liina Pylkkänen and Angelieke van Hout, (eds.) Proceedings of the 2nd Penn/MIT Roundtable on Event Structure, 73–85. Cambridge, MA: MITWPL. Harley, Heidi. 2005. How do verbs get their names? Denominal verbs, manner incorporation and the ontology of verb roots in English. In: Nomi Erteshik-Shir and Tova Rapoport (eds.), The Syntax of Aspect. Deriving Thematic and Aspectual Interpretation, 42–64. Oxford: OUP. Harley, Heidi. (2012). External arguments and the Mirror Principle: On the distinctness of Voice and v. Lingua, 125: 34–57. Harley, Heidi and Rolf Noyer. 1998. Licensing in the non-lexicalist lexicon: Nominalizations, Vocabulary Items and the Encyclopaedia. MITWPL 32: 119–137. Harley, Heidi and Rolf Noyer. 2000. Formal versus encyclopedic properties of vocabulary: Evidence from nominalisations. In: Bert Peeters (ed.) The Lexicon-Encyclopedia Interface, 349–374. Amsterdam: Elsevier. Harris, James W. 1991. The exponence of gender in Spanish. Linguistic Inquiry 22: 27–62. Haugen, Jason D. 2009. Hyponymous objects and Late Insertion. Lingua 119: 242–262. Hay, Jen, Christopher Kennedy, and Beth Levin. 1999. Scale structure underlies telicity in ‘degree achievements’. In: Tanya Matthews and Devon Strolovitch (eds.) Semantics and Linguistic Theory 9. Ithaca, NY: CLC Publications. 127–144. Heim, Irene. 1987. Where does the definiteness restriction apply? Evidence from the definiteness of variables. In: Eric Reuland and Alice ter Meulen (eds.) The Representation of (In)definiteness, 21–42. Cambridge, MA: The MIT Press. Heim, Irene and Christopher Kennedy. 2002. Course materials for 24.979 Topics in Semantics. Fall 2002. MIT OpenCourseWare. [Available at http://ocw.mit.edu/]. Higginbotham, James. 1985. On semantics. Linguistic Inquiry 16,4: 547–593. Higginbotham, James. 1987. Indefiniteness and predication. In: Eric J. Reuland and Alice G.B. ter Meulen (eds.) The Representation of (In)definiteness, 43–70. Cambridge, MA: The MIT Press. Hirschbühler, Paul. 1988. The middle and the pseudo-middle in French. In: David Birdsong and Jean-Pierre Montreuil (eds.) Advances in Romance Linguistics, 97–111. Dordrecht: Foris.

328   

   References

Hoekstra, Teun and Ian Roberts. 1993. Middle constructions in Dutch and English. In: Eric Reuland and Werner Abraham (eds.) Knowledge and Language 2, 183–220. Dordrecht: Kluwer Academic Publishers. Horrocks, Geoffrey and Melita Stavrou. 2010. Morphological aspect and the distribution of the cognate object construction across languages. In: Malka Rappaport Hovav, Edit Doron, and Ivy Sichel (eds.) Syntax, Lexicon and Event Structure. Studies in Honor of Annita Mittwoch, 284–308. Oxford University Press. Huddleston, Rodney and Geoffrey Pullum. 2002. The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press. Hutchison, John. 1981. The Kanuri Language. A Reference Grammar. University of Wisconsin. Institut d’Estudis Catalans, Corpus textual informatitzat de la llengua catalana. http://pdl.iec.es/entrada/paraules.asp Institut d’Estudis Catalans, Diccionari de la llengua catalana DIEC. http://pdl.iec.es/entrada/ diec.asp Intensive Chichewa ©. 1969. Likuni Press and Publishing House. P.O. Box 133, Lilongwe, Malawi. Distributed for non-commercial purposes. Available at http://www.cnr.berkeley. edu/~alyons/chichewa/) Iwata, Seizi. 1999. On the status of implicit arguments in middles. Journal of Linguistics, 35: 527–553. Jackendoff, Ray. 1972. Semantic Interpretation in Generative Grammar. Cambridge, MA: MIT Press. Jaeggli, Osvaldo. 1986. Passive. Linguistic Inquiry, 17: 582–622. Jones, Michael Allan. 1988. Cognate objects and the case-filter. Journal of Linguistics 24: 89–110. Jung, Walter. 1967. Grammatik der deutschen Sprache. Leipzig: VEB Kratzer, Bibliographisches Institut Leipzig. Kayne, Richard S. 1981. Unambiguous paths. In: Robert May and Jan Koster (eds.) Levels of Syntactic Representation, 143–183. Dordrecht: Foris. Kayne, Richard S. 1984. Connectedness and Binary Branching. Dordrecht: Foris Publications. Kayne, Richard S. 1994. The Antisymmetry of Syntax. Cambridge: The MIT Press. Kemmer, Suzanne. 1993. The Middle Voice. Amsterdam/Philadelphia: John Benjamins. Kennedy, Christopher. 1999. Projecting the Adjective: The Syntax and Semantics of Gradability and Comparison. New York/London: Garland. Kennedy, Christopher. 2004. Comparatives, Semantics of. Contribution to the Encyclopedia of Language and Linguistics, Second Edition. Ms, University of Chicago. [Published in 2006 by Elsevier] Kennedy, Christopher and Beth Levin. 2008. Measure of change: The adjectival core of degree achievements. In: Louise McNally and Christopher Kennedy (eds.) Adjectives and Adverbs: Syntax, Semantics and Discourse, 156–182. Oxford: Oxford University Press. Kennedy, Christopher and Louise McNally. 1999. From event structure to scale structure: Degree modification in deverbal adjectives. In: Tanya Matthews and Devon Strolovitch (eds.) Semantics and Linguistic Theory 9, 163–180. Ithaca, NY: CLC Publications. Kennedy, Christopher and Louise McNally. 2005. Scale structure, degree modification, and the semantics of gradable predicates. Language 81: 345–381. Keyser, Samuel Jay and Thomas Roeper. 1984. On the middle and ergative constructions in English. Linguistic Inquiry 15: 381–416.

