Koineization in Medieval Spanish [Reprint 2012 ed.] 9783110901269, 9783110177442

Table of contents :
Acknowledgements
1. Introduction
2. Koines and koineization
2.1. Koine and koines
2.2. Models of koineization
2.3. Elaborating the process model: Conditions of speaker activity
2.4. Observing koineization: The Milton Keynes project
2.5. Koineization and other contact phenomena
2.6. The model and its use
3. The Burgos phase
3.1. Social history
3.2. Previous discussion of language change in early Castile
3.3. Linguistic changes
3.4. Conclusion
4. The Toledo phase
4.1. Social history
4.2. Koineization and language spread/dialect leveling
4.3. Linguistic changes
4.4. Conclusion
5. The Seville phase
5.1. Social history
5.2. Linguistic changes
5.3. Conclusion
6. Conclusions
Maps
Notes
References
Index

Citation preview

Koineization in Medieval Spanish

W G DE

Contributions to the Sociology of Language

88

Editor

Joshua A. Fishman

Mouton de Gruyter Berlin · New York

Koineization in Medieval Spanish

by

Donald N. Tuten

Mouton de Gruyter Berlin · New York 2003

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter & Co. KG, Berlin

© Printed on acid-free paper which falls within the guidelines of the ANSI to ensure permanence and durability.

ISBN 3-11-017744-7 Bibliographic

information published by Die Deutsche

Bibliothek

Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available in the Internet at .

© Copyright 2003 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording or any information storage and retrieval system, without permission in writing from the publisher. Printed in Germany.

Acknowledgements

During the several years that I have worked on this book, I have been fortunate to receive assistance from a variety of knowledgeable and talented scholars. Of these, I must thank explicitly John Nitti, Tom Cravens, Andrew Sihler, and Lesley Milroy, who gave particularly helpful advice at different stages of the project. Jim Milroy and Jeff Siegel volunteered to read and critique parts of the manuscript, and Ralph Penny and Ray HarrisNorthall generously agreed to review the entire manuscript. The final result is much improved thanks to their detailed and thoughtful reactions. All shortcomings, it need hardly be said, are entirely my own. I must also thank others for their generosity: my colleagues at Emory University for allowing me the time off to finish this project; Mikel Valladares for his untiring help with bibliography; Emilia Navarro for proofreading the manuscript; and my friends and family, especially Pepe, for their patience and support.

Contents Acknowledgements

ν

1. Introduction

1

2. Koines and koineization 2.1. Koine and koines 2.1.1. The Greek Koine 2.1.2. Modern use of the term koine 2.2. Models of koineization 2.2.1. Siegel's stage-based model 2.2.2. Trudgill's process-based model 2.2.2.1. Accommodation and salience 2.2.2.2. Interdialect 2.2.2.3. Focusing 2.2.2.4. Mixing, leveling, reduction 2.2.2.5. Reallocation 2.2.2.6. Simplification 2.3. Elaborating the process model: Conditions of speaker activity 2.3.1. Norms, norm enforcement, and social networks 2.3.2. Language acquisition: Adults and children 2.4. Observing koineization: The Milton Keynes project 2.4.1. Interaction of acquisition and social networks 2.4.2. Focusing and the time scale of koineization 2.5. Koineization and other contact phenomena 2.5.1. Pidginization and creolization 2.5.2. Contact between stable dialects 2.5.3. Dialect leveling, homogenization, uniformity 2.5.4. Standardization and standards 2.5.5. Language shift and language death 2.5.6. Convergence 2.5.7. Borrowing 2.5.8. Interacting processes and the study of koineization 2.6. The model and its use 2.6.1. Definition of koineization 2.6.2. Methodological guidelines

9 9 9 13 22 23 28 29 36 39 41 43 45 47 49 54 63 64 67 73 74 79 80 84 86 88 88 89 89 90 91

viii

Contents

3. The Burgos phase 3.1. Social history 3.2. Previous discussion of language change in early Castile 3.3. Linguistic changes 3.3.1. Leveling and simplification of articles and preposition + article contractions 3.3.2. Reorganization and simplification of the tonic vowel system 3.3.3. Koineization and other changes of the Burgos phase 3.3.3.1. The aspiration of F3.3.3.2. The development of /tf/ 3.3.3.3. The development of CL-, PL-, FL3.4. Conclusion

94 94 104 112 114

4. The Toledo phase 4.1. Social history 4.2. Koineization and language spread/dialect leveling 4.3. Linguistic changes 4.3.1. Leveling and reallocation in the development of apocope 4.3.2. Reanalysis and the rise of leísmo 4.3.3. Reorganization and simplification of the possessive system 4.4. Conclusion

145 145 153 159 160

214

5. The Seville phase 5.1. Social history 5.2. Linguistic changes 5.2.1. Previous discussion of the origins of Andalusian 5.2.2. Documentary evidence 5.2.3. The decline of extreme apocope 5.2.4. Simplification of first-person singular possessives 5.2.5. Leveling and the rejection of leísmo 5.2.6. Seseo-. A 13th-century merger? 5.3. Conclusion

215 215 222 223 231 233 238 242 245 256

6. Conclusions

257

Maps

269

119 131 132 136 138 143

173 204

Contents Notes References Index

ix 274 302 332

Chapter 1 Introduction

Approaches to the study of language and language change have long been characterized as following in one of two tracks. The focus might be the external aspects of the status and use of language, or, in line with what was considered more properly linguistic, the internal structural features and the changes they undergo. Little relationship was seen to exist between the external (social and cultural) and the internal (structural), and an exclusive focus on the internal features of language came to characterize linguistic research, whether synchronic or diachronic. Although some scholars questioned the value of this approach, they lacked the theoretical base to counter the views of others who believed and believe still that the sole object of linguistic study should be the internal structure of the language. Belief in the autonomy of language finds its origins in certain metaphors that have long governed scholars' views. In the nineteenth century, the dominant view held that language was a biological organism, which was born, grew, decayed, and died. This belief in the independent life of language led to its study apart from the context of its use. In the twentieth century, this metaphor, though not at all dead, has blended with and been superseded by others, particularly that of language as machine.1 This view, evident in Saussure's definition of a language as a system où tout se tient, has led to ever more precise representations of language as efficient system. While certainly enhancing descriptions of language structure, such a position has left little room for language change, and has led, paradoxically, to the view that language systems must alternate between perfect and flawed states. James Milroy (1992: 23) has argued that such metaphors have actually hindered research on language change. Continuing the machine metaphor, he points out that internal combustion engines are also systems, but they do not change themselves; they can only be changed from without. Still, the understanding that speakers do not, in general, consciously effect changes in the linguistic system has reinforced the idea that change can only be explained system-internally. Therefore, even when attempts have been made to include external factors in explanations of change, the view of

2 Introduction

language as an autonomous entity has tended to impede an accurate conceptualization of how external factors might contribute to change.2 In recent decades, however, sociolinguists have strongly defended the notion that it is not languages that change but rather speakers who change language.3 To the non-specialist, such an observation may appear selfevident, even trivial, but the biological and mechanistic metaphors so dominated linguistic inquiry that the importance of speakers to language change was for a long time largely ignored. Milroy (1992: 24) points out that linguists who have worked within a wholly system-internal approach have made notable headway in defining the linguistic constraints on change, but that they have been unsuccessful in dealing with the Actuation (and Transmission) Problem: "Why do changes in a structural feature take place in a particular language at a given time, but not in other languages with the same feature, or in the same language at other times?" (Weinreich, Labov, and Herzog 1968: 102). In order to answer this question, perhaps the fundamental problem in the study of language change, the researcher must focus on what speakers in society do with their language. To do so is not to reject the idea of language as system, but rather to supplement this idea with a conception of language as a tool or pool of resources that is used by speakers for specific purposes (Milroy refers to this as "speakers' teleology"). From this perspective, linguistic change is seen as the product of the interplay of speakers and systems in specific and changing social contexts. Keller (1994) follows a logic like that of Milroy in arguing for the use of invisible-hand explanations in historical linguistics. Keller points out that languages are social institutions, or phenomena of the third kind, and that they cannot be explained in the same ways that natural phenomena or human artifacts are explained.4 Rather, like other social institutions, languages arise and change as the unintended results of collective and intentional human actions (as do traffic jams, or footpaths across lawns). An invisible-hand explanation of linguistic change will therefore be composed of two levels: a micro-level and a macro-level. The micro-level examines intentional speaker activity and the constraints upon that activity. The constraints, or ecological conditions, include intralinguistic factors such as the linguistic varieties and the specific features associated with each, as well as extralinguistic factors, such as sociocultural, psychosocial, cognitive/learning, and biological conditions that enable and restrict speaker behavior. Speaker (interactions, summed and viewed collectively, then

Introduction

3

lead to consequences on the macro-level, which is constituted by the linguistic structures, processes, or outcomes that need to be explained. Of the extralinguistic factors affecting speaker activity, only those relating to social conditions are normally open to change. For this reason, Milroy emphasizes the link between social change and linguistic change, and argues that the primary task of historical sociolinguistics is the establishment of explicit causal links between these two domains (Milroy 1992: 222). To the extent that patterns of social change and linguistic change cooccur in different situations, models and theories of change can be abstracted from real cases of change. In this study, I critique, elaborate and apply a sociolinguistic model of change, koineization, which provides a means of linking certain kinds of structural changes to a specific type of social change. Koineization is generally considered to consist of processes of mixing, leveling, (limited) reduction or simplification, which occur in social situations of rapid and intense demographic and dialect mixing. The model has been developed primarily from studies within the variationist paradigm of new towns (e.g., Omdal 1977; Trudgill 1986; Bortoni 1991; Kerswill 1996; Kerswill and Williams 2000) and colonial and post-colonial language varieties, often as an extension of research on other contact varieties such as pidgins and creóles. These include discussions on the origins of different varieties of overseas Hindi-Bhojpuri (e.g., Moag 1979; Gambhir 1981; Siegel 1987, 1993; Barz and Siegel 1988; Mesthrie 1993), and, to a lesser extent, colonial English (e.g., Trudgill 1986; Trudgill, Gordon, and Lewis 1999). Unsurprisingly, the model has been applied to language change in other colonial contexts, including Latin of the Roman Empire (Wright 1996) and American Spanish (e.g., Fontanella de Weinberg 1992; Granda 1994; del Valle 1998; Hidalgo 2001). The model of koineization represents a significant theoretical advance for our understanding of language change as influenced by dialect contact and mixing. Traditional historical linguists, heavily influenced by Neogrammarian, structuralist, or generativist/formalist principles, were loath to admit explanations based on dialect contact or mixing in any but anomalous cases of change; the ideal explanation was either internal to the system (and thus little more than a description) or internal to the speech community (if this aspect was considered). Indeed, as we will see in Chapter 2, it is no accident that much early work on koines was carried out by scholars already interested in contact varieties such as pidgins and creóles, or variationist sociolinguists, who were drawn to the study of dialect contact through their study of language/dialect variation and its relationship to

4 Introduction linguistic change. Still, even in sociolinguistics, study of dialect mixing has only now begun to enter the mainstream. 5 Although this model has most often been associated with colonial varieties, it is my contention it should be useful for linking the unique social consequences of the medieval reconquest and repopulation of the area we now know as Spain (phenomena with no obvious analogues in medieval France, Italy, and England) 6 with the particular formative changes of medieval Spanish (or, more precisely, Castilian).7 In fact, as I discovered as I began to research the topic, Ralph Penny had already proposed this idea in his brief study Patterns of linguistic-change in Spain (Penny 1987). Penny suggested that the model of koineization (as defined by Trudgill 1986) could and should be applied to the history of Spanish, not only colonial varieties, as in the work of Fontanella (1992), but also peninsular Spanish itself. Penny pointed out that the medieval expansion of Castile and Castilian had occurred in a series of geochronological stages of population movement and dialect mixing (Penny 1987: 4—7): 1. 2. 3. 4. 5.

6. 7.

Burgos and the early county of Castile (late 9th and 1 Oth centuries) Toledo and surrounding regions (from 1085 and into the 12th century) Seville and the valley of the Guadalquivir (mid- and late 13th century) Granada and surrounding regions (from 1492) The conquest and colonization of America, and the sudden conversion of Seville into the sole port of entry/exit for the colonies (from 1492 and into the 16th century) Judeo-Spanish, in exiled Sephardic communities of the Mediterranean (following the expulsion of 1492) Madrid, following its designation as capital of Spain (from 1561)

According to Penny (1987), there are numerous features of Peninsular Spanish that might be ascribed to koineization, such as the absence of phonemic splits and the small inventory of phonemes relative to other Romance languages. Other such features include: — — — — — — —

reduction from four to three degrees of aperture for tonic vowels reduction of final vowels from four to three merger of Ibi and /β/ loss of voicing in sibilants leveling from four to three verb classes almost complete leveling of second and third verb classes extreme leveling of irregular verb forms

Introduction — — — — — —

5

leveling of perfect auxiliary verb to haber interdialectal solution to third-person object pronouns merger of /j/ and I7J in Andalusia (yeísmo) merger of the antecedents of Isl and /Θ/ in Andalusia (seseo/ceceo) replacement of vosotros with ustedes in Western Andalusia aspiration and loss of l-sl in Andalusia (Penny 1987: 8-17)

However, Penny (1987) kept his comments speculative; he did not attempt to tie most of these changes to any particular stage, provide evidence for such linking, or enter into a detailed explanation of how particular features originated, were selected or spread. Subsequently, Penny has returned occasionally to the topic (e.g., 1992, 1995, 2002) and has published a recent volume, Variation and Change in Spanish (Penny 2000), which includes significant discussion of the effects of dialect mixing on the history of Spanish. In his more recent publications, Penny has begun a more in-depth analysis of some of the problems identified by him, and has begun to answer the question he himself poses in his (1987) essay: "is it possible to observe a correlated series of linguistic levelings and simplifications, in the way the theory predicts?" (Penny 1987: 8). To this end, he has analyzed the origins of Judeo-Spanish (1992) and the rapid propagation of the aspiration of etymological Iii after the naming of Madrid as capital in 1561 (2002). In Penny (2000), he discusses still more features of Spanish that may have arisen as a result of koineization. Still, this volume was designed as a broad overview and introduction to variation and change in the history of Spanish, and Penny therefore did not aim to link specific changes to specific periods (though he does consider some changes that might be associated with the Burgos phase), nor to engage in detailed reconstructions of particular changes. At this point, then, the questions that Penny posed in 1987 have yet to be answered, particularly for the early medieval periods of koineization. Indeed, while the importance of the Burgos phase (or período de orígenes as it is known to most historians of Spanish) has long been recognized, the sociolinguistic significance of the Toledo and Seville phases remains undemonstrated. Moreover, Penny (2000: 5), following Wright (1999), expresses reservations about periodization of the history of any language, including Spanish, and so avoids reliance on the historical schema of geochronological stages of koineization he had proposed earlier (though he continues to suggest them as possibilities).

6 Introduction Penny's original suggestions and unanswered questions form the starting point and organizational basis for much of this study. I have chosen to study only a limited number of features, and, in the interest of exploring its usefulness, I have chosen to maintain the geochronological framework outlined in Penny (1987). The number of features has been limited to only a few for each stage (not all of which were mentioned by Penny), because my primary intent has been to show, in a thorough and detailed manner, how certain changes can best be explained in terms of koineization, and in so doing to offer evidence for the validity of the model as elaborated and defined in Chapter 2. I have limited discussion to the levels of phonology and morphology (or morphosyntax in the case of leísmo), in part because these are the linguistic levels that have received most attention in Hispanic historical linguistics, but also because these have been the components of language least obviously related to social and cultural change. I have also limited my discussion to only the first three stages proposed by Penny those which began and ended in the medieval period. This is so for two main reasons. First, the year 1500, a traditional break-off point between medieval and modern Spanish, marks a convenient break-off point for this study, which would otherwise require entry into the even more complex issues surrounding the origins of American Spanish (though the last chapter includes discussion which will be of interest to those studying this topic). Second, the late 15th century marks the advent of printing and with it the beginning of widespread effects of standardization (Harris-Northall 1996b), a process which may erase or blur the effects of koineization (see Chapter 2). Indeed, even though application of the model represents a clear example of what Labov (1975) has called "using the present to explain the past", medieval Spanish also represents an ideal context and variety on which to test and explore the explanatory power and limits of the model, in that the competing effects of standardization - so difficult to escape in the modern world - were largely absent. And, even though we must still contend with that perpetual dilemma of historical linguistics - relative paucity of data - medieval Spanish is a variety that has been thoroughly studied, and for which we possess fairly plentiful documentary evidence, sometimes (particularly in the Seville phase) much more than that available for some of the colonial/post-colonial language varieties that have been studied as koines up to now. The study is divided into four core chapters (Chapter 2-5). In the first of these, I begin by reviewing the history and use of the term koine, from which the term and concept of koineization is derived. Given the recent

Introduction

1

development of this model, it is only to be expected that consensus about all its defining features has yet to be reached. Even so, a thorough understanding of the model is necessary if it is to be used to reconstruct events at great time-depths (1100 to 700 years ago in this study). Inevitably, my own research has led to re-evaluation of earlier proposals on koineization. I therefore examine key studies that discuss or apply a model of koineization (particularly Siegel 1985; Trudgill 1986; and Kerswill and Williams 2000), synthesizing this earlier work, relating it to more general theories of language change and pointing to possible problems, limits, and refinements. I also propose certain methodological procedures or guidelines that should be adhered to when using the model to analyze and explain past changes. From there, I move to investigate, in three further chapters, the first three periods of possible koineization and rekoineization suggested by Penny for Castilian: the Burgos phase (Chapter 3), the Toledo phase (Chapter 4), and the Seville phase (Chapter 5).8 In each of these chapters, I review the social and demographic history of the period and region, and develop sociolinguistic reconstructions of certain changes that can and should be linked to koineization as it is defined in Chapter 2. But I have also varied the organization of each chapter as needed. For example, the Burgos phase has received a great deal of attention from scholars, so Chapter 3 includes a review and critique of previous discussions of the linguistic significance of this phase, as well as the medieval period in general. In this chapter I argue that there are at least two groups of changes for which koineization was a primary cause, but I also discuss three other changes less clearly related to koineization, and, in the interest of exploring the explanatory limits of the model, consider to what extent koineization may or may not contribute to our understanding of their development. Unlike the Burgos phase, the Toledo phase has not generally been recognized as significant to the development of medieval Castilian, so my primary aim in Chapter 4 is to show that there are several groups of changes that can be attributed to koineization in this period. I also consider the relationship between koineization and the spread of Castilian features in neighboring regions, a phenomenon which appears to have accelerated at this time. Until very recently, the Seville phase had been largely ignored by scholars, but during the past decade the Spanish scholar Frago Gracia has made strident claims that many of the features that today characterize the Andalusian dialect of southern Spain had their origin in the dialectal and demographic mixing of the 13th century. Chapter 5 is therefore, in large measure, a response to the work of Frago Gracia (e.g., 1993), whose

8 Introduction research and views I consider in some detail. In fact, I challenge his arguments that a key modern feature of Andalusian, seseo, arose in the 13th century. On the other hand, I do find convincing textual evidence of other changes not considered by Frago which support his more general claim that the Seville phase was an important period of (rapid) change. This particular case will illustrate the importance of adhering to the methodological guidelines outlined in Chapter 2 when employing the model at a great timedepth.

Chapter 2 Koines and koineization

The model of koineization, of fairly recent development, is based on earlier metaphorical use of the term koine. We therefore begin with an overview of the origins and modern use of the term in the linguistic literature, and of the confusion that its varied meanings have sometimes provoked. The primary aim will then be to define, as thoroughly as possible, what koineization is, and what it is not. Several scholars have sought to answer these questions, though their responses do not agree in all respects, so I have organized the bulk of this chapter as a critical review of previous discussion of koineization, with the goal of synthesizing this earlier work and my own views. Throughout, the various facets of koineization are put in relation to other theories of language use and change, but a special section focuses on the differences between koineization and other processes with which it may interact in real cases of change, or be confused in scholars' discussion of change. The chapter concludes with the definition of a prototypical model of koineization, and the proposal of methodological guidelines for application of the model.

1. Koine and koines Koine is a term with a long history and a wide variety of interpretations. It has its origin in the name of a variety of ancient Greek that became the common language of the eastern Mediterranean. Subsequent metaphorical or technical use of the term has referred to a broad range of language varieties that share some or all of the characteristics of the original Greek Koine.

1.1. The Greek Koine The κοινή (from koiné dialektos or koinè glòssa 'common tongue') was a mixed dialect based largely on the prestigious Attic dialect of Athens. From the middle of the fifth century B.C., when Pericles converted the

10

Koines and koineization

Confederacy of Delos into an Athenian empire, the influence of Attic spread rapidly throughout the Aegean. Most of the other city-states in this empire spoke Ionic dialects (to which the comparatively archaic Attic was closely related)9 and resented the control of Athens, but the emerging Koine, usually referred to as Great Attic at this early stage (Bubenik 1993: 12; Horrocks 1997: 29), was useful for commerce and general intercourse and was also employed as the (written) language of administration (Horrocks 1997: 33). It has been suggested (Thomson 1960: 34; Hock 1986: 486) that a likely birthplace for the Koine was the Peiraieus, or port of Athens, where Attic speakers and Ionic speakers from other parts of the empire interacted, along with Doric speakers from the neighboring Peloponnesus. However, its use as written "standard" and spoken vernacular was never restricted to the Peiraieus, since contact between Attic and Ionic speakers occurred in a variety of contexts. The city-states in the Attic League had to provide soldiers for the Athenian armies, as well as deal with Athenian officials in their territories and Athenian administrative documents composed in official Attic (Horrocks 1997: 31).10 Athens also sent out numerous Attic-speaking colonists to the colonial territories, where they interacted with Ionic speakers. Many speakers of Ionic also took up residence in Athens, and through their interaction with Athenians may have contributed to changes in the speech of "middle-class" residents of the city. Great Attic thus developed in part as a second dialect of Ionic speakers, but it became the native dialect for following generations in some of the Ionic cities. Eventually, Philip of Macedón adopted Great Attic as his language of administration and it later spread throughout the eastern Mediterranean as a result of the conquests of his son, Alexander the Great. The early Koine may have benefited from its ambiguous relationship to traditional Attic; its difference from Attic may have made it more acceptable to the dominated Ionic speakers of the empire (Hock 1986: 486), while its similarity perhaps lent it prestige and made it acceptable to Philip of Macedón. It has been characterized as a "de-Atticized Attic" (Hock 1986: 486) and as a "de-Atticized Ionicized Attic" (Bubenik 1993: 13). It is interesting to note that this mixed and simplified form of Attic was decried from the beginning as being impure and corrupt (Palmer 1980: 175), and centuries later, under the Romans, a campaign of "Atticization" was launched to improve it (Buck 1933: 22). The following features have often been identified as typical of the mixed and simplified nature of the original Koine:

Koine and koines

11

1.

Highly distinctive Attic -it- was largely replaced by the more widespread (Ionic) equivalent -ss-, thus: Attic Koine glötta glossa 'tongue' phulattö phulassô 'guard, watch' tettares tessares 'four' (Hock 1986:486) 2. Distinctive Attic -rr- was replaced by more widespread (Ionic) -rs-: Attic Koine arrën arsën 'male' (Hock 1986: 486) 3. Attic -ä- ( [eno], cum illu > [kono]. Estos fenómenos son un fiel reflejo de las contracciones que se daban (y se dan) en la lengua hablada y que, al parecer, tenían un ámbito de aplicación muy extenso por toda la Península. Posteriormente, la aparición de unas normas de escritura que dieron prioridad a la palabra como unidad gráfica eliminaron esas formas de algunos romances hispanos. (Lleal 1990: 163-164)

However, there are two problems with this view. First, the changes in question clearly predate the establishment of a written Castilian standard, which does not begin to occur until the 13th century. Second, it ignores the fact that contracted forms have been maintained in modern standards such as Portuguese and French. The Portuguese say and write no, pelo, not *en o or *por o\ the French say and write du, au, not *de le or *à /e.81 Literacy and standardization cannot be seen as the cause of loss of such contractions in Castilian.

Linguistic changes

119

Speaker activity in koineization offers a better explanation. The rapid expansion of Castile and the massive demographic mixing that followed led to the removal of speakers from established social networks and their insertion in new, mixed ones. As social and linguistic norms were weakened, speaker-learners were exposed to a wide variety of possible forms of articles and contractions of prepositions and articles. For example, what are now en el and por el would have been part of a pool of competing variants: no, enno, eno, enne, ene, en el and polo, pelo, pollo, pello, por el. In

order to accommodate to their new and dialectally mixed communities, some speakers may have begun to eliminate those forms which were unfamiliar to many other speakers; no and polo could have been replaced with the easily generated en el and por el. More likely still, both adult and child speaker-learners would have tended to search for and reproduce forms which were consistently and frequently produced, such as the easily analyzed and generated en el or por el, whose component parts also appeared regularly in other contexts. In fact, research on contemporary varieties with such contracted forms has shown that they are difficult for children to learn. For example, French-speaking children as old as 5 still do not consistently contract de le to du, de les to des, à le to au, or à les to aux (Clark 1985: 727). All these factors would have led to the production of still more regularized forms (such as en el and por el). With such massive variation and weak norm enforcement, there would have been little need or possibility - given the competing norms - of "correcting" such output, which would have then served as input for other learners.82

3.2. Reorganization and simplification of the tonic vowel system Castilian is well known for its classic five-vowel system. However, this exists in marked contrast to that of both western Ibero-Romance (GalicianPortuguese) and eastern varieties (Catalan), where the seven vowel phonemes of the typical Vulgar Latin/Early Romance system are retained, as is the case in most Romance varieties. This reduction in phonological inventory can be dated from the earliest days of Castilian and is, in fact, the indirect consequence of another of the most characteristic developments of early Castilian: the stabilization of the diphthongal realizations of the Vulgar Latin/Early Romance low mid vowels h i < Ö and /ε/ < Ë (Menéndez Pidal 1964: 486).

120

The Burgos phase

The spontaneous diphthongs that developed from Latin tonic Ö and Ë have long been the source of controversy in Romance linguistics. Past research has focused on the dating and linguistic reasons for their initial appearance, as well as how they should best be analyzed within the phonological system of each language variety affected. The widespread occurrence of such diphthongization throughout the Romance-speaking area indicates that it must have begun in Late Latin/Early Romance, certainly being established in the Iberian Peninsula before the Moorish Invasion. The varieties to the extreme West, Galician and Portuguese, were largely unaffected (though traces of incipient diphthongization can be found in some dialects), as was Catalan in the East. However, the HispanoRomance dialects of the center of the Peninsula were all (eventually) affected by spontaneous diphthongization of Ö and Ë. Moreover, both historical textual and modern dialectal evidence points to tremendous variation in the realization of these diphthongs (and related monophthongs) in the majority of central Iberian dialects, namely, Astur-Leonese, NavarroAragonese and at least some of the principal dialects of Mozarabic. In Leon, early Latinate texts (written in what Menéndez Pidal called latin vulgar leonés) show a mixture of orthographic forms: o, uo, ua, ue < Ö; e, ie, ia < Ë (Menéndez Pidal 1964: 113-115, 144-146). Later 13thcentury texts written in Leonese continue to show this same variation (Staaff 1907: 191-206), though ua declines significantly, leaving uo, ue and o as the dominant forms; ia was less frequent, found principally in documents from western Leon (Menéndez Pidal 1964: 146). With regard to the phonetic value of these orthographic representations, Wright (1994c) has emphasized that the novel combinations of letters are probably more significant than those that correspond to the written (Latin) orthographic norm. This is so because the scribe must depart from an orthographic norm to produce innovations which reflect phonetic realizations. This could mean that there remained only diphthongal realizations in speech, but in Leon and Aragon representations with o or e continued many centuries, so the survival of the monophthongs as variants is plausible. Aragonese texts from the 11th century and later show vacillation similar to that of Leonese texts, between o, uo, ua, ue /e/:

lecho, DERECTU/DIRECTU > derecho, INTEGRU > entero, madera, SUPERBIA > soberbia, CERESIA > ceresa, NERVIU > nervio, PRAEMIU > premio, GREGE > grey, SEDEAT > sea — h! > loi·. OCTO > ocho, NOCTE > noche, OSTREA > ostria, NOViu > novio, HODIE > hoy, MODIU > moyo, PODIU > poyo, FOLIA > hoja, CORDOLIU > cordojo, OCULU > ojo LECTU >

MATERIA >

Menéndez Pidal emphasized that this development set Castilian apart (as a feature of the Castilian cuña or wedge) from other surrounding Romance languages and Iberian dialects, where diphthongization before yod did occur (Menéndez Pidal 1964: 495). It is generally believed that the following glide had an assimilatory effect on a preceding stressed low mid vowel (Lloyd 1987: 185,194), raising it to a high mid vowel (and into the space of another phoneme): [ε] < Ë, but [e] < Ë + yod; [d] < Ö, but [o] < Ö + yod. Supposedly, as the tongue was raised in anticipation of the following glide, the vowel too was raised, thus bleeding the environment of the diphthongization rule. However, this analysis leaves unanswered the perplexing question of why the yod would have a metaphonic effect in Castilian, but not in Leonese and Aragonese.89 In order to explain the Castilian developments, it is necessary to understand the nature of diphthongization before yod in Leonese and Aragonese, which is in fact a rather complex phenomenon. The most convincing evidence of Leonese and Aragonese diphthongization before yod comes from the modem dialects. Zamora Vicente comments on diphthongization before yod in Aragonese: Las vocales breves (Ö, Ë) diptongan ante yod comportándose así en estrecho paralelismo con el leonés y frente al castellano: FOLIA > fuella; OCULU > giiello; ROTULU > ruello·, cueto < COCTU; HODIE > güe (güey en lo antiguo); PODIU > pueyo-, HORDEU > giierdio·, FOVEA > fueva, despuellas, etc. Casos de Ë: TENEO > tiengo; viello, retienga, devienga, etc. (Zamora Vicente 1967: 218)

This gives an impression of regularity that may indeed be true of the modern isolated and conservative dialects. Such regularity is also present in the results of Leonese diphthongization of hi before yod:

128

The Burgos phase

Hoy, esta diptongación es uno de los rasgos más característicos del habla viva. Es general en Asturias central y occidental: nueche o nuechi; ueyu, giieyu; uoy, mueya, muecha; fueya, fuöya; cuechu; adueitar, etc. No se dan casos en Santander y, hacia oriente, vacila ya en Lianes, donde se oye en unos casos y se ignora en otros; en los Picos de Europa, el límite parece estar en Cabrales, donde hay vacilación. En León, casos de este diptongo (cada vez más en retirada ante el empuje de la lengua oficial) se oyen en Sajambre (mueyo, jueya; Güeyo, en topónimos), los Argüellos (ueyo), etc. (Zamora Vicente 1967: 93) Significantly, Zamora views diphthongization as normal and the lack of diphthongization as a result of increasing influence of the Castilian standard. However, modern Leonese diphthongization of /ε/ is less frequent than that of loi, and its lesser frequency does not appear attributable to Castilian influence: La diptongación de Ë + yod no se ha producido con tanto rigor y abundancia como en el caso de Ö. La Ë no se ha diptongado {pecho, entero) ni en los textos antiguos ni en el habla moderna, excepto en el verbo: tiengo, viengo; yes < EXIT, en el Alexandre. Hoy, las formas verbales, con o sin diptongo, se reparten confusamente en el dialecto. (Zamora Vicente 1967: 98) What we observe is a great deal of variability in the appearance of these diphthongs (as opposed to monophthongs). Nevertheless, lack of diphthongized forms could be the result of modern Castilianization. What was the situation in Old Aragonese and Old Leonese? Zamora Vicente is quite certain about the ancient regularity of diphthongization before yod. The following discussion of Leonese diphthongization before yod is indicative: La diptongación de ö ante yod es normal en los antiguos textos: nueche 'noche'; mueyo 'mojo'; cueya < COLLIGAT; ue, vué < HODIE; vuedia < HODIE DIEM; ueyo < OCULU. Todavía Lucas Fernández usa duecho < DOCTU. Otros ejemplos son: Aradue, moderno Araduey (Sahagún, 1096); Pedro Abrueyo < APERIOCULU (Sahagún, 1171); Pedro Redrueyo < RETROCULU (Sahagún, 1253); uecho 'ocho' (León, 1260); etc. La toponimia confirma la vitalidad y extensión antiguas del cambio. Arguelles < ARBOLEIS, 1064, en Oviedo, y Argüellos en La Vedila; Los Fueyos < FOVEU, en Asturias, León y Sanabria; Sigüeya en La Cabrera Baja (frente al castellano Segovia); Cirigiiello, Entruello, en Asturias. (Zamora Vicente 1967: 93) It should be noted, however, that much of this evidence is actually quite late, post-dating the Burgos phase of mixing in Castile by several centuries. For the period of Orígenes, Menéndez Pidal is able to provide only a few

Linguistic changes

129

examples of Leonese, Aragonese and Mozarabic diphthongization before yod (as compared to far more abundant examples in other positions): El caso especial de Ö + yod ofrece pocos ejemplos. (Menéndez Pidal 1964: 139) La Ë diptongada ante yod ofrece casos más oscuros y raros que la Ö. (Menéndez Pidal 1964: 158) Moreover, Menéndez Pidal himself was much more circumspect in his claims. In fact, he claimed that diphthongization before yod was simply far less frequent in Castile than in neighboring (or, in my view, contributing) dialects. The evidence of diphthongization before yod appears to grow in Leonese and Aragonese texts only over time, but lags behind that of diphthongization in other positions. It is only for later stages that Pidal is able to offer significant evidence of Leonese and Aragonese diphthongization before yod (Menéndez Pidal 1964: 139). Even at later stages, it is not clear that diphthongization before yod became regular; Staaff (1907: 207) found only four orthographic examples of diphthongization of loi before yod in 62 13th-century documents from eastern Leon and only 6 examples in 30 documents from Western Leon (described as "tres veces mas abundantes"). Of course, this irregularity in textual evidence could also be due to a lack of phonologization of the diphthongs (and thus lesser inclination to represent them in the orthography), to growing Castilian influence or to Latinizing tendencies, but if diphthongization before yod had been absolutely regular, it seems that Staaff would have found more examples. A plausible explanation for the above situation would be that spontaneous diphthongization spread through the process of lexical diffusion (e.g., Wang 1969), in which speakers adopt features on a word-by-word basis. This adoption is affected by other factors: verb morphology and phonetic environments can favor or constrain the adoption of new sounds, as can word frequency. In early Hispano-Romance, it would appear that words with low mid vowels followed by yod were less likely to be articulated with a diphthong (tautosyllabic nasals also seemed to impede diphthongization).90 In this sense, they were at the "end of the queue" for the change (with a few high frequency exceptions, such as viejo). In areas where mixing was less and social networks more stable (such as Asturias, Leon, and mountain regions of Aragon), speakers retained a large number of variants in the articulation of loi and Id. For example, hi, whatever its position, was probably articulated with a number of these variants: [D, WO, we, wa, woe, wa]. As a result, speakers, over many generations, could continue to

130

The Burgos phase

induce the same phoneme for frequent [a] of oyo (< OCULU) and the frequent [wo, we, wa, woe, wa] of bueno (< BONU), because either word could be articulated with the less frequent variant (i.e., a diphthong for oyo and monophthong for bueno). It is the maintenance of this rich but systematic variation that allows speaker-learners to construct and maintain an abstract phonemic connection between the surface variants, and to continue the analogical process of lexical diffusion (in the Leonese case, this is exemplified by the extension of diphthongization to words with underlying hi and /ε/ and following yod, such as in Leonese ueyo < OCULU). These connections through variation were broken in Castile, and led to an abrupt interruption of the process of lexical diffusion. At the time of the Burgos phase repopulation, phonetic diphthongization was clearly well under way, but it had not yet spread (consistently, at least) to words with low mid vowels followed by yod.91 In the koineizing environment of Castile, however, adult speakers were reducing the non-functional variation in their speech as they accommodated. The most frequent and consistent articulation of words with low mid vowels before yod would have been with a monophthong (e.g., ojo < OCULU), and as adults accommodated out infrequent variants, the monophthong would have become even more prevalent for such words. However, as we have seen, diphthongal articulations were at the same time being reanalyzed as sequences of two phonemes, with a reduction in the number of diphthongal variants. With the loss of a wide range of variants, children could not have perceived any connection between the [D] of ojo (or its equivalent) and the [we] of bueno (or its equivalent). Given such input (the output of other speakers), a reasonable induction for child (and adult) learners to make would be to relate the monophthong with the surviving monophthongs closest to it in phonetic terms: the (formerly high) mid vowels loi (in the case of [o]), and Id (in the case of [ε]). This merger between hi and loi, and between /ε/ and /e/, produced by enough speakers, would not have encountered resistance in the environment of fluctuating norms and weak norm enforcement. It is the combined effects of this merger with the phonological reanalysis of the diphthongs in the koineizing context that leads to the reduction of the 7-vowel Romance system to the simplified 5-vowel system of Castilian. But these phenomena also result in a mixed pattern that is clearly marked in comparison to the surrounding dialects - one of those patterns that Posner (1996) might label "deviant". This may be so, but the changes in question are best understood as the effects of koineizing reanalysis and simplification.

Linguistic changes

131

3.3. Koineization and other changes of the Burgos phase The two groups of changes analyzed above show how the model of koineization can contribute significantly to our understanding of how and why certain changes arose when and where they did. Though we cannot claim to have fully accounted for the actuation and transmission of these changes, the model does allow for a far broader range of factors to be included in our necessarily partial reconstructions. Indeed, the preposition + article contractions and the particular changes in the tonic vowel system are phenomena that remain quite mysterious if not analyzed as koineizing changes. Still, having established that koineization certainly occurred during the período de orígenes, one must also ask: what was the impact of koineization on other changes that are frequently associated with this period? Menéndez Pidal (1964) is well known for having defined a group of ten changes that characterized the early Castilian of Burgos, and then were carried southward in a progressively widening band, creating what is known as the Castilian cuña or wedge that supposedly broke the previous dialectal unity of the center of Visigothic Iberia. These characteristic features include: 1. 2. 3. 4. 5. 6. 7.

Development of F- > [h-] Development of Vulgar Latin [-lj-] > [3] Development of Latin -CT- > [tj] Development of Latin hi > [we] Lack of diphthongization before yod Lack of palatalization of IVI (> [λ-]) Coincidence in a number of features between east and west that are not present in Castilian, e.g., phonetic variation in the realization of the spontaneous diphthongs was frequent to both the east and west, but not in Castilian; many forms of the verb ser are diphthongized in NavarroAragonese and Leonese {tú yes, él ye), but not in Castilian (tú eres, él es). 8. The assimilation of Latin /mb/ > /m/. 9. The monophthongization of Vulgar Latin [aj] > [ej] > [e] 10. The development of Latin consonant clusters PL-, CL-, FL- > [λ]. (Menéndez Pidal 1964: 490-502)

Scholars do not now understand all these changes in the same way that Menéndez Pidal did, but his list of features, restricted though it may be, is widely accepted as representing key changes of early Castilian. Some of them are certainly much easier to understand when analyzed in terms of

132

The Burgos phase

koineization. This is the case of items 4, 5, and 7, which have already been discussed above (Menéndez Pidal did not include the development of the preposition + article contractions in his list). In theory, however, all the changes should be analyzable (and must be analyzed) in terms of koineization, even if koineization merely preserves features found in one or several established contributing dialects. Such changes may be seen as only tangentially related to the process of koineization, but as consistent with it. Indeed, one would not expect to find changes that increase complexity and opacity, and such a finding would weaken the plausibility of other explanations based on koineization. Below, I discuss briefly three of the changes mentioned by Menéndez Pidal in order to highlight three different issues relating to application of the model. First, although the model cannot always be used to construct explanations such as those found above, it may still be useful in judging the relative value of other explanations. Second, some apparent increases in complexity (e.g., the addition of a new phoneme) may in fact result from a decrease in variation (and reanalysis). Third, even changes which seem (at first glance) to fit the model exactly may in fact be better explained as the result of other processes.

3.3.1. The aspiration of F-

One of the best known and most studied features of early Castilian is the change in pronunciation of (normally word-initial) Latin F-. It is generally accepted that by the 10th century the word-initial /f-/ of Latin/Romance developed into [h-] in Castilian. This [h-] then spread south along with reconquest, and in later centuries was itself lost in many varieties of Castilian, including the standard.92 Scholars have tended to take one of two stands regarding the origin of this feature. Menéndez Pidal (1964) and his followers have argued strongly that the change must be understood primarily as a result of a Basque (and Cantabrian?) substrate on early Castilian. Others, including Penny (1972, 1990), believe the change was primarily an internal development of the Romance of Cantabria and/or Castile. Trask (1997) strongly favors an internal explanation, and vehemently rejects the Basque substrate thesis. Lloyd (1987) positions himself between these two extremes. As we will see, an analysis based on the model of koineization tends to support Lloyd's position. Let us consider first the Basque substrate hypothesis and the arguments leveled against it. Since Castilian was clearly in some kind of contact with Basque, many assume that this seemingly

Linguistic changes

133

unusual change (at least within Ibero-Romance) is due to Basque influence, particularly since the same change in found in Gascony, which borders the modern Basque Country to the North, and where a form of Basque was spoken among the Aquitanians when the Romans arrived (Trask 1997: 36). It has long been known that Pre-Basque (the precursor to modern Basque spoken in Roman times) had no Iii, and it is argued that when Basque speakers adopted Romance speech in (or near) Castile, they could not pronounce the [f] and replaced it with [h], the closest phone available in their own speech. Though the existence of [h] has been questioned, Trask (1997: 127, 157) indicates that it is certain that Pre-Basque included a frequent and prominent phonetic aspiration which accompanied stressed vowels (this was later phonologized in many varieties of Basque). Nevertheless, many authors have rejected this hypothesis. Most recently, Trask (1997: 424-429), following Izzo (1977), adamantly rejects the possibility of Basque influence on the change, but his rejection is grounded on a very conservative notion of substrate theory. For instance, he assumes that Basque must once have been spoken in Castile as the primary language in order to speak of direct influence, and he stresses that there is no evidence that Basque was ever a stable language variety of Castile. This is true enough, but it is not necessary that this be so if one considers potential Basque influence during dialect mixing as opposed to language shift by a stable population. Basque settlers most certainly moved into early Castile in significant numbers and would have had to learn Romance. Their linguistic production would then have formed part of the prekoine linguistic pool. We might therefore accept that Basque speakers had some impact on the change F- > [h-]. However, Trask includes other arguments (also made by Izzo and scholars before him) that might call into question even a less rigid view of Basque influence on this change: 1.

