Grammaticalization in Korean: The Evolution of the Existential Verb

Index | 275

Grammaticalization in Korean: The Evolution of the Existential Verb

Grammaticalization in Korean: The Evolution of the Existential Verb

Minju KimClaremont McKenna College

Saffron Korean Linguist ics Series • Number Five

Grammaticalization in Korean: The Evolution of the Existential VerbMinju Kim

ISBN-13 9781872843926 Soft cover

Volume I Number 5

Saffron Korean Linguistics Series ISSN 1740 2956

Series Editors: Jaehoon Yeon and Jae Jung Song

Published by Saffron Books in conjunction with the Centre of Korean Studies, School of Oriental and African Studies, University of London

Cover created by Prizmatone Design Consultancy, a division of EAP

Copyright © 2011. No part of this publication may be reproduced or used in any form (graphic, electronic or mechanical, including photocopying, recording, taping or information storage and retrieval systems) without permission of the publisher. Additional copyright information is available from the publisher

Published by Saffron Books, an EAP imprint EAPGROUP International Media [EAP] P O Box 13666 London SW14 8WF United Kingdom

Commissioning Editor: Sajid Rizvi [Editor-in-Chief, EAP]

Telephone +44-[0]20 8392 1122Facsimile +44-[0]20 8392 1122

E-mail [email protected]

Web www.saffronbooks.com www.eapgroup.com www.eastasianeurope.net www.eapgroupinternationalmedia.co.uk

Designed by Prizmatone Design Consultancy

Manufactured in TurkeyDistributed worldwide by Saffron Distribution, a division of EAP

For Michael Hyun Cho and Russell Schuh

Contents | �

Contents

Series Editors’ Note 8 Acknowledgements 9Abbreviations 111 Introduction 132 Grammaticalization Theory and Explanation of the Source Forms 303 Grammaticalization of is(i)- in Particles: The Role of

Analogy in Grammaticalization 414 Grammaticalization of the Connective -ese: The Iconicity

Principle in the Combination of Clauses and Serialized Verbs 705 Grammaticalization of the Connective myense: Simultaneity

and Concessivity 976 Grammaticalization of the Connective -kose: A Comparison of

-ko, -kose and -konase 1177 The Intersection of the Perfective and Imperfective Domains:

The Grammaticalization of the Aspectual Markers –ess, -e iss- and -ko iss- 139

8 Grammaticalization of the Verb “take” in Korean: Functional Overlaps among -e, -ese, and -e kaciko 173

9 Grammaticalization from Quotation to the Quotative: Similarities among the Three Connectives -e, -ko, and -mye 194

10 Conclusion 226Appendix 1 List of Diachronic Korean Data 238Appendix 2 Abbrevidations for Frequently Used Middle Korean Texts 243Bibliography 244Index 263

� | Grammaticalization in Korean: The Evolution of the Existential Verb | Minju Kim

The Saffron Korean Linguistics Series is devoted particularly to functionally and/or typologically orientated research on Korean language and linguistics.

Volumes in the series, while dealing with specific topics in Korean language and linguistics, will address broadly defined functional and/or typological issues and concerns, rather than matters of abstract theoretical polemics. Theoretical or applied work related to Korean language will also be considered. The series aims to offer an international academic forum for the dissemination of the latest research into Korean linguistics as well as Korean language studies.The Series Editors welcome manuscripts on any aspect of Korean linguistics and language study, including Phonetics, Phonology, Morphology, Syntax, Semantics, Pragmatics, Typology, Discourse Analysis, Sociolinguistics, Language Acquisition, Historical Linguistics, and Korean Language Teaching. Submission enquiries should initially be addressed to Jaehoon Yeon.

Manuscripts or abstracts for book proposals must be submitted simultaneously to both the Series Editors. Contributors whose native language is not English are strongly advised to have their manuscripts read, and revised where warranted, by native speakers.

Jaehoon Yeon and Jae Jung Song, 2011

Series Editors’ postal and e-mail addresses

Jaehoon Yeon Jae Jung SongCentre of Korean Studies, SOAS Linguistics ProgrammeUniversity of London University of Otago, P O Box 56London WC1H 0XG Otago, DunedinUnited Kingdom New Zealand

Email: [email protected] [email protected]

Series Editors’ Note

Acknowledgements | �

Acknowledgements

I would like to express my sincere thanks to the many people who have offered their help and support while I was writing this book. First, I am deeply grateful to my professors at the University of California, Los Angeles: Professor Sung-Ock Sohn, Professor Shoichi Iwasaki, Professor Russell Schuh, and Professor Hongyin Tao. With their nurturance and insightful vision, they guided my spiritual as well as intellectual growth.