References   

   329

Kiefer, Ferenc. 1981. What is possible in Hungarian? Acta Linguistica Academiae Scientiarum Hungaricae 31.1-4: 147–185. Kiparsky, Paul. 1973. ‘Elsewhere’ in phonology. In: Stephen R. Anderson and Paul Kiparsky (eds.) A Festschrift for Morris Halle, 93–106. New York: Holt, Rinehart and Winston. Kiparsky, Paul. 1998. Partitive case and aspect. In: Miriam Butt and Wilhelm Geuder (eds.) The Projection of Arguments. Lexical and Compositional Factors, 265–308. Stanford: CSLI Publications. Kiparsky, Paul. 2001. Structural case in Finnish. Lingua 111: 315–376. Kovacci, Ofelia. 1999. El adverbio. In: Ignacio Bosque and Violeta Demonte eds. Gramática descriptiva de la lengua española, Vol. 1: 705–786. Madrid: Espasa. Kratzer, Angelika. 1981. The notional category of modality. In: Hans-Jurgen Eikmeyer and Hannes Rieser (eds.) Words, Worlds, and Contexts. New Approaches in Word Semantics, 38–74. Berlin: de Gruyter. Kratzer, Angelika. 1991. Modality. In: Arnim von Stechow and Dieter Wunderlich (eds.) Semantics: An International Handbook of Contemporary Research, 639–650. Berlin: Walter de Gruyter. Kratzer, Angelika. 1995. Stage-level and individual-level predicates. In: Gregory N. Carlson and Francis Jeffry Pelletier (eds.) The Generic Book, 125–175. Chicago and London: The University of Chicago Press. Kratzer, Angelika. 1996. Severing the external argument from its verb. In: Johan Rooryck and Laurie Zaring (eds.) Phrase Structure and the Lexicon, 109–137. Dordrecht, Boston: Kluwer. Kratzer, Angelika. 2000. Building statives. Paper presented at Berkeley Linguistic Society. Kratzer, Angelika. 2002. Telicity and the meaning of objective case. University of Massachusetts at Armherst, Ms. [Available at semanticsarchive.net] Krifka, Manfred. 1989. Nominal reference, temporal constitution and quantification in event semantics. In: Renate Bartsch, Johan van Benthem and Peter van Emde Boas (eds.) Semantics and Contextual Expressions, 75–115. Dordrecht: Foris Publications. Krifka, Manfred. 1998. The origins of telicity. In: Susan Rothstein (ed.) Events and Grammar, 197–235. Dordrecht: Kluwer. Krifka, Manfred, Francis Jeffry Pelletier, Gregory N. Carlson, Alice ter Meulen, Godehard Link, and Gennaro Chierchia. 1995. Genericity: An introduction. In: Carlson, Gregory N. and Francis Jeffry Pelletier (eds.) The Generic Book, 1–124. Chicago and London: The University of Chicago Press. Kuno, Susumo and Ken-ichi Takami. 2004. Functional Constraints in Grammar. On the Unergative-Unaccusative Distinction. Amsterdam/Philadelphia: John Benjamins. Kupferman, Lucien. 1991. Structure événementielle de l’alternance un/Ø devant les noms humains attributs. Langages 102: 52–75. Laca, Brenda. 1993. Las nominalizaciones orientadas y los derivados españoles en -dor y -nte. In: Soledad Varela (ed.) La formación de palabras. Madrid: Taurus. Laca, Brenda. 1999. Presencia y ausencia de determinante. In: Ignacio Bosque and Violeta Demonte (eds.) Gramática descriptiva de la lengua española, Vol. 1: 891–928. Madrid: Espasa. Lang, Mervyn F. 1990. Formación de palabras en español. Morfología derivativa productiva en el léxico moderno. Madrid: Cátedra. Larson, Richard K. 1988. On the double object construction. Linguistic Inquiry 19: 335–391.

330   

   References

Lazard, Gilbert. 2003. What is an object in a crosslinguistic perspective? In: Giuliana Fiorentino (ed.) Romance Objects: Transitivity in Romance Languages, 1–16. Berlin, New York: Mouton de Gruyter. Lázaro Carreter, Fernando. 1971. Diccionario de términos filológicos. Madrid: Gredos. Lecarme, Jacqueline. 1999. Nominal tense and tense theory. In: Francis Corblin, Jean Marie Marandin and Carmen Dobrovie-Sorin (eds.) Empirical Issues in Formal Syntax and Semantics 2, 333–354. The Hague: Holland Academic Graphics. Lecarme, Jacqueline. 2008. Tense and modality in nominals. In: Jacqueline Guéron and Jacqueline Lecarme (eds.) Time and Modality. Studies in Natural Language and Linguistic Theory 75. Dordrecht: Springer. Lee, Hanjung. 1999. Aspectual and thematic licensing of grammatical case. Chicago Linguistics Society 35: 203–222. Lekakou, Maria. 2005. In the middle, somewhat elevated. University College London, Ph.D. Dissertation. Leonetti, Manuel. 1990. El artículo y la referencia. Madrid: Taurus. Leonetti, Manuel. 1999. El artículo. In: Ignacio Bosque and Violeta Demonte (eds.) Gramática descriptiva de la lengua española, Vol. 1: 787–890. Madrid: Espasa. Leonetti, Manuel. 2012. Indefiniteness and specificity. In: José Ignacio Hualde, Antxon Olarrea and Erin O’Rourke (eds.) Handbook of Hispanic Linguistics, 285–305. Madrid: Espasa. Leumann, Manu, Johann Baptist Hofmann, and Anton Szantyr. 1963. Lateinische Grammatik, auf der Grundlage des Werkes von Friedrich Stolz und Joseph Hermann Schmalz. München: Beck Verlag. Levin, Beth. 1987. The middle construction and ergativity. Lingua 71: 17–31. Levin, Beth. 2007. LSA Course Materials. [Available at http://www.stanford.edu/~bclevin/lsa07. html]. Levin, Beth and Malka Rappaport. 1986. The formation of adjectival passives. Linguistic Inquiry 17: 623–661. Levin, Beth and Malka Rappaport. 1988. Non-event -er nominals: A probe into argument structure. Linguistics 26: 1067–1083. Levin, Beth and Malka Rappaport Hovav. 1995. Unaccusativity at the Syntax – Lexical Semantics Interface. Cambridge, Ma, London, England: The MIT Press. Levinson, Lisa. 2010. Arguments for pseudo-resultative predicates. Natural Language and Linguistic Theory 28: 135–182. Lieber, Rochelle. 1980. On the organization of the lexicon. Ph.D. Dissertation, Massachusetts Institute of Technology. Lin, Jimmy. 2004. Event structure and the encoding of arguments: the syntax of the mandarin and English verb phrase. Ph.D. Dissertation, Massachusetts Institute of Technology. Luján, Marta. 1972. On the so-called neuter article in Spanish. In: Jean Casagrande and Bohdan Saciuck (eds.) Generative Studies in the Romance Languages, 162–174. Rowley: Newbury House Publishers. Luján, Marta. 2002. Determiners as modified pronouns. Círculo de Lingüística Aplicada a la Comunicación 9: 19–34. Lyons, John. 1977. Semantics. Cambridge: Cambridge University Press. Vol.2. Macfarland, Talke T. 1995. Cognate objects and the argument/adjunct distinction in English. Ph.D. Dissertation, Northwestern University.