2.

Romance /f-/ became [h-] in most cases, but not before the novel diphthong /ue/ (realized as [we]) or before liquids (cf. Castilian fuente 'fountain', frío 'cold'). Why would the Basques pronounce some instances of /f-/ well and others incorrectly? Latin and Romance words with /f-/ were borrowed into Basque, but in such cases the /{-/ was regularly reproduced with Basque [b-] or [p-]. Thus, Latin/Romance festa became Basque besta or pesta. Why would Basque speakers reproduce the words one way within their own communities and another when outside those communities?

134

The Burgos phase 3.

4.

As Menéndez Pidal pointed out, the change /f-/ > [h-] is shared by Castillan with Gascon. On the other hand, the Romance of Navarre, where Basques dominated in demographic terms, retained the /f-/. If Basques were responsible for the change, why did it not occur in Navarre? The change /f-/ > [h-] is found in other varieties of Romance (e.g., in southern Italy) far from the Basque Country. Lloyd (1987) also points out that the change is frequent cross-linguistically, so there seems to be a degree of naturalness to the change. If this is so, why do we need to resort to explanations based on Basque influence?

Such arguments cannot be dismissed, and they oblige us to consider two related factors: the linguistic models presented to learners and the social context of learning. Item 2 should be understood as referring to cases of borrowing, in which speakers often have only limited access to models and the items they learn are for use within their own speech community. They are therefore free to adopt foreign words such as festa and adapt them to their own phonology; this is a normal phenomenon in borrowing. On the other hand, item 3 requires an understanding that Basques learning Romance in Castile or southern Navarre would have needed to learn a variety of Romance for communication with native speakers of that variety (or those varieties). When Basque (and other) learners were presented with clear and consistent models, used by a majority of speakers in the target community, it is probable that they (or their children) acquired the target forms correctly. We know that the rate of expansion was much slower in Navarre than in Castile, and that the degree of demographic and dialect mixing was far less in there than in Castile. Basque learners of Romance were therefore presented with more stable and homogeneous input than they might have received in Castile, where the rapid mixing of speakers from several regions to the north, west, and east, would have made the input more unstable and heterogeneous. Moreover, the relative demographic stability of Navarre would have contributed to much stricter norm enforcement, which would have served to suppress Basque interlanguage features not shared by native Romance speakers.93 If Basque learners of Romance had heard Romance speakers in Castile consistently articulating [f], then they (and, more importantly, their children) probably would have learned this articulation. But the change to [h] did occur, but not in all possible contexts (as indicated in Item 1), so it is likely that at least some native speakers within the early Castilian speech community showed some variation in their speech. As indicated above in Item 4, other regions of the Romance world far distant from the Basque

Linguistic changes

135

country also show the change ItΊ > [h], so it is quite plausible that native Romance speakers introduced the variant [h] into their speech. What sort of variation might this have been? A first consideration is that Latin/Romance /f-/ was probably articulated as both a labiodental [f] and as a bilabial [φ]. The labiodental was favored in those varieties of Romance that also developed a labiodental voiced frivative [v], whereas in central Iberia the bilabial would have been favored through (unconscious) analogy with the bilabial voiced fricative [β] (Lloyd 1987: 213). According to Penny (1972, 1990, 1991a), the articulation of /f-/ (or /φ/) was probably once limited to [φ] in all contexts. Later the appearance of the spontaneous diphthong with [w] would have led to the addition of a rounded variant [M] before the glide [w]. At a very early stage, possibly in Cantabria or primitive Castile, before expansion to Burgos and beyond (but this is impossible to say, for Cantabrian usage might have followed Castilian in this regard), this was complicated by the development of another allophonic variant in complementary distribution with the earlier variants: — [h] before syllabic loi, Ivrf: FURNU > [horno] horno 'oven' — [M] before [w]: [Mwérte] fuerte 'strong' — [φ] elsewhere (before other vowels, the glide [j], /r/, IV)·. FARINA > [eparina] harina 'flour' (Penny 1991a: 80)

Penny suggests that the use of [h-] before back vowels arose from an articulatory conflict between the slit fricative [φ] and the following rounded back vowels. Given that [φ] is lax and inherently unstable (Penny 1972: 466-467), an easy and probably automatic solution in fast speech would have been to simply relax the articulation of the lips even further, which would lead to the articulation (and/or perception) of a silent vowel similar to a glottal aspirate. Once [h] was introduced as an allophone (probably through listeners' reanalysis of the weakened bilabial), it could be generalized to other contexts, leading to the following distribution that was probably characteristic of Old Spanish (and is retained in many rural dialects today): — [M] before [w]: [Mwérte]/weríe 'strong' — [φ] before hi, IV\ frente, frío — [h] elsewhere; before all syllabic vowels and glide [j]: FERRU > [hjéro] (adapted from Penny 1991a: 80)

136

The Burgos

phase

How might koineization have affected this development? If some Romance speakers (perhaps of Cantabrian origen) began to pronounce [h] in at least one phonetic context, then the door was open to the extension of this phenomenon. If the group of innovators formed a sizeable bloc of the early Castilian population, then it is easy to see how the first stage described by Penny could have been favored in the koineizing context. It is possible that overgeneralization of the [h] allophone occurred during Burgos-phase koineization among native Romance speakers, particularly since it is a more natural change. Nevertheless, the maintenance of /f/ was typical of a number of contributing dialects, and this fact might have worked against extension of the phenomenon. Basque learners of Romance may have played a role in extending the phenomenon by increasing the relative frequency of the [h] allophone in the prekoine linguistic pool. Presented with varying realizations of the phoneme /f/, Basque speakers may have tended to reproduce more easily and more frequently the one (perceived) allophone which existed in their native language: [h]. Basque learners of Romance may thus have contributed to "tipping the scales" in favor of the [h] allophone. This reconstruction is just one of several that are possible (see Lloyd 1987: 220-223 for a slightly different version). The extreme lack of evidence means that no definitive explanation of the change /f/ > [h] will ever be possible, and applying the model of koineization gets us no closer to a definitive explanation. However, if we view this change as part of the process of Burgos-phase koineization, then we must reject extreme arguments such as that of Trask, in which even the mere possibility of Basque influence is rejected out of hand.

3.3.2. The development of /t// The development of -CT- to [tj] (= orthographic ch) is a particularly salient feature of early Castilian. In most Ibero-Romance dialects, the syllablefinal [k] in Latin words such as FACTU, NOCTE, LACTE, OCTO, developed first to a fricative [x] or [c], and then voiced in assimilation to the preceding vowel (which itself was raised if a mid-vowel), producing diphthongal forms such as feito, noite, leite, oito, in Galician-Portuguese, western Leonese, and Aragonese (Lloyd 1987: 252-253; Zamora Vicente 1967: 150, 242). Latin words with the sequence -ULT- showed a similar development; for example, MULTU > muito, in Portuguese, western Leonese, and

Linguistic changes

137

Aragonese. However, in Castilian and neighboring eastern Leonese, the Romance group [-jt-] regularly produced [tf] as the realization of the new single phoneme /t//. This is surely one of those Castilian changes that is considered "phonologically deviant" by Posner (1996: 208) and therefore not to be associated with the dialect leveling of koineization. It might also be seen as introducing a new phoneme which existed in none of the contributing varieties and thereby leading to an unexpected increase in the inventory of phonemes. However, I view this change as neither deviant nor contrary to the process of koineization. We have little or no evidence of the articulations of the sequence [-jt-] in most of the contributing dialects in early Castile. Still, assimilation of the following consonant to a neighboring palatal glide is likely to occur at least sporadically in all varieties showing this sequence.94 If, as is often assumed, the dialect of Cantabria had previously developed an allophonic variant [-jt/-], then this would have entered the Castilian prekoine linguistic pool. Still, this variant would have entered into competition with the conservative and common articulation [jt]. In the koineizing environment, however, it may have been favored by a natural articulatory tendency toward (and/or perception of) palatalization of the consonant in this context (especially in fast speech), and norm enforcement mechanisms would not have limited its extension. As such, koineization might actually have favored the extension of this variant and the eventual loss of the original sequence [-jt-] (including loss of the glide). The phonologization of the palatal variant of the original Ν is then no more than the end result of the decrease in allophonic variation that we expect to find in koineization, although in this case we have the addition of a new phoneme. We must therefore conclude that it is possible to occasionally find the development of new phonemes in koineization, but only when this is the result of the rapid elimination of numerous variants. Nevertheless, this is not the perfect case to introduce this argument, for the phoneme /tj/ was almost surely introduced into all Romance varieties of the northwest and northcenter of Iberia (before Burgos-phase koineization in Castile) through the common development of post-nasal consonant clusters (e.g., AMPLU > ancho·, see below). As such, many speakers in early Castile may have perceived (and reanalyzed) a slightly palatalized variant of the consonant in [-jt-] as an exponent of the pre-existing phoneme /t//, and then begun to reproduce it as [tj]. Over a generation or two, this could easily have led to the regular development of [-jt-] to [tj], and then as /tj/. Of course, this is just one potential reconstruction of events. As in the case

138

The Burgos

phase

of the change /f-/ > [h-], the lack of contemporary evidence does not allow us to prove this or any version of events. However, what I hope to show here is not that the change in question is necessarily the result of koineization (though it may be); rather the important conclusion is that this change (and others like it, such as the development of medieval Castillan /g/) are perfectly consistent with the process and expected outcomes of koineization, and therefore do not constitute counter-evidence to my more specific claim that koineization played a key role in the development of the preposition + article contractions and the system of tonic vowel phonemes. The general implication for the model of koineization is that, during the leveling out of phonetic variants, very frequent allophones could be selected by speakers and phonologized. This in turn would actually allow an increase in phonemic inventory, but at the expense (simplification) of phonetic variation.

3.3.3. The development of CL-, PL-, FLThe Latin consonant clusters /kl-/, /pi-/, /fl-/, were all affected by a common change in Late Latin/Early Romance: palatalization of the lateral, which produced the clusters [ρλ-] [kl-] [ίλ-]. These feature-heavy and unstable clusters were nearly everywhere simplified in some way.95 According to Menéndez Pidal, the regular Castilian development of word-initial PL-, CL-, FL- was to a bare palatal lateral /λ/. As we have seen, he identified this as one of the defining features of early Castilian (Menéndez Pidal 1964: 502 and 1962b: 126). There are numerous examples, including the following (from Penny 1991a: 63): —

CLAMARE > llamar

' t o call', CLAUSA > llosa ' e n c l o s e d field', CLAVE > lla-

ve ' k e y ' —

PLAGA > llaga ' w o u n d ' , PLANU > llano ' f l a t ' , PLICARE > llegar ' t o a r r i v e '

—

FLAMMA > llama ' f l a m e ' , FLACCIDU > llacio (later lacio) LA > Lambra

' l a n k ' , FLAMMU-

'personal name'

On the other hand, it has been observed that other words which must belong to the ancient wordstock of Castilian do not show the change (Wright 1980; Anderson 1992; Corominas and Pascual 1981):

Linguistic changes

139

playa 'beach', PLATEA > plaça > plaza 'plaza', PLICARE > plegar 'to fold', PLANGERE >plañir 'to wail', PLANTA >planta — FLACCU > flaco 'skinny', FLORE > flor 'flower', FLUXU > floxo > flojo 'loose', FLOCCU > flueco > fleco 'fringe' — CLAVICULA > clavija 'peg', CLAVU > clavo 'nail' —

PLAGIA >

Menéndez Pidal explicitly rejects these as popular Castilian forms: "en época posterior o por influencia culta, se conservó el grupo" (1962b: 126). However, these words cannot be shown to be simple borrowings (whether dialectalisms or Latinisms), since they often combine "non-Castilian" features (conservation of the cluster) with archetypal popular Castilian features, such as the palatals /·$/ and /ji/ in clavija and plañir. More damaging still for Menéndez Pidal's argument is Badia Margarit's (1972: 148) observation that the empirical base needed to extend the palatalization rule to FL- is lacking; there are only six generic Latin etyma beginning with FL-, of which the majority conserve the cluster (the exceptions are llama < FLAMMA, lacio < FLACCIDU). To further complicate matters, Malkiel (1963-1964: 146-155) and Penny (1991a: 63) argue that the following forms should be considered regular popular developments in Castilian: chato 'snub-nosed'

—

*PLATTU >

—

*PLOPPU (< POPULU) > chopo 'black poplar'

— PLUTEU > chozo > choza 'hut' — CONCLAVARI > conchabar 'to join together' — MACULA > *mancla > mancha 'stain, spot' — AMPLU > ancho 'wide' —

IMPLERE > fenchir > henchir 'to fill u p '

—

INFLARE > flnchar > hinchar 'to inflate, swell'

When one takes into account the fact that Cantabrian and central and eastern Asturian varieties tend to agree with Castilian in the development of CL-, PL-, FL- > /λ/, that eastern varieties (Catalan, Riojan and NavarroAragonese) preserve the initial clusters, and that western varieties (Galician-Portuguese and western Astur-Leonese) show the development CL-, PL-, FL- > /ι//, it is tempting to analyze these changes as the results of mixing during koineization. However, a closer examination reveals that koineization played no more than a secondary role (in preserving these developments), and that the basic patterns arose before the period of Leonese and Castilian koineization.

140

The Burgos phase

The inventory of words affected by palatalization is surprisingly similar across the northwest and north-center of the Iberian Peninsula; words in Castilian that show palatalization to /λ/ tend to correspond to words in Galician-Portuguese that show palatalization to /tj/ (as in Group I of Table 14), while words preserving (more or less) the original cluster tend to correspond as well (as seen in Group ΙΠ). As the examples in Group Π make clear, however, western varieties appear to palatalize more items in the lexical inventory than does Castilian. Varieties of Western Astur-Leonese tend to follow the Galician-Portuguese distribution of forms and specific outcomes, while central and eastern Astur-Leonese and Cantabrian varieties generally agree with the Castilian patterns. There is some evidence that even central and eastern varieties of Astur-Leonese followed the western patterns more than Cantabria and Castile. For example, Asturian retains the form llantón (cf. Castilian planton), which corresponds to Galician (and Old Portuguese) chantao. There is also contemporary dialect evidence which seems to show that a form *llor was once in use in Asturias and Leon (Corominas and Pascual 1981: 917). Nevertheless, the fundamental coincidence in these inventories reveals that the changes in Castilian and other northwestern and north-central varieties must surely be directly related to those in Galician-Portuguese. In fact, what we observe here are almost surely the outcomes of several processes of lexical and social diffusion that occurred throughout northwestern Iberia, but whose epicenter was Galicia itself. The first change certainly occurred before the Burgos phase of koineization, and probably prior to the Moorish invasion. This first regular change to diffuse through the lexicon and the region was the elimination of the consonant preceding the palatal [λ] in many words containing the original Latin clusters. However, some words were left unaffected by this change, and actually preserved or restored the original clusters. This process of lexical diffusion was affected by various factors: CL- and PL- were more often affected than FL- (Malkiel 1963-1964 suggests that this was because the fricative was weakly integrated into the group affected); word-internal post-consonantal position favored the change most strongly (as in Group V), word-initial position less so.96 The process of lexical diffusion was also one of social diffusion across geographical space, and the existing evidence indicates that more forms were affected in the extreme northwest (probably the geographical center of the change) than in areas to the east (Asturias and Cantabria, and areas south that would later be depopulated and repopulated).

Linguistic changes

141

Table 14. Comparative development of words showing CL-, PL-, FL-, in Old Galician-Portuguese and Castilian. Category Group I

Group II

Group III

Group IV

Group V

Galician-Portuguese chave chamar chäo/chan/chao chaga chegar cheio chanto/pranto chuva/chuvia/choiva chama chantagem, chantaxe/chanchagem changer/chanzer chantar/prantar chevella/c(a)ravilla chumbo chôr/frol/flor cravo prazer praça prazo praia pregar fraco froco chato chopo choza ancho inchar encher mancha

Castilian llave 'key' llamar 'to call' llano 'flat' llaga 'wound' llegar 'to arrive' (note doublet: plegar) lleno 'full' llanto 'weeping* lluvia 'rain' llama 'flame' llantén 'plantain' plañid1 'to wail' plantad* 'to plant' clavija 'peg' plomo 'lead' flor 'flower' clavo 'nail' plazer 'to please' plaça 'plaza' plaz(d)o 'term, period' playa 'beach' plegar·" 'to nail' flaco 'flabby, weak' fl(u)eco 'lock of hair' chato 'snub-nosed' chopo 'black poplar' choza 'hut' ancho 'wide' hinchar 'to inflate' henchir 'to fill up' mancha 'stain'

Subsequently, the palatal lateral was devoiced and affricated in some positions, producing [tf]. This occurred with absolute regularity in wordinternal post-consonantal position throughout the region (Group V). In

142

The Burgos phase

Galicia, this voiceless affricate was then generalized to initial position in any words that had already been shifted to initial [λ-] (Lloyd 1987: 226).100 However, in regions to the east, the older palatal lateral was retained in initial position in those words that had been affected by the earlier change. During the Burgos phase, the settlers from Asturias and Cantabria carried [λ] southward, and it survived as the dominant form in Castile and central and eastern Leon. Still, the results in Leon and Castile were not exactly alike, for Galician-Portuguese and Leonese were heavily affected by yet another change: rhotacism of the lateral in remaining conservative clusters. This process led to forms such as cravo and praça (Group ΙΠ). Rhotacism was very regular in Galician-Portuguese (i.e., phonologized), but was restricted (at most) to mere phonetic alternation between [r] and [1] in Asturias and Cantabria.101 Leon, heavily influenced by the norms of its many settlers of Galician origin, followed the western tendency of converting remaining /kl-/, /pi-/, /fl-/, to /kr-/, /pr-/, /fr-/. Castile, on the other hand, shows different results: the remaining conservative clusters in words such as clavo and plaça remained unchanged. Such results are due in large part to lesser influence of westerners among its settlers (particularly in the earlier stages). We may thus conclude that, during the Burgos phase of koineization in Castile, the pre-existing distribution of forms found in Cantabrian, Asturian, and Leonese varieties was used by the majority of settlers and therefore accepted into the koine. On the other hand, the strong western tendency towards rhotacism in conservative clusters was rejected, due to its relative weakness in Asturias, Cantabria, and the east. Finally, we must consider the word-forms chopo, choza, and chato. If we accept that these words do form part of the ancient wordstock of Castillan (as argued by Malkiel 1963-1964 and others), then these forms must be analyzed as the only (surviving) forms affected by the change [λ] > [tj] in the north-center; in other words, they were at the "beginning of the queue" for this change in north-central regions, but no other words were affected by the change. On the other hand, the earliest attestations of these words are from the 13th century or later (Corominas and Pascual 1981), so it is possible that they were introduced by Galician and (western) Leonese settlers to Castile before intentionally phonographic Romance writing became common in the 13th century. This may have happened at any time, even during the Burgos phase (Barrios [1985] shows that Galicians did participate in early attempts at repopulation south of the Duero), but speakers from these regions were probably much more influential during the later

Conclusion 143 Toledo phase of koineization, and even more so in Andalusia (a similar interpretation is suggested by Lloyd 1987: 225).

4.

Conclusion

Though Menéndez Pidal described well both the social and linguistic changes of early Castile, he lacked a means of showing a causal relation between them. Koineization allows us to do just that. During the Burgos phase of Castillan repopulation (9th to 11th centuries), immigrants from Cantabria, Asturias, the Basque Country, Navarre and La Rioja, Leon (and Galicia), as well as Mozarabs from the south and east, settled in the border county of Castile. The bulk of these immigrants spoke different dialects, generally mutually intelligible and considered to be the same language, though the Basques, who spoke an unrelated language, also moved into Castile in significant numbers and there had to learn the Romance speech of the majority. The rapid influx of settlers led to a breakdown in social networks, as seen in the rise of importance of the small landowner, the nuclear family, and customary law. At the same time, the conflicting speech habits of the settlers led to a rapid increase in the number of variants in the linguistic pool, including all the variants of the contributing dialects, as well as innovations in learner language. Accommodation between speakers frequently led to the suppression of variants in favor of some common alternative, but also motivated learning of new features. Cultural and political factors also appear to have favored rapid focusing of this dialect. These conditions had significant linguistic effects on the Romance spoken in early Castile. The most obvious changes include the reduction of the many difficult-to-leam preposition and article contractions, as well as variants of the articles themselves, to invariant and transparent sequences of prepositions and articles. Burgos-phase koineization led to the reanalysis of the originally phonetic spontaneous diphthongs as sequences of two phonemes and the merger of the vowel of undiphthongized forms (i.e., words containing low mid vowels followed by a front glide) with the originally high mid monophthongal vowels, with the larger result that the vowel phonemes were reduced from the seven of Romance to the five of Castilian. Surface analogies, phonological reanalysis, and identity factors led to the choice of the characteristic /ue/ diphthong.

144

The Burgos phase

Given the impact that koineization had on early Castilian, other changes must also be analyzed in relation to koineization (since it is only too easy to attribute all changes to koineization). Discussion of the development of /f/ > [h] allowed us to conclude that Basque influence cannot be ruled out, as Trask (among others) has attempted to do, but the lack of data makes firm conclusions impossible. The development of /tj/ shows that new allophonic variants can be added, and, once allophonic variation decreases, the remaining variants can, in principle, be phonologized. Nevertheless, the fact that this phone or phoneme was probably already used in other contexts and in some contributing varieties means this is probably not a case of introduction of a new phoneme (though it shows how that might happen during koineization). Finally, the varied results of Latin CL-, PL-, FL- might seem to be prototypical effects of mixing, but in fact the broad pattern is better explained as a result of earlier lexical diffusion and areal spread across the northwest, and koineization had only a secondary effect on outcomes in Castile.

Chapter 4 The Toledo phase

The slow advances of the Burgos phase largely ceased at the end of the 10th century with the defeats inflicted by the Moorish general Almanzor, but were resumed and greatly intensified in the Toledo phase, beginning in the late 11th century. Unlike the Burgos phase, the Toledo phase has hardly been considered of any significance to the development of Castilian. This chapter, therefore, focuses on demonstrating that there are indeed changes in medieval Castilian that can best be explained as linguistic effects of the demographic and dialect mixing that occurred in this period. Moreover, the rapid simultaneous advance of Leon and Castile (under one king at the time) across the entire central and western Moorish-Christian frontier favored not only koineization, but the spread of features of the already mixed and reduced Castilian koine.

1. Social history Following the initial occupation of the northern half of the Duero Valley, reconquest and repopulation occurred only slowly and on a relatively small scale, being limited in large part to the attempts to repopulate the southern bank of the Duero. However, several factors would combine to propel the reconquest into a new phase of rapid expansion along the whole Christian/Moorish frontier. Bishko (1975: 356-360) specifies three main factors that contributed to this intensification. First, the Christian kingdoms had attained a level of development that could support such an effort. Significantly, the steady rise in the population allowed the Iberian kingdoms to produce more commodities, to field more men for battle, and to hold reconquered territories through repopulation. Second, the Moors had been weakened by the breakup of their territory into competing taifas, with constant internecine strife. This allowed Alfonso VI - king of both Leon and Castile - to take the central and key kingdom of Toledo in 1085, the fall of which held huge symbolic import for the Christians of all kingdoms since it had been the Visigothic capital and the seat of the Primate of Spain.

146

The Toledo phase

However, this victory caused the remaining Hispanic taifas to invite African forces to aid them against the newly successful Alfonso VI. The Africans, first the Almoravids (from 1086) and later the Almohads (from 1146), came to fight a jihad, took control of the Moorish territory and by uniting it provided the fiercest threat the Christians had seen since the days of Almanzor, a hundred years earlier. This in turn coincided with and provoked the development of a similar attitude on the part of the Christians, who began to view their struggles more clearly as crusades. Indeed, the third factor Bishko cites is the development of the concept of crusade or holy war against the so-called "infidels" not only in Palestine - where the First Crusade took Jerusalem in 1099 - but in Iberia as well. The taking of Toledo, the African threat, and the invention of the crusades turned the Iberian Peninsula into the second front in the war on the infidel. It attracted Europeans from beyond the Pyrenees, particularly Occitans and northern French, zealously dedicated to the ideal of holy war and the extermination of the enemy (unlike many Iberian Christians, who often took a much more secular view of war with the Moors). This coincided with and contributed to a general rise in French (i.e., from the area now associated with the state of France) influence in the Peninsula, with many persons of Gallic origin coming to hold important positions in the government and the church, and large communities of merchants of the same origin being founded in the Spanish cities. Another important effect of this expansion of the reconquest was the constant militarization it imposed. Towns and cities (see below) were required to send militias and cavalry for large armies that had to be fielded again and again, and to a degree that few other European areas would know. These armies promoted the development of numerous weak social ties among those involved. In addition, the constant need for military activity led to the establishment in the mid-12th century of the military orders, seen nowhere else in Europe save the Balkans. These included the Templars and the Hospitalers, but also native orders based on these, such as Calatrava, Santiago, Alcántara, and San Juan. These orders played a key role in defense and repopulation of the new frontier, and are discussed below. The most important effect of the rapid expansion of the reconquest was the necessary repopulation effort that followed in its wake, with concomitant migration and demographic mixing. The taking of Toledo turned it suddenly into the center of (a temporarily) united Castile and Leon, and even of the entire Peninsula, and it attracted new inhabitants from all over

Social history

147

the Peninsula and beyond. The sudden shift of the frontier from the southern rim of the Duero valley to that of the Tagus left a wide, largely empty gulf between Castile proper and the kingdom of Toledo. The area -primitive Extremadura - extending from the south bank of the Duero to the mountain ranges of the Central System, also needed to be occupied and repopulated. During the course of the 12th century, the frontier moved further south and the newly reconquered regions to the east (La Alcarria), the southeast (La Mancha), and the southwest (the Montes de Toledo and Leonese Extremadura), had to be repopulated. Finally, the interior regions of the north, which suffered a drain on their own populations, constituted another important zone of repopulation. In each area, there was great demographic movement and mixing. Toledo itself was one of the principal Moorish cities, with a large population probably numbering in the several tens of thousands, and wellpopulated surrounding areas, including the towns of Guadalajara, Madrid and Talavera. When Alfonso VI conquered the city, many of its residents as well as those of the surrounding region chose to remain. These included the members of the sizable Mozarab community, who were spread throughout the kingdom and who had maintained use of their own dialect of Romance, but who had also become proficient in Arabic. Some Moslems, having received guarantees of protection, also chose to remain (at least for a time). These Mudejars were primarily people of humble position who continued to work the fields or who worked in the building trades (Μοχό 1979: 219). There was also a large Jewish community, and both the Mozarab and Jewish communities saw their numbers increase as Christians and Jews fled persecution under the Almoravids and Almohads in AlAndalus. There is evidence that (parts of) the Mozarab community retained a separate identity up to about 1150, from which time it declined until being completely absorbed by the general community, probably by 1300 (Μοχό 1979: 220). Significant numbers of northern Christians were also needed to defend the new region and replace the many Moors who had left. The new arrivals to Toledo and its surrounding area - who came throughout the 12th century - included a core group of Castilians, but also many Leonese, and significant numbers of Asturians and Galicians: These were also joined by Navarrese, Aragonese, and Catalans (Μοχό 1979: 221). Yet another group must also be counted among the new settlers: Un nuevo y peculiar contingente de pobladores cristianos en Toledo va a estar representado por los francos que hasta allí llegan para establecerse tras

148

The Toledo phase

su conquista por Alfonso VI, pues, aunque no se perciben sus huellas en el ejército cristiano del monarca, gozaban ya de influencia en su corte con el apoyo de la reina Constanza de Borgoña. Los francos establecidos en Toledo debieron representar un número apreciable, para gozar de fuero y barrio propios . . . Dentro de la población franca toledana resalta singularmente por su influjo el elemento clerical, en torno al arzobispo don Bernardo, que llevó a cabo la reorganización de la archidiócesis de Toledo tras su conquista por castellanos, pudiéndose considerar, a su vez, como de origen marsellés la primera generación monástica toledana establecida en San Servando . . . Los francos de Toledo, aunque recibieron su fuero propio, disolvieron pronto su personalidad ante la mayoría castellano-mozárabe. (Moxó 1979: 221)102 In sum, reconquered Toledo (particularly the city itself), though dominated by Castilians, represented a population mix that was far more heterogeneous than that of primitive Extremadura and other regions. Nevertheless, the heavy mixing that occurred in Toledo was echoed, though on a reduced scale, in the cities that were founded to consolidate the new "behind-the-lines" territory of primitive Extremadura. Towns just to the south of the Duero, such as Olmedo, Medina, Coca, Iscar, Cuéllar and Sepúlveda, were apparently repopulated "spontaneously" in the 11th century, as settlers moved in and established themselves under the old presura system (though fueros were also granted). However, the urgent need to secure this region led to an organized effort to found several new cities (or re-establish old ones) as foci for repopulation. These included Salamanca, Ávila, Segovia, and Soria, each of which was founded as an administrative center with control - and linguistic influence - over the villages and towns in the surrounding alfoz. The task of organizing these communities was undertaken by Alfonso's son-in-law, Raymond of Burgundy (a franco), who initiated repopulation of Segovia and Ávila in 1088, just three years after the fall of Toledo. Documentary evidence reveals that the settlers of Segovia included not only Castilians (who represented a clear majority) but also settlers from Galicia, Asturias, Cantabria (montañeses), Leon, La Rioja, Aragon and Gascony (Moxó 1979: 210). Similarly, the Crónica de la población de Ávila reports the arrival of a majority of Castilians, but also of many serranos (from the eastern areas of Castile and La Rioja), Mozarabs, Aragonese, Jews, Mudejars, and Burgundians (followers of Raymond). The repopulation of Salamanca, within the kingdom of Leon, apparently began somewhat later than that of Segovia and Ávila, but it also received a mixed group of settlers, though in this case westerners (Leonese, Galicians, Portuguese) represented the majority; they were joined by many

Social history

149

Castillans, but also Mozarab and Jewish refugees from Al-Andalus, francos, as well as some Mudejars. The definitive repopulation of Soria, to the southeast of Burgos (and directly south of La Rioja), was completed by Alfonso el Batallador of Aragon in order to support his encroachment upon the Moorish kingdoms to its east. Some spontaneous repopulation had already been initiated by Riojans and Navarrese from the immediate north, and to these were added the new settlers attracted by Alfonso, which included a mix of Castilians and easterners. In spite of the evident social mixing that took place in all these cities, Μοχό accords preeminence to the Castilians: Tales gentes procedieron de diversas regiones hispánicas, preponderantemente -claro es- del interior del reino de Castilla y León -como burgaleses, riojanos, 'serranos', vascos, leoneses, palentinos, portugueses, asturianos y gallegos-, pero también procedentes de otros reinos peninsulares, como navarros y aragoneses, así como mozárabes, al igual que los que había llevado Sisnando Davídiz a repoblar Coimbra con Femando I, e incluso - a lo largo del siglo XII- de fuera de la Península, como los 'francos' de que se tiene noticia a través de la documentación y que formaron en conjunto un ingrediente importante de la población peninsular durante la plena Edad Media. Pero ante todo, la nueva población de la Extremadura nació por enjambramiento del castellano situado al norte del Duero. (Moxó 1979: 205)

Still, in general this phase of repopulation shares a key feature with the Burgos phase: in any area of repopulation, the majority of settlers appear to proceed from regions directly north, with the next largest number of settlers coming from neighboring regions. In all cases mixing and contact occurred, and the expansion of Castilian into these areas cannot be seen as the simple extension of the speech of the county of Burgos to regions south. Neither can one argue for the simple political imposition of Castilian speech, since, the expansion of Castilian features does not always coincide with the borders of Castile (see discussion of language spread below). Although the Castilians had pushed the frontier far south in 1085, the counterattack initiated by the Almoravids in 1086 led to a retreat, including the evacuation of existing Mozarab communities. Initial repopulation was consequently limited to areas extending along and behind the Tajo River, from Talavera to Guadalajara, running through Toledo and Madrid. Later this line was gradually extended south and eastward, particularly in the middle part of the century, with the taking of the fortresses of Calatrava (1147) and Uclés (1157), although Calatrava was lost again from 1195 to 1212. The instability and centrality of·this area was to slow repopulation

150

The Toledo phase

and lead to a greater demographic mixing between Castilians, Mozarabs, and easterners (in some areas). Indeed, the eastern frontier of Castile came into direct contact with Aragon in the early 12th century. Zaragoza fell to the Aragonese in 1118, and the Aragonese advance (after merger with Catalonia) reached its southern limit with the founding of Teruel in 1171, but was cut off after the conquest of Cuenca by Castile in 1177. The Fuero de Guadalajara (1133) calls specifically for "pobladores . . . de Castilla . . . de León . . . de Galicia . . . y mozárabes" (Μοχό 1979: 235), but it is known that repopulation of (Castilian) areas east of Guadalajara was organized by the Aragonese king Alfonso el Batallador from 1125, and the Fuero de Guadalajara seems to represent a response to the heavy eastern influence in the region. Nevertheless, Aragonese and Mozarabs from Calatayud were invited to settle Zorita in 1156. Mozarabs were also well represented in the later-resettled regions near Guadalajara and south of the Tajo, with many settlements dominated by them (e.g., the town of Orgaz, just to the southeast of Toledo; González 1975-1976: 119-218). On the southwestern frontier, Plasencia was taken and settled by Castile (separate from Leon from 1157-1230) in 1186. Further to the west, the frontier entered Leonese Extremadura, which remained lightly populated and plagued by warfare, although Leon continued to advance slowly. Coria, for example, was taken in 1143, and, for a limited time, Cáceres fell under Leonese control. At the western extreme, Portugal was recognized as an independent kingdom from the 1140s, and though periodically joined by Crusaders from northern Europe, accomplished much of its reconquest and repopulation on its own, particularly the notable conquest of Lisbon in 1147. Further east and south of Toledo, La Alcarria and La Mancha were for a century the scene of constant warfare, especially in the valley of the Guadiana. Warfare, a lack of natural barriers, and an arid landscape made repopulation of these zones difficult, and repopulation was never as successful here as in Old Castile and interior areas of the kingdom of Toledo. Indeed, repopulation south of the Guadiana was only secured after the battle of Las Navas de Tolosa in 1212 (Μοχό 1979: 230). Elsewhere, detailed lawcodes, fueros extensos, were granted to make settlement here more attractive, and the military-religious orders were given direct administrative control in most frontier zones to facilitate the process. These orders tended to operate autonomously, and their outposts were spread along the frontier from Portugal to Aragon, where they were responsible for defense and repopulation. The Order of Calatrava, for example, was founded in 1158 by the Cistercians of Fitero in Navarre (but with the support of the

Social history 151 Castilian king Sancho ΠΙ), to replace the Templars at Calatrava la Vieja, located on the southern frontier of Castile. It then extended its holdings to Aragon, Leon and Portugal. Similarly, the Order of Santiago developed from the earlier fratres de Càceres (in Leonese Extremadura), sponsored by the Leonese king Fernando Π and the Archbishop of Santiago de Compostela. After the loss of Cáceres (retaken by the Almohads in 1174), the Order was invited to transfer its main frontier operations to the fortress of Uclés in Castile, to the east of Toledo. From then on, the Order was doubly administered from Uclés in Castile and San Marcos in the city of León. Significantly, these orders were composed of members who owed primary allegiance not to any one kingdom or region, but rather to the order, which existed across borders, and they actively recruited settlers from across territories in areas controlled by them, including Cuenca (Bishko 1975: 418; Fernández Ordóñez 2001: 454). Much of the coastal north, including the coast of Cantabria) was not well populated during the Burgos phase, and continuing migration to the south necessitated efforts to restore both rural and urban populations in inland regions. The repopulation of the interior is largely associated, as in other regions, with the founding and development of cities. In particular, cities were founded or repopulated along the northern coast, such as La Corufia in Galicia, Avilés in Asturias, and Santander, Castro Urdíales, Laredo, Santillana del Mar, and San Vicente de la Barquera in Cantabria alone (Lapesa 1951: 206). Raymond of Burgundy also encouraged the growth of Zamora and Valladolid at the time he was founding Ávila and Segovia to the south. As elsewhere, this repopulation was aided by a rise in the birthrate and the granting of fueros extensos, but other important factors were the increased security provided by the then distant frontier, the decline of Norman raids on the northern coasts, and the growing importance of the Camino de Santiago (or camino francés). Early in the 11th century, Sancho el Mayor had changed the Camino de Santiago from its original coastal route to a route skirting the southern edge of the Cantabrian Mountains. This change increased traffic, improved commerce in the cities it traversed (such as Pamplona, Burgos and León), and led to the founding of new cities along the way, such as Estella in Navarre. The importance of the route may have contributed to east-west movement and contact in the region. Toponymie evidence indicates that Catalans and Galicians established villages in Leon and Castile in this period (Μοχό 1979: 266). More importantly, significant numbers of francos, largely merchants, moved into these cities. The francos often established

152

The Toledo phase

separate quarters and received special fueros. Indeed, they appear to have formed a significant portion or even majority of the populations of Pamplona and Estella. Gallic influence in the region was also enhanced as the Cluniac movement took over many monasteries and foreign clerics came to Spain. The large influx of francos helped increase the relatively low urban population of the mountain regions of Navarre. The same appears to have occurred in sparsely populated Aragon (Lacarra 1972: 164), where notable growth only began after 1063, when Jaca was selected as its primary city. As the Aragonese expanded from Jaca to Huesca to Zaragoza, their lower numbers forced them to seek settlers wherever they could, and the francos, many of whom had participated in the taking of Zaragoza, came to make up a significant portion of the new population (Bishko 1975: 404).103 The Mudejars were also encouraged to remain, and Alfonso el Batallador even prohibited them from emigrating. When Alfonso took the Jiloca and Jalón river basins (close to the border with Castile), with the cities of Calatayud and Daroca, he repopulated them (and neighboring areas of Castile) with Aragonese, Navarrese, Catalans and francos. In 1126, Alfonso returned from an expedition to Andalusia with many thousands of Mozarabs. Noteworthy, too, is the presence of Castilians in the south of Aragon. Μοχό (1979: 313) reports that Castilian knights were awarded villas in this area and that the Orders of Calatrava and Santiago were also present in the southern region. Μοχό concludes: la extensión del Derecho de la Extremadura castellana que se efectúa en esta época por la zona del bajo Aragón -partiendo del núcleo propulsor del Fuero de Sepúlveda- encuentra fácil explicación en la presencia de elementos castellanos que le favorecieron. (Moxó 1979: 313).

It is clear that the Toledo phase of repopulation led to a similar kind of demographic movement and mixing as found in the earlier Burgos phase. However, in some areas it involved an even greater variety of groups - and thus language varieties - than those found in the Burgos phase. In particular, the presence offrancos must be stressed, for as a group they possessed tremendous prestige, and could therefore influence prestige norms. Nevertheless, important differences in social organization between the early and Toledo phases have been pointed out by historians. In the Toledo phase, clearly marked hereditary class distinctions appear in Castile, with an aristocracy divided between great nobility (ricos hombres) and lesser nobility {hidalgos). Though such class distinctions often indicate a more rigid society, their appearance during the Toledo phase can be attributed to

Koineization and language spread 153 the opportunity to acquire new wealth as the kingdom expanded, and the penetration of feudal ideology accompanying the arrival of the francos (Μοχό 1979: 403). Along with feudal hierarchization, Castilian repopulation became much more official in nature, with the old system of presuras disappearing, and new settlements organized by the king, nobles, bishops, monasteries, and military orders. Of course, such social class distinctions would have favored the development of social and stylistic linguistic variation. Even with these changes, however, there is evidence that contrary trends towards social leveling and mobility continued. For instance, in order to attract new settlers, fueros promoted equality for non-nobles within new communities; lots apportioned to settlers often were required to be of equal size and in some fueros infanzones/hidalgos were required to give up their special status before being allowed to settle in a new town (Μοχό 1979: 229, 412). The fueros, of course, were associated with cities, and a hallmark of this period is the increased level of urbanization, in which cities were developed as magnets of colonization. But even as the cities served as stable administrative centers, they also served as demographically heterogeneous bases for the development of a widespread network of weak social ties, along which changes initiated in one area could spread to others.