I would like to sincerely thank Professor Sohn for her generous care and support. Through her guidance and scholarly work, I first learned about the fascinating world of grammaticalization. As my advisor, she practically lived my graduate life with me, rejoicing in my small triumphs and comforting me during my failures. I am deeply indebted to Professor Iwasaki for having believed in me even when I doubted myself. For the last twelve years, he has always been there for me. His support and faith has given me great comfort and the courage to keep going. I also would like to express my sincere gratitude to Professor Tao who first introduced me to the field of corpus linguistics. His help at my side gave me the courage to endure all the initial trials of dealing with Middle Korean corpus and computer programs. I am also extremely grateful to Professor Russell Schuh for his generosity and unreserved enthusiasm in my work. His appreciation of my efforts has given me immense positive energy throughout this project. He showed me what a scholar should be. Lastly, I would like to thank Professor John Duncan for his vision and guidance at critical stages of my career.

I have worked on this topic for over eight years. Throughout the process, I received generous help from many people. I would like to thank Professor Kwang-Hae Kim who taught Middle Korean at UCLA. I want to thank those who helped me with the Korean data and concordance programs: Professor Sang-Ock Lee at Seoul National University, Hee-Chang Jung and Han-Saym Kim at The National Academy of The Korean Language, and Jinho Park, Seongcheol Shin, and Jinhyeong Kim at The International Korean Language Foundation. They welcomed me and helped me test various concordance programs. I would also like to thank Hea-Ju Kim for helping me collect such interesting data in Korea. Thanks to her help, I was able to gather conversational data from Korean speakers ranging from teenagers to adults in their 50s.

10 | Grammaticalization in Korean: The Evolution of the Existential Verb | Minju Kim

I also would like to thank Professor Seongha Rhee, Professor Chung-Min Lee, Professor Hyo-Sang Lee, Professor John Whitmann, Professor Masayoshi Shibatani, Professor Sungdai Cho, Professor Sang-Oak Lee, and Professor Ross King. I sincerely thank them for their care and encouragement in my work as a linguist. I am also grateful to my colleague and friend, Mee-Jeong Park, for always being there for me through the good and bad. I sincerely thank my colleagues at Claremont McKenna College. They have been very generous, supportive, and intellectually stimulating. Lastly, I thank the editors of the Saffron Korean Linguistics Series, Professor Jaehoon Yeon and Jae Jung Song, for their patience and support.

Lastly, I would like to thank my family. With their endless support and love, I could focus on my studies. My mother’s boundless love and my sister’s support gave me great confidence and courage to motivate myself to pursue my goals. I am also very thankful to my son, Michael Hyun Cho for his love. My son’s beautiful laughter and kind care brings me so much joy and happiness.

Minju KimClaremont, California

December 2008

AbbreviAtions | 11

Abbreviations

AB AblativeACC AccusativeAG AttentionGetterBENEF BenefactiveCAUS CausativeCIRCUM CircumstantialCOM CommitativeCOMP ComplementizerCONC ConcessiveCOND ConditionalCONN ConnectiveCOP CopulaDAT DativeDEC DeclarativeDEF DeferentialDM DiscourseMarkerEMP EmphaticFP Sentence-FinalParticleFUT FutureGEN GenitiveHSAY HearsayEvidentialHON HonorificHORT HortativeIE InformalEndingIMP ImperativeIND IndicativeMoodINT IntentionalMidK MiddleKoreanNEG Negative

NOM NominativeNOMZ NominalizerONOMAT OnomotopoeiaPASS PassivePAST Past/AnteriorPDK PresentdayKoreaPL PluralPOL PolitesentenceenderPORG ProgressivePOSS Possibility/AbilityPURP PurposiveQ InterrogativeQt QuotativeREL RelativizerRES ResultativeRETRO RetrospectiveSVC SerialverbconstructionTOP TopicTRANS Transferentive

12 | Grammaticalization in Korean: The Evolution of the Existential Verb | Minju Kim

Extended Yale Transliteration SystemThe transliteration system adopted in this study is an extended Yale system developed by Samuel Martin (1992) and Seongha Rhee (1996) for the transliteration of Middle Korean (eg, z for Δ). Also, following Samuel Martin’s (1992) system, capital letters are used to transliterate Chinese characters in original texts.