References   

   331

Maling, Joan. 1993. Of nominative and accusative: The hierarchical assignment of grammatical case in Finnish. In: Anders Holmberg and Urpo Nikanne (eds.) Case and Other Functional Categories in Finnish Syntax, 49–76. The Hague: Mouton de Gruyter. Maling, Joan, Jong Sup Yun, and Soowon Kim. 2001. Case-marking on duration adverbials revisited. Selected Papers from the 12th International Conference of Korean Linguistics, 323–335. Marantz, Alec. 1984. On the Nature of Grammatical Relations. Cambridge, MA: The MIT Press. Marantz, Alec. 1995. ‘Cat’ as a phrasal idiom: Consequences of late insertion in Distributed Morphology. Ms, MIT. Marantz, Alec. 1997. No escape from syntax: Don’t try morphological analysis in the privacy of your own lexicon. Proceedings of the 21st Annual Penn Linguistics Colloquium, UPenn Working Papers in Linguistics 4.2, 201–225. Philadelphia: Penn Linguistics Club. Marantz, Alec. 1999a. Creating words above and below ‘little v’. Hand-out, MIT. Marantz, Alec. 1999b. Reconstructing the lexical domain with a single generative engine. Hand-out, MIT. Marantz, Alec. 2001. Words. Ms, MIT. Marantz, Alec. 2005. Rederived generalizations. Talk held in Paris, 13. 1 2.05. Marantz, Alec. 2012. Verbal argument structure. Events and participants. NYU, Ms. Marchand, Hans. 1969. The Categories and Types of Present-Day English Word-Formation. Munich: C. H. Beck. Marelj, Marijana. 2004. Middles and Argument Structure across Languages. Utrecht, The Netherlands: LOT. Mascaró, Joan. 1983. La fonologia catalana i el cicle fonològic. Bellaterra: UAB. Mascaró, Joan. 1986. Morfologia. Barcelona: Enciclopèdia Catalana. Masolo, Claudio, Laure Vieu, Emanuele Bottazzi, Carola Catenacci, Roberta Ferrario, Aldo Gangemi, Nicola Guarino. 2004. Social roles and their descriptions. In: Proceedings of the International Conference on Principles of Knowledge Representation and Reasoning (KR_04), 267–277. AAAI Press. Massam, Diane. 1990. Cognate objects as thematic objects. The Canadian Journal of Linguistics 35: 2, 161–190. Massam, Diane. 1992. Null objects and non-thematic subjects. Journal of Linguistics 28, 115–137. Mateu, Jaume. 2002. Argument structure: relational construal at the syntax-semantics interface. Ph.D. Dissertation, Universitat Autònoma de Barcelona. Mateu, Jaume. 2010. Conflation vs. incorporation processes and argument structure. Handout of a talk, CCHS-CSIC, October 25th. Matushansky, Ora and Benjamin Spector. 2005. Tinker, tailor, soldier, spy. In: Emar Maier, Corien Bary, and Janneke Huitink. (eds.) Proceedings of SuB9, 241–255. McFadden, Thomas. 2004. The position of morphological case in the derivation: A study of the syntax-morphology interface. Ph.D. Dissertation, University of Pennsylvania. McGinnis, Martha. 2000. Event heads and the distribution of psych-roots. In: Alexander Williams and Elsi Kaiser (eds.) Current Work in Linguistics: University of Pennsylvania Working Papers in Linguistics 6: 107–144. McGinnis, Martha. 2010. Evidence for the syntactic attachment of -able. Handout of a talk, CUNY Conference on the Word in Phonology, New York, January 14–15th.

332   

   References

McIntyre, Andrew. 2006. Get-passives, silent reflexive – middle morphemes and the deconstruction of causation. Hand-out, Universität Leipzig. [Available at www.uni-leipzig. de/~angling/mcintyre] McIntyre, Andrew. 2011. Adjectival passives and adjectival participles in English. Ms, Université de Neuchâtel. [Available at http://www3.unine.ch/andrew.mcintyre/page-300.html] McNally, Louise. 2008. DP-internal only, amount relatives, and relatives out of existentials. Linguistic Inquiry 39.1 : 161–169. Meltzer-Asscher, Aya. 2012. The subject of adjectives: Syntactic position and semantic interpretation. The Linguistic Review 29 : 149–189. DOI 10.1515/tlr-2012-0007 Mendikoetxea, Amaya. 1998. Aspectos sintácticos y semánticos de la oraciones medias. Talk presented at SEL XXVIII, Madrid. Mendikoetxea, Amaya. 1999a. Construcciones inacusativas y pasivas. In: Ignacio Bosque and Violeta Demonte (eds.) Gramática descriptiva de la lengua española, Vol. 2: 1575–1630. Madrid: Espasa. Mendikoetxea, Amaya. 1999b. Construcciones con se: Medias, pasivas e impersonales. In: Ignacio Bosque and Violeta Demonte (eds.) Gramática descriptiva de la lengua española, Vol. 2: 1631–1722. Madrid: Espasa. Menéndez-Benito, Paula. 2013. On dispositional sentences. In: Alda Mari, Claire Beyssade, and Fabio del Prete (eds.) Genericity, 276–292. Oxford: Oxford University Press. Miguel, Elena de. 1986. Papeles temáticos y regla de formación de adjetivos en -ble. Dicenda. Cuadernos de Filología Hispánica 5: 159–81. Miguel, Elena de. 1992. El aspecto en la sintaxis del español: perfectividad e impersonalidad. Madrid: Ediciones de la UAM. Miguel, Elena de. 1999. El aspecto léxico. In: Ignacio Bosque and Violeta Demonte (eds.) Gramática descriptiva de la lengua española, Vol. 2: 2977–3060. Madrid: Espasa. Miguel, Elena de y Marina Fernández Lagunilla. 2000. El operador aspectual -se. Revista Española de Lingüística 30: 13–43. Mittwoch, Anita. 1998. Cognate objects as reflections of davidsonian event arguments. In: Susan Rothstein (eds.), Events and Grammar, 309–332. Dordrecht: Kluwer. Morzycki, Marcin. 2004. Measure DP adverbials: Measure phrase modification in VP. Ms, Université du Québec à Montréal. Müller, Stefan. 2003. The morphology of German particle verbs: Solving the bracketing paradox. Journal of Linguistics 39: 275–325. Munn, Alan and Cristina Schmitt. 2005. Number and indefinites. Lingua 115: 821–855. Musan, Renate. 1995. On the temporal interpretation of noun phrases. Ph.D. Dissertation, Massachusetts Institute of Technology. Nakajima, Heizo. 2006. Adverbial cognate objects. Linguistic Inquiry 37.4: 674–684. Nevins, Andrew Ira. 2002. One -able, two attachment sites. Ms, MIT. Noyer, Ralf. 1992. Features, positions, and affixes in autonomous morphological structure. Ph.D. dissertation, MIT. Ojeda, Almerindo. 1984. A note on the Spanish neuter. Linguistic Inquiry 14.1 : 171–174. Oltra-Massuet, Isabel. 1999a. On the notion of theme vowel: A new approach to Catalan verbal morphology. SM Thesis, Massachusetts Institute of Technology. [I cite from the version appeared as MIT Occasional Papers in Linguistics 19, Department of Linguistics and Philosophy, MIT, Cambridge, MA. in 2000] Oltra-Massuet, Isabel. 1999b. On the constituent structure of Catalan verbs. In: Karlos Arregi, Benjamin Bruening, Cornelia Krause and Vivian Lin (eds.) Papers in Morphology and