2. Koineization and language spread/dialect leveling Migration and mixing during the Toledo phase was far more rapid and widespread than in the earlier Burgos phase. These demographic changes contributed not only to koineization, but also to the rapid and early spread of multiple features which we now identify as Castilian. The spread of Castilian features can be understood as part of a process of language spread or dialect leveling. Significantly, these ways of viewing the spread of features conflict with traditional views. Menéndez Pidal described the expansion of Castilian as assuming the form of a wedge (cuña), which fanned out towards the south from the early county of Castile. However, it is clear from his writings that he did not view this wedge as developing immediately with the expansion southward of the Toledo phase. Rather, he appears to have seen it as developing its full form only over time, largely as a result of the political domination of Castile:

154

The Toledo phase

como en estos siglos XII y XIII la reconquista se activa semejantemente en los otros reinos, no sólo Castilla, sino León, Portugal y Aragón propagan entonces hacia el Sur sus dialectos respectivos mucho más activamente que en las épocas anteriores y los propagan en sus formas meridionales ya algo mezcladas por la reconquista anterior y por las repoblaciones consiguientes. Estos dialectos de invasión rápida, propia de los siglos XII y XIII, comienzan en León al sur del Duero. (Menéndez Pidal 1964: 514) From these comments it is clear that Menéndez Pidal was aware that the differences between some dialects diminished with the initial move southward. However, he clearly defends the idea that the wedge would only appear later as dialect speakers in contiguous territories shifted to categorical use of dominant Castilian (see also Chapter 6). Underlying this interpretation is the image of the column, in which each of the main language varieties spread south with political conquest, assuming the form of parallel columns (as the shape of each of the political territories could be described). Still, if these columns were only vaguely defined for Menéndez Pidal (a position which merits some support), for others they have become clearly defined entities. This conception is clearly represented on linguistic maps in some manuals (e.g., Diez, Morales, and Sabin 1980: 169) and indirectly in others (e.g., Lapesa 1981: 192). Zamora Vicente's comments clearly support the existence of sharply-defined dialect borders that only disappear with the later adoption of the official language of Castile: La primera frontera entre el castellano y sus laterales aragonés y leonés, ha sido o muy alejada o difuminada por la acción secular de la lengua oficial. (Zamora Vicente 1967: 12) However, the massive mixing that took place after the taking of Toledo (and Zaragoza in the east) makes it difficult to argue that sharp divisions for numerous features - to the extent that they ever existed - would have held during the Toledo phase. Indeed, the social environment would not have been able to support them, with norm enforcement mechanisms weakened and exposure to variation increased. In addition, metalinguistic conceptions of dialectal distinction were probably weak at the beginning of the Toledo phase (Wright 1994d: 41). As discussion in the previous chapter reveals, Castillans were probably among the first to begin to define regionally-based norms and a regional conception of their speech variety, but there is significant evidence that this did not exist for northern varieties of Leonese, where locally-based norms predominated. Speakers certainly would have been aware of stylistic and geographic differences, as in all times and places, but a lack of easily categorized difference may have

Koineization and language spread 155 made individual speakers even more permeable to innovation and encouraged them to engage in variant-shifting. This would have made the maintenance of clearly-defined dialect divisions impossible. Indeed, it is at this point in the process of reconquest and repopulation that it becomes clearly obligatory to accept the widespread homogenization that Ridruejo (1995) claims for the early stage. Though this spread can be analyzed as dialect leveling, it is also useful to analyze it from the perspective of the model of language spread, which has received attention from sociologists of language. Cooper (1982: 2122), for example, identified certain factors that generally contribute to the process of language spread. These include low cost, high anticipated profitability, easy trialability, low complexity, high pervasiveness, and positive attitude. During the Toledo phase of repopulation, speakers in the repopulated zones faced minimal cost if they adopted Castilian-like features, since in many cases they already had these features in their own repertoire (exclusively or in variation with others) and did not need to learn any new features, but rather eliminate some of the marked features of their native dialects. The Burgos-phase Castilian koine would have presented less complexity (with one notable exception; see below) than other competing varieties, with, for example, the reduced and stabilized vowel system or the transparent preposition + article combinations. In the new mixed populations, the consistency and frequency with which Castilian speakers produced these features, which were also variably present in the speech of other dialect speakers, would have contributed to their pervasiveness, as speaker-learners were more consistently exposed to these features. Adopting the speech patterns that were most frequently used in the developing speech community would clearly have been socially profitable, and speakers would have been free to use these features without great fear of censure (itrialability), since norm enforcement mechanisms were weakened. In addition, speakers' attitudes may have ranged from positive (prestige of speech associated with victorious Castile and the city of Toledo) to, more likely, at least non-negative, since weak norm enforcement mechanisms would have relaxed any pressure to maintain features from other, possibly more complex, dialects. A particular example of the effects of this process may be found in the 12th- and 13th-century language of Salamanca, a major city repopulated with Ávila and Segovia, but farther to the west within the kingdom of Leon. The early language of Salamanca is often considered to have been a simple southward extension of Leonese, and it is common opinion that the

156

The Toledo phase

speech of the city was not Castilianized until the 14th century. However, several factors might make one suspect this simple characterization. First, at the time of reconquest, Leon and Castile were under one crown, which may have weakened - if only temporarily - perception of collective differences between Leonese and Castilian. Toledo itself had fallen under Castillan hegemony, so, to the extent that prestige was a factor, it would have favored the spread of Castilian features. We also know that the city of Salamanca was settled by people of greatly varied background (see above), so it cannot be argued that colonization by a homogeneous group led to a simple transplantation of Leonese. Finally, Leonese, even in northern Leon, seems to have always shown a great degree of geographic variation, and was apparently less focused as a regional variety than Castilian (i.e., norms were often locally restricted). It is impossible that political factors alone could have supported the simple southward extension of these varieties (to the extent that such variation could be transferred to a new zone). The only extant collection of early Salamanca documents is that transcribed by Martín Martín, Villar García, Marcos Rodríguez, and Sánchez Rodríguez (1977). This collection contains 113 documents from the 12th century and 363 from the 13th, the majority of which were prepared in Salamanca. The editors point out that most of the texts are in Latin or are very Latinate, but that in texts prepared by local notaries and scribes, the Romance (or more phonographic) element appears in ever growing proportions (starting with words and phrases) until the 13th century, when whole sections of some texts are clearly in revised (Romance) spelling (1977: 42). The first entirely phonographic text is from 1240 (Doc. 199). Since the editors' primary intent is to make texts available for historians (1977: 78), the transcriptions are not paléographie. For example, filled-out abbreviations are not indicated in the transcriptions, and the collection is thus not as useful for linguistic study as might otherwise have been the case. Nevertheless, the texts, such as they are, do reveal significant mixing of Leonese and Castilian elements. One finds, for example, either the clearly Latinate spelling pectet or the Castilian (and eastern Leonese) spelling peche (Docs. 144-1222, 201-1242, 211-1245, 222-1247, etc.) with no examples of the Leonese forms with -it-, the same is true of factu/fecho. Little effort is made in the texts to represent diphthongs until the 13th century, but when they do appear, they are the entirely stable and Castilian ie and we: castielo (Doc. 174-1225), Azogue Viello (Doc. 193-1236), sueldos (Doc. 1951237). From early on the preposition and article combinations show fluctuation between the transparent combinations typical of Castilian and the

Koineization and language spread 157 more conservative contractions: en la zoc velo, en la aldea (80-1181), en la Zoc Veio (93-1187), en lo vostro poder, in lo mes (177-1227), in las dos partes (192-1235), en la Azogue Viello (193-1236), con los parientes (199-1240), en el vostro (222-1247), con los (199-1240), con la (2041242), but enno castro (171-1225), enna carera (172-1225), ena collation (179-1228, 180-1228), enna se (181-1229), enno poderío (183-1229), ena, conna (199-1240). Normal apocope is regular in these texts, though there is also a preference for the forms heredade and edade, with final -e maintained as in western varieties. In these texts up to the mid-13th century, I have found only one example of that most characteristic feature of Leonese: syllable-final consonant weakening to a lateral: duldanza (246:1253, where the lateral substitutes b). Still, there are two possible exceptions to these patterns. The Castilian /·$/ ( < CL, LI) was not adopted, at least in the orthography: zoc velo (801181), Azoe Veio (89-1185, where -i- probably represents [j] since we also find Clemens lalfaiat in the same document, which presumably represents the Arabism alfayate 'tailor'), Carvalosa, Carvaiosa for Carbajosa (791181 and 68b-1177), and numerous later examples of filio, fiyo, muler, conceyo, conzello. In addition, count/mass pronoun distinctions and leísmo, which became so typical of Castilian were not established in Salamanca. The rejection of Castilian / j / may be only apparent (i.e., a failure to use a distinct grapheme), but it may be that it arose only later in Castile, and/or became a salient marker of developing Castilian identity (particularly in the later 12th century when Leonese and Castillans were in frequent conflict). The lack of leísmo and count/mass distinctions is first and foremost attributable to lack of a Castilian majority in the original population of Salamanca, but its developing salience as a feature of Castilian identity, may have made it socially difficult for speakers to adopt this feature. In either case, leísmo and count/mass distinctions are among the few features of Castilian which are more complex and marked than equivalent structures in Leonese, and they therefore would have been difficult to learn even if speakers had wanted to learn them. It is interesting to note that today, it is precisely Salamanca that shows the fewest Leonese features, even in rural dialects, while zones to the north and south show stronger retention of such features. Northwestern Extremadura, of course, was reconquered and repopulated later in the 12th and 13th centuries when Leon and Castile were ruled by different monarchs and in frequent conflict. Southern pockets in this zone may represent transplants of Leonese by relocation of more homogeneous groups than those which settled Salamanca.

158

The Toledo phase

Similar arguments can be made for Aragonese south of the Pyrenees. As is the case with Leonese, there are few texts from outside the northernmost areas - Pyrenean areas of Aragon in this case - that are considered as representative of the so-called "pure" dialect. Indeed, Frago Gracia (1991a: 120) has argued that the Aragonese of the Ebro Valley was never simply transplanted to the Ebro valley, where a simplified and mixed language variety - closer in appearance to Castilian - developed, primarily as a result of dialect mixture, but also as a result of simultaneous dialect contact and spread. Significantly, the Toledo phase marks a period of simultaneous demographic mixing for Castile, Leon, and Aragon. In the context of this discussion, it is easy to envision the increase in weak social ties that occurred and the concomitant increase in permeability to innovations that had already been consolidated as changes in the center of the Peninsula. Language spread or dialect leveling of this sort was certainly facilitated by koineization at the same time that the spread to new speakers increased the possibility of koineization. Still, the spread of features that we now associate with the developing Castilian koine was not limited to the southern or lateral areas that were suddenly opened to repopulation. There are two reasons for this. First, as was discussed above, there began a massive program of internal repopulation in the northern areas, focused on though not limited to urban centers. In these areas, traditional norms would have been weakened as social networks were weakened, though quite clearly not to the same degree as in the south, since to this day these northern regions show the greatest conservatism. Still, these areas would have become more permeable to changes that were first consolidated in the newly repopulated zone through the process of spatial diffusion of features. Spatial diffusion has generally been understood as occurring as a kind of wave that spreads geographically. This may be a valid model for the largely rural and agricultural societies of the very early Middle Ages (even early Castile), in which most inhabitants lived in small villages separated from other villages by short distances. In such societies, innovations probably do spread from village to village. However, modern sociolinguistic research has made clear that once larger urban centers appear, particular features do not spread in this way (though historically they may still appear to do so). Rather, features seem to jump from urban center to urban center, and thence to progressively smaller urban centers before slowly spreading into intervening rural areas (Chambers and Trudgill 1980: 189). These authors cite as a particularly stunning example the spread of uvular [R] from France to urban centers in Germany and Scandinavia, across national

Linguistic changes 159 and linguistic borders (Chambers and Trudgill 1980: 189-191). This pattern can also be seen in the spread of front-vowel raising (e.g., bag /baeg/ > [beg]) from Chicago into outlying areas, where some urban centers were participating in the change, but intermediate rural areas remained unaffected (Callary 1975). The Milroys (1985; Milroy 1992: 197) have explained that the impact of geography or urbanity is indirect, and that the real factor underlying this parachute effect is the conglomeration of weak social ties between urban centers that allow the spread of innovations. In 12th- and 13th-century Iberia, the same kind of mechanisms must have been present. Koineization was certainly taking place in the centersouth around Toledo, in a variety of northern areas (though to a lesser degree), and in reconquered Zaragoza, but, unlike cases in distant colonies, contact was maintained with the contributing communities and their linguistic varieties. In particular, we might expect changes to spread back from southern regions (or cities) into northern areas (or cities), even if these areas saw a lower degree of social mixing. The Camino de Santiago must also have promoted the development of weak social ties and thus served as a conduit for innovations across northern Castile and neighboring areas. These considerations must be kept in mind when interpreting the documentary evidence.

3. Linguistic changes We now examine the evidence for three Toledo-phase changes in Castilian and, to some degree, other dialects: apocope (reduction of variants in normal apocope, mixing with stylistic functional reallocation in extreme apocope), early leísmo (mixing showing the survival of interlanguage or interdialect features with distinct solutions for different areas, according to different population mixes), and reorganization of the possessive system, through social and functional reallocation and simplification. Though some of these changes are sporadically attested in Burgos-phase Castilian, and early leísmo certainly originated in the Burgos phase, I argue that each of them was consolidated or significantly altered as an effect of Toledo-phase koineization. With regard to evidence, this period is in some ways easier and in some ways more difficult to investigate than the Burgos phase. Available documentation for this period is more abundant, but no comprehensive study such as Orígenes has been prepared for it, in large part because it has not

160

The Toledo phase

generally been identified as linguistically significant. Moreover, in the 12th century most documentation tends to be highly Latinate, and even less useful as evidence following the reforms of the Council of Burgos in 1086 (Menéndez Pidal 1964: 460; Wright 1982), which helped to reinforce prescriptive norms of writing. Romance only began to be written phonographically with some regularity in the 13th century, so there is little direct evidence available for the Romance of the 12th century. Of course, it is the 12th century that is of greatest interest in this chapter, since it is during this period that the linguistic changes resulting from the Toledo-phase repopulation were being completed. Nevertheless, if we assume that the early 13th century documents from the regions around Toledo preserve the results of the koineizing process that must have occurred during the 12th century, we can, in most cases, deduce from this later evidence changes that must have occurred during the preceding century. Unfortunately, even today relatively few texts can be accurately located both chronologically and geographically. As a result, the richest source of evidence for this period continues to be the texts assembled by Menéndez Pidal in his Documentos lingüísticos de España. In addition, researchers have closely analyzed many of the first texts written in Castilian (often literary or historical works), and their findings will be referred to where appropriate.

3.1. Leveling and reallocation in the development of apocope Apocope in medieval Spanish has been most thoroughly analyzed by Lapesa (1951, 1975, 1982). He divides cases of apocope into two categories: normal and extreme. Normal apocope refers to the permanent loss of word-final /-e/ after the phonemes represented with the letters r, s, I, n, z, d (Lapesa 1975: 13). Lapesa also uses the term to refer to the occasional loss of final /-o/ (or /-u/?) in enclitic or proclitic position, if it left r, s, I, n, z, d in final position (e.g., Martino Gonçalvez > Martín Gonçalvez, solo > sol), whether or not they are maintained in this form to the present day. Extreme apocope refers to the (what historically seems) "temporary" phenomenon of loss of/-e/ after other single consonants (e.g., nuef [< nueve], quiçab, dix, noch, axac) and consonant+dental consonant clusters (e.g., adelant, part, fizist, ond, estonz). It also refers to loss of final /-o/ and even sometimes /-a/ in forms used as clitics after the same consonants and clusters that define extreme apocope of /-e/ (e.g., todo > tot, Lobo > Lop, Diago > Diag, Diac, Fernando > Fernand, Fernant, como > com). Lapesa consid-

Linguistic changes 161 ered extreme apocope to be a phenomenon that occurred intensely from the end of the 11th century to the end of the 13 th, and, in some areas, up to the mid-14th century. Both normal and extreme apocope appear to have been affected by Toledo-phase rekoineization, with this period marking the consolidation of normal apocope (or, better, the rapid reduction of variation through the stabilization of phonological forms) and the functional reallocation of extreme apocope. Normal apocope - as the name might suggest - has received far less attention than extreme apocope. It is true that the loss of final /-e/ following many of the apical consonants became a widespread and regular phenomenon in western Romance (though in Galician-Portuguese it failed to occur after /d/). Menéndez Pidal (1964: 186-187) reports it was probably an ancient phenomenon, and provides the example of a 6th-century coin bearing the inscription LEONES MONETA. Nevertheless, he points out that examples of normal apocope are infrequent even in 10th-century texts, where the few examples he finds occur after n, r, s, z. In the Glosas emilianenses and Glosas silenses the final vowel was regularly conserved: muliere, stiercore, promissione, meretrize, uece, uoluntate, but clear examples of apocope of /-e/ are found in the future form following s in alongarsan (< alongar se han), and following liquids in tal, quai, leuator (Pidal claims that the last of these was influenced by the nominative form being glossed; the others are words whose high frequency would have favored their reduction). Pidal concludes from this that "en la segunda mitad del siglo X la apòcope tenía muy escasa cabida en la lengua común" (1964: 186). If one accepts the later dating now assigned to the Glosas by most scholars (e.g., Wright 1994e: 209; Díaz y Díaz 1978; Bezler 1991), the implications of this evidence would seem to grow in significance, for they would indicate a very rapid change in use of final /-e/. It seems more likely, however, that early full forms hide alternative manifestations that occurred in speech, where these forms regularly suffered phonetic reduction (in a kind of phrasal syncope), but without the phonological reanalysis that underlies adaptations in phonographic spellings. Pidal reports that the process of normal apocope grows stronger in written texts during the later 11th century, when the fluctuating loss or retention of final -e after apicals occurs alongside increasingly frequent manifestations of extreme apocope and syncope. Such forms indicate that phonetic reduction of /-e/ had become so frequent that it was beginning to lead to phonological reanalysis. Such reanalysis would be consolidated during the Toledo phase (although, as we

162

The Toledo phase

will see in the discussion of leísmo, at least one very high frequency form suffered full reanalysis during the Burgos phase). Lapesa links the reasons for the development of normal apocope to those he gives for extreme apocope (see below): En el tránsito del latín al romance español del siglo X, la simplificación, palatalización o asibilación de dobles consonantes y grupos consonánticos latinos había reducido mucho el número de fonemas capaces de cerrar sílaba interior. Así habían desaparecido las consonantes implosivas en SEPTEM > sete o siete-, FACTU > faito o feito; LAKSARE > laisare o lesare; PUGNARE > puñare; CUPPA > copa-, SAGITTA > saeta; BUCCA > boca. Los únicos fonemas consonánticos que podían tener función implosiva eran r, l, s y m/n. Precisamente fue tras estas consonantes donde se inició la apócope de la e latina; recuérdense los ejemplos ya citados qual, tal, segar, aliatoti, o el astoricés de una escritura de 1030. (Lapesa 1951: 192-193) At the phrase level the originally phonetic phenomena of syncope and apocope would have had similar effects. In addition, speakers were under less paradigmatic pressure to maintain final /-e/, since, unlike final /-o/ (or /-a/, to the extent that it was affected), it was not a gender-marking morpheme. Such arguments can explain the origin of the variation between full and apocopated forms in many Romance varieties, but do not explain the chronology of the development in any particular variety. Normal apocope - in its different manifestations - may have developed and been consolidated in different areas at different times and at different rates. When did this happen in Castilian? Lapesa (1951: 204-206) seems to view the mostly regular use of final -e after r, s, I, n, z, d as continuing in Castilian only up to the 12th century, after which its appearance must be ascribed to particular causes. He states: En Castilla, aparte de las rimas en la poesía épica, era muy rara ya [12th century]. No faltan casos sueltos, ora debidos a procedencia dialectal de los notarios, ora a un apego individual por gustos viejos. (Lapesa 1951: 205) Lapesa mentions specific examples of retention of final -e in the Fuero de Valfermoso de las Monjas (1189) and Fuero de Madrid (1202), where they appear mixed with examples of apocope: andadore, menestrare, piede, Madride, but segur, sayal, fiador, Maydrit (1951: 205). Nevertheless, these are both documents that intentionally Latinize many parts of the text, and these particular forms may simply show that Latinizing influence. Lapesa also gives examples of retention of final -e in the Fuero de Avilés (particu-

Linguistic changes

163

larly after d, as in potestade, lide), but ascribes them to conservative Asturian influence (today the conservation of /-e/ after /-d/ occurs only in western Asturias and Galicia). Likewise, the Mozarabs of Toledo adopted use of final -e - with "coherente unanimidad"- during the 12th century, perhaps in an overt conservative reaction against the dominant trends, which may have served to mark a separate identity within the broader koineizing community (Lapesa 1951: 192). Perhaps the most significant evidence of the 12th century shift in status of normal apocope is the disappearance from the texts of a certain kind of hypercorrection: Ultracorrecciones como matode por matod 'mató', kede por ked ( < QUID), bédene por veden (evident), stane < stant, abundan hasta principios del siglo XII. (Lapesa 1951: 188)104

The disappearance of such hypercorrections would appear to indicate the general loss of alternating full and apocopated forms associated with a fundamentally phonetic process of vowel deletion. The end of confusion on the part of scribes indicates the end of the variation produced by the phonetic process of normal apocope, and it reveals speakers' definitive phonological reanalysis of these forms as lacking underlying /-e/. Álvarez Rodriguez (1996), in a study of normal apocope, confirms that it began early after the single consonant phonemes /r, z, 1, n, d7, but argues that it was only extended later to /-d/.105 He adduces as evidence a document from Zamora (dated 1050) that contains within 15 lines numerous tokens of normal apocope (e.g., puçal, aliatoti, pergaminar, sal, segar, caral, segur), but none after -d. This is, of course, a Leonese document, but Lapesa agrees that the same was true of Burgos-phase Castilian (1951: 193): "En la época primitiva no se registran ejemplos de e apocopada tras d; y tras ζ, sólo en patronímicos, donde casi siempre es dudoso que la e fuese etimológica." (in other words, -e in patronyms was a result of hypercorrection). Thus, in the 11th century d maintains some resistance to apocope, but the situation changes abruptly in the 12th: es razonable pensar que la apocope de la /e/ después de /r, z, 1, n, d, dz/ arraigó profundamente en la mayor parte de la población durante los siglos XII y XIII. (Alvarez Rodríguez 1996: 40)

The rapid consolidation of normal apocope following all voiced apicals was apparently favored by the process of Toledo-phase koineization. In the late 11th century the phenomenon had become more frequent throughout northern Iberia (and, according to Lapesa [1951: 190], was probably not

164

The Toledo phase

absent from Mozarabic speech, despite the later reaction against it).106 The mixing of different dialect speakers in Toledo-phase repopulation certainly would have favored their accommodating out of the less frequently occurring forms (in this case, those that maintained the final vowel). Older child and adolescent speaker-learners could then reanalyze the more frequentlyoccurring apocopated forms as what they seemed to be: transparent realizations of phonological forms lacking final /-e/. This is a process that probably would have occurred slowly with or without koineization, but in the koineizing environment, it occurred much more rapidly and was thus incorporated into the new speech norms of the Toledo phase, particularly as they spread north along the Toledo-Burgos axis, but also west into Leon.107 In effect, koineization was not critical to the origin of the process itself, but it very likely did contribute to its sudden completion in the early 12th century. The development of extreme apocope is more difficult to understand, though koineization was critical to its constitution as a Toledo-phase norm. In some respects, such as the chronology, its development parallels that of normal apocope. Menéndez Pidal (1964: 190-191) gives a few examples of loss or hypercorrect use of -o, -a, and -e in proclitic forms from the 10th century: don Paulo apostolo vs. dueno Cristo ( < duerno or dueño, with the nasal possibly articulated as a conservative geminate or as a palatal; examples from the Glosas Emilianenses)·, Seemen Didaci (< Ximeno), Fredenande Albarez (< Ferdinando), Bal carcere (examples from eastern Leon); Lope Garsea (< Lop < Lobo < LUPU), Fortuni Sancio (< Fortunio), ual de ripa Hibre (< valle) (examples from Castile). These kinds of examples - showing loss or confusion - become more frequent from the second half of the 11th century, though alternation with full forms continues: Nun Sanciz, Fernán Gonsalbez, DuenfaJ Elo, font de caballos (from a document of 1009), allend presa / allende parte del aqua (1084), quern quadra (1091), (examples from Leon); Fortun Sanz, Lop Garcez, duanne Ezo (< DOMINU), al mont Sancì Mikael (1097), Ajerb (1118, from Aragon), Nun Albariz, Fortun Sangiz, ual de uascones, la sierra adelant (1057), duos uasos de argent (1082), sachestme (1107) (examples from Castile). For word-final -o, Menéndez Pidal (1964: 173) suggested that the apocopated forms spread analogically from the frequent proclitic use to stressed forms, as in de comité don Lop et sua uxore (1107). Apocope of enclitic pronouns also appears more frequently in 11th century texts: lexol, levos, quem quadra, quet dare (Lapesa 1951: 194). It is worth remembering, however, the apparent increases in frequency of apocope between the 10th and 11th

Linguistic changes

165

centuries may simply be due to the greater number of texts retained from the 11th century. The causes of extreme apocope have long been the topic of debate. Lapesa (1951: 193) argued that there was a direct causal relationship between syncope and apocope. Continuing the arguments he applied to the case of normal apocope, in which he suggested that normal apocope could occur because the consonants that it occurred after were already acceptable in syllable-final position, he suggests that the results of syncope opened the door to apocope by regularizing the appearance of many new consonants and consonant clusters in syllable-final position: La caída de las vocales intertónicas convirtió en implosivas muchas consonantes que antes precedían a la vocal interior elidida. Los primeros ejemplos se dan ya en el siglo X (limde, año 934; Fretnando, 940; semdero, 964; Mamlas, 972; trepde, 984); pero la síncopa no logra afirmarse hasta fines del siglo χι. Entonces quedaron como nuevos fonemas aptos para función implosiva la ζ (plazdo, aztor); la d (judgar, cadnadó)·, la ρ (trepde, riepto); la b (cobdo, cibdad)·, la m {semdero, comde); la t (setmana); y hasta grupos poco estables, como los de comptar, Antfossus, *antnado, *vendgar, sangne. (Lapesa 1951: 193)

There are problems with this argument, however. Harris-Northall (1990) argues cogently that syncope was a process that began in Latin in a limited number of phonetic contexts (e.g., between /s/ and /t/, as in POSITU > POSTU) and which gradually affected other phonetic contexts (and vowels) over the course of centuries. While it is likely that syncope was still operating in some of these contexts at the time that apocope began, it is impossible to argue that it led to a sudden change in the phonotactics of the language. This is evident in Lapesa's own examples. The starred forms are not attested and cannot serve as evidence of new phonotactic norms. These items are only attested in forms - like their modern reflexes alnado, vengar - which show adaptation of the novel consonant clusters produced by syncope to the existing phonotactic constraints of the language. The other cases were highly unstable - probably representing sporadic speaker innovations - and they were replaced by phonotactically more acceptable alternatives: contar for comptar, sangre for sangne, candado for cadnado, sendero for semdero. While it is true that some of these forms lasted for centuries (e.g., cobdo, cibdad), these may have been the last affected by syncope and they too were reformed over time (Harris-Northall 1990a: 153). As Harris-Northall (1990a: 152) argues, syncope did challenge the

166

The Toledo phase

phonotactic constraints of the language, but most of its products were modified in the end. On the other hand, syncope and apocope cannot be seen as unrelated: they are both processes that represent a normal effect of fast or relaxed speech. In the case of syncope, the output of speakers (or its effects, accumulated over centuries) was reanalyzed by learners without the original underlying vowel, but also with consonant sequences conforming to the phonotactic constraints - or norms - of the language. They could do this because the same speakers who produced the syncopated forms also articulated a much larger number of forms that continued to reflect the phonotactic constraints of the language. As a result, the phonotactic norms of the community were not greatly altered by syncope (except perhaps for the voiced apicals affected by normal apocope), but the results of syncope were greatly altered by the phonotactics. Like syncope (and normal apocope), extreme apocope in Castilian began as a phonetic process. Indeed, this is argued for by Lapesa: La fonética sintáctica contribuiría a este desbordamiento de la apocope, omitiendo como intertónicas las vocales finales de palabras agrupadas en torno de un solo acento o constitutivas de una sola unidad significante. La proclisis favoreció, sin duda, la pérdida de la e en el topònimo «Font de Caballos» (1009, León, Arch. Episc., 84), y en el «allend presa» que otro diploma leonés de 1084 emplea en alternancia con «allende parte». Los pronombres enclíticos apoyados en palabras con terminación vocálica empezaron a perder su e (feto/, 1055; leuós, 1100). (Lapesa 1951: 193-194)

But in Castilian, what distinguishes extreme apocope from syncope is the degree to which extreme apocope was able to alter the phonotactic norms of the community - at least for a time - in a way that syncope did not. So syncope cannot be seen as a cause of extreme apocope, but it is a similar kind of process that disturbed the phonotactic norms of the community by permitting phonetic innovations in the output of many speakers. Only to the extent that it weakened norms (with respect to syllable-final consonants) can it be seen as a cause of apocope.108 However, as HarrisNorthall's arguments make clear, it is by no means certain in what phonetic contexts syncope was still operating in the 11th century, and to show a direct causal link one would have to show that all the consonants affected by extreme apocope were also affected by syncope and preserved in final position. In at least one case this was clearly not so: extreme apocope is frequent after the cluster -nt, but the same cluster is exceedingly rare wordinternally (see quotation from Lapesa below).

Linguistic changes

167

It seems likely that at the earliest stage (corresponding to what I have called the Burgos phase, though examples of early extreme apocope were not geographically limited to the immediate region of Burgos), phonetic factors were essential in allowing and determining the appearance of apocopated forms. However, in many studies phonetic factors have been seen not only as contributing to the origins of extreme apocope, but also of continuing to govern its use in the later periods of the 12th and 13th centuries. In this koineizing environment, we might have expected either a) full maintenance of regular phonetic constraints, b) the selection and stabilization of apocopated forms, as in normal apocope, or c) the selection and stabilization of full forms of words that were prone to extreme apocope, (particularly given the existing phonotactic constraints that remained so strong in cases of syncope, and the lower frequency of extreme apocope). What we observe in the documentation of the 12th and 13th centuries a period of some 200 years - is the very frequent appearance of forms showing extreme apocope as well as those that do not, with some variation in the relative frequency of apocopated forms from text to text. However, this phenomenon has escaped convincing description and explanation. Most scholars have seen the continued use of apocope as governed principally by phonetic factors. Catalán argued, along the lines of Lapesa before him, that the development of extreme apocope - from the end of the 11th century to the 13th (Catalán 1971: 81) - was simply the extension and regularization of the closed syllable structure that he saw as dominating at the time, concluding: No veo razón alguna para seguir considerando a la 'apócope extrema' como extraña al genio del idioma. De otra parte, la geografía del fenómeno favorece el carácter autóctono que creemos que hay que conceder a la apócope, pues su intensidad disminuye gradualmente según pasamos del catalán al aragonés, del aragonés al castellano, del castellano al leonés y del leonés al gallego-portugués. (Catalán 1971: 79)

While both his assertions - that extreme apocope was "native" to Castilian and that there is a gradual decrease in intensity from east to west - are certainly true, Catalán oversimplifies in two ways. First, like Lapesa he assumes that syncope and apocope are one and the same. Second, he incorrectly equates the early phonetic conditions that allowed the rise of apocope with the conditions that determined its later use in the 12th and 13th centuries. Surprisingly perhaps (given the above discussion), Lapesa (1975: 17) responded to Catalán by pointing out that there were important

168

The Toledo phase

mismatches {"desajustes") between word-internal syllable-final consonants and word-final consonants left by extreme apocope: No todos los fonemas y grupos registrados como finales de palabra a causa de la apócope extrema se encuentran como finales de sílaba interior ni en conglomerados fonéticos-sintácticos de palabras. Así ocurre con la Iti... la /g/ o /k/. . . / n s / . . . M . . . En otros casos la correspondencia es excepcional o escasa: la /-f/ . . . y la / § / . . . en los pocos frecuentes frexno, trexnar. La cronología de los fonemas y grupos implosivos es distinta según estén en interior o en final de palabra. El grupo /-nt/ + consonante (antparar, *antnadó) se resolvió muy pronto (amparar, atinado, alnado) en posición interior, mientras que en final de palabra perduró largamente, con ejemplos hasta muy avanzado el siglo xiv. El paso de *astmar ( mi madré). Lyons (1993) suggests that numerous atonic preposed possessive adjectives were reanalyzed by speakers as determiners/specifiers, on the model of the atonic articles. Such reanalysis favored even more consistent use of short atonic possessives when preposed to nouns, which carried the primary stress accent of the resulting noun phrase.126 At the same time, the disyllabicity of the stressed forms tuo and suo was reinforced with the insertion

Linguistic changes 213 of an intervening glide: tuyo, suyo and related forms (the earliest attestations of suyo are from the 12th century). As has often been suggested, this was favored by analogy with the form cuyo < CUIUS 'whose', which paired pragmatically with possessives (e.g., question: ¿cúyo esΡ; answer: es suyo). Such analogical patterning is strongly favored in koineizing contexts. At the same time, the many variant forms such as tue, tua, sue, sua, were largely rejected, and so, to were generally dispreferred for most functions of the possessives. So and to were left in only one context/function: atonic and preposed, where they were supposed to contrast with su and tu. However, the increasingly unstressed articulation of these forms tended to weaken the phonemic opposition. The tendency for unstressed vowels to raise and form diphthongs with following vowels obscured input for learners. In many prevocalic contexts, so was articulated as [sw], as in so amigo, and this [sw] could be (mis)interpreted by adult and child learners as a realization of an underlying form /su/ (rather than /so/, the correct option in earlier stages). Moreover, the form so, which contains an apparent masculine morpheme o, might have been perceived as more marked than su, which had no overt gender marking. For generic use, the less marked form may have been preferred. Finally, and most importantly, morphological analogy with the stressed forms suyo, tuyo, certainly influenced speaker-learners, who would have sought transparent and regular relations between the different possessive forms and functions. Modeling the short atonic forms on the first and stressed syllables of the tonic suyo and tuyo highlighted those relations. As speaker-learners began to base their own production on incorrect abductions about the functions of su and tu, they increased the frequency and consistency of occurrence of su for all speakers attempting to learn the community norms (some of these causal factors are also discussed in Méndez García de Paredes 1988: 536; Penny 1991a: 127).127 Thus, it is likely that koineization played a key role in the definition of the Spanish possessive system and the timing of its development. In this reorganization we see the interaction of leveling, functional (grammatical) reallocation, and simplification. The development of the possessives during the Toledo phase also reveals clear stylistic reallocation of variants, along the lines of that of extreme apocope.

214

The Toledo phase

4. Conclusion Though the expansion of Castile into Toledo is not normally considered to have had an impact on the history of Castilian, there is significant evidence for making such a claim. The late 11th and 12th centuries show a renewed and enlarged process of demographic and dialect mixture within Toledo (or New Castile) in particular and the Peninsula in general. In New Castile, this mixing was greater than in the Burgos phase, with the novel elements of francos, Mozarabs, Mudejars, and colonists from a greater variety of regions (though Castilians dominated in most central regions). As in the Burgos phase, there occurred a breakdown of social networks and norm enforcement among newcomers, at the same time that linguistic variation peaked. Three significant linguistic effects have been identified: the establishment of extreme apocope as a native Castilian norm through reallocation as a stylistic variant, apparently in most parts of Castile; the reorganization of the possessive system, a process that was first completed in the region around Toledo and only later spread north to Burgos and Old Castile; and the extension of Burgos-phase leísmo to much of New Castile, but also its reanalysis and simplification in some areas of New Castile. The discussion of leísmo also involved exploration of its origins in the Burgos phase; this change shows that an apparently marked feature in the resultant koine can arise from speaker-learners' attempts to make sense of variable and opaque input. A significant feature of Toledo-phase rekoineization is its occurrence in many cities at once, and its contribution to the spread of Castilian features (those of the Burgos phase) to lateral varieties and the spread of new features from the south (Toledo and New Castile) to the north (Burgos and Old Castile). This spread, in turn, probably served to accentuate the effects of koineization. Nevertheless, the period is also marked by the development of internal dialect divisions, partly between North and South, since the North adopted some changes more slowly (as indicated by early 13thcentury differences in use of the new possessive system), but also new dialect divisions within New Castile due to different demographic mixes, timing, and organization of settlement (seen in the differences between pronouns systems H and E, and between these and the etymological system). Finally, it should be noted that the resultant koine of the Toledo phase is marked not only be geographic variation, but also by significant social/stylistic variation, a fact which may be due to the greater social class/caste distinctions which characterized this era and area.

Chapter 5 The Seville phase

The reconquest and repopulation of Betic Andalusia in the 13th century represents the last great phase of population movement in Castile begun and completed during the medieval period. As in preceding chapters, I first address the social and demographic history of this period in order to identify the sources of variation in the new koineizing environment. Unlike the earlier Toledo phase, language historians have commented on the possible linguistic consequences of the original repopulation of Betic Andalusia, though most of this attention has been recent and no consensus has yet been reached. Some see this phase as representing the mere transplantation of Castilian to the south while others view it as marking the very origin of Andalusian as a distinct dialect. I therefore examine this recent debate and how the model of koineization as here defined and applied may impact upon it. From there, I move to offer evidence for three linguistic changes that can reasonably be attributed to 13th-century koineization in Andalusia. Finally, I discuss seseo, a typical feature of modern Andalusian which has been attributed to 13th-century dialect mixing, show that this is unlikely, and consider the value of this negative evidence in helping understand the limits and usefulness of the model of koineization.

1. Social history By the late 12th century the Toledo phase of reconquest and repopulation had slowed. Though military activity continued all along the frontier, the Guadiana valley had clearly become a stable dividing line between Christian north and Moorish south. In fact, the Moors made important incursions into Christian territory later in this century, including, for example, their victory at the battle of Alarcos in 1195. Still, this was their last great victory, and the later defeat by the Christians at Las Navas de Tolosa in 1212 marked a decisive reversal in their fortune. As the Almohads fell into civil war, the Christians reinitiated their advance. This third phase of the reconquest, in which the Christians eventually conquered the whole of the valley

216

The Seville phase

of the Guadalquivir - Betic Andalusia - along with Murcia in the east, took place from the 1220s up to 1264.128 Fernando ΙΠ of Castile initiated this phase of the reconquest as he pushed south of the Sierra Morena into the Moorish kingdom of Jaén, taking and repopulating several strongpoints along the upper Guadalquivir valley: Andújar in 1224, Martos in 1225, and Baeza in 1226. In 1230, Fernando ΙΠ inherited the crown of Leon, which allowed him to combine the might of both kingdoms. Before this, however, Alfonso IX of Leon had already taken Valencia de Alcántara (1221), Cáceres (1227), as well as Mérida and Badajoz just before his death in 1230. Fernando added to these Extremaduran advances with the victories of Trujillo (1233) and Medellin (1234). In Jaén, the Archbishop of Toledo took Cazorla in 1231, while the King took the city and surrounding lands of Ubeda in 1233. However, these successes paled in comparison to the taking in 1236 of Córdoba, former capital of the Omayyad Caliphate. From Córdoba, Christian forces advanced both east and west. In 1243, Fernando sent his his son and heir, Alfonso, to negotiate the capitulation of Murcia, where the local king declared himself a vassal of Fernando and allowed the placement of Castilian garrisons in the main fortresses of the kingdom. In 1246 the city of Jaén itself, one of the best fortified of the Andalusian cities, was added to the list of Christian victories and left Fernando in control of the entire upper part of the Guadalquivir valley, which could then provide a secure base for the siege of Seville, the economic and cultural capital of the region. When its Moorish inhabitants finally surrendered in 1248, other cities belonging to the kingdom of Seville - equally exposed to attack along the plain of the river valley - ceased to resist, and the kingdom of Niebla (in modern Huelva) became a vassal state of the Castilian king. For a short time the reconquest slowed, as the victors went about consolidating and repopulating their newly gained territory. However, in 1262 Alfonso X took direct control of the kingdom of Niebla and in the following year initiated repopulation of Cádiz, closing off access to the sea from Jerez and other Mudejar cities just to the north. These actions contributed to the Mudejar Revolt of 1264, in which the remaining Moors in Betic Andalusia and Murcia rose up against their Christian overlords. Alfonso X quelled this uprising, then expelled the Moors from most of Andalusia and initiated repopulation of the newly vacated zones. The limits of the territory conquered up to this point would remain roughly the same until the 15th century, though the Christian-Moorish frontier shifted slightly as the

Social history 217 Christians took and repopulated Tarifa (1292), Gibraltar (1309), Algeciras (1344), and Antequera (1410). The conquest of these new territories in so short a time necessitated yet another massive repopulation effort. This was focused, as in the Toledo phase, on large urban centers. Indeed, the newly conquered and vacated cities were so large that they were difficult to repopulate. The concejos of the main cities (Jaén, Córdoba, Seville) administered a surrounding alfoz, in which they controlled the founding and repopulation of surrounding villages. In this sense and others, Andalusian repopulation simply continued the traditions developed in the Toledo phase, but the size and speed of the needed repopulation required the development of new procedures: Resultaba claro que ya en el siglo XIII no cabía aplicar las viejas formas de la presura castellano-leonesa o de la aprisio catalana, realizadas en buena parte de manera espontánea e informal y cuando lo era dirigida, se efectuaba ordinariamente con volumen de población reducido y con objetivos muy locales y predominantemente dentro de un ámbito rural y en tierra habitualmente abandonada, donde se efectuaban asentamientos de índole en buena parte familiar. Tampoco las primitivas cartas-pueblas aparecían ya suficientes por sí solas para las nuevas necesidades, y ni aun siquiera los propios fueros municipales, puesto que las primeras resultaban fórmulas lentas para la atracción de pobladores y con objetivos modestos, en tanto que los segundos se hallaban encaminados fundamentalmente a regir la organización y la vida ciudadanas una vez establecida la población cristiana que acudiera para establecerse en las tierras meridionales. (Moxó 1979: 396)

In order to facilitate the new round of repopulation, the Castilian/Leonese monarchy organized repartimientos, in which all properties abandoned by the Moors were inventoried and reassigned to new settlers. This was coordinated through commissions named by the king (known as the junta de partidores in Seville). This close control of the distribution of property heightened the official nature of the repopulation by allowing the monarchy to determine who would receive rewards of land. Fernando ΠΙ ordered his first repartimiento in 1224, when he took Andújar. Afterwards, nearly all the larger towns, such as Baeza, Jaén, Córdoba, Seville, Jerez and Cádiz, received fuero and repartimiento (Moxó 1979: 353). This system was so successful that it was also used in Murcia and Lorca, as well as outside of Castile in newly reconquered Valencia (1238). One might assume that the new system of repartimientos prevented social mobility in Andalusia (and thus would have retarded or limited the effects of koineization). In Seville, for example, the upper nobility or ricos

218

The Seville phase

omnes (including members of the royal family), royal officials, and various branches of the Church, received some of the largest and best properties, generally in the form of donadíos (González 1951: 257; González Jiménez and González Gómez 1980: xxx).129 Many of these lands were those most in need of repopulation, and they were given out as donadíos in the hope that their recipients would oversee their repopulation. However, many ricos omnes never actually resided in Andalusia, and some eventually sold their properties to the new local nobility (González 1951: 323), known as caballeros de linaje (in Seville). These were nobles of originally lesser rank who received larger properties in the repartimiento, but who were obliged to provide military service and to reside in the city. The families of these caballeros de linaje were destined to dominate the social hierarchy in Andalusia, and, indeed, the powerful Guzmán family of Seville had its roots in this group. The military concerns of this frontier society were especially significant in determining social status, which was often assigned not on the basis of previous or existing status but rather on ability to contribute military service. A very clear example can be found in the caballeros villanos (or ciudadanos), who, though non-noble, formed the most privileged group among the majority of the population, and whose privileges were directly linked to their military capacities. The importance of this and other newlycreated groups is patent in the Repartimiento de Jerez (a frontier city constantly exposed to attack), in which the most important social distinctions are cast in socio-military terms. The relative importance of these newlyprivileged groups can be seen in Table 29. Table 29. Social Structure of the Population of Jerez. Source: González Jiménez and González Gómez (1980: liv). Social Category Caballeros del feudo (linaje) Caballeros ciudadanos Other military groups: Peones

Number 42 212 109 1467

% of total 2.3 11.6 5.85 80.25

In effect, this widespread and rapid (re)assignment of social status to significant numbers of settlers represents a tremendous fluidity of social class at the beginning of repopulation (though paradoxically it would lead to a rigid stratification). Given its occurrence at this initial phase, it would have

Social history 219 contributed to the spread of koineizing linguistic changes throughout the social hierarchy in spite of subsequent stratification. Few of the libros de repartimiento have survived, but those that do, such as those of Seville and Jerez de la Frontera, are rich sources of information on the varied geographical origins of the first settlers, most probably the primary source of variation. Julio González' (1951) study of the Repartimiento de Sevilla, considered one of the best studies on the Andalusian repopulation, includes a detailed description of the origins of the settlers. As one might expect, the early Christian inhabitants of Seville were of quite varied origin. Still, the exact nature of the mixing was different from that of the earlier Toledo phase. The Moorish inhabitants of the city were exiled, being provided passage to Africa or allowed to move south to Jerez or east to Granada (González 1951: 308). Elsewhere in Andalusia, areas that were forced to capitulate through military actions were also emptied of their Moorish inhabitants. Only in those areas which were acquired through pacts - such as La Campiña of Córdoba, or regions south and west of Seville - were the Moors allowed, at first, to remain (González Jiménez 1982: 131). However, after the Mudejar Revolt of 1264, nearly all the Mudejars abandoned Andalusia for Granada or North Africa (only in Murcia did a sizable Mudejar population remain). Still, a few Moors probably those that had served the king in some way and were therefore referred to as moros del rey in the repartimientos - were allowed by the king to remain in the new Christian communities of Andalusia (González Jiménez and González Gómez 1980: lxxxi). In addition, a small morería appeared in Seville, probably due to the extended contact and trade with neighboring Granada (González 1951: 310). Another factor which differentiated Andalusian repopulation was the lack of surviving Mozarab and Jewish communities in Almohad Seville. Gonzalez (1951: 303-305) finds no textual evidence whatsoever to indicate the survival of a Mozarab community in Seville at the time of its defeat. This is not entirely surprising, since the Almohads had decreed their expulsion under pain of death. The Jews had also been expelled from the city by this time (1951: 311), though a new judería appeared after the reconquest. According to González, the absence of Mozarabs and Jews was also characteristic of other Andalusian regions reconquered in the 13th century (1951: 308). Once again, however, Murcia was different, in that it retained ancient juderías that remained isolated from the rest of society until the time of the expulsion in 1492 (Μοχό 1979: 373).