EXTENDED YALE TRANSLITERATION SYSTEM

The transliteration system adopted in this study is an extended Yale system developed by Samuel Martin (1992) and Seongha Rhee (1996) for the transliteration of Middle Korean (e.g., z for ). Also, following Samuel Martin's (1992) system, capital letters are used to transliterate Chinese characters in original texts.

k kkn ttt lL mM pv pps ssz Ø (syllable-initial)ng (syllable-final) ngY Hc ccch khth phf hhh

a ayya yaye eyye yeyo waway oyyo yoywu wewey wiyu yuyu uyi AAy

Chapter 1 | Introduction | 13

1Introduction

1.1 Purpose of the StudyCross-linguistically, verbs denoting existence have been one of the most fertile sources for grammatical markers (Heine, Claudi, and Hünnymeyer 1991, Hopper and Traugott 1993, eg, English “be”). In particular, Heine and Kuteva (2002: pp97-102, pp127-128) listed the three paths presented in table 1-1 as cross-linguistically common developmental patterns of the verbs denoting existence.

Table 1.1 Historical Development of Verbs Denoting Existence across Different Languages

A. Locative

Ewe le “be at” > “at” preposition

Supyire na “be at” > “at,” “on” postposition (Carlson 1991: pp207-209)

Kikuyu kwĩ “be at” > “to,” “from” preposition (Barlow 1960: p200)

Chinese zài “be at” > “at,” “in” preposition (Hagège 1975: p154, Peyraube 1996: p182-185)

Yao Samsao yi↔m “be at” > “in” preposition (Matisoff 1991: pp417-8)

Hmong nyob “be at” > “verposition” (Matisoff 1991: p418)

B. Continuous

Godié kù “be at” > progressive (Marchese 1986: p63)

Tyurama na “be at” > progressive (Prost 1964: p105, Heine and Reh 1984: p117)

Maninka yé ... lá “be at” > progressive or durative (Spears 1972: pp15-16)

Lingala –zala “be at” > durative (Mufwene and Bokamba 1979: pp244-246)

14 | GrAmmAtiCAlizAtion in KoreAn: the evolution of the existentiAl verb | minju Kim

Basque egon “be in” > continuous (King 1994: p384)

Burmese nei “be at” > continuative/progressive (Matisoff 1991: p416)

Thai jùu “be at” > continuative/progressive (Matisoff 1991: p416)

Chinese zài “be at” “be at” > continuative/progressive (Matisoff 1991: p416)

Kongo kala “to be” “exist” “remain” > ka(la) progressive

(Laman 1912: pp159-180; Heine and Reh 1984: p88)

Yagaria hano “exist” “be” > no’-/ne’- progressive aspect, prefix

(Renck 1975: p90)

Ghanaian PE dèy locative/existential copula > progressive/habitual

(Huber 1996, Turchetta 1998)

C. Verbs denoting possession

Lezgian gwa locative copula> temporary possession (Haspelmath 1993: p318)

Lezgian awa “be in” locative copula > predicative possession

(Haspelmath 1993: p317)

Chinese yŏu “exist” > verbal possession (Li and Thompson 1981: p513)

North !Xun gè “exist” > have construction (Bernd Heine, field notes)

Turkish var “existent” > have construction

Similarly, the Korean existential verb is(i)- (in Present Day Korean form iss-) has played a pivotal role in shaping the Korean language. First of all, the Korean existential verb can denote the meaning of possession.

(1) Ku-nun cip-i iss-ta He-TOP house-NOM exist-DEC “He has a house.”

In addition, the existential verb is(i)- has participated in the creation of eleven different grammatical markers in association with various linguistic structures; see Table 1.2.