References   

   333

Syntax, Cycle One, MIT Working Papers in Linguistics 33, 279–322. Cambridge, MA: MITWPL. Oltra-Massuet, Isabel. 2010. On the morphology of complex adjectives. Ph.D. Dissertation, Universitat Autònoma de Barcelona. Oltra-Massuet, Isabel. 2012. Notes sobre els adjectius passius modals en alemany i català. In: Michael Pfeiffer, Teresa Vinardell and Anna Montané. (eds.). Was mich wirklich interessiert. Homenatge a Jordi Jané, 177–187. Girona: Documenta Universitaria (Edició Forum). Oltra-Massuet, Isabel. to appear. On the status of impossible words systematically derived by the grammatical system. Proceedings of the 47th Annual Conference of the Chicago Linguistic Society. Chicago: Chicago Linguistic Society. Oltra-Massuet, Isabel and Karlos Arregi. 2005. Stress-by-structure in Spanish. Linguistic Inquiry 36.1: 43–84. Oshita Hiroyuki. 1994. Argument structure template and formation of English deverbal adjectives. MITWPL 23: 247–265. In: Proceedings of the Sixth Annual Student Conference in Linguistics, University of Rochester, edited by Chris Giordano and Daniel Ardron. Partee, Barbara. 1970. Opacity, coreference, and pronouns. Synthese 21: 359–385. Partee, Barbara. 1974. Opacity and scope. In: Milton Munitz and Peter Unger (eds.) Semantics and Philosophy. New York: NYU Press. Partee, Barbara and Vladimir Borschev. 2003. Genitives, relational nouns, and argument-modifier ambiguity. In: Ewald Lang, Claudia Maienborn, and Catherine Fabricius-Hansen (eds.) Modifying Adjuncts Interface Explorations 4, 67–112. Berlin, New York: Mouton de Gruyter. Payne, John and Rodney Huddleston. 2002. Nouns and noun phrases. In: Rodney Huddleston and Geoffrey Pullum (eds.) The Cambridge Grammar of the English Language, 323–523. Cambridge: Cambridge University Press. Pereltsvaig, Asya. 1999. Two classes of cognate objects. Proceedings of the 17th West Coast Conference on Formal Linguistics, 537–551. Pereltsvaig, Asya. 2000. On accusative adverbials in Russian and Finnish. In: Artemis Alexiadou and Peter Svenonius (eds.) Adverbs and Adjunction. Linguistics in Potsdam 6 : 155–176. Pérez Jiménez, Isabel. 2006. La gramàtica de las cláusulas absolutas de predicación en español. Ph.D. Dissertation, Universidad Complutense de Madrid. Pérez Jiménez, Isabel and Norberto Moreno Quibén. 2003. Argumentos a favor de la centralidad de las nociones aspectuales en la interfície léxico-sintaxis: la correlación telicidad-inacusatividad en español. In Juan Cuartero Otal and G.erd Wotjiak (eds.) Algunos problemas específicos de la descripción sintáctico-semántica, 197–212. Berlín: Frank and Timme. Pérez Jiménez, Isabel and Norberto Moreno Quibén. 2005. ¿Son todos los verbos inacusativos aspectualmente télicos en español? El papel de la telicidad en la interficie léxico-sintaxis. In: Pablo Cano López (ed.) Actas del VI Congreso de Lingüística General, 1807–1820. Santiago de Compostela. Pérez Saldanya, Manuel. 1999. El modo en las subordinadas relativas y adverbiales. In: Ignacio Bosque and Violeta Demonte (eds.) Gramática descriptiva de la lengua española, Vol. 2: 3253–3322. Madrid: Espasa. Perlmutter, David M. 1978. Impersonal passives and the unaccusative hypothesis. Berkeley Linguistics Society 4: 157–189. Pesetsky, David. 1987. Binding problems with experiencer verbs. Linguistic Inquiry 18: 126–140. Pesetsky, David. 1995. Zero Syntax. Experiencers and Cascades. Cambridge, MA: The MIT Press.