220

The Seville phase

Even without these groups, however, the size of Seville and the forced exit of its entire Moorish population must have made it the single largest repopulation project that the Christians had ever undertaken. As was the case of Toledo, much of the new population was made up of settlers from nearly all of the northern regions. González (1951) was able to determine the primary areas of origin of the settlers in Seville by analyzing indications of place of origin accompanying the surnames of beneficiaries listed in the repartimiento. He found that Leonese Extremadura and Valencia were weakly represented, since both had recently been reconquered and were themselves in need of new inhabitants. The core group of settlers were from the region centered on the cities of Burgos, Palencia and Valladolid, in Old Castile. Leonese and Galicians were also prevalent, though González warns: "Hay que tener en cuenta que en el repartimiento a veces se llaman gallegos a los del reino de León" (1951: 321). Navarrese - lacking new territories in which to expand - also came in significant numbers. Many more Catalans than Aragonese (limited to a small group of nobles) appear to have participated in the repopulation (1951: 317). There were significant numbers of settlers from New Castile (Toledo and the more recently repopulated Cuenca, Huete and Alcaraz), as well as previously reconquered areas of Andalusia itself. Some areas of the city were characterized by a heavy predominance of colonists from particular areas, as was the case of the barrio de la mar, where many natives of Santander (Cantabria) settled (Μοχό 1979: 361). The francos were also represented in the new mix, though far less significant in terms of absolute numbers and relative influence. On the other hand, the repopulation of Seville also included at least one group that had been absent from Toledo: Italians, especially Genoese, Lombards and Pisans, who came to set up commercial ventures (González 1951: 312). The mixing found in Seville also seems to have characterized smaller communities, though in each community the proportions from different regions were different. Jerez, for example, underwent repopulation when the Mudejars were forced to abandon the city after the uprising of 1264. Alfonso then ordered the city's repopulation and the preparation of a Libro de repartimiento. In their study of this text, González Jiménez and González Gómez (1980) prepared a quantified analysis of the origins of its earliest inhabitants, basing their counts on surnames and toponyms used as surnames (see Table 30).

Social history

221

Table 30. Origins of the Settlers of Jerez de la Frontera. Source: González Jiménez and González Gómez (1980: xlvii). Region Old Castile Leon New Castile Galicia Extremadura Catalonia Jaén Navarre Portugal Aragon Seville Basque Country Córdoba Asturias France Valencia Italy England

Number of settlers 307 155 137 82 61 44 42 34 30 27 25 22 20 13 7 4 3 1

% of total 30.27 15.28 13.57 8.08 6.01 4.33 4.14 3.35 2.95 2.66 2.46 2.16 1.97 1.28 0.69 0.39 0.29 0.09

As in the case of Seville, the Castillans form a sizable block, but they represent less than half the total population. There are some differences in the proportions for Seville and Jerez; in Jerez, New Castile and Extremadura appear more heavily represented, while the non-Iberian presence is practically nil, as we might expect in this much smaller border town. A few Moorish families were included in the repartimiento, though they were apparently new arrivals under royal protection. Jews of northern origin also established a judería in the town. Mixing of this sort must have occurred throughout Andalusia, though in each locale there may have been particularly large or small contingents of certain groups. This appears to have been the case of Cádiz: Parece que el mayor número de los repobladores gaditanos procedió de las villas cántabras -Laredo, Santander, San Vicente de la Barquera y Castro Urdíales- y Vizcaya, pero aunque prevaleciera este contingente norteño -al que hay que agregar un cierto número de gallegos y astures- encontramos también en buena medida castellanos e incluso algunas gentes de la recién repoblada Andalucía. No faltaron tampoco individuos de fuera de la Corona de Castilla, especialmente francos, catalanes y genoveses. (Moxó 1979: 364)

222

The Seville phase

There were also small towns that in their beginnings were settled by fairly homogeneous groups. For example, Alfonso X gave land grants to 100 Catalan ballesteros in Camas (Seville) and 150 omes de Catalunna in Coria del Río (González Jiménez 1988: 48). As in early stages of the reconquest, placenames reveal a certain regional dominance in some newly founded towns. Alvar (1979: 1870), for example, reports 5 towns in Huelva which include de León in their names, and a Zamoranos in Córdoba. Even though the main effort of reconquest and repopulation ceased following the 1260s, Andalusia was not left fully populated. Enormous areas along the frontier and even in the interior remained empty or underpopulated. Indeed, many of the earliest settlers abandoned their properties during the second half of the 13th century and returned to northern homes. As a result, efforts at repopulation did not cease entirely, continuing weakly into the 14th century. These efforts were divided between the internal rural zones, where the great nobles and the church especially sought more peasants to farm the land, and the frontier, where the military orders and the Crown sought more residents to defend the borders. However, most of those who participated in this repopulation appear to have been peones already settled in Andalusia, particularly Seville. While this later phase of repopulation did not really increase the overall population of the region, it did maintain a certain degree of demographic movement within Andalusia (González Jiménez 1975: 41-47) and perhaps extend the influence of the speech then developing in key urban centers, most notably Seville, and perhaps Córdoba as well.

2. Linguistic changes Given the kind of demographic and dialectal mixing that clearly occurred in 13th-century Andalusia, one must assume that koineization did in fact occur. The results of this koineization may logically have been of two kinds: changes which were restricted to the repopulated region and which became characteristic features of the newly-formed dialect, and changes which also affected the dialects of Castilian to the north through the process of language spread and dialect leveling. Numerous scholars have mentioned or discussed the first (though frequently only with a tangential interest, since only recently has medieval Andalusian come to be considered an object itself worthy of study), while the second of these possibilities has rarely been considered.130

Linguistic changes

223

2.1. Previous discussion of the origins of Andalusian The first serious discussion of the dating of the typical changes of Andalusian arose as a result of the famous polemic on the origin of American Spanish between Max Wagner (1920), on the one hand, and Amado Alonso (e.g., 1953) and Henríquez Ureña (e.g., 1932), on the other. Wagner held that the base of American Spanish was the Andalusian dialect; Alonso and Henríquez Ureña defended the separate development of American Spanish, and Alonso in particular tended to date the characteristic changes of Andalusian to periods following the founding of the American colonies. But Alonso also proposed that it was a process of linguistic leveling (nivelación) that led to the unique development of both Andalusian and the American Spanish dialects. For him, leveling seems to have implied mixing and homogeneization - concepts that are certainly fundamental to koineization - but did not necessarily include a notion of simplification. Interestingly, he indicates that such leveling had also occurred in Betic Andalusian: En realidad, los andaluces anticiparon tres siglos la misma operación lingüística niveladora que estamos explicando en los americanos: después de reconquistada Andalucía se pobló con castellanos, leoneses, asturianos y gallegos. El castellano dio el tono (o fue la base, según la tradicional imagen), por análogas razones que en América. (Alonso 1953: 54-55)

This, of course, is consonant with the belief in the separate but similar "polygenesis" of American and Andalusian, though the early dating implied for changes in Andalusian is surprising within the general context of his work. In another discussion on what he considered the early merger of /z/ and /d z / and the supposed later extension of this merger to /s/ and /ts/,131 he refers to leveling once again, this time even briefly referring to processes like leveling and simplification: Y esto ocurrió con mucha más profundidad y extensión en aquellas regiones donde el castellano había sido transplantado: en el siglo XIII, a la Andalucía occidental y Córdoba; en los siglos XIV y XV, al resto de Andalucía y a Canarias . . . a América y Filipinas. En estas regiones el castellano no sólo fue transplantado con la reconquista o la conquista, sino que los repobladores o colonizadores procedían de regiones dialectalmente heterogéneas y tuvieron que llegar a una nueva unidad idiomàtica por nivelación de las diferencias, con renuncias y aceptaciones, si bien siempre a base del 'castellano' oficial al que todos tendían. La formación de conglomerados humanos, de componentes dialectalmente heterogéneos, tiene entre otros el efecto de

224

The Seville phase

aflojar o de agravar la flojedad de ciertos resortes lingüísticos hasta entonces funcionantes, una especie de renuncia económica a ciertos 'primores' del sistema: en las nuevas condiciones sociales se ven como de lujo y prescindibles signos, funciones, valores y oposiciones que en la homogénea área originaria se tiende a guardar como naturales. En este sentido Andalucía tiene historia ejemplar, porque no sólo fue repoblada en los siglos XIII al XV con gentes de distintas regiones norteñas, sino que en los siglos XV y XVI otra vez se vio excepcionalmente agitada y demográficamente revuelta, primero por los ejércitos españoles que acudieron a la conquista de los reinos de Málaga, Almería y Granada; después con la formación de los ejércitos del Gran Capitán, y sobre todo con la extraordinaria afluencia de españoles de todas partes y de movimiento de andaluces cuando el descubrimiento, las guerras de conquista y las mareas de la colonización de las Indias hicieron de Sevilla la ciudad de más atracción y de más brillo de toda España y el corazón de aquel movimiento de poblaciones. (Alonso 1969: 108-109) Though issue may be taken with Alonso's view of continuous repopulation from the 13th to the 15th centuries, this passage does make the very important point that Andalusia was affected by large-scale demographic mixing not only in the 13th century but also in the late 15th and 16th centuries. Alonso, however, could not afford to continue the parallel between the origins of Andalusian and American Spanish, since one of his most precious theories was the separate and unique development of American Spanish. If Andalusian had completed many of the mergers that its shares with American Spanish before the founding of the American colonies, Alonso's anti-Andalusian conclusions would have proved unsustainable. He was thus forced to backtrack with regard to the supposed effects of 13thcentury repopulation: Los dialectalismos son mucho más difíciles de rastrear porque en el siglo XVI los dialectos andaluces no habían desarrollado aún, o habían desarrollado muy poco, los rasgos peculiares que hoy los separan del castellano . . . [in footnote] En fonética sólo el seseo de la -z y de la -z-, no aún el de la c, hasta entrado el siglo XVI; luego el trueque anárquico c-s, que arrastra en seguida a la z, y que produce el ceceo en Andalucía la Baja al cabo de larga gestación. (Alonso 1953: 57-58) Though this statement would appear to allow some limited effects of 13thcentury mixing (which remain unspecified), we know that Alonso associates even these few phonetic changes with the 15th century at the earliest (Granda 1994: 93).

Linguistic changes

225

Alonso, as he himself made clear, did not have access to medieval Andalusian manuscripts or to paléographie transcriptions of such manuscripts (Alonso 1969: 79). This had serious implications for his theories, since these were based almost exclusively on evidence obtained from the less precise transcriptions done by and for historians. As a result, numerous linguists have questioned Alonso's stance on quite a range of issues, not least among them the dating and sequencing of the changes that characterize Andalusian. Even in the 1950s, Catalán (1956) and Lapesa ([1957] 1985) began to question the late dating assigned by Alonso to the changes of Andalusian, pushing the earliest dates for such phenomena as seseo (or, in the terms of Catalán, çeçeo-zezeo) as early as the 15th century. Menéndez Pidal (1962c) also sought evidence for typical changes (such as seseo) in the 15th century, and argued for the existence of an estado latente in which changes of this sort were incubated for centuries before fully developing in the 16th century. In recent years, however, a polarization has developed in the dating assigned to many Andalusian phenomena,132 pushing the assigned dates to much earlier periods or to much later periods than those previously assigned. Perhaps no one has assigned later dates to the general rise of Andalusian features than the Granadan scholar José Mondéjar, who states: Puesto que la mayoría de los rasgos señalados desde el XIV hasta el XX han sido comunes al castellano (lengua medieval) y al español (lengua clásica y actual) de Andalucía y del resto de España, y puesto que los más característicos y diferenciadores son los aparecidos entre los siglos XVIII y XIX, creo que sólo es histórica y estructuralmente posible hablar de andaluz o de hablas andaluzas a partir del siglo XVIII. (Mondéjar 1991c: 231)

In line with this late dating, Mondéjar clearly views the 13th-century repopulation as constituting no more than a simple extension of northern Castilian: Todos los rasgos que presentan los documentos del castellano medieval, en la Andalucía del siglo XIII, son los mismos que se encuentran en otras regiones del dominio lingüístico peninsular. (Mondéjar 1991c: 227)

Although it seems likely that most of the features of the early Andalusian koine were found (at least as incipient phenomena) in other contributing varieties, it is not true that no changes occurred in Andalusia, or that no differences developed between the new Andalusian koine and northern (varieties of) Castilian (see below). Indeed, Mondéjar assumes that northern Castilian was a homogeneous and uniform variety, capable of simply

226

The Seville phase

being transplanted to the south. Moreover, he offers no evidence for his claim. As we will see in the discussion of seseo, Mondéjar is not totally consistent in his own claims, but the forcefulness of his assertions have led to the articulation of an antithetical view, according to which the very origin of Andalusian is to be found in its 13th-century repopulation. The leading advocate of this theory is Juan Antonio Frago Gracia, who, from the 1980s and into the 1990s devoted the bulk of his efforts to the study of the history of Andalusian (and American) Spanish, particularly as revealed through philological study of surviving medieval manuscripts. This effort led to a long series of articles (e.g., 1983, 1984, 1989, 1991b), culminating in his synthesis and magnum opus, Historia de las hablas andaluzas (Frago Gracia 1993). He explained the origin of his thinking on this matter in the following passage: Pues bien, son ya diez años bien cumplidos los que llevo a vueltas con el acervo documental de Andalucía, y desde los primeros contactos con estos corpus escritos y con los indicios lingüísticos que de ellos emanaban se fue alumbrando en mí la sospecha de que los distintos modos de hablar andaluces tenían una antigüedad mucho mayor de la que solía suponérseles, sospecha que, conforme se hacía más acendrado e intenso mi trato con los viejos textos del mediodía español, se ha ido convirtiendo en afirmación cada vez más rotunda, lo cual no es sino el resultado de reposadas reflexiones dentro de una concepción historicista, cuyo horizonte he procurado ensanchar y profundizar lo más que he podido. (Frago Gracia 1993: 7) Frago Gracia's insistence on the early origin of a broad spectrum of features associated with Andalusian goes far beyond the claims of any other scholar who has written on the topic (and he has continued to maintain them; cf. Frago Gracia 1999: 65). His fundamental thesis in this regard is more clearly defined in the following passage: Indudablemente, es el habla de la Andalucía primeramente reconquistada vista en su conjunto y con sus variedades internas, y no por referencia a la de Sevilla en particular o en exclusiva- el fundamento del dialecto que en el cuatrocientos se extendería por el reino de Granada en competencia con otras modalidades castellanas. Por consiguiente, a este inicial arranque dialectal es al que obligadamente se debe acudir en los comienzos de la indagación científica, a fin de determinar en la medida de lo posible el surgimiento de los distintos rasgos del peculiarismo andaluz, así como el proceso de afianzamiento social. Parece lógico pensar, pues, que si el andalucismo termina triunfando con la totalidad de sus modismos en buena parte del territo-

Linguistic changes

227

rio nazarí tomado en el siglo XV, y con varios de ellos en la completa extensión de ese dominio oriental -algo en cierto modo similar ocurriría en la siguiente centuria en la colonización de las Indias - , es porque tal diferenciación dialectal se había producido antes en la vieja Andalucía cristiana. (Frago Gracia 1993: 68)

While issue must be taken with some of the assumptions revealed here (see below and Chapter 6),133 Frago's emphasis on the history of the medieval language in Andalusia is laudable, for until recently this was an almost untouched domain. Lapesa (1981), for example, does not even mention the language of Andalusia in the 13th century, focusing exclusively on the language of the Alfonsine corpus. More importantly, Frago's fundamental hypothesis is thoroughly plausible within the model of koineization. Though he does not mention koineization as such, he attributes the early development of Andalusian features in the valley of the Guadalquivir to external causes: the process of demographic mixing and concomitant dialect mixing and leveling that he assumes must have occurred. Also, like Amado Alonso before him, he indicates the importance of at least two great periods of movement: Si en el Nuevo Mundo americano el español moderno experimentó una nivelación sobre la base lingüística de variantes peninsulares entrecruzadas, las que en sus hablares llevaban emigrados salidos de todos los rincones de España, este fenómeno, que inevitablemente acompaña a una lengua de colonización, se había producido antes en Andalucía, y repetidamente, además, en dos sincronías perfectamente marcadas por los momentos claves de la expansión castellana en el sur de la Península, a saber, el siglo XIII, cuando el primer impulso guerrero arrebata a los musulmanes el señorío de la Andalucía bética y el XV, sobre todo en su último tramo . . . Una y otra situación traería consigo la mezcla de gentes venidas de muy diversas partes, con modos de hablar en mayor o menor medida, según los casos, divergentes, y en ocasiones con lenguas distintas, lo cual traería consigo una serie de entrecruzamientos culturales y de compromisos lingüísticos. (Frago Gracia 1993: 55)

An even-handed consideration of the two periods of demographic movement in Andalusia is essential (as a minimum) to an understanding of linguistic change that occurred there. Despite paying lip-service to this important axiom, Frago Gracia again and again returns to the 13th-century repopulation as the source of nearly all the characteristic features of Andalusian:

228

The Seville phase

Puede decirse sin temor a exagerar que el andaluz ha sido el más acendrado crisol de las distintas variedades que el castellano medieval tuvo . . . Efectivamente, a los territorios andaluces reconquistados en el siglo XIII afluyen castellanohablantes de todas las procedencias, pero también asturianoleoneses y gallegos, así como, aunque en menor medida, vascongados y navarros, aragoneses y catalanes, junto al aporte demográfico de portugueses, franceses o genoveses, entre otras partidas de extranjeros. Y todos estos elementos no castellanos, peninsulares y ultrapirenaicos, de aquella naciente sociedad meridional en su seno hubieron de castellanizarse y de alguna manera tendrían que influir en la germinación dialectal muy pronto planteada en la novísima Castilla. (Frago Gracia 1993: 54)134

As we will see below, this tendency to place priority on the 13th century comes to amount to an article of faith in the work of Frago, and it sometimes leads him to twist his interpretation of the data to make it fit this claim. Nevertheless, it must be emphasized that Frago's fundamental thesis that the most significant features of Andalusian can be traced to 13thcentury demographic/dialect mixing in Betic Andalusia- remains fundamentally sound as a starting hypothesis, albeit an extremely strong one. One of the major problems with Frago's development of this thesis is his refusal to modify it in the face of the evidence, or lack thereof. There is no a priori reason to assume that this hypothesis is true, particularly since he himself reminds us that Andalusia saw two great periods of demographic movement: the 13th and the 15th/16th centuries. Other problems stem from the assumptions that he makes. First, he assumes that the changes of Andalusian (primarily phonetic in his studies) occurred throughout Betic Andalusia: "el andalucismo fonético se fraguó al mismo tiempo en su completa extensión geográfica" (Frago Gracia 1993: 228). This claim apparently arises in response to the traditional insistence on the primordial importance of Seville in the development of Andalusian, an importance which is clearly greater during the 16th century, when Seville was the very center of the great population movements. In reality, it is possible that particular areas - even particular towns or villages- saw unique developments in the early period that failed to spread until later (or which disappeared under the effects of dialect leveling). Such may have been the case with villages founded predominantly by Catalans or Leonese, for example, but this is a far cry from the more extreme claim made by Frago. More importantly, however, such an assertion ignores the fact that Jaén, also repopulated in the same period, is characterized to this day by features more typical of the

Linguistic changes

229

center-north of Spain (e.g., contrast of /s/ and /Θ/, loss of /h-/). I will return to this issue in the discussion of seseo. Another problem may be found in Frago's conceptualization of the process of language change in the context of dialect mixing. To begin with, Frago never enters into detailed discussion of the effects of dialect and demographic mixing, restricting himself to vague references to the terms mixing and leveling. He never mentions the term koineization and cites none of the relevant bibliography, and his conceptualization differs from koineization in at least one very significant and damaging way: his assumptions about the rate of language change. His view is made patent in a passage in which he criticizes the belief of some earlier scholars in the relative suddenness of the "great phonological readjustment" of the 16th century: En realidad, a veces parece como si con demasiada facilidad se olvidaran los grandes principios que todos confesamos aceptar, entre los cuales está la creencia en la enorme lentitud con que se verifica la difusión geográfica y social del cambio lingüístico, particularmente cuando se trata de innovaciones de tanta complejidad como las que han cuajado en la fonética andaluza. ¿Cabe tal diversificación lingüística en los estrechos márgenes que apenas dos centurias marcan? (Frago Gracia 1993: 6)135

This statement assumes that there is always an extremely slow, perhaps even constant rate of linguistic change (one is reminded of Swadesh's glottochronology), but why should we believe any such thing? Indeed, the model of koineization predicts a temporarily faster rate of change for a community suffering simultaneous demographic/dialect mixing and weakening of norm enforcement (as compared to a stable, homogeneous community). If koineization can in fact occur in the space of two or certainly three generations (Trudgill 1986: 95; Granda 1994: 100; Trudgill 1998; Kerswill and Williams 2000), two centuries becomes a very long period of time. The conceptualization of change as necessarily slow has significant effects on the interpretation of documentary evidence. In what way? First, as is revealed in the quote above, Frago repeatedly emphasizes the need to return to medieval documents in search of evidence of linguistic change, even lambasting other scholars for not doing so (e.g., Frago Gracia 1993: 37). There is no question but that Frago is right to make this claim. In this regard, his practice is consistent with his belief, and in his studies of the language history of Andalusia he has examined untold number of manuscripts. In fact, so new is the close philological study of medieval Andalusian that no collection of paleographically-transcribed 13th-century Anda-

230

The Seville phase

lusian documents exists, making Frago Gracia's contributions in this regard irreplaceable (even some ten years after the publication of his book; see discussion below). However, by prioritizing the value of textual evidence, Frago obliges himself to provide such evidence for his primary thesis: that the (nearly entire) set of features that characterize Andalusian had their origin in 13th-century Andalusia. As stated above, this is a reasonable starting hypothesis, and Frago is right to seek evidence in its support in the documentary record. However, the doctrine of the slowness of change means that even meager evidence can be interpreted as significant, since one might not expect changes stemming from the 13th century to appear in the documentation until much later. I provide specific examples of the problems occasioned by this belief in the discussion of seseo below. Scholars currently working on these issues must situate themselves somewhere on the continuum between Mondéjar on the one hand and Frago Gracia on the other. Among the most outspoken of these is Manuel Ariza, who continues to see the 16th century, the period traditionally assigned to the reajuste fonológico, as key to the formation of Andalusian: Parece hecho indiscutido que la consolidación de los fenómenos lingüísticos que conforman básicamente lo que se suele llamar andaluz se debe en gran medida al hecho histórico del descubrimiento de América y, por ende, a la importancia de todo tipo -cultural, política, económica- que adquirió Sevilla. (Ariza 1992: 15) This, of course, stands in direct opposition to Frago, who sees the 13thcentury repopulation and dialect mixing as the most significant. Ariza has argued against many of Frago's views. We will see many of his arguments below in the discussion of seseo, but there are two general criticisms he makes of Frago that must be addressed here. First, he objects to Frago's "colonizing" theory because Frago sees such phenomena as sibilant devoicing, seseo, and final sibilant aspiration as being originally northern phenomena that must have been carried south by 13th-century reconquistadores and then spread quickly in Andalusia (Ariza 1996: 44).136 Ariza believes that Galician and Leonese minorities (seen by Frago as the source of many changes, including early seseo) were too small to affect the Castillan majority, who would have provided the prestige norm. While it is true that the minority status of the Galicians and Leonese would have disfavored the selection of features of their speech (even simplified features), it is incorrect to assume that Castilians constituted a majority (the available evidence would indicate the contrary), or that their speech served as a prestige norm or target for other speaker-learners. As indicated in Chapter 2,

Linguistic changes 231 young learners do not seem to aim at established "targets" or pre-existing varieties in the linguistic pool (cf. Siegel 2001: 182), and absolute frequency has been shown to be a much stronger determiner of outcomes (cf. Trudgill, Gordon, Lewis, and Maclagan 2000). Moreover, neither Frago nor Ariza recognize that learner language can also be a source of innovations in the prekoine linguistic pool. Ironically, Ariza also rejects Frago's arguments because he believes that Frago views change as occurring too rapidly, asking, with regard to 13thcentury seseo/ceceo: Si Sevilla se reconquista en 1248 y la zona de Huelva en 1262, ¿es posible que en el período de menos de un siglo se haya conformado uno de los rasgos más peculiares del andaluz? ¿Es que, nada más reconquistarse estas zonas, los repobladores castellano-leoneses decidieron hablar de forma distinta a como lo hacían en sus lugares de origen? (Ariza 1994: 229)

While adult speakers may or may not change their speech habits in a koineizing environment (although the answer seems to "yes" with respect to certain features of Andalusian), the effects of learning in this environment of great linguistic variation can easily lead to quite significant changes in the space of 100 years (or much less). Although Ariza's comment points to some interesting inconsistencies in Frago's arguments, Ariza's a priori rejection of Frago's hypothesis must itself be rejected. The hypothesis - qua hypothesis - is sound. On the other hand, whether or not particular changes can be attributed to 13th-century koineization is another matter, and such issues can only be decided by returning to the documentation, as Frago and Ariza both argue, and interpreting it appropriately.

2.2. Documentary evidence Returning to the documentation may seem a simple matter, but in fact, there are no collections of completely paléographie transcriptions of early medieval Andalusian texts. Those collections that do exist have been transcribed primarily for use by historians, and thus attention to many details has often been ignored. The difficulties that the language historian encounters can be illustrated by considering one well-known and well-received collection: the Diplomatario andaluz de Alfonso X (González Jiménez 1991). Several factors combine to eliminate this collection as a useful source of data on medieval Andalusian linguistic changes. First, no attempt was made in the transcription to indicate occurrences of the notoriously

232

The Seville phase

ambiguous sigma(s), which have been resolved as s or z, according to the etymology. As a result, the collection is of little value for study of the development of seseo. In addition, the editors filled in word-internal abbreviations (without indications) and often modernized the spelling of proper names (González Jiménez 1991: ix). Nevertheless, the collection could be used to trace the Andalusian development of apocope, leísmo, possessives, or intervocalic distinctions between -ss- and -s- (expressly preserved for their possible linguistic interest, González Jiménez 1991: ix), were it not for another problem. The Diplomatario is restricted to documents that were prepared by the Royal Chancellory (González Jiménez 1991: vi), excluding private, municipal, and ecclesiastical documents, as well documents of noble households and the military orders. The scribes of royal documents were associated directly to the royal household through the Notaría mayor del rey or the notaría de cámara, or less directly through one of the traditional notarías mayores of the kingdoms of Leon and Castile, or through the newly created Notaría Mayor de Andalucía. These scribes often followed the king, and, though all the documents in this collection pertain to Andalusia, many were prepared as the king traveled and resided outside of Andalusia. In fact, if one eliminates all documents prepared outside of Andalusia and those listed without text, only 277 of the total 539 remain potentially useful. Of these, the potentially most useful are the 25 documents ascribed to the Notaría Mayor de Andalucía (Sanz Fuentes 1991: cc), but only 13 of the 25 were actually prepared in Andalusia. Clearly, these scribes did not reside regularly in Andalusia, but rather held an administrative responsibility over the preparation of Andalusian legal documentation. They certainly would have been less likely to conform to the newly developing norms of Andalusia, being more directly influenced by supra-regional or more traditional sociolinguistic and scribal norms. On the other hand, there are document collections which do satisfy, albeit partially, the needs of the language historian. This is the case of the document collection prepared by Ostos and Pardo (1989). As with the Diplomatario, the texts were not prepared with linguistic ends in mind; the editors resolve sigmas and fill out word-internal abbreviations in the documents in order to facilitate their reading. On the other hand, it has seemed reasonable to use them as sources of data in this study for two reasons. First, the transcriptions are sufficiently precise to provide reliable evidence on certain phenomena, including three which are discussed below (extreme apocope, possessives, and object pronouns). More importantly, however, these documents are restricted to texts prepared by scribes who

Linguistic changes

233

lived in 13th-century Seville, who worked for the municipal concejo, who prepared (and of course dated) the documents in Seville, and who consistently identified themselves in the texts as escríbanos de Sevilla. As a result, it is much safer to assume that these scribes were directly affected by the developing norms of 13th-century Seville and that the texts prepared by them therefore reflect those norms. Below, I discuss and provide textual evidence for three developments that can reasonably be attributed to 13th-century koineization in Andalusia: the end of extreme apocope, the simplification of the medieval first person singular possessives to mi(s), and the restitution of the etymological object pronoun system. The first two are changes that arose first in Andalusia, but later developed in the north as well, probably under the intense influence of weak social ties with the south, though the structure of the language itself already favored the change to a certain extent. The third is a feature of Andalusian that has remained restricted to Andalusian and other southerly dialects (and the lateral dialects of Aragon and Leon). Finally, in response to Frago and in order to demonstrate the need for constraints on claims of change caused by dialect mixing, I analyze his arguments and evidence for the 13th-century origin of Andalusian seseo.

2.3. The decline of extreme apocope As explained in the previous chapter, it appears that extreme apocope became established in Toledo-phase koineization through reallocation of a phonetic and regionally restricted variant to a stylistic function in Castilian. This is evident in the fact that, long after French influence declined, extreme apocope continued to appear in Castilian texts. Lapesa marks the zenith of this phenomenon - as represented in texts - as the first third of the 13th century (Lapesa 1982: 173), but we know that it continued quite strongly until the end of the 13th century. One is thus led to ask why extreme apocope began to decline and full forms began to appear more regularly. Lapesa points to what he considers a primary reason: desde 1230 aproximadamente su predominio se atenúa un poco, coincidiendo con el creciente afán de la iglesia castellano-leonesa por independizarse de Cluny y el Císter, y con la mayor acomodación de los oriundos ultrapirenaicos a la sociedad española. (Lapesa 1982: 173)

However, as was made clear in the preceding chapter, the decline in French influence began long before the decline of extreme apocope, and it is cer-

234

The Seville phase

tainly safe to claim that extreme apocope was decoupled from French influence by 1200. A more immediate cause, according to Lapesa, can be found in the influence exercised upon the language by King Alfonso X (Lapesa 1951: 219-220). Lapesa pointed out that the prologues to many of the Alfonsine texts were supposedly written by the king himself, and these prologues are notably lacking in extreme apocope, while the following texts generally show heavy apocope. Eventually, the king - reacting negatively to French political moves after 1276 - even decided to impose his non-apocopating preference as a court norm. However, even if Alfonso did effect some editorial prescription against use of apocopated forms, this cannot be viewed as a cause of the change. Rather, it must have been a mere reflection of a change already in progress in a wider community. What then might have provoked the decline of apocope? Certainly its limited status as a formal and therefore marked linguistic variant could have been a factor. But this fact alone cannot explain why extreme apocope declined at a particular time. I suggest that the decline is rooted in the process of koineization that was initiated in Andalusia from the very beginning of reconquest in 1224. In this military environment, the combination of popular Castilian norms with the generally conservative western Galician-Portuguese and Leonese speakers and scribes (Azevedo Maia 1986: 524; Staaff 1907: 213), would have combined to drastically lower the frequency with which extreme apocope was employed, and therefore the chances of its being learned by succeeding generations.137 In fact, Lapesa himself made reference to the possible influence of demographic mixing in Andalusia: La reconquista de Andalucía da lugar a los últimos grandes desplazamientos de población; la mezcla de gentes no se resuelve allí en dominio de la apócope, que nunca llega a igualar el número de formas plenas en las escrituras andaluzas. (Lapesa 1951: 214)

Still, Lapesa mentions this fact simply as a passing comment, and sees in it little causal significance. Other scholars have also made passing reference to this phenomenon. Frago, much more recently, includes only this comment in a footnote: "No cabe duda, pues, de que en el castellano de la época alfonsi implantado en Andalucía la apocope apenas tuvo repercusión" (Frago Gracia 1993: 86). Beyond these vague remarks, is there hard evidence that full forms first began to be consistently restored in Andalusia? Lapesa attempted to investigate the regional distribution of apocope in its final stages by using the texts in the Documentos lingüísticos de España of Menéndez Pidal. From those notarial documents not granted by the Royal

Linguistic changes

235

Chancellory during the reign of Alfonso X (1252-1284), he was able to obtain the following data: El computo de ejemplos recogidos en documentos no precedentes de la cancillería regia arroja los siguientes resultados: Campó, 4 apócopes extremas contra 2 casos de vocal conservada; Castilla del Norte, 39 apócopes y 17 extranjerismos sin vocal contra 14 ejemplos de vocal conservada o añadida; Álava, 5 apócopes extremas sin contradicción; Rioja Alta, 15 apócopes contra 29 casos de conservación o vocal subsidiaria; Rioja Baja, 7 contra 14; Burgos, 2 contra 7; Ávila, 2 contra 8; Toledo, 19 contra 30; Andalucía, 25 contra 38; Murcia, 19 (incluyendo muchos nombres catalanes o catalanismos como agost) contra 25. (Lapesa 1951: 216)

These data, if superficially interpreted, appear to indicate that during the reign of Alfonso, the phenomenon of apocope was both at its weakest and its strongest in the extreme north of Castile: weakest in Burgos and La Rioja, strongest in Castilla del Norte and Álava. However, it must be remembered that Lapesa's only data on Burgos are reduced to just four short texts that Menéndez Pidal happened to include in the DLE. Indeed, if we look generally at the north (e.g., La Rioja, Castilla del Norte, Álava and Campó), it seems that apocope was relatively strongly retained there, maintaining rates of occurrence above 50% even in local notarial documents during the reign of Alfonso. The data on Andalusia, based on only eleven texts, indicate that apocope was relatively weaker in this region than in the extreme north, though it appears as roughly similar, in proportional terms, to that of Toledo (39.6% in Toledo vs. 36% in Andalusia). However, this too may simply be an effect of the relatively small number of documents studied by Lapesa. In my own investigation of the 13th-century notarial documents transcribed by Ostos and Pardo (1989), the results appear quite different. Following Harris-Northall (1991: 31), I analyzed the forms of the preposition delant(e) and the adverb adelant(e). These forms were chosen because they are native Castilian forms frequently subject to apocope in texts of Old and New Castile, and they therefore represent a fairly reliable gauge of the general intensity of apocope. Since, as we saw before, HarrisNorthall (1991: 35) reports that the phonetic context was only a minor factor in determining the appearance of apocopated forms in Alfonsine texts, I have counted the number of occurrences of apocopated and full forms in all contexts. The Ostos and Pardo collection contains a total of 134 documents prepared by scribes of Seville between the years 1253 and 1300. The great majority of these documents contain at least one occur-

236

The Seville phase

rence of (a)delant(e) (the results appear in Table 31). Indeed, the very last occurrence of the apocopated form is in the year 1276 (Document 65).138 These data contrast markedly with the very strong preference for apocopated forms in the Alfonsine corpus. For example, in the General Estoria, the percentage of apocopated forms in the first part (GEI) reaches 64.4%, a tendency which grows stronger in the later fourth part (GE4), with apocopated forms rising to 95% (Harris-Northall 1991: 35). Table 31. Occurrences of (a)delant(e) in 13th-century texts from Seville. Form

(a)delant (a)delante

Number of occurrences 8 117

% of total 6.4 93.6%

Still, the Andalusian figures given by Lapesa do not clearly indicate an overwhelming majority of apocopated forms compared to contemporary notarial documents in Toledo. In part, this might represent the lingering influence of northern norms, since the first Andalusian scribes of the 13th century were all settlers who had moved from areas in the north. More important was the probably spread back north of the consistent use of full forms (or more accurately, the effect that contact with the novel and consistent Andalusian usage had on northern sociolinguistic norms). It must be remembered that massive demographic movements began with the extended military mobilization of the second quarter of the 13th century (1224—1248). In addition, repopulation of the province of Jaén dates to 1224 and that of Córdoba to 1236. From this time forward, the effects of these movements must have been felt in the north as well, and, as revealed in the roughly similar figures given by Lapesa for Toledo and Andalusia, were probably having an impact on Toledo, the primary northern urban center with which the communities of the newly reconquered Andalusian cities were in contact. From Toledo, the rejection of extreme apocope could have then jumped north to Burgos and other cities, leaving apocope only in isolated areas. Lapesa may have inadvertently identified the pattern of spread northward from Andalusia in the following passage: Tras el reinado de Alfonso X se precipita la decadencia de la apocope extrema, que a fines del XIII y principios del XIV sólo se mantiene con alguna intensidad en La Montaña, Álava, la Rioja Baja y Murcia, en las dos últimas regiones es indudable el influjo aragonés y catalán. Castilla del Norte pasa al dominio de las formas plenas. En Burgos, Osma, Valladolid y Segovia, así como en la parte castellana de la Extremadura actual y en Andalucía, los ejemplos son excepcionales o no se dan ya. (Lapesa 1951: 220)

Linguistic changes

237

The spread of the complete preference for full forms in the latter part of the 13th century was facilitated as the massive numbers of soldiers needed for the extended reconquest of Betic Andalusia returned to their homes in the north. This would have occurred after each compaign, but we also know that many of the earliest settlers sold the properties they had obtained in the repartimientos after a few years and returned to extended families and easier lives in the north. In so doing, they would have acted as language missionaries, spreading the new features that were being negotiated as norms in the demographically-mixed and socially-leveled south.139 King Alfonso X himself may have been one of these language missionaries. As indicated above, the royal prologues to many of the texts in the Alfonsine corpus are characterized by an almost total lack of apocope (Lapesa 1982: 180). Lapesa, of course, believed that the King's influence was decisive in bringing about the decline of apocope. Whether or not Alfonso actively discouraged the use of apocope, it is certain that he did not invent a new "norm" as a reaction to French political and military manoeuvres, as Lapesa suggested. More likely is that Alfonso - if he actually did have any influence on certain segments of these texts - was simply reflecting the rejection of apocope that gained currency first in Andalusia. It should be remembered that Alfonso, though raised in the outskirts of Burgos, spent much of his youth in the Andalusian military environment, which he apparently entered at the age of 10. Alfonso's padrino, in charge of training the prince in military matters, had been named the commander of the strongholds of Martos and Andújar from 1225, and in 1231 he took the young Alfonso with him on an attack into Cordobán territory. Alfonso later participated in the battle for Córdoba, led a delegation and troops to take control of Murcia (1243), and participated in the siege of Jaén. Troops entered Seville on his birthday and he was crowned in the same city in 1252, where he would also eventually die (González Jiménez 1993: 1522). Moreover, Alfonso was made king of Castile and Leon, in a city populated by many non-Castilians (at least half the population), and it may have been socially (and politically?) effective to reduce use of such a marked and localized Castilian feature when interacting with subjects from other areas. Under such circumstances, Alfonso and those around him may have learned to prefer the less socially-marked forms of language - more or less as Lapesa claimed (1982: 189) - that were more frequently employed in early Andalusia.