Table 1.2 Historical development involving Korean existential verb is(i)-A. Postpositions-eyse “from,” dynamic locative “at”-eykeyse “from” an animate referent-hantheyse “from” an animate referent-lose “from” > qualification “as”-kkeyse “from” > honorific subject marker


B. Clausal Connectives-ese sequential, causal connective-myense simultaneous, concessive connective-kose sequential connective

C. Tense-Aspect Markers-ko iss- progressive-e iss- resultative-ess- anterior/ past tense

As shown in Table 1.2, the remnant of the existential verb is(i)- is found in particles and connectives in the form of se, while in tense-aspect markers it is found either in the form of ss or in the current existential verb form iss-. This study will examine the multiple grammaticalization pathways of the Korean existential verb is(i)- by analyzing corpus data from the 15th century to the present day. In particular, the study will explain these changes on the basis of the roles of discourse context and frequency, unlike earlier approaches that focused more on cognitive and metaphoric explanations. By doing so, this study will demonstrate the importance of situating language changes in actual language use and explaining these changes in terms of both cognitive and communicative mechanisms.

1.2 Roles of Discourse Context and Frequency in the Emergence of Grammar

There have been several definitions of grammaticalization; some of the important ones will be discussed in Chapter 2. One recently modified definition is from Traugott and Hopper’s second edition of Grammaticalization1: “The change whereby lexical items and constructions come in certain linguistic contexts to serve grammatical functions and, once grammaticalized, continue to develop new grammatical functions” (2003:xv). Research in grammaticalization has demonstrated that, contrary to the structuralist view of language as an autonomous and stable system, grammatical markers emerge and evolve. Furthermore, for the last three decades, grammaticalization theories have been successful in describing regularities in morphosyntactic and semantic changes involving the emergence of various grammatical markers and their subsequent evolution.

Researchers in this area generally agree with the basic premise that changes in grammaticalization are generated by the cognitive and communicative use of language. Nevertheless, until recently, explanations of the grammaticalization processes have focused more on the cognitive aspects rather than on the communicative aspects. In accordance with this emphasis, grammaticalization has commonly been accounted for as a result of metaphorization - the mapping of one domain onto another through a cognitive leap or conceptual shift based on the inherent semantics of the source item (Diehl 1975, Genetti 1986, Svorou 1988, Heine et al 1991, Matisoff 1991, Matsumoto 1998). For instance, the common grammaticalization path from the verb “go” to the future marker (eg, English “be


going to”) is explained as a cognitive leap from the source concept in the spatial domain (directional movement from the deictic centre in space) to the temporal domain (directional movement from the deictic centre in time) (Emanatian 1992); see (2a) below.

One of the reasons why grammaticalization studies have readily turned to cognitive explanations lies in the fact that similar patterns of grammaticalization are recurrently observed in genetically and geographically unrelated languages. Heine et al (1991: pp32-38) show that the source items of grammaticalization are usually the most basic and concrete concepts of human experiences (eg, “be,” “exist,” “go,” “come,” and body parts); hence, these source items tend to be indifferent to cultural differences and language specifics. Furthermore, in many unrelated languages, similar source items entering grammaticalization result in similar target items. For instance, the verb “go” often develops into the future marker and the verb “exist” or “be” often develops into the locative “at.” In such situations, it would appear appropriate to look at human cognition for explanations, as cognition operates independently of language specifics.

Recently, however, such a heavy reliance on cognitive and metaphoric explanation has been called into question (Hopper and Traugott 1993, Hopper 1998, Traugott and Dasher 2002). The most decisive evidence disputing the metaphoric explanation is that what is grammaticalized is not a lexical item, but rather is a construction with a lexical item in it (Hopper and Traugott 1993, Traugott and Dasher 2002). For instance, the English future meaning is delivered not by the lexical verb “go,” but by “be going to,” a construction that contains the lexical “go.”

In addition, closer examination of the gradual changes in grammaticalization has shown the critical importance of discourse context. In-depth investigation of the grammaticalization of “be going to,” for example, has revealed that this construction could have obtained its future meaning through frequent use in purposive contexts such as “I am going to marry Jane”; see (2b). Through frequent use in this “local context” marking purpose (Hopper and Traugott 1993), the progressive construction could have gained the pragmatic inference of intention. In turn, this inference of intention had led to prediction, which is the prototypical future meaning (Bybee and Pagliuca 1987, Bybee et al 1991, Bybee et al 1994).