334   

   References

Picallo, Carme. 1991. Nominals and nominalizations in Catalan. Probus 3: 279–316. Piera, Carlos and Soledad Varela. 1999. Relaciones entre morfología y sintaxis. In: Ignacio Bosque and Violeta Demonte (eds.) Gramática descriptiva de la lengua española, Vol. 3: 4367–4422. Madrid: Espasa. Piñón, Christopher. 1997. Achievements in an event semantics. In: Aaron Lawson and Eun Cho (eds.) Proceedings of Semantics and Linguistic Theory VII, 273–296. Ithaca, NY: CLC Publications, Cornell University. [Available at http://pinon.sdf-eu.org/covers/aes.html] Piñón, Christopher. 2000. Happening gradually. In: Lisa J. Conathan, Jeff Good, Darya Kavitskaya, Alyssa B. Wulf, and Alan C. L. Yu (eds.) Proceedings of the Twenty-Sixth Annual Meeting of the Berkeley Linguistics Society, 445–456. Pollock, Jean Yves. 1983. Sur quelques propriétés des phrases copulatives en français. Langue Française 58: 89–125. Polzin, Claudia. 2000. Observaciones sobre la potencia textual de los adjetivos acabados en -ble. In: Gerd Wotjak (ed.) En torno al sustantivo y adjetivo en español actual, 369–377. Frankfurt am Main: Vervuert/Madrid: Iberoamericana. Portner, Paul. 2009. Modality. Oxford: Oxford University Press. Pustejovsky, James. 1995. The Generative Lexicon. Cambridge, MA: The MIT Press. Putten, F. van der. 1997. Matter and Mind in Morphology: Syntactic and Lexical Deverbal Morphology in Dutch. The Hague: HIL Dissertations. Pylkkänen, Liina. 1998. On individual-level predicates. Ms, MIT. Quer, Josep. 1998. Mood at the Interface. The Hague: Holland Academic Graphics. Quer, Josep. 2000. Licensing free choice items in hostile environments: The role of aspect and mood. SKY Journal of Linguistics 13 : 251–268. Quer, Josep. 2001. Interpreting mood. Probus 13: 81–111. Quer, Josep. 2002. Subordinació i mode. In: Joan Solà, Maria Rosa Lloret, Joan Mascaró, and Manuel Pérez Saldanya (eds.) Gramàtica del català contemporani, Vol. 3 : 2799–2866. Barcelona: Empúries. Quine, Willard. 1961. From a Logical Point of View. New York: Harper and
 Row. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1972. A Grammar of Contemporary English. London and New York: Longman. RAE and Asociación de Academias de la Lengua Española. 2010. Nueva gramática de la lengua española. Madrid: Espasa. [NGRALE] RAE y Asociación de Academias de la Lengua Española. 2001. Diccionario de la lengua española. Madrid: Espasa. [DRAE]. [Available at http://rae.es/rae.html] Rainer, Franz. 1993. Spanische Wortbildungslehre. Tübingen: Niemeyer. Rainer, Franz. 1999. La derivación adjetival. In: Ignacio Bosque and Violeta Demonte (eds.) Gramática descriptiva de la lengua española, Vol. 3: 4595–4644. Madrid: Espasa. Ramchand, Gillian. 2008. First Phase Syntax. Cambridge: Cambridge University Press. Ramos, Joan Rafel. 1998. Presentational sentences in Catalan dialects. Catalan Working Papers in Linguistcs 6: 41–58. Ramos, Joan-Rafel. 2002. El SV II: La predicació no verbal obligatòria. In: Joan Solà, Maria Rosa Lloret, Joan Mascaró, and Manuel Pérez Saldanya (eds.) Gramàtica del català contemporani 2: 1951–2044. Barcelona: Empúries. Randall, Janet H. 1985. Morphological Structure and Language Acquisition. New York and London: Garland Publishing. Randall, Janet H. 1988. Inheritance. Syntax and Semantics 21: 129–146. Rapoport, Tova R. 1999. The English middle and agentivity. Linguistic Inquiry 30: 147–155.

References   

   335

Rappaport Hovav, Malka. 2006. Lexicalized meaning and the internal temporal structure of events. Ms, The Hebrew University of Jerusalem. [I cite for the published version in Susan Rothstein (ed.) 2008 Theoretical and Crosslinguistic Approaches to the Semantics of Aspect, 13–42. Amsterdam/Philadelphia: John Benjamins.] Real Puigdollers, Cristina. 2006. La classe dels verbs de mesura: Un cas de verbs fronterers. UAB, Master Thesis. Real Puigdollers, Cristina. 2008. The nature of cognate objects. In: Proceedings ConSOLE XVI, 157–178. Reiβe, Christiane. 2006. Die Negation von bar-Adjektiven – eine korpuslinguistische Untersuchung mithilfe von Data-Mining-Methoden. Magisterarbeit, Technische Universität Chemnitz. Reuland, Eric. 2000. (ed.) Arguments and Case. Explaining Burzio’s Generalization. Amsterdam/Philadelphia: John Benjamins. Riehemann, Susanne. 1993. Word Formation in Lexical Type Hierarchies: A Case Study of bar-Adjectives in German. Master Thesis, Universität Tübingen. van Riemsdijk, Henk. 2012. Discerning default datives: Some properties of the dative case in German. In: Günther Grewendorf and Thomas Ede Zimmermann (eds.) Discourse and Grammar. From Sentence Types to Lexical Categories, 247-288. Boston/Berlin: Walter de Gruyter. Rigau, Gemma. 1994. Catalan presentational sentences and the properties of Agr nodes. In: Guglielmo Cinque et al. (eds.) Paths Towards Universal Grammar: Studies in Honor of Richard S. Kayne, 343–359. Washington: Georgetown University Press. Rigau, Gemma . 1999. La estructura del sintagma nominal: Los modificadores del nombre. In: Ignacio Bosque and Violeta Demonte (eds.) Gramática descriptiva de la lengua española, Vol. 1 : 311–362. Madrid: Espasa. Rigau, Gemma. 2002. Els complements adjuncts. In: Joan Solà, Maria Rosa Lloret, Joan Mascaró, and Manuel Pérez Saldanya (eds.) Gramàtica del català contemporani, Vol. 2: 2045–2110. Barcelona: Empúries. Riutort Riutort, Macià. 1998. Íslensk-Katalónsk Orđabók. Diccionari Islandès-Català. [http://usuaris.tinet.cat/mrr/islandes/islandes1.html] Roach, Peter; James Hartman; Jane Setter; Daniel Jones. 2006. Cambridge English Pronouncing Dictionary. Cambridge: Cambridge University Press. Roberts, Ian G. 1987. The Representation of Implicit and Dethematized Subjects. Dordrecht: Foris. Rodríguez Ramalle, Teresa M. 2003. Los objetos cognados como expresión de la manera verbal. Verba 30: 317–340. Roeper, Thomas. 1987. Implicit arguments and the head-complement relation. Linguistic Inquiry 18: 267–310. Roeper, Thomas and Angeliek van Hout. 1999. The impact of nominalization on passive, -able and middle: Burzio’s generalization and feature-movement in the lexicon. In: Papers from the Upenn/MIT Roundtable on the Lexicon, MIT Working Papers in Linguistics 35: 155–211. Rosen, Sara T. 1999. The syntactic representation of linguistic events. Glot International 4: 2, 3–11. Rothstein, Susan. 2008. Two puzzles for a theory of lexical aspect: semelfactives and degree achievements. In: Johannes Dölling, Tatjana Heyde-Zybatow, and Martin Schäfer (eds.) Event Structures in Linguistic Forms and Interpretation, 175–198. Berlin: Mouton de Gruyter.