238

The Seville phase

2.4. Simplification of first-person singular possessives Andalusian rekoineization also seems to have been the catalyst for the final step in the reorganization of the possessive system: the simplification of the mìo/mi contrast (e.g., mio padre, mi madre).140 Of course, this change did not remain limited to Andalusia, but it only spread in the north after its establishment in Andalusia. As we have seen, the Toledo phase of koineization led to the selection and reallocation of long and short possessive forms in all Toledan styles, and simplification of the 2nd- and 3rd-person preposed possessives to tu(s) and su(s) in the Toledan popular norm. Toledan formal varieties maintained gender distinctions (but not with complete success), while conservative northern regions, which maintained more close-knit social networks, took longer to accept all the changes. Oddly, however, the first person possessives showed very strong resistance to leveling and simplification. Méndez, like other scholars, claims that the first-person possessives are regularly maintained even into the 14th century: En el masculino, hasta la primera mitad del siglo XIII, sólo existia la forma mio con una alternancia acentual mio ~ mió. A partir de entonces, la forma femenina mi comienza a desplazarse hacia el masculino; estas confusiones son algo más tardías que las ocurridas en segunda y tercera personas: en la época de Alfonso el Sabio son raras y don Juan Manuel sigue prefiriendo la forma mio aunque son numerosas las confusions. (Méndez García de Paredes 1988: 539)

Nevertheless, although the mio/mi contrast did survive until the 14th century, this was not so for all regions. In an analysis of the collection of documents from 13th-century Seville prepared by Ostos and Pardo (1989), I find occurrences of mi(s) + masculine noun from the very beginning, with extremely rapid growth in the frequency of usage (mi(s) is used exclusively before feminine nouns in these documents, with no examples of mio(s) + feminine noun or of the older forms mia(s), mie(s)). Table 32 shows the developing preference for the use of mi(s) over mio(s) with masculine nouns in percentages and absolute counts (in parentheses), broken down by decades. Many tokens of mio, however, occur in the final formulae of these documents, especially in the sequence mio signo (e.g., Et yo Roy Perez, escriuano publico de Seuilla, la fìz escriuir e so testigo e fìz y mio signo, from Doc. 83-1285). In fact, by the end of the century use of mio seems to survive only in formulaic passages, and removal of these examples makes

Linguistic changes

239

much more dramatic the sudden reduction to exclusive use of mi(s), as the statistics in Table 33 reveal. Table 32. Frequency of use of mi(s)/mio(s) + masculine noun in documents from Seville (1253-1300). Period (total tokens) 1253-1260(36) 1261-1270 (34) 1271-1280(20) 1281-1290(51) 1291-1300(119)

mi(s) 41.7% (15) 58.8% (20) 70% (14) 60.8% (31) 62.2% (74)

mio(s) 58.3% (21) 41.2% (14) 30.0% (6) 39.2% (20) 37.8% (45)

Table 33. Frequency of use of non-formulaic mi(s)/mio(s) + masculine noun in documents from Seville (1253-1300). Period (total token) 1253-1260(36) 1261-1270(32) 1271-1280(16) 1281-1290(39) 1291-1300(83)

mi(s) 41.7% (15) 62.5% (20) 87.5% (14) 79.5% (31) 89.2% (74)

mio(s) 58.3% (21) 37.5% (12) 12.5% (2) 20.5% (8) 10.8% (9)

These data make clear the rapid dominance of mi(s) in the newlyrepopulated community of Seville (along with nearly categorical use of su(s) for masculine and feminine nouns, even in formulaic phrases). The early evidence of shift to mi(s) shows the initial importance of adult usage and accommodation, but that overall progression underscores the importance of the first generation of children to the definitive selection and stablization of the change. These data contrast sharply with roughly contemporary documents from the north, even those from the early 1300s. For example, though sporadic confusions do appear, there is generally regular maintenance of gender distinctions in the DLE documents that show lstperson possessives: — Burgos (205-1315): mjos tíos, mjos tutores, mjos tutores, mjos cogedores, mi carta, mjo vassallo, mjo mayordomo, mj carta, mjo seello — Burgos (206-1367): mjo fijo (no other examples) — Toledo (289-1302): mj don Johan, mjo vassallo, mjos ganados, mjo sseello

240

The Seville phase — Toledo (290-1302): mj don Johan, mjo vassallo, mjs vacas, mj cabanna, mi carta, mjo seello — Toledo (291-1306): mj guarda, mj acomienda, mjo deffendjmiento, mis cosas, mjo sseello

Evidence for Andalusian-style reduction can be found in the following documents: — Toledo (292-1324): mj muger, mj muger, mj poder, mj muger, mj marido, mj signo — Burgos (207-1414): mjplaser, mj otorgamiento, mjo signo We can conclude that reduction of masculine mio(s) and feminine mi(s) to gender-neutral mi(s) was completed as a general change first in Andalusia, from where it then spread north. Two questions remain to be answered: a) why did this pair resist reduction until the Andalusian phase of koineization?; and b) what factors contributed to the final simplification? Méndez García de Paredes (1988: 533) suggests that it was the larger number of competing variants that led to quick reduction of 2nd and 3rd person adjectives, while the fairly regular masculine mio(s) helped to preserve distinct input for learners. The very high relative and absolute frequency of mio(s) probably did help to preserve it during Toledo-phase koineization, but other factors played important roles as well. The original monosyllabic articulation of mió made it less susceptible to phrasal syncope or elision (a significant factor in the earlier changes), which also served to protect the clarity of the input for language learners. The reallocation of forms in the Toledo phase led to a strong association of monosyllabic unstressed forms as preposed possessives, and disyllabic stressed forms with other functions of the possessive: — Preposed unstressed: mio(s), mi(s), tu(s), su(s) — Stressed: mio(s), mia(s), tuyo/a(s), suyo/a(s) The monosyllabic mio was partly consistent with this pattern, and therefore was less likely to be eliminated. Nevertheless, innovations did creep in. Although the masculine form was originally pronounced as a monosyllable, mió(s), rhyme and assonance in medieval poetry reveal that with time it came to be pronounced more and more frequently like the disyllabic feminine mia(s), to give mio(s)

Linguistic changes

241

(Menéndez Pidal 1962b: 256; Malkiel 1976: 472). This is unsurprising, since even speaker-learners in the more stable social context of 13th century Burgos and Toledo would have tended to seek and produce forms of greater paradigmatic regularity and transparency. Speaker-learners certainly compared mio with mía, though by itself this analogy would not have led to change of the articulation. Also influencing speaker-learners was the larger possessive paradigm, in which all other stressed forms were disyllabic, and all other unstressed forms were monosyllabic. Mío may have appeared first in stressed contexts (in unstressed contexts there was more likelihood of a diphthongal articulation), particularly when paired with the other possessives (e.g., lo tuyo y lo mío), since it was then that the evident analogy was most likely to be made. By the beginning of the Andalusian phase at least, mio(s) had come to be pronounced as disyllabic mio(s) much of the time (Lloyd 1987: 279). Nevertheless, this change in articulation of the mio/mio did not lead to loss of the mio/mi distinction (i.e., through reallocation of mio to stressed position only), since the stable social conditions of post-koineization Toledo and Burgos then favored overall resistance to innovations.141 However, during the Seville phase, the less constrained tendency of speaker-learners to seek and produce greater regularity and transparency led to the final simplification of the mio/mi distinction in preposed contexts, and to the reallocation of mio to stressed contexts only. Some of the factors that affected the simplification of so/su and to/tu may have played a part in this change. As with su(s) and tu(s), the form mi(s) was favored for generic use because it was less-marked morphologically (Méndez García de Paredes 1988: 536, 539). The definitive loss of stress on preposed possessives probably favored some coalescence of the final -o of mio with initial vowels in following words (e.g., mio honor > mi honor). Undoubtedly, however, the dominant factor affecting speaker-learners' analyses and production was the overwhelming analogical pressures created by the stable Toledan popular norm (in even the earliest Andalusian texts su(s) is used with stunning regularity for masculine and feminine). In Andalusia, speakers were able to develop the innovative overgeneralization of mi(s) to masculine use, and to reserve disyllabic mío for use in stressed contexts. With the weakening of social networks and of norm enforcement, the scales were tipped towards those factors favoring the reduction. Significantly, at this stage the opacity of surface phonetic realizations played at best a minor role. Rather, it was the establishment of clear morphological patterns that most influenced these changes. Once the change was estab-

242

The Seville phase

lished in Andalusia, the multiple weak social ties functioning along the Seville-Toledo-Burgos conduit facilitated its subsequent spread, and, along with the existing structural influence, ensured that the change was regularized. Finally, it is worth noting that the structural reorganization of the possessive system is normally perceived as a unitary phenomenon. Lleal (1990: 242), for instance, simply postulates that the gender distinctions for all persons (save 1st and 2nd person plural) began to disappear in the mid13th century. However, from a sociolinguistic perspective, the changes of the Seville phase are quite distinct from those of the Toledo phase. Moreover, in the constellation of phonetic/phonological, morphological, cognitive and social factors contributing to the changes, phonetic/phonological factors played a far more important role in the Toledo phase than in the Seville phase.

2.5. Leveling and the rej ection of leísmo Lack of leísmo has long been seen as an unusual feature of Andalusian Spanish (as well as either inexplicable or in no need of an explanation). Indeed, in Ariza's discussion of the origins of Andalusian, he lists it as one of its most distinctive morphosyntactic features: Dejando fuera las neutralizaciones morfológicas derivadas de la pérdida de /s/ final, la principal diferencia del andaluz se produce en el sistema pronominal. Efectivamente, en el andaluz no suele darse el leísmo, que tanta difusión tiene por el resto de la Península, lo que viene a ser un mantenimiento del sistema medieval castellano. (Ariza 1997: 64)

He also indicates that this is one of the most widespread features of Andalusian. However, aside from the unlikely claim that Andalusian continues the pronoun use of medieval Castilian (see Chapter 4), he makes no reference to either the causes or the chronology of its development, or lack thereof. Elsewhere, scholars who have studied the medieval origins of leísmo have paid little attention to its lack in Andalusian. This may be due to the perceived normality of the lack of leísmo, since Andalusian preserves the etymological system of object pronoun reference (see Chapter 4). At first glance, therefore, this appears to be a phenomenon in need of no explanation, for it is exactly what is expected from the perspective of the modern standard language. However, if one remembers that personal leísmo and object leísmo were well established phenomena by the early

Linguistic changes

243

13th century (see Chapter 4), then the lack of leísmo and the stable use of the etymological pronoun system becomes a phenomenon in need of explanation. Why did Castilian leistas not impose their usage as the norm in Andalusia? Or, if they did, why has it disappeared? It would appear that only two of the scholars who have studied leísmo have also considered its absence from Andalusian. Fernández-Ordóñez (1994: 123) first addressed this issue in her sociolinguistic study of leísmo and the referential systems, in which she located the origins of leísmo in the Basque substrate. In that scenario, the Basque influence was considered strongest in the north, weaker in the center (Toledo), and absent in the south, due to a tendency of syntactic features to resist borrowing. The problems with this argument are evident, and more recently, Fernández Ordóñez (2001: 462) has suggested that dialect mixing is likely to be the cause, as I argue here. Lapesa (1968) too addressed the issue of leísmo in Andalusia, though only tangentially (he refers to Andalusian in his discussion of the possible causal relation between apocope and leísmo).142 In his interpretation, leísmo is first extended to Andalusia, but is later extinguished: No es que falte algún ejemplo de le acusativo en la Andalucía medieval. Registro 'este pedazo de tierra calua, todo uos le uendemos . . . e nos vos somos fiadores. . .de todos los omnes del mundo que uos le quieran demendar . . . e os fe fagamos sano' 1263 Córdoba . . . 'el bien e las vebras que el omne faz en este mundo le seguirán e le serán prueua . . . ante la faz del Sennor' 1270, Jaén . . . Pero en 1303, Sevilla, encuentro 9 lo acusativo . . . sin ningún le. (Lapesa 1968: 527)

Lapesa was hard pressed to provide evidence, and the examples le seguirán e le serán prueua actually contradict his point, since seguir takes the dative regularly in many dialects, as does ser in this kind of construction. Still, Lapesa's basic argument is valid: the elimination of leísmo in Andalusian dates from its very origins. It is clear that what is lacking is data on the development of pronoun use in early Andalusian. In order to address this problem, I have carried out an analysis of the use of object le(s) and lo(s) in the Ostos and Pardo (1989) collection of 13th-century documents from Seville. The results are clear in their implications. In the 134 documents of this collection, there appear a total of 403 tokens of unapocopated lo or los. These appear with both neuter and masculine accusative reference (though few refer to masculine persons):

244

The Seville phase

— — — — —

como la carta ... lo dize e lo departe (11-1254) todo uos lo uendo (1-1253) dos maravedís alfonsis... que los paguedes cada anno (22-1263) el heredamiento .. . que me lo de (28-1264) aquel donadío ... que lo tengades & lo lauredes (69-1278).

On the other hand, there are only 26 cases of les, 10 cases of le, and 18 cases of apocopated / ' . I n the great bulk of instances, these forms are employed with clearly dative reference: — — — — —

una mezquita quel diemos (68-1277) que uos el cabildo quel fagades cada anno aniuersario (45-1266) assi les do poder (78-1284) la pregunta que les fiz (90-1289) etpreguntel si le plazia... dixo quel plazie (90-1289)

Nearly all the possible cases of leísmo involve verbs or constructions that alternate in taking the dative or the accusative (all of which are discussed in Lapesa 1968, Cuervo 1895, and Fernández Ordóñez 1999): — Elvira . . . que ninguno de nuestros herederos noi puedan demandar (131255) — Belenguer... α pleyto quel quitemos desta fiadora sin danno (24-1263) — que cunpliere mandado del arçobispo segunt el le enuiaua mandar (971285) — vn capellan ... et... vn sacristan quel ayude (127-1300) Indeed, there is only one token of le(s) in all these documents that may be a clear example of leísmo', los heredamientos . .. que les metades (84-1285). However, even this example is questionable, since the exact reference may not have been totally clear to the scribe: some 73 words appear between the antecedent and the anaphoric pronoun. Moreover, a similar document from the same year gives: los heredamientos . . . que los metades (87-1285). This leaves the few examples from Córdoba as Lapesa's only clear evidence of leísmo in 13th-century Andalusia. Clearly, however, that evidence merely reflects the usage of a scribe who must have moved to Córdoba as an adult and who maintained his own leísta usage for a time. In fact, the tokens from the Cordobán document point to the contemporary vitality of object leísmo not in Andalusia but rather in more northerly regions.

Linguistic changes

245

With this evidence it becomes relatively easy to reconstruct the factors which contributed to the establishment of the etymological pronoun system in Andalusia. First, even though large numbers of the new settlers must have been leistas, they do not appear to have represented a majority (settlers from Old and New Castile seem to have represented not quite half the total population, and not all New Castillans were leistas). In most koineizing situations, the results of dialect mixing tend to reflect simple majorities (i.e., if there were more users of the etymological system in the mix, then their usage would be selected, no matter how varied their origin), as opposed to the potential (social and cultural) impact of one particularly large and influential group (cf. Trudgill, Gordon, Lewis, and Maclagan 2000). Moreover, teista usage represented an added complexity to all those speaker-learners who used the etymological system. Without an absolute majority of leista speakers, it was even less likely to survive into the new koine. Still, one might ask why an interdialectal solution like system E did not appear in Seville, as had occurred in or near Toledo. Such an outcome might even have been expected given the level of mixing and the frequent presence of the Castilian court in the city. However, the social situation had changed somewhat. Previously, Castilians had represented a clearly defined and relatively homogeneous high-prestige group. However, by the time of the conquest of Seville, the combined kingdom of Castile and Leon had come to encompass large non-leista areas to the west (Galicia and Leon), to the east (Cuenca, La Alcarria) and areas south and east of Toledo controlled by the military orders (La Mancha). As such, the clearlydesirable target that leista usage had represented in Toledo had lost much of its social value. Indeed, the evidence from the Seville documents seems to indicate that leista usage had become so marked that it was consciously avoided even by those who had grown up using it.

2.6. Seseo\ A 13th-century merger? As we saw above, Mondéjar (1991c: 227) has claimed quite insistently that 13th-century Andalusian is a unaltered transplant of the Castilian of Toledo and Burgos (under the assumption that these were uniform). However, the data on the above phenomena provide fairly certain evidence to the contrary. Moreover, the first two (the elimination of extreme apocope and the reduction of lst-person singular possessives) eventually affected dialects to the north, and the last (the elimination of leísmo) survived as

246

The Seville phase

one of the key characteristics of the Castilian of Andalusia (as well as southern Spain). On the other hand, we are left with the dilemma of deciding the dates and causes of many other features, particularly those that now characterize Andalusian varieties. There can be little doubt but that the environment of social change and demographic/dialect mixing in 13th-century Andalusia was propitious to linguistic change, and certainly other changes might reasonably be attributed to the mixing that occurred during this period.143 However, we must consider the strong claim articulated by Frago Gracia: was Andalusian indeed "born" - with the great bulk of its features - in the 13th century? Principal among the features that Frago believes can be traced to the 13th century is seseo, perhaps the single most characteristic feature of Andalusian (and American Spanish) vis-à-vis northern varieties of Castilian (1993: 308).144 This is both a most daring and most reasonable claim: daring, because it directly contradicts the much later dating for the phenomenon given by many eminent scholars, and reasonable, because the type of merger represented by seseo is exactly the kind of simplification or reduction of phoneme inventory that one expects to find in a case of koineization. The only reasonable response to this claim is to seek to evaluate the textual evidence in its favor. As we have seen, however, this is not an easy thing to do, particularly in the case of seseo, where there is little option but to return to the manuscripts themselves. To his great credit, Frago Gracia, in accordance with his own oft-repeated claim (e.g., Frago Gracia 1993: 37), has done just these things. Therefore, his studies should offer us the best available textual evidence for the early appearance of typically Andalusian features in general, and seseo in particular. It is largely this evidence that I examine here. Frago Gracia addresses the issue of seseo primarily in his large (1993) monograph and an earlier article (1989), in which he argues not only that seseo must have already been well-developed in medieval Andalusia for it to have been transplanted to America (an idea itself worth questioning; see below and Chapter 6), but that its spread must be included among the effects of the 13th-century demographic movement and mixing within Andalusia. In these two studies, he presents his own evidence as well as that of other researchers. Of particular importance here is the reference he makes to studies of Menéndez Pidal, Mondéjar, and Lapesa, all of whom have also offered apparently valid textual evidence of seseo from medieval Andalusian texts. Paradoxically, Frago rejects the evidence offered by both Pidal (1962c: 112-113) and Mondéjar (1991b, 1991d: 114-115), since

Linguistic changes

247

much of this is based on forms appearing with the famously ambivalent sigma(s) (a symbol or symbols often used to represent more than one sibilant phoneme): Resulta peligroso, sin embargo, fiar a la igualación de s y ζ en un solo dibujo, e incluso al uso indiscriminado de la sigma con la zeta en un mismo corpus, la atestiguación del fenòmeno seseo-ceceoso, cosa que Menéndez Pidal hace con grafías andaluzas de finales del siglo XV como aseyte, aseytuno, disen, faser, fiso, hasen, honse, entre otras. Del mismo modo continúa argumentando J. Mondéjar cuando propone como «los más antiguos testimonios del seseo en Andalucía» las formas -de 1302, dice- alguasil, arros, cafis, nues, ves, veses, vesino, contrastadas con registros de cafizes y vezino, con el apoyo de un quinse de 1423 que parece serle sumamente significativo. Curioso planteamiento metodológico: por la misma regla de tres dicho seseo habría estado extendido a lo largo y a lo ancho de todo el dominio castellano y hasta por otros ámbitos peninsulares, y, ni que decir tiene, en fuentes andaluzas ejemplos como los precitados podrían amontonarse ad nauseam. (Frago Gracia 1993: 332-333) On the other hand, Frago apparently accepts much of the evidence from 15th-century Andalusia offered by Lapesa; though he cites only some of this evidence directly (Frago Gracia 1993: 282), the generally positive tone in which he refers to Lapesa's studies (e.g., Frago Gracia 1993: 332) leads one to conclude that Frago assumes that the reader is already familiar with the examples of 15th-century Andalusian seseo that were presented by Lapesa ([1957] 1985: 252-3). These include: — 1419 (Sanlúcar o Niebla, Huelva) : dief, diefmo — 1460s (Cancionero de Baena)·. escaçeza for escaseza, çatan for Satan(as), çedal for sedal, çenado for senado, azaz for assaz, bruçelas for Bruselas, Amadiz for Amadis. (Frago Gracia 1993: 282, from Lapesa 1981: 283) — 1475 (Pero Guillén de Segovia, Seville): çemençera, çenzilla, çenzillo — 1487 (Pedro de Toledo, Seville): Roblez, inglez, fijoz, Andrez, Blaz, viscayno, Sanches, Gomes, durasnos, Beatris — 1492 (Seville): sirios for cirios, fiçieçe for fiziesse — 1495 (Seville): çufriendo for sufriendo Assuming, as Lapesa and Frago do, that the forms with s and ζ are correctly transcribed (and not resolutions of sigmas), they may represent significant evidence of some kind of change. Lapesa viewed these merely as evidence of early confusion resulting from the deaffrication of /t7 and /d2/,

248

The Seville phase

seseo as it existed in its supposed estado latente. His goal was merely to

provide roots for a phenomenon that can only be well-documented from the mid-16th century, and which Lapesa clearly linked to the second great phase of Andalusian population movement: ¿Qué había ocurrido en Andalucía para que el cambio [incubado seguramente durante siglos] lograse tan amplio desarrollo en el que va de 1470 a 1570? Hubo una doble sacudida en la vida andaluza: primero con los desplazamientos acarreados por la guerra y conquista de Granada: después con el descubrimiento de América, la intensa emigración y, a la vez, el crecimiento de Sevilla en pobladores y riqueza. (Lapesa 1985: 255)

On the other hand, Frago's arguments are much more extreme than Lapesa's. Since it would clearly be difficult to use this evidence alone to argue for the 13th-century origin of seseo, Frago provides further evidence from older texts, often original manuscripts. Frago Gracia (1989) and (1993) include what appears to be the sum total of textual evidence discovered in his analysis of the medieval Andalusian corpus. However, Frago presents this evidence in such a way that it is difficult to see the evidence in its entirety and, thus, to evaluate it. This is so for several reasons. First, he mixes evidence for seseo from different time periods, jumping, for example, from the 14th century to the 17th without hesitation (Frago Gracia 1989: 282). Second, since he is convinced that devoicing of sibilants was brought into Andalusia by new settlers in the 13th century and that the old system of four sibilants (-ss- Iii, -s- Iii, ç Iti, ζ /dz/) began to collapse to one (Isl) from that time, he mixes evidence for seseo with evidence for devoicing. Since there is in fact substantial orthographic evidence for confusion of -ss- with -s-, as well as confusion of ç with z, it becomes difficult to accurately gauge the quantity of the evidence for seseo.H$ Third, he reports this evidence in different places, without providing a clear summary. Even in his (1993) monograph, some orthographic evidence of seseo is provided in chapters devoted to other topics, and some is simply not included, such as when Frago refers the reader to a previous article (i.e., Frago Gracia 1989) for further examples. It therefore seems prudent to list all the tokens found by Frago of confusion of originally apicoalveolar -ss- Isl and -s- Iii with originally dental affricate ç Iti and ζ ldzl. Below I include the sum total of tokens from before the year 1500,146 presented in roughly chronological order, that I have collected from Frago Gracia (1993) and (1989). Frago is quoted where this is necessary for inclusion of pertinent details:

Linguistic changes

249

1272 (Seville): 3 ençenz, 1 ençienz for censo, encenso (Frago Gracia 1993: 74) 1275 (Seville): 2 ençenz (Frago Gracia 1993: 74) 1293 (Seville): ensensarios for incensarios, susepçores for sucessores (Frago Gracia 1993: 46). Frago points out that, elsewhere in the work of this scribe, there is: "una constante atestiguación del trueque de c por ss en el ant. suçessores: sosçepçores, soçepçores, sopçeçores, sopçepçores, susçepçores, suçepçores" (Frago Gracia 1993: 226). 1338 (Repartimiento de Jerez): Almança for Almansa (Frago Gracia 1993: 228) 1347 (Huelva): fessieron for fezieron (Frago Gracia 1989: 282) 1352 (Santa Clara de Moguer, Seville): hesebçion for excepción (Frago Gracia 1989: 282) 1353 (Santa Clara de Moguer, Seville): exssebçiones (Frago Gracia 1989: 282) 1366 (Seville): exepçion, esepçion (Frago Gracia 1993: 233) 1366 (Palma del Río, Córdoba): exepçion, efepçion (Frago Gracia 1993: 340) 1375 (Palma del Rio, Cordoba): conno/co for conozco (Frago Gracia 1993: 340) 1379 (Seville): beçola for besóla (Frago Gracia 1993: 230) 1369-1421 (Palma del Río, Cordoba): "En dos de estos textos es significativa la existencia del cultismo excepción con las grafías seseosas esepcion (Doc. de 1369) y essebcion-esebcion (Doc. de 1374). Algunas décadas después el análisis de dicha fiiente documental continuará arrojando los mismos resultados, no siendo raros los escritos que por completo desconocen la ss, por ejemplo imo de 1403, que también incluye tres casos de esepcion, otros dos de 1412 y 1413, respectivamente, con sendas muestras de exsebçion, y uno más de 1421 en el que se repite la forma excepçion." (Frago Gracia 1993: 233) 1370-1400 (Alcalá de Guadaira, Seville): asas for assaz, pescueso for pescuezo (Frago Gracia 1993: 278) 1402-1426 (Morón de la Frontera): "en el corpus ACM I (años 14021426), aparecen los lapsus calami y variantes asas-aças, caliçes-caliaejo, caliçes-caliaes, ençençario "incensario", ençienfo al lado de dos ençienço, enpefca, e/quierdo, ofiçiales-ofiaiales, Velajques, vifcayno, ÇaharaSahara ["Zahara"].' (Frago Gracia 1993: 343) 1407 (Seville): "En carta de donación dictada en Sevilla el año 1407 se pone eglecia junto a indusimiento, y dos ejemplos de eglecia se encuentran en

250

The Seville phase

— — — — — — — — — — — —

— — — — — — — —

otro documento nobiliario dado en el mismo año y lugar." (Frago Gracia 1993: 342) 1411 (Carmona, Seville): resebryades for recibiríais (Frago Gracia 1993: 342) 1413, 1414, 1420 (Peñaflor, Seville): esebçion-esepçion (Frago Gracia 1993: 233) 1417 (Puerto de Santa María, Huelva): fesiese (Frago Gracia 1993: 234) 1419 (Homachuelos, Córdoba): esepçioti (Frago Gracia 1993: 233) 1425 (Santa Clara de Moguer, Seville): ençenço 'censo' (Frago Gracia 1989: 282) 1430 (Santa Clara de Moguer, Seville): exebçion, ecleçiasticas (Frago Gracia 1989: 282) 1449 (Santa Clara de Moguer, Seville): ençienço (Frago Gracia 1989: 282) 1460 (Segovia, Andalusian copy): contumasia, yndusido, perteneserpertenesçen, 5 exepçion (Frago Gracia 1993: 342) 1467 (Huelva): Medina Çidonia for Medina Sidonia (Frago Gracia 1989: 282) 1470 (Huelva): Çiguença for Sigüenza (Frago Gracia 1989: 282) 1478 (Santa Clara de Moguer, Seville): juysio (Frago Gracia 1989: 282) 1480 (Seville): quatroçientos-quatrosientos, Vyçente-Visente, aluaseas for albaceas, desendientes for descendientes, mayorasgo for mayoradgo, mayorazgo (Frago Gracia 1993: 235) 1484 (Huelva): Medina Çidonia (Frago Gracia 1989: 282) 1484 (Seville): Aluaref for Alvarez (Frago Gracia 1993: 337) 1486 (Seville): syçion for cesión (Frago Gracia 1993: 235) 1489 (Seville): desernimiento for discernimiento, preçion for prisión (Frago Gracia 1993: 235) 1489-1500 (Málaga): Ançurez for Ansurez, Lezmez for Lesmes, Yçrael for Israel (Frago Gracia 1993: 350) 1490 (Córdoba): aniz for anis (Frago Gracia 1993: 342) 1492 (Granada): Yzmael for Ismael (Frago Gracia 1993: 234) 1494 (Pilas, Seville): "Haznalcaçar-Haanalcaçar-Hasnalcaçar, el ceceo gráfico de Çeçilla "Secilia" -tal vez lexicalizado, en vista de su frecuencia en muchos textos de la época-, y las formas Vifcaya y efsebçiones "excepciones", siendo digno de notarse el hecho de que la formación escolar del pendolista, sin duda apreciable, se manifieste en la corrección que introduce sobre un posision "posesión", en primera instancia con la segunda sibilante escrita ç (posiçion) y luego retocada con ese sigma" (Frago Gracia 1993: 344).

Linguistic changes

251

— 1495 (Seville): 5 sedula for cédula, asebtada for aceptada (Frago Gracia 1993: 235) — 1495 (Francisco de Cisneros, Sevilla): sojusgar, Calis for Cáliz, topasios for topaçios, Topasia (Frago Gracia 1993: 235) Frago employs this evidence to argue that seseo was brought to Andalusia in the 13th century, where it then spread rapidly. Of particular interest is the following assertion: Antiquísimos son los inicios de las transformaciones fonéticas conducentes al seseo y al ceceo; las hemos visto nítidamente apuntadas en diplomas notariales sevillanos del último tercio del siglo XIII, algunos de ellos redactados por escribanos públicos de procedencia noroccidental, y no es descartable que un posible primitivo seseo y ceceo gallego hubiera colaborado en el desencadenamiento del cambio andaluz. Pero también . . . gentes de lengua catalana, que ya, desde tiempo atrás practicaban el seseo . . . y con ellos . . . vasconavarros . . . portugueses, probables apoyos asimismo de la tendencia confimdidora. (Frago Gracia 1993: 362) This sounds convincing, until one filters out the true examples of 13thcentury seseo: six cases of enc(i)enz for encens(o) and one of susepçores. Given the massive numbers of documents that Frago Gracia claims to have examined, are we to assume that a couple of examples of confusion show the rapid spread of a merger? According to Frago, yes: el que en un diploma hispalense de 1293 figure la grafía susepçores 'sucesores' es indicio irrefutable de la iniciación, insospechadamente temprana, del proceso seseo-ceceoso. (Frago Gracia 1993: 34-35) I will return to this particular example below. Here it must be emphasized that in general Frago has little to offer for 13th-century seseo. Indeed, this couple of examples of supposed 13th-century seseo stands out in opposition to the much better evidence he is able to provide for confusion between -ss- and -s-. Overall, the relatively low number of examples of medieval orthographic representations of seseo that Frago (and others) are able to provide is perhaps the most telling flaw in the argument (particularly when compared to the greater evidence for devoicing).147 Frago himself is aware of the problem, and offers several explanations. For example, he argues that it was easier for scribes to remember distinct spellings with ç, ζ and -s-, even when the phonemes they represented had merged (Frago Gracia 1993: 281). This, of course, leads one to ask why scribes suddenly stopped remembering in the 16th century (when evidence of seseo becomes

252

The Seville phase

abundant). Perhaps in reply to this question, Frago offers another explanation for lack of evidence of seseo in the medieval documentation: Lo que sucede es que entre los últimos decenios del siglo XV y los primeros del XVI se opera en la documentación andaluza una transición bastante brusca en cuanto a la manifestación de las grafías incorrectas, que pasan en poco tiempo de ser relativamente dispersas a abundar extraordinariamente, en lo cual mucho tuvo que ver, quizá, el relajamiento de las normas ortográficas más introducidas en la scripta medieval, progresivo distanciamiento a su vez favorecido por la misma evolución lingüística, que acaba siendo asumida como una nota de afirmación regional. (Frago Gracia 1993: 355)

It may be that there was a relaxation of spelling norms at this time (just as other social norms probably were relaxed during this period of movement), but Frago cannot make clear why the norms seemed to prevent indications of seseo while allowing significant examples of confusion of voiced and voiceless sibilants to appear.148 More significantly, Frago is forced to argue against himself in making the above affirmation: Por otro lado, es muy arriesgado hablar de normas escriturarias de semejante rigidez en relación a la ortografía antigua, pues el trato directo con los textos manuscritos no da esa visión de los hechos: en no pocos casos de lo que se trata es de tendencias, cumplidas con mayor o menor exactitud, pero susceptibles de ser rotas incluso por los autores más cultos. (Frago Gracia 1993: 155-156)

Clearly then, the lack of medieval orthographic evidence is a problem for Frago, a problem which he is forced into the difficult position of having to explain away. Another problem with Frago's insistence on the early origin of Andalusian seseo is that it forces him into another dilemma already commented upon. Since he insists repeatedly that the characteristics of Andalusian, including seseo, appeared simultaneously throughout Betic Andalusia in the 13th century (i.e., they did not spread from Seville), he has to explain why certain areas of Betic Andalusia, such as Jaén and the north of Huelva, have never accepted seseo. His response is to suggest that this may be due to the (small minority of) Galicians who participated in the reconquest of Córdoba and Seville after 1230 (Frago Gracia 1993: 361), but who were not well represented in eastern Jaén.149 Unfortunately, the problems do not end here. As Ariza has pointed out in a series of articles (1992, 1994, 1996, 1997), Frago appears to misinterpret - or, at least, over-interpret - many of his data. One the greatest prob-

Linguistic changes

253

lems is that evidence similar to that found by Frago for Andalusian crops up in the north of Spain. Ariza cites evidence from the 14th-century Fuero de Huete: fisieron, asension, consejo 'conçejo' (Ariza 1994: 230, from Diaz Montesinos 1987). 150 Ariza (e.g., 1994: 228-229) also offers a few other examples from northern regions, seeing in these very sporadic confusions signs of deaffrication of the affricate sibilants, which became more prone to confusion as they became distinguished by only one distinctive feature.151 Indeed, the clear tendency of marginalized groups, such as Black Africans and Gypsies (who arrived in Spain in the late 15th century), to acquire Spanish with seseo (Ariza 1996: 62) would indicate a situation already conducive to merger. Other over-interpretations and misinterpretations on Frago's part can be grouped into several categories: 1.

2.

Syllable- or word-final confusion. Ariza (1996: 53) suggests that sporadic confusion of -s and -z in implosive position probably indicates not seseo but rather deaffrication and devoicing (or, rather, the scribes' attempts to represent a phone that was not clearly assignable to only one phoneme). Harris-Northall (1992) has argued that final -z was used to represent a devoiced and deafiricated sibilant in 13th-century Alfonsine texts, so it would come as no surprise to find examples of orthographic confusion in this period. Indeed, these are the very cases that also show up in the nonAndalusian Fuero de Huete. Lapesa, of course, offers us dies, diesmo, and many of Frago's forms fall into this category: encenz, connosco, asas, enpesca, esquierdo, Velasques, viscayno, mayorasgo, Yçrael, Yzmael, Lezmez, Hasnalcaçar, sojusgar. For later periods, such as the end of the 15th century, it does seem reasonable to hypothesize an initial neutralization of -s and -z in implosive position.152 Lapesa offers some of the best: Roblez, inglez, fijoz, Andrez, Blaz, viscayno, Sanches, Gomes, durasnos, Beatris. Post-nasal confusions. Confusion in this position appears frequently in Frago's examples: encenz, Almança,153 ençienço, Ançurez, conçejoconsejo (and in many other examples after 1500). Terrado Pablo, commenting on the forms seguransa and fiansa in a 1380 document from Barcelona offers the following explanation: "Acerca del valor fricativo o africado de esta s no vamos a emitir juicio alguno, y ello por una razón de fonética general: la nasal η supone una oclusión en la cavidad oral, que seguida de una fricación puede equivaler a (§). El uso de s en lugar de ç no creemos que posea aquí valor probatorio en favor del ablandamiento de las sibilantes africadas, pues el rasgo oclusivo puede estar implícito en la nasal que precede a i . . . [footnote] Se pudo comprobar esto mediante sonogramas realizados por el autor de este artículo en el laboratorio de fonética Pere

254

The Seville phase

3.

4.

5.

Barnils. En una pronunciación enfática, una secuencia como 'con sus' parecía poseer una realización (§): era patente la barra de explosión que precedía a la zona de turbulencia fricativa" (Terrado Pablo 1986: 177). Given such phonetic similarity, we should not be surprised to find occasional confusions in post-nasal position (cf. English speakers' confusion of prince and prints, mince and mints). Metathesis, assimilation, dissimilation. In discussing the examples çemençera, çenzilla, çenzillo, Lapesa commented that they were "casos todos explicables por asimilación o disimilación" (Lapesa 1985: 252). Other examples which show the appearance of two sibilants can easily be explained in the same manner, or by reference to the related phenomenon of metathesis. Forms used by Frago that can be explained as resulting from these phenomena include: susepçores, encenz (among Frago's most significant tokens), ensensario, asas, pescueso, ençençario, ençienço, fesiese, Çiguença, Hasnalcaçar, ecleçiasticas, syçion, desernimiento, Ceçilla. In addition, the numerous variants of the Latinism excepción are clearly affected by a combination of these phenomena (consider the number of sibilants in [e(k)stseptsjón]), as well the strong tendency to adjust this alien surface form to the phonotactics of the language.154 Influence of etymological forms. Ariza (1996: 57) points out that the form connosco actually corresponds to the etymological form of the word, and is thus best not included as evidence oí seseo. Misreadings. Correct interpretation of the data depends, first and foremost, on correct transcription of original manuscripts. In the case of the sibilants, this is especially difficult, since the sigma (or sigmas) was often used to represent both ζ and s in even the most conservative dialects and, in the case of devoicing, could represent all four sibilants. As we have seen above, Frago chastises Mondéjar for using forms with sigmas as evidence of early confusion. However, Frago himself falls to the same temptation. To wit, he offers the following forms as evidence of seseo, even though in each case he explains that they appear with sigma: caliaejo, caliaes, oficiales, Sahara (Frago points out that the initial sibilant could be a sigma), Calia, topamos, Τορασΐα (Frago indicates that the last three are all transcribed with sigma). Ariza has also returned to some of the same manuscripts used by Frago in order to check the transcriptions. He suggests that Frago has made two important misreadings. First, Frago suggests that the formposision (actually posiaion) 'posesión', was originally written with ç (posiçion) and then corrected with the sigma. Ariza does not agree: "Es verdad que hay una sigma, pero no es claro que debajo haya una ç" (Ariza 1996: 47). More importantly, Ariza also checked the form beçola from a 1379 document. Given that the use of the ç would provide incontrovertible evidence of confusion, it is an especially clear example. Ariza includes a

Linguistic changes

6.

7.

255

photocopy of the original in his paper, in which he reads a clear s (Ariza 1996: 54). Substitutions. Alonso (1947) pointed to the frequency of substitutions of one sibilant phoneme for another. Though there are claims in his article that we must now question, his assertions about confusion of s and ç are validated by the survival in northern peninsular dialects of lexicalized forms showing such confusion (e.g., çurcir 'zurcir' for original surcir). With regard to Frago's claims, the following observations are of particular interest: "ençienço alternaba con encienso en la Edad Media . . . sencillo; cenzillo en Mingo Revulgo . . . sedaço hasta el siglo XV, luego cedaço . .. acechar alterna con assechar (*ASSECTARE) desde muy antiguo . . . reçuçitar aparece algunas veces. . . macizo sobre MASSA . . . çoçobra (SUBSUPRARE) alternaba todavía con soçobra en el Cancionero de Castillo . . . bisma-bizma, lesna-lezna, mezquino-mesquino, brosno-brozno, biscochobizcocho, respe-rezpe, chosne-chozne, Velasco-Velázquez, cascorvocazcorvo, mescolanza-mezcolanza." (Alonso 1947: 1-2). Pascual (1991) has also pointed out that sufrir-çufrir is a form that shows a long history of phonemic vacilation, with lexicalization of çufrir in some dialects. Other forms reported by Frago can also be interpreted as lexicalized substitutions, particularly such repeated examples as egleçia, seduta and Medina Çidonia. Errors. Ariza (1996: 47) indicates that some forms may simply be scribal errors. He points to the case of conçejo for consejo, in which the scribe had just written conçejo 40 times and then should have written consejo, but perhaps wrote conçejo by dint of repetition.

After considering the alternative explanations, Frago's use of the limited number of forms he can supply to support his thesis appears unjustified. Of course, the evidence here reviewed may point to a very limited and gradual development of neutralization of final sibilants (or perhaps even merger) in the 15th century (as Lapesa suggests), but it clearly does not serve to support Frago's more extreme claims of 13th-century origin for this phenomenon.155 Given the small number of forms offered as evidence, and the existence of very plausible counter-explanations, Frago's hypothesis that Andalusian seseo originated during the 13th-century repopulation of Andalusia must be rejected. On the other hand, it may very well be related to a later stage of demographic movement and koineization in the 16th century (see Chapter 6). Moreover, the rejection of this particular hypothesis does not imply the rejection of other hypotheses linking particular features to 13th-century koineization in Andalusia. The point is that each case must be investigated,

256

The Seville phase

and confirmed or denied on the basis of the historical record. The model of koineization makes no predictions about such particular developments. To the degree that it is predictive, it is only so in the broadest sense: given certain sociohistorical conditions, including but not limited to demographic mixing, dialect mixing and breakdown of close-knit social networks, one can expect to find the co-occurrence of changes that show mixing and simplification. In this sense, the model can be compared to a guide map, telling the researcher where there is likelihood of discovering certain types of change. Of course, future research may be able to identity more specific types of change as characteristic of koineization. Beyond that, only historical research and analysis of the particular social and linguistic situation can provide further illumination.