336   

   References

Roy, Isabelle. 2004. Predicate nominals in eventive predication. USC Working Papers in Linguistics 2: 30–56. Ruwet, Nicolas. 1989. Weather-verbs and the Unaccusative Hypothesis. En Carl Kirschner y Janet DeCesaris (eds.) Studies in Romance Linguistics, 313–345. Amsterdam: John Benjamins. Sadler, Louisa and Andrew Spencer. 1998. Morphology and argument structure. In: Andrew Spencer and Arnold M. Zwicky (eds.) The Handbook of Morphology, 206–236. Oxford: Blackwell. Sánchez López, Cristina. 1999. Los cuantificadores: Clases de cuantificadores y estructuras cuantificativas. In: Ignacio Bosque and Violeta Demonte eds. Gramática descriptiva de la lengua española, Vol. 1: 1025–1128. Madrid: Espasa. Schäfer, Florian. 2007. Middles as voiced anticausatives. Ms, Universität Stuttgart. Selkirk, Elisabeth O. 1982. The Syntax of Words. Cambridge, MA, London, England: The MIT Press. Siegel, Dorothy. 1979. Topics in English Morphology. New York/London: Garland. Siewierska, Anna. 1984. The Passive. A Comparative Linguistic Analysis. London, Sydney, Dover: Croom Helm. Smith, Carlota. 1991. The Parameter of Aspect. Dordrecht: Kluwer. Solà i Pujols, Jaume. 2002. Modificadors temporals i aspectuals. In: Joan Solà, Maria Rosa Lloret, Joan Mascaró, and Manuel Pérez Saldanya (eds.) Gramàtica del català contemporani, Vol. 3: 2867–2936. Barcelona: Empúries. Sorace, Antonella. 2000. Gradients in auxiliary selection with intransitive verbs. Language, 76: 4, 859–890. Sorace, Antonella. 2004. Gradience at the lexicon-syntax interface. Evidence from auxiliary selection and implications for unaccusativity. In: Artemis Alexiadou, Elena Anagnostopoulou, and Martin Everaert (eds.) The Unaccusativity Puzzle. Explorations of the Syntax-Lexicon Interface, 243–268. Oxford: Oxford University Press. Sproat, Richard W. 1985. On deriving the lexicon. Ph.D. Dissertation, Massachusetts Institute of Technology. Steriade, Donca. 1999. Lexical conservatism. In: Linguistics in the Morning Calm, Selected Papers from SICOL 1997, Linguistic Society of Korea, 157–179. Seoul: Hanshin Publishing House. Steinbach, Markus. 1998. Middles in German. Ph.D. Dissertation, Humboldt-Universität zu Berlin. Steinbach, Markus. 2002. Middle Voice. A Comparative Study in the Syntax-Semantics Interface of German. Amsterdam/Philadelphia: John Benjamins. Stowell, Tim. 1991. Determiners in NP and DP. In: Katherine Leffel and Denis Bouchard (eds.) Views on Phrase Structure, 37–56. Dordrecht: Kluwer. Stroik, Thomas. 1992. Middles and movement. Linguistic Inquiry, 23: 127–137. Stroik, Thomas. 1999. Middles and reflexivity. Linguistic Inquiry 30: 119–131. Stump, Gregory T. 2001. Default inheritance hierarchies and the evolution of inflectional classes. In: Laurel Brinton (ed.) Proceedings of the XIVth International Conference on Historical Linguistics, 293–307. Amsterdam: John Benjamins. Sullivan, Thelma D. 1983. Compendio de la gramática náhuatl. México: Universidad Nacional Autónoma de México. Svenonius, Peter. 2001. Case and event structure. Ms, University of Tromsø.

References   

   337

Svenonius, Peter. 2002. Case is uninterpretable aspect. In: Proceedings of Perspectives on Aspect, Utrecht. Swart, Henriëtte de, Yoad Winter, and Joost Zwarts. 2005. Bare predicate nominals in Dutch. In: Emar Maier, Corien Bary, and Janneke Huitink (eds.) Proceedings of SuB9, 446–460. Swart, Henriëtte de, Yoad Winter, and Joost Zwarts. 2007. Bare Nominals and Reference to Capacities. Natural Language and Linguistic Theory 25: 195–222. Swart, Peter de. 2007. Cross-linguistic variation in object marking. Ph.D. Dissertation, Radboud Universiteit Nijmegen. Tenny, Carol. 1987. Grammaticalizing aspect and affectedness. Ph.D. Dissertation, Massachusetts Institute of Technology. Tenny, Carol. 1994. Aspectual Roles and the Syntax-Semantics Interface. Dordrecht: Kluwer. Tenny, Carol and James Pustejovsky. 2000. A history of events in linguistic theory. In: James Pustejovsky and Carol Tenny (eds.) Events as Grammatical Objects. The Converging Perspectives of Lexical Semantics and Syntax, 3–37. Stanford: CSLI Publications. Toman, Jindřich. 1987. Wortsyntax. Eine Diskussion ausgewählter Probleme deutscher Wortbildung. Tübingen: Niemeyer. Trips, Carola and Achim Stein. 2008. Was Old French –able borrowable? A diachronic study of word-formation processes due to language contact. In: Richard Dury, Maurizio Gotti, and Marina Dossena (eds.): English Historical Linguistics 2006, Volume II: 217–241. Amsterdam: John Benjamins. Val Álvaro, José Francisco. 1981. Los derivados sufijales en -ble en español. Revista de Filología Española LXI:185–198. Van Oosten, Jeanne. 1986. The Nature of Subjects, Topics and Agents: A Cognitive Explanation. Bloomington: Indiana University Linguistics Club. Varela, Soledad. 1990. Fundamentos de morfología. Madrid: Síntesis. Varela, Soledad. 2002. Active or subjective adjectival-participles in Spanish. In: James F. Lee, Kimberly L. Geeslin, and J. Clancy Clements (eds.) Structure, Meaning, and Acquisition in Spanish. Papers from the 4th Hispanic Linguistics Symposium, 304–316. Somerville: Cascadilla Press. Varela, Soledad. 2003. Lexical morphology revisited: Form/meaning correspondences in psych adjectival participles. In: Geert Booij, Janet DeCesaris, Angela Ralli, and Sergio Scalise (eds.) Topics in Morphology. Selected Papers from the Third Mediterranean Morphology Meeting, 51–71. Barcelona: IULA – UPF. Varela, Soledad. 2005. Morfología léxica: La formación de palabras. Madrid: Gredos. Vendler, Zeno. 1967. Linguistics in Philosophy. Ithaca: Cornell University Press. Vendler, Zeno. 1968. Adjectives and Nominalizations. The Hague: Mouton. Villalba, Xavier and Anna Bartra-Kauffman. 2009. Predicate focus fronting in the Spanish determiner phrase. Lingua. 120.4: 819–849. Volpe, Mark J. 2005. Japanese morphology and its theoretical consequences: Derivational Morphology in distributed morphology. Ph.D. Dissertation, Stony Brook University. Wanner, Anja 2000. Intransitive verbs as case assigners. In: Hero Janβen (ed.) Verbal Projections, 85–103. Tübingen: Max Niemeyer. Wasow Thomas. 1977. Transformations and the lexicon. In: Peter W. Culicover, Thomas Wasow and Adrian Akmajian (eds.) Formal Syntax. New York: Academic Press. WEBSTER. 2001. Webster’s Encyclopedic Unabridged Dictionary. New York: Random House.