3. Conclusion The Seville phase represents the last great phase of koineization begun and completed in the medieval period. During this phase, there occurred, once again, great demographic and dialect mixing in the valley of the Guadalquivir. This mixing, amply attested in the repartimientos, presented an ideal social context for the occurrence of koineization. Analysis of a collection of 13th-century documents prepared by municipal scribes of Seville reveals three changes that can plausibly be attributed to this rekoineization: the elimination of extreme apocope (which probably had an impact on subsequent northern elimination of the phenomenon, particularly through weak ties between south and north); the elimination of the minority feature of leísmo (with some evidence that adult leistas suppressed this marked feature in their speech); and the swift reduction of the first person singular possessives to the invariant form mi(s) (which also began to affect northern usage). These changes are in some respects less "dramatic" than those of earlier phases, but this is unsurprising given the extensive leveling or spread of features that had already occurred in the north by the time of the Seville phase. The evidence of such changes serves to rebut the claims of some who see no significant effects in the 13th-century repopulation of Betic Andalusia. On the other hand, analysis of the case of seseo reveals that it cannot be plausibly linked to 13th-century koineization in Betic Andalusia. This case in particular shows that efforts to link dialect mixing and language change must be grounded on appropriate application of the model and careful interpretation of the evidence.

Chapter 6 Conclusions

This study has had two fundamental objectives: first, to apply the model of koineization to the study of early Castilian, in order to gain a clearer understanding of the nature and causation of characteristic changes of this variety (or group of varieties), and, second, to refine our understanding of the model itself and to explore its potential uses in the explication of language change, particularly in the reconstruction of changes which occurred in the distant past. For this reason, I have focused on the earliest phases of development of Castilian, beginning at a time-depth of over a thousand years. To complete this task it was necessary to start with a well-defined and coherent model. I therefore examined the literature on koineization, with particular attention to the key studies of Siegel, Trudgill, and Kerswill and Williams, and arrived at a definition of the model which is a critical synthesis of existing research. It is important to note that this model is explicit in distinguishing a micro-level of analysis, that of individual speakerlearners in a particular type of social and linguistic context, from a macrolevel of analysis, which consists of a description of the predominant or systemic linguistic results of collective speaker activity (i.e., the community norms, or the abstraction known as a dialect or a language). The prototypical social context of koineization is characterized by relatively free and unfettered interaction of speakers of mutually intelligible dialects (which by definition share most lexical and structural features), with rapid increase in variation accompanied by equally rapid decline in norm enforcement. As adult and especially child/adolescent speaker-learners accommodate to one another and re-establish social networks, they develop new linguistic norms, which most often favor the features which have the highest frequency in the prekoine linguistic pool. In addition, speaker-learners sometimes favor features because of their relative salience. Salience depends necessarily on some structural factor (e.g., transparency or regularity) or cognitive/perceptual prominence (e.g., stress or initial position in an utterance, constituent, or word), but also depends, often crucially, on the attribution of sociocultural significance to a particular variant (e.g., identitymarking and stereotyping). Some items (e.g., morpholexical items) are inherently more salient than others (e.g., some phonetic variants). In ac-

258 Conclusions commodating to others, adult and child speaker-learners must analyze the varied output of other speakers (converting learner input into intake) and base their own production on these inductions or analyses. In most cases they overgeneralize frequent, regular, or transparent variants and patterns to the detriment of others. Opaque features in others' speech can lead learners to misanalyses, leading in turn to innovations not present in any of the established contributing varieties. When this is done by large numbers of speakers (either through simultaneous innovations, or through spread from a smaller group), then even such misanalyses can be expected to enter the resultant koine. Koines can generally be characterized by simplification and mixing/leveling, but we need not expect all features of a koine to be directly inherited from existing contributing varieties, particularly those which can be shown to have represented simplification from the speakerlearners' perspective, or those which are exaggerated and/or selected in order to mark new identities. This model has allowed the identification of Burgos-phase demographic and dialect mixing as a key causal factor for some, but not all, of the characteristic changes of the período de orígenes, changes which Menéndez Pidal described long ago, but for which he was unable to offer any but circular explanations. One of these, the leveling out of the variant forms of articles and of contractions of prepositions and articles, reveals clearly the tendency of koines to avoid fusional opacity, which results from the weakening of social constraints on speaker-learners' tendency to perceive and reproduce the most frequent, transparent, and regular variants. Another change, the decrease from the seven vowel phonemes of Romance to the five of Castilian, reveals the effects of speaker-learner reanalyses of the most frequent variants in the surface output of other speakers. In addition, the reanalysis of the Romance low mid vowel phonemes as either twophoneme diphthongs or single-phoneme monophthongs reveals how koineization can lead to abrupt interruption of lexical diffusion. When speakers begin to abandon non-functional variants in the new social context (or simply favor one to the exclusion of others), the rich variation required for the continuation of the analogical process of lexical diffusion is lost, and even changes in progress shared by most contributing dialects are likely to be frozen by the rapid selection and reanalysis of the most frequent forms. This, of course, leads to results that seem irregular when compared to (usually) more stable contributing dialects, where lexical diffusion is more likely to progress. Another significant finding is that the rather marked development of Romance h / into Castilian /ue/ (as [we]) is

Conclusions 259 best understood as the combined effect of surface analogy, reanalysis of frequent forms, and most important, its exaggeration and use as a marker of generation and local/regional identity. Examination of the Burgos phase also obliged discussion of other changes which Menéndez Pidal identified as hallmark features of early Castilian. In the case of /f-/ > [h-], koineization may have favored the overgeneralization of a pre-existing Cantabrian variant, but the lack of evidence makes more definitive claims impossible. The development of Latin -CT- > /tj/ (= orthographic ch) was analyzed as a possible introduction of a new phone thanks to the overgeneralization of a natural assimilatory/perceptual variant, and of a new phoneme thanks to the reanalysis of the new palatal phone. However, close analysis of dialectal evidence reveals that this phone was a pre-existing, possibly phonemic, variant in other contributing dialects (though in different phonetic contexts). Nevertheless, in more conservative dialects these innovations were resisted, so koineization probably did favor the natural phonetic variants of -CT- (or [jt]) by allowing them to be freely produced and quickly stabilized by speakers, as well as associated with pre-existing /tj/ (a similar argument can be made for the development of -C'L- > ! γ). Finally, the development of the Latin obstruent + lateral consonant clusters, PL-, CL-, FL-, shows mixed results in Castilian and would seem to be a prototypical result of mixing. However, comparison with contributing dialects, including Galician, reveals that the patterns of development in Castilian follow closely those of western varieties of IberoRomance, and that the changes probably occurred before the Moorish invasion. The development of these clusters, so often cited as a unique feature of Castilian, is not in fact so typical of Castilian as it is of the entire northwest of the Peninsula. Koineization in Castile merely preserved the majority features of the contributing dialects (though it is possible that the particular demographic and dialect mix favored preservation of the lateral in those words where the original consonant clusters had not been altered). While koineization may not have played the key role in most of these developments, it is important to note that these changes, sometimes perceived as marked or even "deviant" (Posner 1996), are not in fact contradictory to the tendencies of koineization, and therefore do not serve as counterevidence to my arguments (as specified in the methodological guidelines outlined Chapter 2). At some time or another, nearly all of the hallmark features of early Castilian have been attributed to the impact of Basque learners of Romance. In Chapters 2 and 3, I discussed some of the proposals that have

260 Conclusions been made, and examined them in light of the model of koineization. Overall, it seems clear that koineization would indeed allow Basque learners of Romance to exercise some influence in the prekoine linguistic pool. Larry Trask's complete rejection of Basque influence on Castilian must itself be rejected. However, it is equally clear that marked features of the Basques' interlanguages were unlikely to survive the selection process of koineization, since minority features are generally leveled out. Only when features produced by the Basques were also produced by native Romance speakers were they likely to be selected in the resultant koine. In every potential case of Basque influence examined in this study (e.g., vowel simplification and stabilization of diphthongs; /f/ > [h-]; the rise oí leísmo), the influence of Basques is not strictly speaking necessary to explain the changes in question, but in most cases it is possible to see their linguistic output favoring or disfavoring particular Romance features in the prekoine linguistic pool. Therefore, each case of putative Basque influence must be reconsidered in relation to koineization.156 A major objective of Chapter 4 was the establishment of the Toledo phase as a significant period of koineization in the history of Castilian, through analysis of three groups of changes. It was also a period of widespread dialect leveling and language spread (but also dialectal differentiation; see below). For example, typically Castilian features such as the /ue/ diphthong and transparent preposition + article contractions appear to have became common in parts of Leon and Aragon in this period. This occurred as these regions to the west and east of Old and New Castile underwent nearly simultaneous koineization at the same time as they were in contact with Castilian(s). Such outside influence certainly favored Castilian-style variants in the new and less distinctive Leonese and Aragonese dialects.157 Though the specific changes of this period include such simple effects as the stabilization of normal apocope, they also reveal other effects of koineization which I did not find in the Burgos phase. For instance, the development of the rich variety of early Castilian possessive forms shows functional reallocation and simplification of (some) atonic forms, apparently as a result of reanalysis of pre-posed possessives as determiners and their simultaneous deaccentuation, in contrast to the continued salience of stressed pronominal and post-posed adjectival possessive forms. This is an interesting case because frequent but less salient forms suffered simplification (loss of gender differentiation), while stressed forms were actually built up and their regular differences reinforced. In addition, the reallocation of possessives affected all styles in Toledo, but the simplification of

Conclusions 261 preposed possessives developed only in more popular registers. Yet another case of stylistic reallocation can be found in the development of extreme apocope, in which a phonetic variant typical of fast (and probably informal) speech in early Castilian, but which was also a regular feature of the speech of eastern Iberians and francos, suffered reallocation to a formal stylistic (and possibly social) function. This retention of extreme apocope as a stylistic variant was influenced by the heavy presence of prestigious francos in the early Toledo demographic mix. One of the most interesting changes analyzed in this work is the rise of leísmo, which began in the Burgos phase, but which spread (and was reanalyzed) during the Toledo phase. Its origins reveal the impact of apocope, which obscured the input for speaker-learners, and allowed the reanalysis/overgeneralization of dative le for use as a marker of singular masculine accusative count reference (it must be recognized that the reanalysis represented, from the learners' perspective, a simplification of complex input). The conversion of this innovation (probably shared by many speakerlearners) into a change was favored by the koineizing context, in which such innovations were not suppressed or corrected. Such marked use of le later spread north into Cantabria (with internal repopulation) and was carried south into Toledo. However, in areas of Toledo (including Guadalajara near Aragon, and Toledo itself, which saw very high levels of dialect mixing), this original extension of le was itself reanalyzed. In these areas, speakers constructed their grammars based on the perceived most frequent and consistent use of le: for reference to masculine persons. This reanalysis, along with loss of count/mass distinctions, also represented a simplification, and in the koineizing context it too became established as a norm in some areas. Leísmo is clear evidence of my claim that koines should not be viewed as mere reductions to a "least common denominator" of contributing dialects. This idea, prevalent from the days of Meillet, but alive and well today (e.g., Posner 1996; Mufwene 2001: 5), relies on a view of koineization which does not take into account the role of learning and the importance of induction/reanalysis. Leísmo (or better, leísmos) also appears to reveal the development of internal dialectal variation during the Toledo phase of koineization (though further research of the textual record needs to be done to confirm some of the details). This is found in the different norm (personal leísmo) that developed in eastern (but not central or western) regions repopulated during the Toledo phase. Though it has often been assumed that koineization necessarily leads to widespread uniformity or homogene-

262 Conclusions ity, as in Australia, this is not necessarily so for all features, nor for all cases of koineization. In this case, the causes are fairly clear: the mix of settlers was distinct, and most of the areas that developed personal leísmo were settled not immediately but rather slightly later in the early to midnth century. Indeed, even more mixed settlements settled in the late 12th and 13th centuries abandoned leísmo and favored the etymological system. Leísmo also presents an interesting example of the long-term effects that koineization can have. Leísmo itself is clearly tied to Burgos- and Toledo-phase koineization. However, in the following centuries, the introduction of leísmo into the system presented succeeding generations of speaker-learners in Old Castile and Toledo with a structural anomaly. Even in the more stable social environment of those centuries, speaker-learners appear to have innovated, and to have done so to such an extent that over time new norms were established. For example, by analogy, speakers who used le for both (masculine) dative and masculine accusative count reference, began to use les for plural reference. This was followed by use of la(s) for both feminine accusative count and feminine dative reference. In some varieties, such analogies eventually led to the regularization of les or the extension of los for plural masculine datives and accusatives. Finally, singular lo was extended to the dative, thus eliminating the last remnant of case distinctions. We can see that the initial change provoked by koineization set up a structural irregularity that speaker-learners sought to regularize by altering other components of the pronoun system (in a kind of chain reaction), until slowly arriving at the most advanced system we know today, which lacks case reference. This example points to the potential importance that "catastrophic" processes of change such as koineization can have even after social networks resolidify. A parallel claim is to be found in Dixon (1997) and in Labov's recent volume Principles of Linguistic Change: Social Factors. Though Labov is primarily interested in "changes which emerge from within the linguistic system" (Labov 2001: 20), or changes in stable communities, he mentions several times that the catastrophic social mobility and dialect contact occasioned by World War Π (similar to - though different from - prototypical koineization) may have served as a trigger for many of the sound changes under way in Philadelphia (Labov 2001: 227, 318, 509). Not unrelated to the preceding discussion is the fact that many changes that are initiated in a koineizing region later spread to regions where contributing dialects are spoken. In such cases, koineization is once again a catalyst for changes that do not progress in other regions until after the

Conclusions 263 original phase of koineization is well under way. Examples of this can be seen in the spread of the reorganized possessive system from Toledo back north to Burgos, as well as the spread north from Betic Andalusia of regular simplification of mio/mi and the decline of extreme apocope in all styles. Several factors can be cited that favor such spread: the pre-existence of the innovations in contributing varieties (originally resisted by stable norms); the probability that the structure of the system already favors the change; the effects of numerous weak ties between innovative and conservative regions (including the impact of language missionaries). Still, these are only partial explanations for the spread of innovative koineizing features, since resistance is maintained to others, such as when leísmo was preserved in the north even after its loss in La Mancha, and, more importantly, Andalusia. Questions of identity certainly come into play, and in the case of leísmo, both its sudden disappearance in Andalusia and its surprising preservation in the North (with subsequent analogical changes in the pronoun system) argue for its high sociocultural salience (stereotyping). Changes that do not spread outside the original place(s) of koineization become new regional dialectal features. Frago Gracia has claimed that a whole range of hallmark features of Andalusian can be traced to 13thcentury dialect mixing in Andalusia. In my analysis of his data relating to the sibilant merger known as seseo, I have been forced to conclude that the evidence offered by Frago does not support the hypothesis. The analysis of Frago's claims and data shows once again that the establishment of causation in cases of dialect mixing must be subjected to the methodological constraints outlined in Chapter 2. On the other hand, I have found solid evidence for the rapid loss of leísmo in Andalusia (as compared to Old Castile and Toledo), so it is plausible to speak of the early formation of a distinct dialect and some notion of regional identity in Andalusia (although this same rejection was apparently occurring in other southern varieties as well). The findings on seseo have further implications. Given the analysis of available evidence, it seems likely that seseo - as a regular feature of the dialect - dates from the 16th century, and was probably initiated by the massive demographic movements which followed the fall of Granada and the discovery of America in 1492. If so, then the common claim that seseo (along with some other Andalusian features?) was already present and simply transported to America needs to be re-examined. It seems that at the end of the 15th century the difference between the phonemes in question had been reduced to a minimum (based only a small difference in place of

264 Conclusions articulation), and seseo may have existed in some communities as an incipient syllable-final neutralization. These features would have been carried by colonists to all parts of the Americas. As such, it is possible that the regularization of seseo as a feature of Andalusian and all American varieties of Spanish occurred as part of a particularly far-flung process of roughly simultaneous koineization(s). Indeed, a similar kind of analysis has recently been proposed by Trudgill, Lewis, and Maclagan (2000), who have argued that some features of southeast England and of SouthernHemisphere Englishes developed simultaneously as a result of "inheritance" of common tendencies or "drift". In the case of seseo, the inherited tendencies or structural features were the weakly-marked phonemic distinction and incipient neutralization. In the koineizing context(s), speakerlearners everywhere could easily have generalized the merger and thereby established it as a new norm. The changes of the Seville phase are significant for other reasons as well. For example, the better textual evidence of this period makes it possible to identify the effects of both adult and child accommodation/acquisition. Adults appear to have rejected use of stereotyped Castillan leísmo from the very beginning of repopulation. They also dropped use of extreme apocope fairly quickly (though the formal written texts probably maintain it longer than it was maintained in speech). On the other hand, the data for the simplification of the mio/mi contrast show that it began with adults, but was only carried to completion and stabilized by the first child generation. Indeed, the changes for this period seem not only to confirm the rapid time scale of koineization suggested by Kerswill and Williams (2000), but suggest that adults can quickly negotiate new norms for certain very salient features from the very beginning of the process. Of course, by the time of the Seville phase, many of the contributing varieties had already suffered significant leveling and simplification, so the dialects in contact were probably not as distinct (or opaque to learners) as in earlier phases. It is also useful to consider the three stages of koineization in medieval Castilian in relation to each other. One very significant difference is the tendency in the Toledo phase to establish markedly different stylistic norms. For example, both extreme apocope and maintenance of gender distinctions in to/tu and so/su were features of the conservative Toledan norm, while weakening or elimination of both marked the popular norm. Examples of such marked stylistic distinctions are not present in my analyses of the Burgos and Seville phases, though this may simply be an effect

Conclusions 265 of the data and/or the features chosen for analysis. Still, the Toledan tendency to maintain a conservative norm even in the face of the usual koineizing tendencies may reflect the unique combination of factors found in the repopulation of Toledo: the arrival of the court to the symbolically important city; the great institutional presence, and with it, the arrival of aristocrats, royal officials and servants, and ecclesiastical bureaucrats, including many francos in high positions; as well as the maintenance of an influential and tightknit Romance-speaking Mozarab community, which may have shared and favored some of the more conservative features of Castilian and other varieties. The perspective across different phases of koineization also allows us to note that, on the one hand, changing conditions can lead to very different results (cf. the establishment vs. loss of leísmo and extreme apocope in the Toledo and Seville phases), while, on the other, what might be considered unitary processes from a structural perspective are in fact completed only over the course of several periods of punctuated or catastrophic change. From a sociolinguistic perspective, therefore, they are not unitary changes. This is well exemplified by the reallocation and simplification of the possessive forms, which begins in the Toledo phase, but is not "completed" until the Seville phase with the loss of the mio/mi contrast. Beyond this, it is not unreasonable to claim that the repeated series of koineizations in Castilian/Spanish (including those not studied here) are at the root of the long-recognized "drift" of Spanish toward more analytical, transparent, and simplified structures.158 Such tendencies have also been identified in other "central" or urban varieties, and would seem to warrant further investigation of dialect mixing in the history of other language varieties, even those less clearly affected by prototypical koineization (cf. Milroy 2002).159 Throughout this study I have assumed that the periods of koineization originally suggested by Penny are in fact valid and useful periods for the history of Spanish. Indeed, one of my principal aims has been to substantiate Penny's broad claim that the repeated periods of demographic and dialect mixing had a significant effect on the development of medieval Spanish. Still, there is presently general disagreement on the proper division of the history of Spanish into periods (e.g., Eberenz 1991; Harris-Northall 1996b; Martinez Alcalde and Quilis Merin 1996), and even over whether any periodization of the language is possible or desirable (Wright 1999; Penny 2000: 5). Of course, as Penny and Wright argue, if one establishes temporal divisions in the history of a language and assumes that these represent absolute boundaries between the language of one period and that of

266

Conclusions

another (rather like absolute divisions between dialects), then periodization must be abandoned. Wright argues that the lack of such absolute differences means that any periodization based on internal features is nearly meaningless, and he suggests that periodization of a language be based exclusively on external, metalinguistic factors. He argues that there is probably only one really important period division in the history of Castillan: the 13th century, when the distinction between Latin and Romance penetrated most Iberian societies, along with sharp conceptions of difference between Romance "languages" (as a result of a change from a more logographic "Latin" script to a more phonographic one). He also shows some willingness to accept Harris-Northall's (1996b) demonstration of the standardizing effects of printing, from about 1500. Otherwise, nothing counts. I would argue, however, that to establish such an extreme dichotomy between internal and external factors is to fall into, in part, the very "structuralist fallacy" that he criticizes in his essay. Of course, Wright's rejection of periodization has to be understood as a rejection of traditional and largely arbitrary divisions such as "Old Spanish", "Golden Age Spanish", "Modern Spanish", as well as Eberenz' slightly less arbitrary alternatives, a "fase expansiva" from 1050-1250, a "fase antigua" from 1200-1450, and a "castellano medio" from 1450-1650 (Eberenz 1991: 94, 105). Interestingly, however, Eberenz asks if "la evolución de la lengua se produce a un ritmo siempre igual o si, por el contrario, los cambios se acumulan en determinadas épocas. Si la segunda hipótesis se revelara como exacta, se resolvería el problema de la periodización" (Eberenz 1991: 93). I would argue that the analysis presented in this study implies just this: periods of rapid change punctuate periods of slow change in the history of Castilian/Spanish. Koineization allows us to establish a periodization that links both internal and external factors (cf. Eberenz 2000: 17), and which can improve our understanding of the patterning and causation of changes in ways that previously were not possible. Of course, this periodization is rather different from earlier types of periodization, since it shuns absolute divisions, and is sensitive not only to temporal but also geographical variation. As such, a periodization of Spanish based on koineization can help avoid certain pitfalls of historical linguistic research, since researchers will be required to specify not only when and where changes took place, but also when and where they did not. The failure to comment on such variation in the past has been explained by Milroy as yet another effect of the

Conclusions

267

standard ideology, or tendency to project the modern conception of uniform standard back onto earlier language states: when dates are suggested for changes . . . they are usually given, without comment, as the dates at which the changes took place in this standard variety or its unilinear precursor, and not the dates at which they might have taken place in some other variety. Despite this, however, they are then usually cited as the dates at which the changes took place 'in English'. (Milroy 1992: 125)

Although Milroy focuses on the study of English, the discussions of previous research contained here reveal that similar problems plague the study of Spanish (cf. also Fernández Ordóñez 2001: 399). Surprisingly, however, there remain scholars who continue to subscribe to the belief that language change is never rapid or sudden, but rather is necessarily gradual. We have seen evidence of this belief in the work of Frago Gracia and even Ariza, but they are not alone; recently the creolist Salikoko Mufwene (2001) has come out in defense of this position. Mufwene rejects completely the idea that the rate of language change can vary: "Creoles did not develop more rapidly than other languages.. .All of them developed gradually" (Mufwene 2001: 130). For Mufwene, all linguistic change is necessarily gradual. This is a stunning conclusion, given that throughout this work Mufwene argues that scholars must investigate "ethnographic ecologies" in order to explain language change, and many of his ideas are similar to those I defend here (e.g., the tendency of linguistic contact to favor cognitive and linguistic processes of simplification, transparency, and regularity). Why then does he insist on the gradualness of change? Mufwene rejects the dominant prototype model of creolization (i.e., sudden change or language creation caused by children's learning of a pre-existing adult pidgin or pre-pidgin); instead, he argues that creóles developed slowly over many generations, gradually developing away from their lexifiers as speaker-learners were socially distanced from nativespeaker models (he also completely rejects the importance of children to the development of creóles).160 Although this may well have occurred in some cases, Mufwene does not present this as another possible model of creole formation; rather he presents his as the only model (and though he argues for gradual distancing, he rejects the gradual rapprochement implied by the concept of post-creole continuum). In addition, he argues throughout the book that creolization, koineization and all contact-based change works in fundamentally the same way (certainly true at deep levels of abstraction, but I address the value of distinct prototypes of koineization and creoliza-

268

Conclusions

tion in Chapter 2). As a result, his belief in the gradualness of creole formation forces him to view all language change as gradual. Nevertheless, these claims strike me as untenable, for the research of many other scholars (not least that of Kerswill and Williams) stands as strong counter-evidence to Mufwene's view, as does the research presented here (although the value of the evidence depends on how "gradual" is defined; in language change, its meaning can only be relative). It is clear that the rate of language change can vary along with change in the sociocultural context, and it is equally clear that both adults and children influence language change, though not necessarily in the same ways, and perhaps in different ways in different social contexts. Still, in every case, children's cognitive and psychosocial development play key roles in stabilizing new norms and new linguistic varieties. These questions lead back to the more fundamental goal of this study: to respond to the Actuation problem, through the establishment of explicit causal links between social conditions, speaker activity, and linguistic changes. This is increasingly recognized as a fundamental goal of research on language change. However, not all scholars are agreed on the value of the approach adopted here. For example, Roger Lass agrees that "the most important act of theoretical integration that could be performed in historical linguistics is somehow establishing . . . a clear and intelligible nexus between short-term individual behavior and long-term linguistic evolution" (Lass 1997: 336). Though this is exactly what Milroy (1992), Keller (1994), Mufwene (2001), and others set out to do, Lass argues that the "hermeneutic approach" of these scholars can only be effective for explanation of non-structural lexical change. For structural changes, he suggests that all attempts to establish such a nexus are doomed to failure. However, it is my view that analyses of change which take into consideration structural, cognitive, psychosocial, and sociocultural factors, such as those included in this study, show that Lass' pessimism is not completely warranted, and that so-called hermeneutic approaches do indeed offer scholars a promising means of developing cogent and plausible explanations of linguistic change.

Maps

270

Maps

Maps

271

I .s aCO •43 U •υ O m O m

!

272

Maps

IS o oU 0 (O

1 1

Β[β •ν

Λ ο

CQ

Ι

Maps

273

H 1

3 §o

υ 41 S O δ ^ O α 0

1Vi o &

&

oυ 00 10 1υ I υ

£

Notes

1.

Substrate theory in historical linguistics was based on a geological metaphor. Though the approach did acknowledge the importance of external factors through its emphasis on language contact, such contact was viewed as contact between systems, rather than between speakers. 2. Among historians of Spanish, Menéndez Pidal and Lapesa stand out as advocates of the close relation between external and internal history. However, their views ranged from reasonable claims of foreign influence in the case of borrowings, to broad and sometimes questionable claims of substrate influence, to clearly Humboldtian ideas of language as embodying the spirit of the nation. Lapesa, in particular, sometimes employs concepts that are similar to those employed here (see Chapters 4 and 5), but my overall approach and conclusions differ markedly from his. 3. A particularly clear and well-known example of the impact of speakers on language change was provided in Labov's (1963) study of Martha's Vineyard. 4. Recently, scholars such as Lass (1997), Croft (2000), and Mufwene (2001) have developed models of language change based on biological evolution. These rely on yet another biological metaphor, but in this case languages are seen not as organisms but rather as species. This work has had a significant influence on recent discussion of language change. For example, references to "competing variants" within a "linguistic pool" are influenced by this metaphor. The comparison of linguistic and biological evolution is clearly an intellectually fruitful endeavor, and I refer to this work where appropriate (particularly as discussed in Croft 2000). Nevertheless, I remain wary of the evolutionary metaphor, since it is difficult to reconcile with factors such as human learning, human intentions, and human culture. 5. For example, Labov' (2001) recent volume Principles of Linguistic Change: Social Factors includes little discussion of dialect mixing, and focuses instead on change within stable communities, though he does at times acknowledge the potential impact of "catastrophic" social mobility and dialect contact (see the Conclusion for further discussion). On the other hand, there is growing emphasis on study of dialect contact/mixing and language change (e.g., the special issue of the Journal of Sociolinguistics on dialect contact and mobility edited by Lesley Milroy (2002), and Paul Kerswill's (2002) recent article on "Koineization and Accommodation" in the Handbook of Language Variation and Change). 6. Note, however, that Nevalainen (2000), Mufwene (2001), and Milroy (2002: 6) suggest that dialect mixing in London had a significant effect on the development of English English. Lodge (1999) argues similarly for Parisian French. Smith and Sneddon (2001), following in part Tuten (2000), argue that

Notes

275

Smith and Sneddon (2001), following in part Tuten (2000), argue that dialect mixing in medieval Paris led to the loss of some of the preposition + article contractions of French (in medieval Spanish they were completely eliminated). The model of koineization is applicable to these cases, but Castilian presents a more nearly prototypical case of (repeated) koineization. 7. Spanish is used in the title of this book for two reasons. First, it is the name most frequently used to refer to the modern standard language which developed from the medieval Hispano-Romance variety known as Castilian (castellano). Second, in informal English usage, Castilian (spoken in central northern Spain) is sometimes contrasted with Andalusian (spoken in southern Spain), so use of Spanish avoids any possible confusion. Nevertheless, I use both labels in this book, though I employ Castilian whenever it is necessary to distinguish the variety under study from other varieties of Hispano-Romance, such as Leonese and Aragonese. 8. In each case, I have used the name of the principal urban center of the region/period as a label, both for ease of reference and since these urban centers probably played key roles as centers of mixing and of diffusion of change to surrounding regions, and thus in determining the features that would make up the regional varieties in each period. Note that I have given placenames in their English form only if those forms are frequently used and likely to be recognized (Seville, Lisbon, Castile, Catalonia). Otherwise, I have used Spanish forms of placenames (particularly for cities and rivers). 9. It now seems that the Ionic dialects, particularly Eastern Ionic, were themselves the results of dialect mixing and leveling, for the Ionic coastline had earlier been colonized by speakers of varied origin (though early Attic speakers probably dominated). Attic is thus often identified as a conservative or archaizing form of Ionic (Horrocks 1997: 7, 27). 10. Horrocks (1997: 29) suggests that the new written standard of Great Attic strongly affected speech. 11. Horrocks (1997: 36) rejects the supposed Doric origins of this change (frequently claimed by other scholars) since the early Koine (Great Attic) was based almost exclusively on Attic and Ionic varieties. Instead, he argues that the forms selected in the Koine were favored because they allowed speakers to avoid the morphological irregularity which resulted when the Attic or Ionic forms were inflected (the propagation of such regularizing innovations is frequent in koinezation). But Horrocks also believes that speakers may have opted for the minority Doric forms precisely because they allowed avoidance of a "parochial morphological anomaly"; in other words, the avoidance of marked dialectal forms through a "strategy of neutrality" (see below). Still, Horrocks (1997: 19-20) grants that these forms may have been recognized and accepted as a result of their use in the heavily Doric choral lyric of Athenian drama. It probably is not necessary to choose any one of these factors,

276

12.

13. 14. 15.

16.

17.

Notes since all would have contributed to increasing the frequency of the forms that survived. Some regional differences in the Greek homeland can also be attributed to imposition of the Koine and to "substrate" effects of dialects such as Boeotian or Doric (indeed, Bubenik [1989] and Horrocks [1997] use the term Koineization to refer to the spread of the Koine into the regions where traditional dialects were spoken). It is also worth noting that these same factors can be seen as crucial to the development of Latin American varieties of Spanish (though with different factors weighing more heavily in some regions than in others). See Wright (1996) for a more recent discussion of dialect mixing and the development of Vulgar Latin. Siegel (1985) cites brief definitions of these terms proposed by Mühlhäusler (1980: 21) for pidginization (see below). The very existence of the medieval Provençal koine as a uniform linguistic variety has been brought into question (Blanchet 1992, Zufferey 1987). Its perceived uniformity is quite relative, since a multitude of dialectal variants are attested in the medieval chansonniers. More importantly, the appearance of uniformity, to the degree that it exists, probably reflects the fact that surviving texts were copied largely between 1250-1350, towards the end or after the great period of Troubadour production. Moreover, they were copied by scribes in Northern Italy who would have known Occitan or Provençal only as a foreign language, particularly as it existed in the poetry. As a result, they may have perceived it as more uniform than it actually was, and thus regularized in their copies dialectal variants they perceived as incorrect. The Andalusian form juerga is not useful to Hall's argument unless one assumes that the development of Andalusian was clearly distinct from that of Castilian. This view cannot now be accepted. Hall (1974) also describes numerous peninsular dialects as koines, but it is clear that he uses the term only to mean shared or common language of a region: "In the twelfth and thirteenth centuries, the dialects of Aragon, Asturias, León, Castile, and Galicia were about equal to each other in prestige and administrative and literary use . . . Concomitantly with the Reconquista, the Castilian dialect became the standard for the regions which came under Castilian rule, gradually overlaying the other regional koines such as Asturian, Leonese, Aragonese, and the conservative Mozarabic spoken in the central area" (Hall 1974: 121). Only some of these varieties could be considered koines in the technical sense of the term developed here. In light of Mufwene's (1997) criticisms of the second criterion, Siegel (2001) restricts the definition of contributing varieties to those that are mutually intelligible (a criterion which allows for the inclusion of second-language varieties). Of course, speakers in a contact situation may at first believe that their

Notes

18.

19.

20.

21.

22.

23.

24. 25.

277

varieties are mutually unintelligible, but later change their attitudes as they become more familiar with different speech forms and their users. The assumption that native speakers need to be present (and dominate) in the mix excludes Israeli Hebrew from the category of "prototypical koine" (see discussion of prototype models below). However, that is not to say that the koineization is not useful in explaining its development, a fact which is recognized in Blanc's (1968), Siegel's (1997) and Kerswill's (2002) discussion of it as a koine. Still, Israeli Hebrew escapes (and will continue to escape) easy classification, for its development (i.e., its resurrection as a native spoken language based on use as a learned literary language) is surely unique in history. Other cases in which non-native speakers dominate are probably best classified and studied primarily as cases of language contact (and only secondarily as cases of dialect contact). Such varieties will, of course, show clear multiple influences of the source language(s) on the target language. In the discussion of Basque and Castilian in Chapters 3 and 4,1 show how contact between two languages in a prototypical case of koineization differs from that in other situations. More recent work (e.g., Milroy 2002) tends to analyze the development of regional varieties (or standards) as the result of dialect leveling (see below for further discussion of the relation between koineization and dialect leveling). Definition of a technical sense for koine (and koineization) implies that some varieties often referred to as koines will no longer be classified as such in specialist discourse. Thomason and Kaufman (1988: 170) also question Mühlhäusler's developmental model of pidgins and creóles, since the expansion of a pidgin depends on social factors that may or may not be present. In other words, there is no necessary developmental model of pidgins into creóles. In his explicit discussion Siegel himself indicates that reduction must be less frequent in koines (1985: 370-371), and Siegel (1987: 187) does not consider radical reduction as a feature of koineization. I discuss it here because the (1985) the stage-based model seems to imply its importance. Koineization is not the only type of new dialect formation, for new dialects certainly arise through slow accretion of (opacity-producing) changes and the development of new community identities in socially-isolated communities. However, koineization is certainly the most rapid. Speakers may assimilate to indexical features other than social or geographical dialect, such as age, gender, and emotional state. These two processes are not necessarily mutually exclusive, nor completely conscious. Yaegor-Dror (1993) describes a situation in modem Israel in which ethnic minority speakers of Hebrew mark their identity by diverging from mainstream Israeli koine norms for /r/ in salient positions, but accommodating unconsciously towards the Israeli koine for /r/ in non-salient positions.

278

Notes

26. Subsequent research confirms many early hypotheses about (short-term) accommodation. For example, studies have demonstrated that indexical cues (including regional or social accent) can influence listeners' perception of the speaker's attractiveness and intelligence, their classification of the individual into a particular group or class, and even the persuasive power of the speaker (e.g., Oskenberg et al. 1986, DePaulo 1992, Gallois and Callan 1988, Gallois et al. 1992). Interestingly, women have consistently been shown to accommodate more than men, and both men and women tend to accommodate more to men than to women (e.g., Willemyns et al. 1997). This greater tendency to accommodate fits well with the fact that women are more often leaders of linguistic change (see Labov 2001). Nevertheless, Communication Accommodation Theory is now often subsumed within Bell's (1984) Audience Design model. For many cases of face-to-face accommodation, accommodation theory generally remains an adequate framework for explaining speaker actions and motivations. However, Bell's emphasis on "initiative style shifts", in which speakers change their language in order to alter the situation and/or their presentation of themselves (accommodating to a perceived outside norm as opposed to accommodating to any particular interlocutor) is necessary for analyzing cases in which speakers attribute a social value to a variant and then adopt/exaggerate its use (Trudgill's hyperdialectalism). It also shows that cultural factors will affect short-term accommodation, thus making the results of accommodation less predictable (see Kerswill [2002] for further discussion). 27. It is well known that the efforts of speakers to imitate and/or learn second dialects are are plagued by errors of analysis and production. For example, speakers of Southern American English often notice the mis-analysis that nonSourtherners make of the 2nd-person plural pronoun y 'all·, this can be interpreted by northerners as simply a southern variant of singular/plural you. NonSoutherners attempting to accommodate to a Southerner will sometimes greet a potential interlocutor with a perplexing "How are y'all?", when no one else is in sight and the speaker has no previous knowledge of the interlocutor's family and social relations (which might otherwise justify such usage). 28. This seems similar to the case described by Yaeger-Dror (1993). Avoiding the stereotyped pronunciation would allow Northern English English speakers to both accommodate and continue to mark their northern identity. 29. Trudgill's original list of examples of accommodation influenced by salience does not constitute a theory of salience, and, though interesting, remains slippery as first presented. Speakers can adopt or avoid a feature because of its salience. For example, speakers often do not adopt the contrast mentioned in item 1. Items 2 and 5 seem to conflict, because if English English speakers do not acquire post-vocalic /-r/, then there is homonymie clash and speakers pronounce hot and heart the same way, since in item 2 we are told that speakers adopt the American pronunciation of hot. Homonymie clash itself does not

Notes

30.

31.

32.

33.

34. 35.

279

seem to be something that speakers worry about (with any consistency), since any conversational breakdown can be repaired with a repair strategy. Trudgill, Gordon, Lewis, and Maclagan (2000: 309) also reject Lass' (1990) thesis of swamping, according to which South African English developed along the lines of southeast-of-England English because colonial speakers consistently selected forms from the English southeast (as a prestige target variety) when faced with a number of options. They find, rather, that majority features tended to win out, but that majority features shared by all contributing dialects (in Southern-Hemisphere Englishes) tended to coincide with southeast-of-England features. Kerswill and Williams (2002: 103) also attribute the decline of clause-final like to its lack of pragmatic or interactional salience, since it is customarily used in propositions that are not foregrounded. This claim is founded on proposals made by Cheshire (1996), who argues that certain variables are favored in interactionally prominent or foregrounded constructions. For example, she claims that interrogative and negative clauses are inherently interactive syntactic environments where non-standard forms tend to occur with greater frequency. Wolfram and Schilling-Estes (2003: 722) argue, however, that a fudged lect is something of an anomaly, since speakers in an area where an innovation is diffusing generally show alternation (mixing) between two or more variants, and they suggest that (some) apparent cases of fudged vowels simply show intermediate stages in the movement of the vowel (at least in English). Mesthrie (1994: 1866) also suggests a key role for children: "While accommodation is a characteristically adult process, selection of accommodated forms, and stabilization, are more likely to be associated with child acquirers of the koiné." However, Kerswill and Williams (2000) show that older children accommodate to and learn from peers, and, for psychosocial reasons, probably do so more frequently and consistently than do adults. Trudgill also uses reduction as a cover term for the processes of leveling and simplification. Trudgill (1996: 12-13) also defends this analysis, but this is particularly surprising since in this article he is at pains to demonstrate that small isolated communities with closeknit social networks are likely to develop and retain complex changes, while larger communities with frequent contact with other communities and looseknit social networks are more likely to favor simplifying changes (including reductions in inventories). Of course, the Belfast communities described by the Milroys fit neither of these categories. They are small communities with closeknit social networks, but they are in constant contact with other varieties in the city. It is far more likely that the many different allophones of /a/ have been adopted from outside the community

280

36.

37.

38.

39.

Notes through weak ties and that the closeknit social networks have allowed their retention in a relatively complex system. Siegel (1985) and Mesthrie (1994) have both questioned whether simplification necessarily occurs in koineization. When contributing varieties show significant differences in linguistic structure, simplification is sure to occur (at least in comparison to some of the contributing varieties). When differences between varieties are limited to allophonic variation (as Kerswill and Williams found in Milton Keynes) then only leveling will be found. Note that transparency is conceived of here as an effect of speaker activity and learning, and as an influence on speaker-learners, but not as a systemic objective, as in the work of Lightfoot (e.g., 1979) and Kiparsky (1982). In fact, koineization is dependent on the co-occurrence of "increased interaction among speakers of various dialects" and "decreased inclination to maintain linguistic distinctions". Chambers (1992) presents a very informative analysis of the acquisition of English English by six Canadian children/adolescents who had recently moved to England with their families. Chambers defines and provides evidence for eight principles of dialect acquisition (which focus primarily on the learning of phonetics/phonology and lexicon, precisely those components of the grammar that tend to differ most between dialects). There are some differences between SLA and SDA, but these are fairly easily explained by differing contextual needs of speakers. For example, Principle 1 (Lexical replacements are acquired faster than pronunciation and phonological variants) is exactly in line with SLA, in which learners begin by learning vocabulary items, which are by definition more salient than phonological items. Principle 2 (Lexical replacements occur rapidly in the first stage of dialect acquisition and then slow down) indicates a difference, but this is partly due to the fact that learners do not have to learn more in order to communicate. Other aspects fit neatly in line with the predictions of SLA theory: Principle 3 (Simple phonological rules progress faster than complex ones) is expected if learners overgeneralize common rules and delay the learning of exceptional rules. Principle 4 (Acquisition of complex rules and new phonemes splits the population into early acquirers and later acquirers) reflects the fact that children will normally learn "perfectly" any variety to which they have sufficient exposure and interaction, no matter how complex it may be. Principle 5 (In the earliest stages of acquisition, both categorical rules and variable rules of the new dialect result in variability in the acquirers) reflects the fact that all interlanguages are highly variable, and that learners often switch between a "careful" style and a "vernacular" style. Principle 6 (Phonological variants are actuated as pronunciation variants) reflects the fact that adult and child learners begin by learning isolated lexical items, and only when knowledge of a critical mass of instances has been attained do they become rule-governed. Principle 7 (Eliminating old

Notes

40.

41.

42.

43.

281

rules occurs more rapidly than acquiring new ones) is dependent on the structural similarities between dialects and is predicted by accommodation theory, but it is also the most controversial of Chambers' principles. This is so because in order to explain his evidence, he is forced to argue that "absence" of a historically novel merger (=acquisition of a lexically unpredictable phonological split) is an example of elimination. This reasoning seems difficult to accept, and it directly contradicts the findings of Kerswill (1996), who argues that new phonological oppositions and lexically unpredictable phonological rules are among the most difficult features to acquire. Principle 8 (Orthographically distinct variants are acquired faster than orthographically obscure ones) reflects the impact of salience (provided in this case by the writing system) on learning, common to both SLA and SDA. It should be noted, however, that adults may overgeneralize more than children for some features. Thus, Bybee and Slobin found that adults tend to "double-mark" past tense and produce forms such as hitted, which small children tended to avoid, perceiving the final -t of hit as a past tense marker itself. Ravid's study of psycholinguistic processes and their relation to language change in modern Hebrew is especially pertinent to the model of koineization, since many of her observations about change in Hebrew take for granted the quasi-koineizing social situation of Israeli Hebrew (cf. Blanc 1968). In this context, Ravid shows literacy (and/or standardization) is the primary conservative force that preserves irregularities. It is not certain whether this limitation is due to cognitive or social factors, or a combination of both. Payne's sociolinguistic study (1976, 1980) of children in King of Prussia, Pennsylvania revealed that certain features of the Philadelphia dialect (fronting of /ow/ and /uwΓ) were only acquired regularly if children moved to the area by age 8 (and the complex short a pattern was only acquired if the children and both of their parents were from Philadelphia). This is regularly cited as evidence for a cognitive or maturational constraint on the age of language learning, but Labov (2001: 430) carried out a multiple regression re-analysis of Payne's data and found that the most significant independent variable was the degree of integration in the peer group. Other evidence indicates that frequency of occurrence remains a strong predictor. For example, Trudgill, Gordon, Lewis, and Maclagan (2000: 310) report that the English phonemes /w/ and /MJ had merged in southeast England and probably in Australia at the time of the main immigration to New Zealand in 1840-1880. Although many colonists came from southeast England and Australia, they were far outnumbered by colonists from Scotland, Ireland, and northern England. These speakers consistently preserved the distinction, and it survived as a norm of the New Zealand koine. On the other hand, Britain (1999) analyzes a minority complex feature propagated during dialect mixing in New Zealand English. This is the split in pronunciation of (-own) words

282

44. 45.

46. 47.

48.

49.

50.

51.

52.

Notes into a monosyllabic groan class and a disyllabic grown class. Britain argues that this change was favored because the novel disyllabic articulation increased the degree of correspondence (transparency) between a grammatical category (participle) and its expression. The discussion of psycho-social stages of development is based on Giddens (1989: 82-85). As the discussion in Kerswill and Williams (2000) indicates, the term generation is best understood as a convenient shorthand for reference to the succession of overlapping cohorts of children and adolescents. This sort of lingering competition between two strong variants may be exactly what is required for reallocation to occur. Mashlum (1992) studied the linguistically "chaotic" society of Spitsbergen, Norway (where no permanent community exists), and found that the children employed a variety of "strategies of neutrality" to negotiate the mixed and unstable sociolinguistic context. Such strategies include code-switching, mutual accommodation and code-mixing, and use of the relatively neutral standardlike East Norwegian accent. There may have been incipient processes of koineization within each group, which if maintained would have led to the appearance of two koines, one middle class and the other working class. Nevertheless, Mufwene (2001) betrays a certain ambivalence: he criticizes the distinction between koineization and creolization, but he also contrasts koines and creóles throughout the work. See Chapter 6 for further discussion of these issues. This passage reflects Thomason and Kaufman's larger concern with the circularity of appeals to markedness as a kind of linguistic universal. Here, I follow Trudgill in assuming that markedness is dependent on the structure of the particular language varieties in question, and can be equated with "exceptional" or "less frequent". Kerswill and Williams' (2000) describe the different results for koineization in a new town (Milton Keynes) and strict dialect leveling in a relatively stable and established old town (Reading). The actual results seem to be the same in the long run, but the changes are more gradual in the old town, and relatively sudden in the new town, where the first generation of children has already defined new norms and is not familiar with many of the traditional features of the region. In the old town, the youngest generation retains some features typical of the oldest generation. Population movements toward the west tended to group settlers according to their place of origin, so that the North (Great Lakes region) was populated primarily by New Englanders, the Lower South by lowland Southerners, and the Midlands by a great mix of people from all areas. This created three horizontal dialect tiers stretching from the eastern dialect zones to about the Mis-

Notes

53.

54.

55.

56. 57.

58.

283

sissippi River, where differences between them begin to disappear (Wolfram and Schilling-Estes 1998: 103-108). The development of dialect areas in medieval Spain looks very much like this, but shifted to a north-south axis. See Granda (1994), Rivarola (1996), and the studies in Hidalgo (2001) for other factors that may have played a role in koineization and regional differentiation of American Spanish. It seems that these new communities were founded with the very intention of weakening resistance to the standard. Versteegh describes a case of settlement of the reclaimed land of the Zuider Zee (in 1936); the Dutch government decided to select new inhabitants from a variety of regions, religions, professions, and dialects: "It was thought that a representation of all Dutch regional accents would gradually lead to the disappearance of the regionalisms and the emergence of a new interregional standard, which would be better suited to serve as the means of communication of these carefully selected inhabitants" (Versteegh 1993: 65). The social history given here represents a synthesis of earlier research. I have consulted Sánchez Albornoz (1966), Pérez de Urbel (1945), Menéndez Pidal (1964), Menéndez Pidal (1960); however, given the extreme positions adopted by some of these, I have also relied heavily on Μοχό (1979), still recognized as an excellent and balanced account of the social history of medieval Spain (cf. Mitre 1999: 103), as well as more recent manuals and reviews: Riu Riu (1989), Minguez (1989), Iradiel et al (1995), Torres Fontes et al. (1990), Tusell et al. (1998), and García de Cortázar (1991). I have included some useful references from Pastor (1996), but his central thesis - that Castile was repopulated by natural internal growth and that immigration played only a minor role - remains unconvincing in the face of significant evidence to the contrary. Pastor's larger aim seems to be to reject the tradition of Castilian exceptionalism by emphasizing parallels between early Castilian history and that of other parts of Europe. However, to do so he is forced to ignore the large demographic movements that are so characteristic of Iberian history. I leave aside the questions of how and why use of Basque persisted. See Trask (1997) for an excellent history of this language. Menéndez Pidal (1960, 1964: 442-443) provides toponymie evidence of the mixed origins of settlers, gleaned from contemporary documents of the period, such as the name of the village of Toldanos, near the city of León (and others like it), indicating the arrival of a group of Mozarabs from Toledo, or similarly, placenames such as Gallegos, Galleguillos, Astureses, and even Castellanos. One might also ask what attracted settlers to this region rather than to the safer lands to the west. Certainly proximity to home was key for settlers from the east and north. However, Castile's proximity to the Moorish regions played a role, since the Moors represented both a threat - for their military power - and

284

59.

60.

61.

62.

63.

Notes an attraction - for their wealth and sophistication. Soldiers also had the opportunity of gaining booty in raids on Moorish communities, and it is known that at least one Castilian count dressed in Moorish style (Menéndez Pidal 1964: 472-482). Menéndez Pidal lists names of towns, found in contemporary documents, that attest to the Basque migration: Báscones, Basconcillos, Billabáscones, Bascuña, etc. (1964: 473), as well as placenames of Basque origin, such as Zalduendo. These tend to appear in the northeastern part of the county. Such toponymie evidence is difficult to interpret. It may be that some such homogeneous communities resisted the dominant regional norms and developed their own (e.g., maintenance of Basque, which we know occurred in the nearby Riojan valley of Ojacastro). However, it appears that relatively few communities were founded by homogeneous groups of immigrants (including nonBasques): Pastor (1996: 76) reports that only 2% of village names in southern Castile indicated a place or group of origin. Moreover, in the case of the Basques, Romance names such as Villabâscones highlight their minority linguistic status. Sánchez Albornoz (1966) refers to serfs, but some recent scholars argue that workers tied to the land (and without the rights of hombres libres) in this early period were probably slaves in a system carried over from Roman and Visigothic times, while serfdom only appears later, with the penetration of feudal ideology (cf. Pastor 1996: 280). The infanzones would rise further in social status in 11th-century Castile. For example, the Cid was an infanzón, but in his youth he was a close advisor of the king of Castile and, after his conquests, he married his daughters to royalty. The social effects of the frontier have often been overestimated (see the collection in Hofstadter and Lipset 1968), but its effects are (in part) those which accompany mixing and the breakdown of social networks. The frontier is therefore a specific manifestation of more general kinds of social change, which may interact with the cultural values of the immigrants (Lipset 1968: 12). Lee (1968: 69-70), for example, pointed out that frontier societies are simply examples of the process of migration, which often leads to a decline in the importance of hereditary elites and extended families. However, the isolation produced by migration to a frontier is a feature that may further social change and contribute to the formation of new identities. With regard to Castile, the linguistic features which came to be seen as prototypically "Castilian" may actually have begun as variants which distinguished the frontier people of the Duero Valley (including much of the flat, open zone between Burgos and León) from other Christians who lived to the hilly west, north, and east. Wright (1994b) has argued that some ballads may be much older than previously thought, and many perhaps survive from the early Middle Ages.

Notes

285

64. The general lack of literacy also favored a weakening of influence of written prestige norms. Even among the literate, the writing system of these centuries had become highly logographic (see the comments in Wright 1994c) and was thus less likely to exercise a conservative influence on pronunciation. 65. Wright asserts that it is anachronistic to visualize separate Ibero-Romance dialects before the 13 th century, since speakers probably did not conceive of each other as speaking different language varieties (e.g., Galician, Leonese, Castilian, etc.) during this time (Wright 1994f: 160, 1994d). When I use such terms I generally do so for the sake of convenience. Nevertheless, speakers certainly would have been able to identify differences between their own speech and those of other communities, whatever label they did or did not give them. Moreover, the development of a regional (linguistic) identity may have occurred at different time in different places. The oppositional identity of early Castile along with its political separation would have easily favored an earlier development of Castilian dialect identity (based on a Burgos-area norm) than for other varieties such as Leonese (assuming Leonese as a regional variety has ever been anything more than a creation of modern dialectologists). 66. Alarcos Llorach (1982: 18) had earlier made a more tentative claim that the rural dialect of early medieval Cantabria was almost a creole or lingua franca that was utilized by Basque-Romance bilinguals. This, along with López García's work, reflects a general trend in the 1970s of applying early pidginization/creolization theory to the development of Romance (e.g., Whinnom 1980; Schlieben-Lange 1977). 67. Surprisingly, López García tries to use the fact that ser and estar exist in all Ibero-Romance varieties as proof of the broader existence and use of vascorrománico beyond the bounds of Castile. 68. López Garcia (1985b) takes a decidedly more linguistic approach to discussing the linguistic features of Basque that the author claims appear in vascorrománico. 69. Torreblanca's main point is that spellings with only i and u did not reflect pronunciation as well as ie and ue, but his attribution of this to carelessness and omissions implies some kind of written norm which at the time was lacking for these innovations. I am therefore inclined to agree with Menéndez Pidal's interpretation. 70. Penny (2000) identifies other features that can reasonably be seen as the results of koineization in early Castile. However, he tends not to enter into detailed reconstruction of their development, as I do here. 71. Mozarabic is excluded. Since Mozarabic forms are generally limited to singleword citations or brief lines of text, there is little evidence of how Romance articles were treated. Moreover, the Arabic article al came to be used before both masculine and feminine Romance nouns.

286

Notes

72. The forms without the lateral appear with much greater frequency. Also attested in Old Galician-Portuguese texts are the forms elle, ele, el, apparently derived from the nominative ILLE. 73. Also found are contractions with the indefinite article: cun, cunha, cuns, cunhas and contractions with the preposition ca: co, cós, cá, cás. 74. Per was found only in western and central Leonese. 75. Normally, the forms ele, el (< ILLE) appear as nominatives; elo, lo as obliques. This indicates an added degree of complexity that was lost from Castilian. The great variety of forms reported for Leonese reflects (in part) the great variation in local norms within Leon. 76. Though many of the 13th-century documents consulted by Staaff contain only contracted forms, he points out that the combinations in question are not numerous relative to text size. Moreover, while most long documents show a mixture of contracted and full forms, some documents contain only iull forms. One must note, however, that the documents consulted by Staaff were from the 13th century, a period in which Castilian influence was increasing. Indeed, Staaff (1907: 256) points out that the overall variation in forms increases in the east (near Castile) and decreases in the west (where contracted forms were preferred). 77. The contracted forms of both Old and Modern Aragonese are often considered to be restricted to the combination of en + o. However, in the mountain dialects of Aragonese forms such as the following are found in speech: do 'del' , dos 'de los', to 'al', tos 'a los', da 'de la', das 'de las', no 'en el', ñas 'en las'. 78. The forms del and al are the only contracted forms to survive in Castilian. Note, however, that they are found in a majority of the contributing dialects (a rough measure of their frequency in the linguistic pool), are among the most frequently-used preposition + article combinations in any variety (and thus more likely to suffer reduction), and reflect the coalescence of vowels (common throughout the history of Castilian) rather than the assimilation of consonants. It is true that the coalescence of a+e normally leads to loss of the first vowel (at least today), but in the koineizing context speaker-learners probably expected and produced a combination that preserved transparency. In contrast, Galician phonetic reductions allowed the (near) erasure of transparency in some combinations. 79. The article el was used before feminine words beginning with a vowel: el estrella, el alma. This feature was shared with nearly all contributing dialects and, unsurprisingly, it was maintained in Castilian. Throughout the centuries it has suffered progressively greater morphosyntactic and lexical restrictions on its occurrence. See Janda (2003: 407-408) for a clear and succinct explanation of the reanalyses that characterize its development. A more conservative form, ell, the precursor to el (both masculine and feminine), also survived in the Burgos-phase koine. This form preserved a geminate or palatal articulation be-

Notes

80.

81. 82.

83.

287

fore words beginning with a vowel. Like "feminine" el, it was found in all contributing dialects, and its frequency made possible its survival. Nevertheless, by the 13 th century, its use had become restricted to certain high-frequency lexical items. See Pensado (1999) for a detailed description and explanation of its development. In forms of the type cono, eno, the article has suffered apheresis and assimilation of the lateral to the preceding nasal of the preposition, resulting in the production of an opaque form, in which the underlying representation must differ from the surface realization, at least initially. Thus, cono, articulated [kono], would have had an underlying representation of /kon elo/ or /kon lo/, depending on whether the particular speaker had come to consider /elo/ or /lo/ the base form. Speakers of the time would have varied in their perceptions of underlying structure; some may have learned /kono/ as a separate form that corresponded to the lexical combination con + lo. Whether learned as forms generated by rules or as suppletions, the contracted forms still differ from their conservative or underlying representation. As a result, the analysis made by learners of the surface output might vary. Smith and Sneddon (2001) have also related the loss of (some) preposition + article contractions in French to dialect mixing in Paris. In his discussion of conservative dialect features of Cantabria, Penny (2000: 84-88) suggests (directly in some cases, indirectly in others) that many of these were lost in early Castile. One of these is very similar to the change I have discussed here. In conservative Cantabrian dialects the final /-r/ of infinitives is deleted when followed by a clitic pronoun, so medirla is realized as [me5ila]. This feature is shared along the northern dialect continuum, from Galicia, through Asturias and Santander, to La Rioja and Aragon (and was more frequent in the past). According to Penny it was lost in Burgos, most certainly as part of the general tendency to reproduce and learn easily-analyzed transparent units. There is evidence of phonetic conditioning for these different articulations. The disyllabic [ie], [ia] are found primarily in normally monosyllabic words (e.g., diez > díaz), but disyllabic [úe], [úo] are found in a wide range of contexts. The different vowel qualities of the second element are also (partly) conditioned by the phonetic context. For example, a following /-r/ favors [ia] and [ua], following labials and velars favor [uè] and [ie], a front glide favors [ie] and [ue]. Unsurprisingly, some varieties show a narrower range of variation, having lost variants with [a], for example, while others, such as those of Sanabria (in western Zamora) show the full range. Catalán finds three separated areas in Leon and Asturias maintain very similar patterns of complex variation and concludes that these dialects probably preserve the ancient variation that medieval scribes struggled to represent.

288

Notes

84. Lloyd (1987: 184) also adduces the developments CORIU > cuero, QUO MODO > cuerno as evidence of the fluctuating realizations of the diphthongs, where the earlier diphthongs [oj] and [wo] respectively alternated with other diphthongal realizations and were later stabilized along with the spontaneous diphthongs. 85. Zamora Vicente (1967: 90) indicates that orthographic o and e continued to dominate in early Leonese texts, for which he suggests the following causes: 1) the proximity and influence of culturally prestigious Galician; 2) influence of Latin orthographic norms; 3) inability of scribes to represent the diphthongs (which could, of course, be related to their uncertain phonological status). These arguments prefigure those offered so insistently by Wright (e.g., 1994c), who emphasizes that conservative spellings are often opaque to interpretation, while novel spellings imply a conscious effort on the part of the scribe to adapt the orthography to the developing sound system. Still, it is not clear that monophthongal articulations, at least in conservative dialects, need have disappeared entirely, given the kind of phonetic variation that has been found by modern researchers in conservative dialects, and, more specifically, the evidence of Menéndez Pidal cited above. 86. "Pre-Basque had just five vowels i e a o u. So far as we can judge, it probably also had the modern set of diphthongs, ai ei oi ui au eu. These diphthongs were not distinguished from sequences of the corresponding vowels, but the point is that they counted as single syllables for such processes as aspiration assignment, whereas the other vowel sequences counted as two syllables" (Trask 1997: 149). Note that there are no falling diphthongs, as in Romance and Castilian, so Basque speakers may have been especially likely to interpret these as sequences of two phonemes and produce them as such in their interlanguages. 87. The vowel [a] is considered the most perceptible while [i] is the sharpest. Perceptibility is based on the position of the first formant and sharpness is based on the position of the second formant. 88. What cannot be decided from the evidence is the exact order of these events: was the articulation of the phonetic diphthong exaggerated before phonologization (perhaps in primitive Castile), or was the diphthong selected from the pool of variants at the very time it was being reanalyzed by speaker-learners? 89. Perplexing too is Pidal's insistence that diphthongization before yod is the first or most ancient form of diphthongization in Romance: "la diptongación ante yod de e y o abiertas (pueyo . . . ) . . . es la diptongación más general a la Romania y sin duda de fecha más antigua que la diptongación de e y o en otras circunstancias . Esa diptongación de y + palatal, muy arraigada en Galia . . . se extendía por España ininterrumpidamente desde Cataluña hasta Asturias, a través del mozárabe" (Menéndez Pidal 1964: 495). This view was subsequently adopted and defended by Schürr (e.g., 1970), who saw all Romance

Notes

90.

91.

92.

93.

94.

95.

289

diphthongization as resulting from metaphony with subsequent generalization to other phonetic contexts. This does appear to describe the development of diphthongization in many varieties of Gallo-Romance and Italo-Romance, but it does not fit the Hispano-Romance data (Alarcos Llorach 1996: 17). It is well known that Castilian shows mixed results for diphthongization in this context as well: cf. monte < MONTE and puente < PONTE. Old Spanish sometimes shows competition between such variants: cf. comde and cuemde (< COMITE), with the first winning out in the modern language. Catalán (195657) shows that this phonetic environment tends to impede diphthongization in western Astur-Leonese as well, but it is worth noting that both monte and conde were often used as preposed (atonic)forms, and in such contexts diphthongization would not have been favored. The exceptional Castilian form viejo 'old' (< VETULU) indicates that diphthongization before yod had at least begun as a variable phenomenon in Hispano-Romance at the time of Burgos-phase koineization. Penny (2000: 45-46; 2002) analyzes this spread of the loss of/h/ and its acceptance as a prestige form as a result of dialect mixing during and following the move of the Court to Madrid in 1561. Lloyd (1987: 222) interprets these situations rather differently: "In Navarre, where the Basques remained at home and did not mingle actively with Romance speakers to form a new linguistic community as in Castile, the Romance dialect was not impelled to adopt the innovation." This tendency also favored the development of [-jt-] to [tf] in other regions. Commenting on evidence of medieval Mozarabic forms, which most often show forms with [-xt-] and [-jt-], Galmés de Fuentes (1996: 107) claims: "en otras areas de los dialectos mozárabes, en donde no es pensable un influjo castellano, encontramos, a veces, la palatalización: felech, felecho, felecha, /election < FILICTU, ejemplos del cordobés del siglo X Ibn Juljul, de Pedro de Alcalá y del andaluz Ibn al-Baytar, que prueban, sin duda, la autoctonía del inicio de la palatalización en los dialectos mozárabes". Of course, the developments in Castile and eastern Leon were much more regular. The history of changes of CL-, PL-, FL- (and GL-, BL-) begins with a widespread alteration in the articulation of the lateral. It is generally agreed that speakers of Late Latin/Romance began to produce a palatalized articulation of the lateral in some or all of the obstruent + lateral clusters (this articulation is preserved today in some Pyrenean dialects). One of the most widespread palatalizations was that of CL-, and it is often hypothesized that palatalization began as a reciprocal assimilation of the velar stop and the dentoalveolar lateral in CL- ([kl-] > [kft.-]), with subsequent partial or complete extension through analogy to other clusters. In some dialects, such as Tuscan, all the obstruent + lateral Latin clusters CL-, PL-, FL-, GL-, and BL- were affected similarly. In others (e.g., Rumanian), only the clusters with velar consonants palatalized, or

290

Notes

only the clusters with voiceless/tense consonants (as seems to be the case in Hispano-Romance). It is also generally assumed that nearly everywhere these "feature-heavy" clusters were eventually "reduced" in some way (Repetti and Tuttle 1987: 54). In Tuscan, for example, this was done through regular delateralization of the second element (e.g., CLAVE > *[kXáve] > [kjáve]). 96. Coraminas and Pascual (1981: 724) also report Ilaner in some medieval texts. This could be a Leonese form, however, given the western-style conservation of -er. 97. The form llantar is preserved today in Asturian, but Coraminas and Pascual (1981: 573) find few medieval attestations of the form. They also report that the etymologically-related placenames Llantada and Llantadilla are found in Cantabria and the province of Palencia (between old Leon and Castile). The related noun forms pranta and planta are the normal outcomes in Portuguese and Castilian, respectively. However, there are medieval attestations of llanta meaning 'plant', though planta was normally used in this sense, while llanta was used in Castile with the narrower meaning 'cabbage'. Doublets of this sort are a common outcome of incomplete lexical diffusion. 98. Plegar and pregar meant 'to nail' in early medieval texts. These forms appear to be doublets of llegar/chegar 'to arrive'. The conservative forms were once explained away as upper-class or urban forms (e.g., Piel 1931), or as later borrowings from Latin. Wright (1980) and Anderson (1992) have argued that doublets of this sort are a normal result of the incomplete lexical diffusion which seems to underlie the development of these clusters. 99. Malkiel (1963-1964) also suggests that particular results may have been influenced by a host of minor factors, including (unfortunately) avoidance of homonymie clash (clavo/cravo vs llave/chavé), and (more likely) dissimilation between palatals in the same word (llave vs. clavija, llanto vs plañir). 100. Lloyd's interpretation (i.e., that the change progressed from CI > CX > λ > tj) is the only one that is consistent with the linguistic outcomes presented here. The theories put forward by Malkiel (1983) and Repetti and Tuttle (1987) must therefore be rejected. Torreblanca (1990) argues similarly, but suggests that the initial consonant in each cluster first aspirated and was then lost. This seems unlikely in word-internal post-consonantal position. 101. Today Cantabria and eastern and western Asturias (but not central Asturias) show phonetic alternation between [1] and [r] in /CI-/ and /Cr-/ clusters (Zamora Vicente 1967: 138; Ñuño Álvarez 1996: 188). 102. Franco was a term used to designate anyone from outside the Peninsula, primarily persons of Gallic origin, but also including people from the areas of present-day England, Germany and Italy (Μοχό 1979: 263). 103. Alvar (1968) presents a detailed discussion of the linguistic effects of French settlers in Aragon. See Frago Gracia (1991a) and Imhoff (2000a) for discus-

Notes

291

sion of the effects of dialect mixing in Aragon. Imhoff (2000b) also compares possible effects of dialect mixing in Aragon and Castile. 104. Excepting, of course, forms with paragogic -e used in verse. 105. This argument broaches the interesting side question of the merger of /d/ and /δ/ (assuming these were indeed separate phonemes in Hispano-Romance). Alvarez Rodríguez (1996: 33, 39) argues that normal apocope occurred only in ungrouped apical consonants that were "originally voiced" and "not folly occlusive". Thus, the loss of /-e/ following /-d/ (< τ) could not take place until the consonant had spirantized. He hypothesizes that this is a 12th-century change, since he finds no evidence for apocope following this consonant until 1146 (in a document from Toledo). Torreblanca (1986a: 15) also points out that, up to the 12th century, Hispanoarabic transcribers of Spanish used the letters dal and δαί to represent the sounds [d] and [δ], respectively. In the 12th century, however, the transcribers begin to use these two symbols interchangeably. Torreblanca believes that this signals the loss of distinction between the two sounds. Certainly, such a merger would have been favored by the massive movement and mixing of the 12th century, but there is no orthographic evidence available, unless one sees this as part of a chain, the second lenition. In this case, the loss of intervocalic -d- ( nief, Lobo > Lop), this might lead one to conclude that the -d was still - at least variably - occlusive. Still, at the time there was no means of indicating orthographically the devoiced equivalent of /δ/. 106. Galmés de Fuentes argues against significant apocope in Mozarabic, since he finds that scribes normally made an effort to represent Romance words with the final vowel, at least in Toledo and Seville (1983: 71, 192). However, Peñarroja Torrejón (1990: 241) argues that apocope was prevalent in Valencian Mozarabic, which would support the pre-Invasion origin of apocope, at least in the east. 107. Some areas were slower to accept the new norm. Staaff (1907: 210) points out that in Leonese the tendency to maintain final -e after apical consonants was retained into the 13th century (e.g., sacare, abere). 108. Lapesa (1951: 188-190) also points out that Arabic borrowings and Mozarab names ending in final consonants could have contributed to these developments in the 10th and 11th centuries. However, he finds that most of the terms were regularized to the dominant phonotactic structure through the addition of final -e. Still, such borrowing continued unabated and may have also contributed to opening up the system to new word-final consonants and clusters.

292

Notes

109. There were clearly other constraints on appearance of apocopated forms. For example, there was an almost total lack of apocope in the 1st and 3rd persons singular of the present subjunctive of -ar verbs. Luquet (1992) offers a possible explanation for this restriction. 110. These texts date from the latter half of the 13th century, but they probably represent a traditional usage within the area of Toledo, what Hernández (1991 : 330) refers to as the norma culta of the time - although this norma was clearly variable. 111. It seems reasonable to conclude that such alternation would have represented the penultimate stage to the loss of apocopated forms, for, as Harris-Northall points out: "In such circumstances, it is questionable from a phonetic point of view whether apocope had actually taken place; we might just as accurately speak of coalescence of vowels, a common enough occurrence in the history of Spanish, and, of course, in the modern spoken language, though not normally reflected in the standard orthography" (Harris-Northall 1991: 34). The results of Harris-Northall's analysis discontinued this hypothesis. 112. Harris-Northall excluded apparently phrase-final tokens from his tabulations. 113. Moreno Bernal (1993), in his study of Escorial Bible 1-1-6, reports a significant correspondence between phonetic context and apocope: "dans des conditions favorables (devant une pause ou une voyelle), nous trouvons plus de 75% des cas d'apocope, tandis que, dans ces conditions moins favorables (devant une consonne), nous relevons précisément le contraire: 75% des cas de conservation de la voyelle finale" (1993: 195). This still leaves in need of explanation some 25% of instances of apocope or lack thereof. Going beyond Lapesa's observation that apocope was probably more frequent in writing, Moreno claims that apocope was completely restricted to written language and the variation between full and apocopated forms simply represents conscious manipulation of the rhetorical device of variatio (1993: 201). While this last may be true (and is conceivable within the approach I take), his rejection of apocope as a phenomenon of the spoken language stands in clear contradiction to his own assertions that phonetic context was an important factor in determining the appearance of apocopated forms. In addition, such a claim fails to consider an important phonetic effect of apocope: the devoicing of final consonants (e.g., nief for nieve). 114. Staaff (1907: 213) observed in 13th-century Leonese documents that apocope occurred principally in eastern regions near Castile, but was little used in central and western Leonese. Lapesa pointed out, however, that "en los comienzos de la apocope extrema los ejemplos leoneses no son menos que los castellanos . . . aunque luego escaseen" (Lapesa 1951: 186). It may be that a simple relation (French influence > extreme apocope) was valid for northern Leon, but not for Castilian zones.

Notes

293

115. Personal leísmo is the only one of those related to leísmo, laísmo or loísmo that today receives the approval of the Royal Spanish Academy (Abad 1985: 14). The prestige associated with this acceptance by the Academy has apparently contributed to its modern spread, at least in formal registers, to regions which traditionally maintain the etymological pronoun system (Quilis et al. 1985: 37). 116. Interestingly, the narrow area of primitive Castile showing system H (including the town of Espinosa de los Monteros) once formed an administrative unit with the neighboring Cantabrian Valle de Pas, which also shows system H (Penny 1969: 27). 117. System A is modified even further by some speakers (Fernández Ordóñez 1994: 86). This system, labeled system A', results from the loss of marked use of lo for feminine mass reference (a tendency of variable strength in nearly all the systems where it exists), and a concomitant loss of its use for masculine mass reference. System A' thus loses completely both case and count/mass distinctions, and the pronoun lo disappears except for neuter reference. This is the most altered or advanced of any of the pronoun systems, but given that it clearly arises as an adaptation of System A, it will not be further discussed here. 118. Examples of variable constructions include those in which differing degrees of agentivity are expressed by use of the accusative or dative: Aquellos amigos lo interesaron en la política 'Those friends interested him in politics.' vs. A Juan le interesa la política 'Politics interest Juan'. Certain constructions with infinitive clauses can also affect pronoun use, depending on whether or not the infinitive itself takes a direct object: Su madre no la dejó subir 'Her mother didn't let go up' vs. Su madre no le dejó conducir el coche 'Her mother didn't let her drive the car'. Some verbs can omit their direct object, though it is understood from the context: Los enseña '(s/he) teaches them (=the facts)' vs. Les enseña '(s/he) teaches them (= the children) (the facts)'. In Northern Spain the verb llamar takes the accusative for permanent names, but the dative for nicknames and such: Cuando nació mi hija, la llamamos María 'When my daughter was born, we called her Maria' vs. Aunque se llama María, todos le llaman Irene 'Although her name is Maria, everybody calls her Irene'. Significantly, Fernández Ordóñez (1999: 1335) indicates that use of the accusative has been generalized to both contexts in Zamora, Salamanca and all of southern Spain. 119. The best potential source of this information (at least for geographical distribution in the late 12th and 13th centuries) is the Documentos lingüísticos of Menéndez Pidal. However, as Echenique Elizondo (1981: 135) and Sanchis Calvo (1991: 811) have pointed out, these texts offer relatively few examples of /', lo or le, and are thus unusually poor sources of information on leísmo. This probably stems from the need for clarity of reference in legal texts, which

294

Notes

leads to an extreme redundancy in which full names and nouns are repeated and the pronouns le and lo avoided. 120. The exact source of this form is not known. Atonic articulations of the proposed adjectival forms may have favored the weakening of the final vowel. But raising of final -a is a well-known feature of central Asturian (la vaca > les vaques), and it may have retained a place in the earliest Castilian koine (giving mies, then mie as a backformation). On the other hand, the medieval variation between -ia and -ie in the imperfect verb forms may reflect a broad phonetic trend extended even to possessive forms, producing variants mía-míe, which in turn may have affected the development of túa-túe, súa-súe. 121. The relative lack of early documentation of second person forms requires that discussion of these forms be based on the evidence for the third person. Presumably the second person forms would have followed analogically the development of the third person. 122. An interesting question, not yet investigated, is the origin of the loss/weakening of phonological distinction between non-high vowels /e-i/ and /o-u/. It may be that this was yet another result of Burgos phase koineization but which Latin orthography, literacy/standardization, and the pattern of tonic vowel distinctions have tended to erase in recent centuries. See Penny (2000: 133-134) for discussion. 123. For the documents in Orígenes Pidal commented: 'No he recogido ejemplos de la preferencia por la forma femenina mi, su que prevaleció en castellano' (1964: 346). The leveling to su(s) and tu(s) appears to be a Toledan innovation. 124. The abbreviations are those established by the Hispanic Seminary of Medieval Studies (manuscript production dates are included in parentheses) GE4 = General Estoria IV (1280); MOA = Moamyn - Libro de las animalias (1250?); EE1 = Estoria de España I = Primera Crónica General (1270-1284); GEI = General Estoria I (1272-1275); LEY = Libro de las leyes (12567-1265?); CRZ = Libro de las cruzes (1259); LAP = Lapidario de Alfonso Χ (1250?1279?). 125. For example, LEY includes 5 tokens of mio + masculine noun, and 9 of tokens oîmi(s) + feminine noun. There are no cases of confusion. 126. Scholars have recognized but do not agree on the causal relation between shortening and deaccentuation. With specific reference to the 1st person, Malkiel comments: "The shortening was, in turn, conducive to loss of stress in tone-setting New Castile (hence, in standard Spanish, proclitic mi-padre), with the further consequence that the definite article before the possessive adjective became dispensable, except in archaic formulas." (1976: 473). Penny, on the other hand, implies that the loss of tonicity was key: "When used adjectivally before a noun, the possessives lose their tonicity, shed their final vowels, and (in the case of to(s), so(s)) suffer raising of their newly atonic vowels' (1991a:

Notes

295

127). These phenomena may have been simultaneous and mutually reinforcing in the koineizing context. Penny (1969: 114) provides some evidence of the impact that stress could have. In Cantabria (Valle de Pas), where stress on the possessives was maintained, all the possessives were eventually leveled to long forms for all functions: míyu, míya, míyu, túya, túyu, túyif, etc. 127. The value of the reconstruction and explanation for Castillan possessives can be highlighted by briefly comparing them to developments in neighboring Aragon and Leon. In Aragon, where massive demographic mixing also occurred during the 12th century, there is documentation that reveals complete leveling to so; the Liber Regum (Cronicón Villarense), the manuscript of which dates from before 1211 (Cooper 1960: 7), contains only the form so(s), used for both masculine and feminine reference, with 33 cases of so + feminine (Cooper 1960: 88-89). Gallic and Catalan influence may have been the deciding factor in Aragon. These immigrants, arriving in the 11th century, would have used forms such as masculine son/ton, whose vowel corresponded to that of the Hispanic form so, even as the feminine forms they employed favoring -a - tended to weaken the frequency and consistency with which the equivalent Hispanic forms were used. For Leon, Staaff (1907: 273-274) reports that 13th-century northeastern Leonese documents reveal limited evidence of leveling: 3 occurrences of sos + feminine, but 4 cases of sues + feminine and 5 cases of sus + feminine. The leveled forms to so(s), to(s) later predominated in central Leonese and spread to central and eastern Asturian, where they survive today (Menéndez Pidal 1962a: 96). In Asturian and northern Leonese, the broader association of -u with masculine reference and -o for neuter reference may have contributed to the selection of so and to as a lessmarked variant. 128. Basic events and dates referred to here can be found in the standard histories on the topic. I have relied primarily on Μοχό (1979: 349-382), González Jiménez (1982: 97-127), and Bishko (1975: 422-439). 129. Properties were generally distributed as donadíos or as heredamientos. Donadíos were generally made by the king, primarily to noble lords or ecclesiastical institutions as vassals, who then became responsible for overseeing the repopulation of these areas. Heredamientos were generally smaller, went to those lower on the social scale ( /λ/ en español. Revista de filología española 70: 317-327. Torres Fontes, Juan, Julio González González, Salvador de Moxó, and María Paz Alonso Romero 1990 Historia de España Menéndez Pidal: La expansión peninsular y mediterránea (c. 1212 - c. 1350). Vol. I: La Corona de Castilla. Madrid: Espasa-Calpe. Trask, R. L. 1997 The History of Basque. London: Routledge. Trask, R. L., and Roger Wright 1988 El 'vascorrománico'. Verba 15: 361-373. Traugott, Elizabeth 1977 Pidginization, creolization and language change. In Pidgin and Creole Linguistics, Albert Valdman (ed.), 70-98. Bloomington: Indiana University Press. Trudgill, Peter 1986 Dialects in Contact. Oxford: Blackwell. 1989 Language contact and simplification. Nordlyd 15: 113-121 1992 Dialect typology and social structure. In Language Contact: Theoretical and empirical studies, Ernst Hâkon Jahr (ed.), 195— 211. Berlin/New York: Mouton de Gruyter. 1994 Language contact and dialect contact in linguistic change. Dialektkontakt, spràkkontakt och spräkßrändring i Norden: Föredrag frân ett forskarsymposium, Ulla-Britt Kotsinas and John Helgander (eds.), 13-22. Stockholm: Stockholm University. 1996 Dialect typology: Isolation, social network and phonological structure. In Towards a social science of language: Papers in Honor of William Labov, Vol. 1, Gregory Guy, Crawford Feagin, Deborah Schiffrin, and John Baugh (eds.), 3-21. Amsterdam/Philadelphia: John Benjamins. 1998 The chaos before the order: New Zealand English and the second stage of new-dialect formation. In Advances in Historical Sociolinguistics, Ernst Hâkon Jahr (ed.), 197-207. Berlin/New York: Mouton de Gruyter.

References 2002

329

Linguistic and social typology. In The Handbook of Language Variation and Change, J. K. Chambers, Peter Trudgill, and Natalie Schilling-Estes (eds.), 707-728. Oxford: Blackwell. Trudgill, Peter, Elizabeth Gordon, and Gillian Lewis 1999 New-dialect formation and Southern Hemisphere English: The New Zealand Short Front Vowels. Journal of Sociolinguistics 2: 35-51. Trudgill, Peter, Elizabeth Gordon, Gillian Lewis, and Margaret Maclagan 2000 Determinism in new-dialect formation and the genesis of New Zealand English. Journal of Linguistics 36: 299-318 Trudgill, Peter, Gillian Lewis, and Margaret Maclagan 2000 The role of drift in the formation of native-speaker southern hemisphere Englishes: Some New Zealand evidence. Diachronica 17(1): 111-138. Tusell, Javier, Carlos Martinez Shaw, and José Luis Martin 1998 Historia de España. Madrid: Taurus. Tuten, Donald Ν. 2000 Linking social change and linguistic change: Koineization in early Castile. In New Approaches to Old Problems: Issues in Romance Historical Linguistics, Steven N. Dworkin and Dieter Wanner (eds.), 97-105. Valle, José del 1998 Andalucismo, poligénesis y koineización: dialectología e ideología. Hispanic Review 66: 131-149. Versteegh, Kees 1993 Leveling in the Sudan: From Arabic creole to Arabic dialect. International Journal of the Sociology of Language 99: 65-80. Wagner, Max 1920 Amerikanospanisch und Vulgärlatein. Zeitschrift fur Romanische Philologie 40: 286-312, 385^04. Wang, William S-Y. 1969 Competing changes as a cause of residue. Language 45: 9-25. Weinreich, Uriel 1968 Languages in Contact. The Hague: Mouton. Weinreich Uriel, William Labov, and Marvin Herzog 1968 Empirical foundations for a theory of language change. In Directions for Historical Linguistics, Winfred P. Lehman and Yakov Malkiel (eds.), 95-189. Austin: University of Texas Press. Whinnom, Keith 1980 Creolization in linguistic change. In Theoretical Orientations in Creole Studies, Albert Valdman and Arnold Highfield (eds.), 203-212. New York: Academic.