338   

   References

Wechsler, Stephen. 2005. Resultatives under the event-argument homomorphism model of telicity. In: Nomi Erteschik-Shir and Tova Rapoport (eds.) The Syntax of Aspect. Deriving thematic and aspectual interpretation, 255–273. Oxford University Press. Wechsler, Stephen and Yae-Sheik Lee. 1996. The domain of direct case assignment. Natural Language and Linguistic Theory 14: 629–664. Wells, J.C. 2000. Longman Pronunciation Dictionary. Edinburgh, Harlow, Essex: Pearson Education Limited. Wheeler, Max. 2002. Flexió verbal irregular I verbs defectius. In: Joan Solà, Maria Rosa Lloret, Joan Mascaró, and Manuel Pérez Saldanya (eds.) Gramàtica del català contemporani. Barcelona: Empúries, Vol. 1 :  647–729. Williams, Edwin. 1981a. On the notions ‘lexically related’ and ‘head of a word’. Linguistic Inquiry 12: 245–274. Williams, Edwin. 1981b. Argument structure and morphology. The Linguistic Review 1: 81–114. Williams, Edwin. 1983. Against small clauses. Linguistic Inquiry 14.2: 287–308. Winter, Yoad. 2005. On some problems of (in)definiteness within flexible semantics. Lingua 115: 767–786. Wyngaerd, Guido Vanden. 2001. Measuring events. Language 77: 61–90. Zribi-Hertz, Anne. 1993. On Stroik’s analysis of English middle constructions. Linguistic Inquiry 24: 583–589. Zwart, Jan Wouter. 1997. On the generic character of middle constructions. Ms, NWO/University of Groningen.

Index Subject (a)telicity 168, 176–181, 234, 240 ––variable telicity 177n, 208, 223, 237, 241 adjective ––deverbal 8–9, 47, 128, 254, 261, 283 ––opaque 35, 38, 41 ––vs. participle 8–12, 17, 60, 130 adjectivehood, degrees of 10, 60 affixation ––above vs. below first categorizer 5, 32, 145–146, 150–155 ––-ly 94–95 ––negative see negation ––superlative 95–96 agentivity 78–80, 96, 248–250 allomorphy ––of the root/stem 30–35, 98, 174–176 ––of the affix 35n amount relative 235n, 311 argument ––dative 51–55 ––degree 239–244 ––external 62, 66–69, 80–93, 119–120, 124–127, 147, 151–154, 246–248, 303 ––incremental theme 189–190, 200, 228, 235, 240–242, 245–246, 251, 309–311 ––initiator/originator 78–80, 115, 248–250, 294–296 ––internal (theme) 44–46, 78, 132, 190, 200–206, 240, 242, 269, 310 ––measure 219–221, 249 Aronoff’s Unitary Base Hypothesis 45, 256 aspect ––Aktionsart 237–239 ––shift 176–181, 205 -bar ––vs. -abel and -wert 32–33, 111–114, 156–157, 315–316 -ble ––active 13, 74–75n, 76–78, 101n, 133 ––causative 13, 36, 76 ––decomposition of 101–104, 150–155

––denominal 32n, 256–269, 282–301, 312, 315 ––evaluative or non-potential 13, 105–111, 153–154 ––high 42, 62–63, 97, 145–148, 151–153, 173–174, 259, 268 ––low 42, 97, 110, 145–148, 153–155, 259–260 ––negative 32, 59–64, 140n, 154–155, 234n ––passive 18, 64–65, 69–76, 78, 103–104, 129–133, 151–152, 160–162, 257, 260 see also passive ––potential 18, 30–31, 83, 96, 99, 102–104, 151–153 ––properties 24–25 ––vs. middle 139–144 ––vs. se-constructions 27–28 ––vs. tough-construction 27–28, 74n Burzio’s generalization 151, 219 Case 217–220, 245–246, 249n ––exceptional case-marking (ECM) 75 characterizing predicate 55, 58, 98–99, 120, 126 ––see also genericity class marker, see theme vowel cognate object 188–192, 197n, 213–214 ––adverbial vs. argumental 194–195, 218–223 ––degree 192n, 225, 244–245, 309–311 ––nominal vs. adjectival 239–240, 306, 307–308 ––vs. hyponymous object 188–189, 191n ––root identity in 191n, 174–176, 197–198 ––with unaccusative 213–226, 240, 245, 251 concealed definite 276–277, 299, 302, 314 definiteness effect 302–303 degree phrase 184, 186–187, 209, 219–220, 225, 228, 244–245

340   

   Index

Distributed Morphology 3–4, 17, 30n, 41n, 145–147, 156–157, 175, 283–284 duration 77, 169, 180, 219, 245n, 248–250 event structure 168, 176, 180–181, 225, 231 eventivity 11, 25 ––in nouns 276, 279–280, 283–286, 299 ––vs. negation see negation extent phrase, see degree phrase free choice any 121, 123 genericity 83, 98–99, 119–120, 126, 137n ––dispositional predicate 107n, 120, 134 impossible words 3, 20–21, 167, 250–251 instrumental phrase 90–93 late insertion 4, 157, 316 lexicon 3, 20, 166, 251, 255, 286 lo 165n, 200–201, 306 ––individuative 200, 209 ––quantitative 232–233 locality 39, 175 meaning ––compositional 5, 17, 28–29, 146, 228 ––idiosyncratic 17, 19, 30–32, 153–154, 159–161, 282 measure ––function 227, 245 ––phrase, see degree phrase middle 27–28, 134–144 modality 101–105, 110, 257n, 292, 310n ––ability/abilitative 117–121, 150n ––circumstantial 115–121, 142, 232 ––deontic 73n, 115 modification ––adverbial 55–59, 140n, 171n, 288–291 ––agent-oriented 87–88, 136–137 ––aspectual 56–57, 60, 286–288, 293, 297–298 ––attributive 62–64 ––degree 96, 221 ––manner 88–90 ––postponed 55–56, 62–64 morpheme 3, 25