330

References

Willemyns, Michael, Cynthia Gallois, Victor J. Callan, and Jef&ey Pittan 1997 Accent accommodation in the job interview. Impact of interviewer accent and gender. Journal of Language and Social Psychology 16: 3-22. Wolfram, Walt, and Natalie Schilling-Estes 1998 American English: Dialects and Variation. Oxford: Blackwell. 2003 Dialectology and linguistic diffusion. In The Handbook of Historical Linguistics, Brian D. Joseph and Richard D. Janda (eds.), 713-735. Oxford: Blackwell. Wright, Roger 1980 Linguistic reasons for phonetic archaisms in Romance. In Papers from the 4th International Conference on Historical Linguistics, Elizabeth Closs Traugott, Rebecca Labrum, and Susan Shepherd (eds.), 331-337. Amsterdam/Philadelphia: John Benjamins. 1982 Late Latin and Early Romance in Spain and Carolingian France. Liverpool: Francis Cairns. 1994a Early Ibero-Romance. Newark, Delaware: Juan de la Cuesta. 1994b How old is the ballad genre? In Early Ibero-Romance, 290-299. Newark, Delaware: Juan de la Cuesta. 1994c Logographic script and assumptions of literacy in tenth-century Spain. In Early Ibero-Romance, 165-180. Newark, Delaware: Juan de la Cuesta. 1994d Metalinguistic change in medieval Iberia. In Early IberoRomance, 31-44. Newark, Delaware: Juan de la Cuesta. 1994e The purpose of the glosses of San Millán and Silos. In Early Ibero-Romance, 209-219. Newark, Delaware: Juan de la Cuesta. 1994f Sociolinguistics in Spain (8th—11th centuries). In Early IberoRomance, 155-164. Newark, Delaware: Juan de la Cuesta. 1996 Latin in Spain: Early Ibero-Romance. In The Origins and Development of Emigrant Languages: Proceedings from the Second Rasmus Rask Colloquium, Hans Nielsen and Lene Schosler (eds.), 277-298. Odense: Odense University Press. 1999 Periodization and how to avoid it. In Essays in Hispanic Linguistics Dedicated to Paul M. Lloyd, Robert J. Blake, Diana L. Ranson, and Roger Wright (eds.), 25-41. Newark, Delaware: Juan de la Cuesta. Yaeger-Dror, Malcah 1993 Linguistic analysis of dialect "correction" and its interaction with cognitive salience. Language Variation and Change 5: 189-224. Zamora Vicente, Alonso 1967 Dialectología española. 2nd ed. Madrid: Gredos.

References

331

Zufferey, François 1987 Recherches linguistiques sur les chansonniers provençaux. Geneva: Droz.

Index

Abad, Francisco, 293 abduction, 58-60, 125 accentuation, 4,11, 12, 28, 34, 113, 119, 120, 122, 123, 127, 131, 133, 138, 152, 164, 168, 169, 171, 196, 205-208, 212, 213, 240, 241, 257, 260, 289, 294 accommodation, 22, 28-33, 35, 37, 38.41.47, 48, 50, 53-55, 57, 59, 62, 68-70, 75, 78, 88, 90, 112, 119,124, 130, 143, 164, 172, 187, 188, 201,202, 239, 257, 264, 274, 277-279, 281, 282 acquisition, 2, 13, 21, 26, 28, 30-34, 36, 38,40, 43,45, 46, 48, 50, 54-66, 68, 70, 74, 80, 85-87, 90, 92, 96, 101, 105, 112, 117, 119, 123, 125, 126, 133, 134, 143, 153, 155, 157, 178, 179, 183, 198,213,219, 231,234, 237, 253, 261, 264, 267, 268, 274, 277-281, 287 actuation, 113, 131, 191, 192, 207 adolescents, 31, 34, 56, 62, 64, 65, 67, 68, 71, 75, 90, 164, 257, 280, 282 adults, 18, 26, 27, 31, 32, 34, 35,40, 41.47.48, 54, 56-58, 61-71,74, 75, 77, 90, 93, 118, 119, 124, 125, 130,213, 231,239, 244, 256, 257, 264, 267, 268, 279281, 301 Alarcos Llorach, Emilio, 123, 124, 285, 289 Albalá, María José, 293

Alcarria, La, 147, 150, 200, 245 Alfonso X, 183, 207, 216, 222, 231, 234-237, 294 Alonso Romero, Maria Paz, 283 Alonso, Amado, 223-225, 255 Alvar, Manuel, 116, 222, 290, 298 Alvar, Manuel, 174 Álvarez Maurin, Maria del Pilar, 113 Álvarez Rodriguez, Adelino, 163, 291 Allen, J. D. H., 168, 169 allophone, 46, 65, 135, 136, 296 Amery, Rob, 71 analogy, 60, 125, 130, 135, 143, 182, 190, 191, 194, 203,205, 206, 208, 209,212,213,241, 258, 262, 263, 289 Andalusia/Andalusian(s), 5, 7, 17, 68, 85, 121, 143,152, 175, 183, 185, 188, 195,202,215-256, 263, 275, 276, 295, 296, 297, 298, 299, 300 Andersen, Henning, 58-60, 80 Anderson, James M., 138, 290 animacy, 181-184, 186, 187, 190192 apocope, 157, 159-173, 184, 186, 194-199, 205, 212-214, 232, 233-237, 243-245, 256, 260265, 291, 292, 297, 298, 300, 301 Arabic, 15, 18, 22, 70, 120, 147, 285,291 Aragon/Aragonese, 97, 104, 110, 114, 116, 117, 120-122, 125-

Index 129, 131, 136, 139, 147, 148, 150, 152, 154, 158, 164, 173, 186,188,195,199,200,209, 220, 221, 233, 260, 261, 275, 276, 286, 287, 290, 295, 301 Ariza, Manuel, 230, 231, 242, 252255, 298-300 articles, 94, 113-119,131, 132,138, 143, 155, 156, 175, 203, 205, 208, 212, 255, 258, 260, 275, 285-287, 294, 301 aspiration, 5,47, 110, 132, 133, 135, 230, 288, 290, 296 assimilation, 35, 118, 131,136,137, 171, 254, 277, 286, 287, 289 Asturias/Asturian(s), 95-99, 103, 105, 121, 123, 128, 129,139, 140, 142, 143, 147, 148, 151, 163,176, 177, 193,195, 206, 221, 276, 287, 288, 290, 294, 295 Azevedo Maia, Clarinda de, 234 Badia Margarit, Antoni, 125, 139 Bailey, Charles-James N., 15 Bamgbose, Ayo, 15 Barrios, Ángel, 99, 142, 200 Bartsch, Renate, 49, 50 Barz, Richard K., 3 Basque(s), 87, 93, 95-98, 100, 103, 108, 109, 123, 124, 132-134, 136, 143, 144, 175, 177-179, 181, 182, 191-193, 198,221, 243, 259, 277, 283-285, 288, 289, 300 Beebe, Leslie M., 30 Bell, Allen, 278 Bemba, 15 Berceo, 184, 185, 188

333

Berruto, Gaetano, 16 Bezler, Francis, 161 Bidwell, Charles E., 15 Bishko, Charles Julian, 145, 151, 152, 295 Blanc, Haim, 15,18, 22, 70, 89, 277 Blanchet, Philippe, 276 borrowing, 22, 24, 25, 29, 62, 86, 88,91, 108, 133, 134, 170, 171, 191, 192, 243, 291 Bortoni, Stella Maris, 3,40 Britain, David, 43, 44, 46, 71-73, 281 Bubenik, Vit, 10-13, 276 Buck, Carl Darling, 10, 12 Burgos, 4, 5, 7, 93-145, 149, 151153, 155, 159, 162-164, 167, 170, 177, 179-181, 188, 193, 195,197-199, 204-207, 210212,214, 220, 235-237, 239242, 245, 258-264, 284-287, 289, 294, 297, 299, 301 Bybee, Joan, 57, 61, 196,281 Callan, Victor J., 278 Callary, Robert E., 82, 159 Canadian Raising, 44, 72 Cannell, C., 278 Cano Aguilar, Rafael, 301 Cantabria/Cantabrian(s), 93, 95, 96, 98, 99, 103, 105, 107, 132, 135137, 139, 140, 142, 143, 148, 151, 176-180, 186, 191-193, 195, 196, 198, 220, 259, 261, 285, 287, 290, 293, 295 Cantarero, Margarita, 293 Cardona, Giorgio Raimondo, 13, 14, 16

334

Index

case, 6, 8, 11, 20, 23, 25-27, 29, 33, 35, 37,40, 42-46, 49, 50, 53-55, 58, 59, 66, 67, 69-77, 80, 82, 85, 86, 88, 89,91,93,105-107, 110, 112, 113, 119, 122, 126, 130, 132, 137, 144, 148, 156-158, 164-166, 169, 172-183, 185192, 194,195,197-199, 201203, 220, 221, 228, 232, 243, 244, 246, 254-256, 259-264, 268, 274, 275, 277, 278, 281, 283,284, 290, 291,293, 294, 300 Castile, primitive, 96, 98, 99, 105, 116, 135, 159, 181, 191-193, 195-198, 288, 293 Catalán, Diego, 122, 167,199, 225, 287, 289 Catalonia/Catalan(s), 95, 97, 150, 221, 275 catastrophic/punctuated change, 14, 262, 265, 274 ceceo, 5, 224, 231, 250, 251, 296, 298, 299 Clark, Eve, 119, 198 Clavería Nadal, Gloria, 108 code-mixing, 78, 282 code-switching, 55, 69, 282 Coleman, L., 278 consonant clusters, 139, 140, 166, 290 consonants, 11,12, 35,44,47,131, 137, 138,140, 141, 157, 160, 161, 163, 165, 166, 168, 169, 197, 205, 223, 259, 286, 289, 290-292 constraints, 2, 18, 31, 32,48, 50, 78, 91, 92, 165-167, 169, 233, 258, 263, 281,292

contact between stable dialects, 28, 38, 79 convergence, 29,76, 88 Cooper, Louis, 199, 295 Cooper, Robert L., 155 Corbett, Noel, 125 Coraminas, Joan, 138, 140, 142, 290 count/mass distinction, 157, 175177, 179,180, 181, 182, 184, 186,187, 188, 191, 192, 197, 199, 201,203, 261,280, 293 creole/creolization, 22, 24-27,40, 67, 74, 76-78, 87, 109, 267, 282, 285, 301 Croft, William, 37, 49, 75, 76, 274 Crystal, David, 17 Cuervo, Rufino José, 185, 189, 194 cuñalwedge, 127, 131, 153, 154, 301 Chambers, J. K., 31, 61, 83, 158, 159, 280 Cheshire, Jenny, 32, 279 children, 25-27, 31, 34, 36, 40, 48, 54, 56-58, 60-72, 75, 78, 82, 86, 87, 90,91,93, 118, 119, 125, 130, 134, 164, 171, 197, 198, 213, 239, 257, 264, 267, 268, 279-282, 293, 301 Chomsky, Noam, 49 deduction, 58, 60 DePaulo, Bella M., 278 Deutscher, Guy, 59, 60 Díaz Montesinos, Francisco, 253, 299 Diaz Roig, Mercedes, 103 Diaz y Diaz, Manuel C., 161 Diez, Miguel, 154 difusión (vs. focusing), 27, 39, 67 Dillard, J. L., 15, 17, 22, 81

Index dipbthong/diphthongìzation, 11, 38, 44, 69, 105, 113, 120-131, 133, 135,143,156,205,213,258, 260, 288, 289, 301 diphthongizatíon before yod, 126— 129,131,288, 289 dissimilation, 205, 254, 290 Dixon, R. M. W., 262 Domingue, Nicole, 43, 48 Dorian, Nancy, 87 drift, 35, 92,117, 264, 265, 301 Eberenz, Rolf, 265, 266 Echenique Elizondo, Maria Teresa, 184-188, 194, 293 Elcock, W. D., 16 Ellis, Rod, 31 English, Australian, 35,44, 82, 262, 281 English, Belfast, 44,45, 50-52, 279 English, Canadian, 16, 31,44,72, 81, 83, 280 English, English, 4, 15, 31-33, 37, 63, 65, 67, 69, 72, 80-82, 221, 264, 279-281, 290, 296 English, New Zealand, 34, 35, 68, 71,81,281 English, Scottish, 44, 281 English, South African, 23, 35, 279 English, Southern Hemisphere, 34, 264,279, 296 English, U.S., 15, 81, 278 Ervin-Tripp, Susan M., 15 Estoria de España, 184,186, 211, 294 etymological system, 173-178, 181, 182, 186-188, 199, 200, 202, 214, 242, 245, 262, 297

335

evidence, 5, 6, 8, 22, 32, 34, 36,43, 44, 56, 60, 68, 71-73,79-81, 95, 98, 99, 106, 107, 111-114, 120122,125, 127-129, 133, 136138,140, 147, 148,151, 153, 154, 159, 161, 163, 165, 171, 182,183, 186, 188,190, 193, 203,207, 209,210,211,214, 215, 219, 225, 226, 228-234, 239, 240,243-248, 251,252, 253,254, 255, 256, 259, 261, 263, 264, 267, 268, 280, 281, 283-289, 291, 294, 295, 297299 exceptionalism, 104, 283 explanatory limits, 7, 93, 131-143, 255-256 external factors, 1, 189,191, 193, 266, 274 Extremadura, primitive, 147, 148, 200 F azienda de Ultramar, 185 Fehderau, Harold W., 15 Ferguson, Charles Α., 15, 17, 18, 22, 45 Fernández Ramírez, Salvador, 189, 201 Fernández-Ordóñez, Inés, 151,175195, 199-203, 243, 267, 293 Ferreira, Manuel, 115 feudalism, 153, 284 Flores Cervantes, Marcela, 190 focusing, 18-20, 22-24, 26-28, 36, 39-41,45,48,49, 53, 54, 57, 63, 64, 66-74, 92, 100, 103, 105107, 113, 114, 119, 123, 126, 134, 143, 155, 161, 167, 170, 172, 198, 204, 205,207,211,

336

Index

227,259, 260, 264, 279, 288, 300, 301 Fogny, 15 Fontanella de Weinberg, María Beatriz, 3, 26,47, 85, 89 Frago Gracia, Juan Antonio, 7, 8, 158, 226-231, 234, 246-255, 263, 290, 296-299 France/French, 4, 16, 17, 107, 118, 119, 146, 158, 170, 171, 173, 205, 221,233,237, 274, 287, 290, 292 francos, 147, 148, 149, 151, 152, 153, 170,171, 172, 214, 220, 221, 261, 265, 290 frequency, 16-18, 20, 26, 27, 30, 32, 33, 35, 37, 41-43, 4 5 ^ 8 , 51, 54, 55, 57-61, 63-65, 67, 68, 70,71, 74, 75, 77, 79, 80, 82, 86, 89, 90, 96, 100, 104, 105, 111, 112, 114, 119, 120, 122,124, 125, 128131, 133, 134, 136, 138, 139, 142, 143, 148, 149, 152, 155157, 161, 163, 164, 166-169, 172, 174, 181, 182, 184, 185, 187, 193, 196-198, 199, 202, 206, 207,211-213,218, 222, 230, 234-240, 245, 253, 255, 257-261, 275, 277, 279, 281, 282, 286, 287, 292, 295, 296, 298, 300 frontier, 96, 102, 126, 145-147, 149-151,202,215,216,218, 222, 284 fudging, 36, 38, 279 Fuero de Madrid, 162, 186 Galicia/Galician(s), 95-97, 99-103, 105, 115, 119, 120, 122, 124,

136, 139-143, 147, 148, 150, 151, 161, 163,200, 220, 221, 230,234,245,252,259,276, 285-288 Galmés de Fuentes, Alvaro, 120, 289, 291 Gallois, Cynthia, 278 Gambhir, Surendra K., 3, 15, 16, 23, 42, 87, 89 García de Cortázar, José Ángel, 283 García de Diego, Vicente, 107, 108 García González, Francisco, 191 García, Èrica, 190 gender, 162, 174-179, 192, 197, 204-207, 209-211, 213, 238240, 242, 260, 264, 277 General Estoria, 169, 186, 211, 236, 294 generation, 10, 21, 26, 27, 35,40, 50, 53, 60, 62, 64, 67-72, 80, 83, 84, 88, 90,91, 104, 125, 129, 137, 229, 234, 239, 259, 262, 264, 267, 282, 296 German, 15, 107 Giddens, Anthony, 282 Giles, Howard, 29, 30 Gimeno Menéndez, Francisco, 110, 111

González Gómez, Antonio, 218, 219, 220, 221, 295 González Jiménez, Manuel, 200, 218-222, 231,232, 295 González, Julio, 150, 201, 218, 219, 220, 283 Gordon, Elizabeth, 3, 33, 35, 42, 68, 71,81,245, 279,296 Graff, Willem Laurens, 16 grammarian, 13, 201, 202 Granda, Germán de, 3, 224, 229, 283

Index Granovetter, Mark, 51 Greek, 9, 11, 12, 13, 21, 24, 77, 276 Guaraní, 175 Guerra, Rafael, 293 Gumperz, John J., 88 Hall, Robert Α., 17, 276 Haller, Hermann W„ 15 Hancock, Ian F., 16 Hardwick, Margaret, 83 Harris, John, 47 Harris-Northall, Ray, 6,165,169, 170,172, 196, 235, 236, 253, 265, 266, 292, 299 Hartmann, R. R. K., 15 Haugen, Einar, 26, 84 Hebrew, Israeli, 15, 17, 22, 58, 70, 89, 277, 281 Henríquez Ureña, Pedro, 223 Henry, Alison, 60 Hernández, César, 292 Herzog, Marvin, 2 Hidalgo, Margarita, 3, 283 Hill, Trevor, 15 Hindi-Bhojpuri, 3, 15, 20, 22-25, 27, 38,41-43, 87, 89 Historia troyana, 186 Hock, Hans Heinrich, 10, 11 Hofstadter, Richard, 284 homogenization, 80, 82, 83,123, 155, 262 homonymie clash, 32, 192,278, 290 Hopper Paul, 61, 196 Horrocks, Geoffrey, 10-13,18, 275, 276 Heyanger, 40, 45, 67-69 Huber, Joseph, 115 hypercorrection, 37, 163, 197, 209 hyperdialectalism, 38, 278

337

identity, 13, 28, 29, 38, 39,45, 49, 50, 52, 54, 69, 70, 75, 77, 78, 83, 88,92,102-104, 111, 124,126, 143, 147, 157, 163, 198, 257, 259, 263, 277, 278, 284, 285 Imhoff, Brian, 290, 291 immigrant koine, 20, 21 impoverishment, 25, 87 Indonesian, 15 induction, 58, 60, 80, 90, 130, 258, 261

innovation, 23, 31, 36, 50, 51, 54, 56-61,78, 80, 83, 87, 111, 120, 143, 155, 158, 159, 165, 166, 170, 178, 181, 182, 194, 197, 198, 201,205, 207-211,231, 240, 241, 258, 259, 261-263, 275, 279, 285, 289, 294, 297 input/output, 18, 25,46, 49, 56, 58, 60-62, 63, 70, 74-77, 80, 81, 87, 90,118,119, 124-126, 130, 134, 166, 195, 197, 198, 213, 214, 240, 258, 260, 261, 287 intelligibility, mutual, 17, 24, 27, 29, 75, 77, 79, 90, 106, 143, 257, 268, 276 interdialect, 28, 36-38,42, 54, 61, 90, 106, 159, 202 interlanguage, 18, 38, 54-56, 70, 90, 134, 159,178 internal factors, 33, 191, 192, 301 invisible hand, 2, 47-50, 63, 64, 9092, 257 Iradiel, Paulino, 283 Italian, 4, 15, 16, 26, 134, 177, 221, 276, 290 Izzo, Herbert, 133 Jakobson, Roman, 14

338

Index

Janda, Richard D., 126, 286 jargon, 16,24 Jews, 147,149,219 Jonxis-Henkemans, Wilhelmina 207, 211 Kannada, 88 Karlgren, Bernhar, 15 Kasa, 15 Kasten, Lloyd, 207,211 Kauffeld, Cynthia J., 299, 300 Kaufman, Terrence, 25, 74, 75, 79, 86, 88,277, 282 Keller, Rudi, 2, 268 Kerswill, Paul, 3, 7, 19, 26, 27, 3135,40,46, 53, 61-73, 76, 80, 86, 89, 92,93, 229, 257, 264, 268, 274, 277, 279, 281,282 Kikongo, 15 Kiparsky, Paul, 57, 58, 280 Klein-Andreu, Flora, 175, 202 Knecht, Pierre, 17 Koine, Greek, 9-14, 16, 18, 20, 24, 275, 276 Krüger, Fritz, 121 Labov, William, 2, 6, 38, 112, 126, 262, 274, 278, 281 Lacarra, José Maria, 152 laísmo, 173, 174, 179, 180-183, 185, 187, 188, 192, 203,293 language death, 86, 87, 89 language missionary, 297 language shift, 21, 86, 89, 133, 301 language spread, 21, 140, 149, 153, 155, 158,219, 222, 260 Lantolf, James, 300 Lapesa, Rafael, 151, 154, 160, 162168, 170,171, 184, 185, 187-

189,194, 197, 225, 227, 233237, 243, 246-248, 254, 255, 274,291,292,298,299 Lapidario, 211, 294 Lass, Roger, 192,268, 274, 279 Latin, 3, 14, 15,87, 95, 119, 120, 122, 131, 132, 133, 135, 136, 138-140, 144,156,165, 174, 183, 185, 189, 196, 204-206, 259,266, 276, 288-290, 294, 296 least common denominator, 14, 38, 261 Lee, Everett S., 284 leísmo, 6, 68, 106, 157, 159, 162, 173-203, 214, 232, 242-245, 256, 260-265, 293, 297, 298 Lenneberg, Eric, 61 Leon/Leonese, 96-105, 110,111, 114, 115, 117, 120-123, 126129, 131, 136, 139, 140, 142, 143,145-151,154-158, 163, 164, 166, 167, 181, 187, 188, 197, 199, 200, 202, 206, 209, 216, 217, 220-223, 228, 230234, 237, 245, 260, 275, 276, 283-292, 295, 297, 298, 300, 301 LePage, Robert B., 18, 39,40, 83 leveling, 3-5, 28, 29, 41,45-47, 64, 68,74, 76, 86, 87, 90,91, 112, 114, 138, 160, 207-211,213, 223, 229, 238, 242, 258, 264, 275, 279, 280, 294, 295, 297 leveling, dialect, 13, 22, 33, 67, 68, 79-84, 86, 89,107, 126, 137, 153-159, 222, 223, 227, 228, 256, 260, 275,277, 282

Index Lewis, Gillian, 3, 33, 35,42, 68,71, 81, 245, 264, 279, 296, 301 lexical diffusion, 36, 62,129,130, 140, 144, 258, 290 lexicon, 26, 29, 32, 74, 75, 77, 88, 92, 140, 280, 298 Liber Regum, 199, 295 Libro conplido en los judizios de las estrellas, 186, 187 Libro de Buen Amor, 187, 188 Libro de las cruzes, 186, 211, 294 Libro de las leyes, 211, 294 Lightfoot, David, 57,280 Lingala, 15 lingua fianca, 14-21, 109, 285 Lipset, Seymour, 284 literacy, 55, 57, 84, 85, 118, 123, 173, 281,285,294 Lodge, Anthony, 26, 275 loísmo, 173, 174, 179, 181, 183, 185,187, 188, 203, 293 López Bobo, María Jesús, 187 López de Aberasturi Arregui, Ignacio, 298 López García, Ángel, 108, 109, 285 Lüdtke, Helmut, 118 Lunt, Horace G., 15 Luquet, Gilles, 292 Lyons, Christopher, 204, 205 Lleal, Coloma, 118, 242 Llórente Maldonado, Antonio, 175 Lloyd, Paul M., 122, 127, 132,134136, 142, 143, 241, 288-290, 299, 300 Macedonian, 15 Maclagan, Margaret, 33, 35,42, 68, 71,81,245, 264, 279, 296, 301

339

Madrid, 4, 5, 147, 149, 179, 182, 200, 289 Mahlum, Brit, 282 Malkiel, Yakov, 139, 140,142, 241, 290, 294 Mancha, La, 147, 150, 202, 245, 263 Manessy, Gabriel, 15 Marathi, 88 Marcos Rodríguez, Florencio 156 markedness, 25, 28, 38, 41, 42,46, 64, 77, 79, 80, 87, 119, 124, 130, 152, 155, 157, 172, 181, 186, 190,192, 193,197, 202, 213215,234, 237, 241,245, 256, 258-261, 264, 275, 282, 293, 295, 297, 301 Martin Martin, José Luis, 100, 156, 283 Martínez Alcalde, María José, 265 Martínez Shaw, Carlos, 100, 283 Marzys, Zygmunt, 17 Matute Martínez, Cristina, 186 McMahon, April M. S., 59 Meillet, Antoine, 13, 14, 111 Melanesian Pidgin, 15 Méndez Garcia de Paredes, Elena, 204, 206, 207, 213, 238, 240, 241, 297 Menéndez Pidal, Ramón, 100, 103107, 112-117, 119-123, 127, 128,131, 132, 138, 139, 153, 154, 160, 161, 164, 195-197, 204-207, 225, 241, 246, 258, 259, 274, 283-285, 288, 293295, 300 merger, 4, 5, 11, 17, 52, 130, 143, 150, 181, 199, 223, 245,246, 251,253, 255,263,264, 281, 291,295, 296, 298, 300

340

Index

Mestone, Rajend, 3, 18, 20, 21, 23, 26, 36, 279, 280 Mestre Campi, Jesús, 98 metaphor, 1, 9, 13, 14, 274 metathesis, 254 methodology, 7, 8, 9, 72, 89, 91, 93, 259, 263 military-religious orders, 146, 150, 153, 202, 222, 232, 245 Milroy, James, 1-3, 29, 45,49-52, 54, 105, 159, 268 Milroy, Lesley, 50-52, 54, 117, 124, 159, 265, 274, 277 Milton Keynes, 36,45, 46, 63-65, 67-69,78, 82, 89, 99,280,282 Minguez, José Maria, 283 Mitchell, J. Clyde, 53 Mitre Fernández, Emilio, 104, 283 mixing, 3,4, 5, 7, 12-14, 16-19, 21, 23-25, 27, 28, 35, 36, 38, 40,41, 44,45, 47, 50,51,72, 74, 79, 83, 86, 88, 90-93, 103, 105-108, 112, 113, 119, 128, 129, 133, 134, 139, 144-146, 148,149, 150, 152-154, 156, 158, 159, 164, 173, 184, 193, 199-202, 214,215,219-224, 227-230, 233, 234, 243, 245, 246, 256, 258, 259, 261, 263, 265, 274276, 279, 281,284, 287, 289, 291, 295, 296-298 Moag, Rodney F., 3 Moamyn, 211, 294 mobility, 52, 53, 80, 82, 101, 102, 153, 217, 262, 274 model, 3,4, 6-9, 22-28, 31, 38,43, 47,49, 51, 52, 56-58, 60, 65, 68, 71-73, 76, 77, 79, 89-91, 93, 113, 123, 131, 132, 136, 138,

155, 158, 203,212,215,227, 229, 256-258, 260, 267, 275, 277,278,281 Mohan, Peggy, 15 Mondéjar, José, 225, 230, 245,246 Monge, Félix, 190 monophthong, 69, 120, 130 Montgomery, Michael, 81 Montgomery, Thomas, 196 Moors, 93, 96, 101, 102, 104, 145147,215-217,219, 283 Morales, Francisco, 154 Moreno Bernal, Jesús, 292 Moreta, Salustiano, 283 Morgan, Michael, 299 morphosyntax, 6, 12,23, 26, 29, 32, 33, 35, 37, 46, 49, 58, 60, 62, 68, 70, 74-76, 85, 86,88, 111, 129, 193, 199, 207,213, 241,242, 243, 275, 279 Μοχό, Salvador de, 97, 98, 101, 102, 147-153, 200, 217, 219, 220, 221, 283, 290, 295 Mozarab(s), 96, 98-100, 105, 108, 120, 129, 143, 147-149, 152, 163, 164, 195,200, 201,214, 219, 265, 276, 283, 285, 289, 291, 298 Mudejars, 147, 148, 152, 214, 216, 219, 220, 296 Mufwene, Salikoko, 76-78, 261, 267,268, 274, 276, 282, 301 Mühlhäusler, Peter, 16, 24, 25, 45, 79, 276 nativization, 15, 16, 18, 23, 24, 26, 109 naturalness, 2, 21, 32-34,47, 49, 75, 134, 136, 137, 150, 259, 283

Index Navarre/Navarrese, 95, 97, 98, 100, 102,103, 110, 134,143, 147, 149-152,178,193,195,200, 220, 221, 289 Navarro Carrasco, Ana I., 298 Navarro Tomás, Tomás, 206 neutralization, 36, 206, 253, 255, 264, 296, 300 Nevalainen, Terttu, 274 new town, 3, 28,40, 63, 83, 90,153, 282 Nida, Eugene Α., 15 Nitti, John, 207, 211 non-prevocalic /-r/, 32, 35, 81 Nordenstam, Kerstin, 31, 32 norms, 18,20, 23, 27, 38^0,44, 48-54, 56-63, 65, 67, 68, 71, 72, 82, 85, 89, 90, 92,100, 102-106, 112, 113, 118,119, 120,123, 125, 130, 134, 137, 142, 152, 154-156, 158, 160, 164-166, 171-174, 182-184, 187, 193, 196, 198, 199, 201, 202, 206211,213,214, 229, 230, 232234, 236-238, 241, 243, 252, 257, 261-264, 268, 277, 278, 281, 282, 284-286, 288, 291, 301 Norway/Norwegian, 31, 37, 40, 282 nouns, 11, 98, 168, 175, 177,180, 182, 186, 192,199, 203, 206, 209, 210, 212, 238, 239, 285, 290, 294, 298 number, 6, 11,12, 32, 37,42,45, 51, 52, 54,71,73, 75,81,99, 100, 102, 110, 124,125, 129-131, 136, 143, 149, 165, 166, 171, 176,210,218, 221,229, 235, 236, 240, 251,254,255, 279

341

Ñuño Alvarez, Maria Pilar, 290 Ohala, John J., 35 Omdal, Helge, 3,40, 69 orthography, 12, 85, 113, 120, 122, 129, 136, 156, 157, 206, 232, 248,251-253, 259, 288, 291, 292, 294, 299 Oskenberg, L., 278 Ostos, Pilar, 232, 235, 238 overgeneralization, 37, 55-57, 59, 60,128, 136, 152, 198,210,212, 223,227, 228, 241,258, 259, 261, 280, 281 palatalization, 131, 137-140, 289 Palmer, J. M., 278 Palmer, Leonard R., 10,11,12 paraphrase, 11, 54, 55 Pardo, Maria Luisa, 232, 235, 238 Pascual, José Antonio, 138, 140, 142, 255, 290 Pastor Díaz de Garayo, Ernesto, 99, 283, 284 Payne, Arvilla, 281 Pei, Mario, 15, 17 Pellegrini, Giovanni Battista, 16 Penny, Ralph, 4-6, 46, 76, 106, 111, 132, 135, 138, 139, 195,213, 265, 285-287, 289, 293, 294, 297, 301 Pensado, Carmen, 287 Peñarroja Torrejón, Leopoldo, 121, 291 Pérez de Urbel, Justo, 283 periodization, 5, 265, 266 personal a, 80, 110, 178, 186, 188, 189, 198 Petrini, Dario, 19, 26

342

Index

phoneme, 32, 52, 55, 66, 123-125, 127, 130, 132, 136, 137,144, 246,247,253,255,258,259, 298,300 phonologizatìon, 123, 125,126,129, 133, 137, 138, 142, 144, 288, 301 phonology, 6, 23, 26,29, 32, 33, 37, 43,44,46,47, 62, 69,70,72, 74-76, 86,91, 112, 119, 120, 122, 123, 126, 130, 134,143, 161, 163, 164, 168, 169, 205, 229, 242,280, 288, 294, 296, 298 phonotactics, 32, 165-167, 172, 197, 254, 291 pidgin/pidginization, 22, 24-27, 67, 74-78, 89, 267, 276, 277, 285, 301 Piel, Joseph, 290 Pittan, Jeffrey, 278 Portuguese, 40, 118-120, 136, 139, 140-142, 148, 161, 234, 286, 290 Posner, Rebecca, 111, 130,137, 259, 261 possessives, 43,159,204-214, 232, 233, 238-242, 245, 256, 260, 263, 265, 294, 295 Pottier, Bernard, 174 prekoine linguistic pool, 18, 23, 24, 26, 27, 29, 35, 38, 39, 43,47, 52, 55, 56, 60, 65, 76, 80, 85-87, 90, 100, 124, 133, 136, 137, 143, 172, 193, 198, 231,257, 260, 274, 286, 300 prepositions, 11, 93, 113-117,119, 131, 132, 138, 143, 155, 156,

210, 235, 258, 260, 275, 286, 287, 301 prestige, 9,10, 52,53, 85,101,104, 106, 123, 152, 155, 156, 171, 172,174, 176,182, 189, 201, 230, 245, 261, 276, 279, 285, 288, 289, 293 pronouns, demonstrative, 189,190, 191,198, 203 pronouns, object (see also leísmo), 5, 173-203, 232, 233, 242, 294 prototype, 9,18, 24, 27, 54, 74, 77, 79, 83, 84, 86, 89, 92,102, 144, 190, 257, 259, 262, 265, 267, 275, 277, 296 Provençal, 16, 276 Quechua, 175 Quilis Merin, Mercedes, 265 Quilis, Antonio, 293 rate of change, 229, 267, 268 Ravid, Dorit Diskin, 57,70, 281 reallocation, 29, 3 6 , 4 3 ^ 5 , 68, 69, 72, 91, 159-161,172, 173, 204, 207-209,211,213,214, 233, 238, 240, 241, 260, 265, 282 reanalysis, 37, 60, 123, 125, 126, 130, 132, 135,137, 143, 161, 163, 164, 166, 173, 183,191, 192, 194,195, 197,199, 202, 203, 212, 214, 258-261, 288 reconstruction, 22, 91,136, 137, 199, 202, 257, 285, 295 reduction, 3, 4, 14, 16, 19, 21-23, 25, 28, 38,41,45,46, 52, 60, 70, 74, 87,91, 113, 114, 119, 123, 125, 126, 130, 143, 159, 161,

Index 196, 206, 209, 239-241, 245, 246,256,277,279,286 referential system, 175,176,178180, 182, 183, 186, 190,199, 203, 243 regional koine, 16, 19, 82, 84, 276 regional standard, 16 regularity, 11, 25, 35,44-^7, 57, 58, 71,76, 90,91, 114, 119, 127129, 137-142, 157, 160-162, 167, 172, 173, 176, 181, 188, 190, 196, 199, 202-204, 206208, 213, 239-241, 257, 258, 260, 262, 263, 267, 276, 289, 290,291,299 rekoineization, 7,28, 161, 214,238, 256 Repetti, Lori, 290 Ridruejo, Emilio, 95, 106, 111, 155 Rioja/Riojan(s), 98, 104,108, 114, 116, 118, 139, 143, 148, 182, 184,188, 193,200, 202, 235, 236, 284, 287 Riu Riu, Manuel, 283 Rivarola, José Luis, 283 Roland/rolling distinction, 46, 65 Romance, 4, 16, 87, 93, 95, 98,100, 103, 107, 108, 110-112, 114, 118-120, 122-125, 127, 129, 130, 132-138, 142, 143,147, 156, 160-162, 172, 173, 177179,185,190-193,195, 196, 203-205, 207, 210, 258,259, 265, 266, 275,284, 285, 288, 289, 291 Sabaté, Flocel, 98 Sabin, Ángel, 154

343

salience, 28, 29, 32-35,47, 50, 5860, 62, 65, 68, 73, 77,90, 125, 136,157,183,198, 205, 206, 257, 260, 263, 264, 277-279, 280 Salvador Salvador, Francisco, 296 Samarin, William J., 15, 22 Sánchez Albornoz, Claudio, 100, 101, 104, 283, 284 Sánchez Rodríguez, Marciano, 156 Sanchis Calvo, María del Carmen, 184-186,194 Sandve, B. H., 40 Sanga, Glauco, 17 Sanz Fuentes, María José, 232 Sarasa, Esteban, 283 Saussure, Ferdinand, 1,49 Schilling-Estes, Natalie, 81, 82, 279, 283 Schlieben-Lange, Brigitte, 285 Scholtmeijer, Harm, 86 Schiirr, Friedrich, 288 Selinker, Larry, 38, 54, 55 ser/estar, 109, 131, 285 Serbo-Croatian, 15 seseo, 5, 8, 215,224-226, 229-233, 245-256, 263, 296, 298-300 Seville, 4-7, 121, 195, 204, 207, 215-256, 264, 265,275, 291, 296, 298-300 sibilants, 4, 230, 247, 248, 252-255, 263, 295, 296, 299, 300 Siegel, Jeff, 3, 7, 14,15-20, 23-28, 33, 34,42-44,47,48, 51, 74, 77, 89, 257,276, 277, 280 sigma, 232, 247, 250, 254, 300 Silva-Corvalán, Carmen, 87, 301 simplification, 3, 10, 15, 16, 19, 22, 23, 25,26, 28, 29, 35, 45^18, 52,

344

Index

53, 56, 58, 61, 62, 64, 65, 68,70, 73, 74, 80, 85-87, 90, 91, 93, 109,112-114,118,119,126, 130, 132, 138, 155, 158, 159, 203, 204, 207, 210, 213, 214, 223, 230, 233, 238, 240, 241, 245, 246, 256, 258, 260, 261, 263-265, 267, 279, 280,286 Slavish, 15 Slobin, Dan, 57, 281 Smith, John Charles, 274, 287 Sneddon, Clive, 274, 287 social networks, 39,41, 45,48-53, 56, 57, 60, 63, 64, 66, 67, 71, 76, 79, 80, 82, 83, 88, 90,92, 96, 100-103, 111, 117, 119,129, 143, 158, 171, 173,214, 238, 241, 256, 257, 262, 263, 279, 284, 297 Spanish, American, 3, 6,47, 85, 87, 92, 223, 224, 246, 283 speaker activity, 2, 3, 29,47-50, 91, 92, 99,112, 119, 257, 268, 280 Staaff, Erik, 115, 120, 129, 234, 285, 291,292, 295, 297 stages, 4-6,23, 26-28, 54, 81, 100, 101, 105, 118, 122, 129, 142, 178,192, 206, 213, 222, 234, 264, 279, 280, 282, 301 standard ideology, 105, 267 standard/standardization, 6, 10, 12, 13, 15-18, 20,21,23,26, 27,31, 37, 39,41,50, 55, 60, 70, 79, 83-86, 89, 105-107, 114, 118, 123, 128, 132, 176, 242, 266, 267, 275, 276, 279, 281-283, 292, 294, 295 Steinsholt, Anders, 297

stereotype, 32, 33, 50, 178, 257, 263, 264, 278 Stork, F. C., 15 strategy of neutrality, 69, 275, 282 style, 23, 29-31, 36,43, 44,49, 69, 80,91, 153,154, 159, 173,204, 208, 210-214, 233, 238, 240, 260, 263, 264, 278, 280, 284, 290 substrate, 18, 21, 70, 86, 87,109, 110, 123, 124, 132, 133, 243, 274, 276 Swahili, 15 swamping, 279 syncope, 161, 162,165-168,196, 240 Tabouret-Keller, Andrée, 39,40, 83 Tagliavini, Carlo, 14 target, 25, 31, 34,42, 54, 56, 67, 74, 77, 86, 134, 230, 245, 277, 279 Tarone, Elaine, 55 teleology, 2, 192 Terrado Pablo, Javier, 253, 254 Thomason, Sarah, 25, 74-79, 86, 88, 277, 282 Thomson, George, 10-12 Tilander, Gunnar, 199 time depth, 7, 8,22,42, 89, 257 Tok Pisin, 24, 26,74 Toledo, 4, 5, 7, 97, 117, 121, 126, 143, 145-217, 219, 220, 233, 235, 236,238-243, 245, 247, 260, 261-265, 283,291,292, 297, 301 Torreblanca, Máximo, 112, 113, 291 Torres Fontes, Juan, 283 transfer, 52, 55, 57, 79, 86, 87,151, 178, 189

Index transitivity, 185, 190 transmission, 2, 51, 67, 75, 76, 83, 113, 131 transparency, 11, 33,46,47, 57, 58, 65, 90,91, 114, 118, 123,124, 132, 143, 155, 156, 164, 197, 213,214, 241,257, 258, 260, 264, 265, 267, 277, 280, 282, 286-288 Trask, R. L., 109, 110, 132, 133, 136, 260, 283, 288 Traugott, Elizabeth, 15 Trudgill, Peter, 3, 7, 27, 2 8 ^ 8 , 51, 54, 56,67, 68,71,76, 80,81, 117, 158, 159, 172, 229, 245, 257, 264, 279, 296, 297, 301 Tusell, Javier, 100, 283 Tuten, Donald N., 275 Tuttle, Edward, 290 uniformity, 11-13, 22,49, 52, 80, 81,225,245, 261,267, 276 Urdu, 88 Valle, José del, 3 verbs, 4, 5, 11, 37, 57, 74, 129, 131, 168, 175, 185, 189, 190, 191, 198, 199, 244, 292-294 Versteegh, Kees, 283 Vidal Mayor, 199 Villar García, Luis Miguel, 156 vowels, 4, 11, 12, 34, 42,44,46, 57, 65, 69, 93, 105, 113, 119, 122, 124-127, 129, 130, 131, 133,

345

135, 136, 138, 143, 155, 159, 161, 163-166, 168-170, 177, 178, 195-197, 205, 206,212, 213, 241, 258, 260, 278, 279, 286-288, 291,292, 294, 295 Wagner, Max, 223 Wang, William S-Y., 129 weakening, 32, 33, 35,47, 52, 85, 87, 100, 157, 171, 179, 205,206, 212, 229, 241, 258, 264, 283, 285, 291, 294 Weinreich, Uriel, 2, 55 Whinnom, Keith, 285 Wilson, Robert, 88 Willemyns, Michael, 278 Williams, Ann, 3, 7, 19, 27, 33-35, 46, 53, 63-73, 76, 80, 86, 89, 92, 93, 229, 257, 264, 268, 277, 279, 282 Wolfram, Walt, 81, 82, 279, 283 Wright, Roger, 3, 5, 95, 106, 109, 110, 113, 120, 138, 154, 160, 161,265, 276, 284, 285, 290 Yaeger-Dror, Malcah, 34, 35, 70, 277, 278 yeísmo, 5, 295, 296 Yoruba, 15 Zamora Vicente, Alonso, 121, 122, 126, 127, 128, 136, 154, 288, 290, 298 Zufferey, François, 276