ne-cliticization 66–69, 131 negation 32, 140n, 154–155, 234n ––and eventivity 59–64 noun ––inalienable 278, 297n ––post-denoting, see social roles ––predicative 274–277, 284–286, 293–294, 302–303, 312 ––referential 274–277, 292–293 ––relational 278–280 operator ––generic 119–120 ––maximality 235n ––modal 116–121, 151–152 participle 8–12, 34, 60, 62–64, 128, 130, 148–149, 159–161, 307–308 passive 69–76, 82, 84–86, 129–133 ––interaction with potential 73–75, 101–105 ––periphrastic 84–85, 247n ––se-passive 27–28, 103–104, 247, 274 pleonastic structure 173–176, 191n, 235 predicate ––bare 271–273, 275–281 ––gradable 168, 184n, 226–230, 233 ––individual-level 62n, 99–100, 118n, 120, 138 ––stage-level 62n, 99–100, 258, 275–276, 283 productivity 172, 267–269, 312–314 quantification 169–170, 176, 200, 203, 234–235 ––see also operator Randall’s Inheritance Principle 50, 93n reduced participial clause 79n, 195–197 resultative construction 68, 130–131, 135, 143, 309–310 roots ––(un)bounded 190n, 205, 233–246 ––uncategorized 4–6, 189n ––see also cognate objects, root identity in scale 227–228, 230–235, 240–244, 309 situation delimiter 245

Index   

social roles 255, 280–281, 292 ––post-denoting noun 261–269, 286, 294–296, 298–299, 312–314 ––titles 272–274 specificity 53, 83–84, 122 ––and subjunctive 122–127 stativity 84–86, 88, 142, 151–152 stress shift 28–29, 50, 154 theme vowel 6–7, 34n, 38–41, 160–161, 292n ––default 36, 261n, 282n ––multiplicity of 157–159 truncation 29–31, 32–33, 154 variation ––crosslinguistic 22–23, 155–157, 304, 306 ––intralinguistic 112–114, 151–155, 315–316

verb ––accomplishment 142, 168, 178–180, 200, 205, 243n ––achievement 142, 178, 180, 231 ––activity 180, 186, 200, 205, 243n ––degree achievement 207–209, 225, 227–230, 233–236, 239, 241–242 ––change of state 45–46, 79–80, 119, 189–190, 209, 228, 234, 293–294 ––psychological 106, 112, 142–143, 154, 156–157 ––reflexive 104, 210 ––semelfactive 177n, 180, 186, 238 ––weather 211–212 verbhood, degrees of see adjectivehood, degrees of well-formedness condition 6–7, 158–159, 195

Languages Arabic 244, 307 Alguerese Catalan 302–303

Icelandic 102n, 111, 112n Italian 66–67, 306, 312, 315

Chichewa 78, 102–103

Japanese 102n

Dutch 82n, 111, 312–314

Kanuri 104

French 16n, 43n, 52, 136n, 141n, 271, 273, 277, 306, 312 Georgian 104 German 25, 32–33, 48–49, 53–54, 58, 59n, 65, 69, 78, 80–81, 83, 111–114, 116–117, 125n, 156–157, 219, 312–314, 315–316 Greek 144, 161, 219, 314n

Latin 40n, 160n, 161, 219

Hebrew 47n, 218, 223n, 307 Hungarian 103, 110–111, 162

   341

Nahuatl 161 Portuguese 306, 312 ––Brazilian 276 Swahili 103 Turkish 102, 110–111, 162

342   

   Index

Authors Alexiadou 50, 55, 151, 272–273 Anderson 28–31, 79n, 132 Aronoff 17, 28–32, 45, 52, 128, 256–258, 259n Beavers 309, 310 Borer 1, 26, 78 Bosque 9–10, 27, 55–56, 62–63, 74n, 91n, 122, 165n, 168, 200–201, 228–230, 232, 236–237, 274, 288–289, 294 Chapin 43, 69–72, 128 Chierchia 98, 107, 120n, 151 Chomsky 5, 145, 147 Cinque 56, 66–67 Demonte 255, 275, 308 Di Sciullo 46, 52, 58, 99–100, 119–120, 152, 258–260 Dowty 88, 127n, 138, 178, 207, 248 Embick 11–12, 148–150, 292 Fabb 59, 62, 68, 80, 93 Fábregas 30n, 34n, 60–61, 140n, 230n, 292n Gràcia 27, 45–46, 51, 59, 76, 128, 140–143, 262 Gutiérrez-Rexach 21, 165n, 169, 201, 248 Hackl 117–119, 150n, 152 Hale 79n, 136n, 188–189, 190n, 204, 239 Halle 3–5 Harley 147, 190n, 237 Kayne 52, 68, 80, 93, 128 Kennedy 227–228, 231, 233, 236 Keyser 79n, 87, 136n, 188–189, 190n, 204, 239 Kratzer 5, 103, 115–117, 147n, 150n, 219–220, 228, 283 Krifka 83, 98, 107n, 190n Kuno 214–223, 239n Leonetti 83, 122–123, 200n

Levin 21, 138, 217, 227–228, 231, 248n Macfarland 190, 191n, 221, 222n, 231 Marantz 3–5, 26, 137, 145, 147, 152–153, 154, 155n, 189–190, 259, 310–311 Massam 138, 190, 222n Masullo 205n, 228–230, 236–237 McGinnis 17n, 75, 146–147 Mendikoetxea 27–28, 84–86, 137n Menéndez-Benito 120 Miguel, de 44, 79n, 84, 128, 133, 179–180, 195–196 Moreno 165n, 168, 200, 232 Munn 273, 276–277, 280, 296, 299 Nakajima 218–223, 245 Nevins 17n, 29n, 42, 146 Oltra-Massuet 6–7, 38–39, 157–158 Pereltsvaig 194, 197, 221–223 Pesetsky 76, 80, 127, 152 Quer 121, 123 Randall 50–51, 90–93 Rappaport Hovav 21, 187, 227–228, 241 Riehemann 32–33, 53–54 Roeper 80n, 87, 128, 133 Roy 270, 294, 312 Schmitt 273, 276–277, 280, 296, 299 Steriade 29n Takami 214–223, 239n Tenny 187, 228n, 231n, 233, 240, 310 Val Álvaro 22, 106, 254, 261–262, 292, 312n Van Hout 80n, 128, 133 Varela 9n, 94, 115, 234n Vendler 13n, 36n, 105, 130, 148 Volpe 17, 47, 145–146, 158 Wechsler 245, 309, 310 Winter 277, 280, 296