(2) Different Explanations of the Grammaticalization from “go” to the Future

Marker (a) Space (“go”) > Time (future)

(b) “Go” in purposive context > Intention > Prediction > Future

Clearly, the path in (2a) proposed in the metaphoric explanation is intuitively more plausible and simpler than the path in (2b). Nonetheless, the actual language change occurred by way of the more complex path. Traugott and Dasher made the following observation in this regard: “The ‘metaphor’ often appears to be primarily a function of the fact that we are looking at lexical entries in their ‘before’ and ‘after’ stages, and out

Chapter 1 | Introduction | 1�

of context. …As soon as we think about change in terms of its syntactic and ultimately discourse contexts, associations arising out of the context can be construed as playing a major role in change” (2002: pp80-81).

The appearance of similar paths of change also can be explained without relying so heavily on a cognitive, metaphoric account. In her recent study, Bybee (2003a: p157) states that the common paths of change would not be attested “unless users of these languages made very similar inferences under similar conditions.” I would like to elaborate upon this important statement. Research in grammaticalization has indicated that the source items entering the grammaticalization process are usually those concepts that are semantically general and most basic to human experience (Heine et al 1991). Then, across different languages, similar lexical items tend to be recruited for grammaticalization and these similar words tend to be used in comparable discourse contexts. From the use of these source items in similar contexts, human cognition tends to make similar inferences (eg, inferring the meaning of intention from the use of “be going to” in purposive contexts), which eventually results in a grammatical device that is similar across different languages. As shown in this account, when changes of grammaticalization are situated in discourse contexts, the interpretation of the role of cognitive mechanisms is also modified.

Previously, an extensive number of studies have been devoted to detecting common or universal grammaticalization pathways with a focus on what kinds of source concept lead to what kinds of target concepts. These efforts recently culminated in a seminal work, Heine and Kuteva (2002), which listed the common target concepts for each source concept across hundreds of different languages.2 At the same time, as Heine (2002) pointed out, it is essential to closely examine “intermediate stages” leading from the source to the target. Previous studies of grammaticalization quite often were forced to turn to metaphoric explanations due to the lack of diachronic data that could illustrate these intermediate stages. In that sense, the present study is very fortunate because it can make use of diachronic corpus data from the first Korean alphabet documents in the 15th century.

This study inherits its basic premise from studies that situate grammaticalizational changes in the use of language in communication. In order to better understand how grammar emerges, we have to look not only at the function of cognition, but also at the less discussed communicative mechanisms—the mechanisms of change situated in discourse context. More specifically, the present study will closely examine the intermediate stages between the lexical source and the grammaticalized target marker. On the basis of this examination, it will show that grammaticalization typically takes place in specific discourse contexts of use: that is, a pragmatic inference form a local context becomes semanticized into a grammatical meaning through frequent use (Traugott and Dasher 2002). In particular, some of the Korean case studies of grammaticalization which will be discussed in this book follow the universal pattern of development (eg, “be at” > “at,” “take” > instrumental marker, and “say” > quotative marker). In the subsequent discussions, these common paths of grammaticalization, which have been

1� | GrAmmAtiCAlizAtion in KoreAn: the evolution of the existentiAl verb | minju Kim

previously explained on the basis of metaphoric extension, will be re-examined with special reference to discourse contexts.

When language change is studied within communicative language use, in addition to discourse context, frequency emerges as another important contributor to the process (Barlow and Kemmer 2000, Bybee and Hopper 2001, Bybee 2001, 2002). From its beginning, grammaticalization has been understood as a change whereby the semantic specifics of a lexical item are “bleached out” leaving only the general core meaning that becomes a grammatical function (Givón 1973, Lehmann 1982, Heine and Reh 1983, Bybee and Pagliuca 1985, Bybee et al 1994). In the process, an increase in frequency has been taken for granted. In the first edition of Hopper and Traugott’s book Grammaticalization 1993, frequency was considered as concomitant of the grammaticalization (1993: p103) or evidence for the degree of grammaticalization (1993: p110). However, with the advancement of the corpus linguistics and the recent literature emphasizing the role of frequency in grammaticalization, the second edition of their book Grammaticalization (2003) includes a new section 5.6 devoted to the discussion of frequency.

Drawing on construction grammar, Bybee (2003a, 2003b) has begun a systematic explication of the role of frequency in grammaticalization. She compares aspects of language production to neuromotor concepts such as “habituation” and “a single processing unit.” According to her, the mechanism behind semantic bleaching is “habituation,” the process by which a stimulus loses its impact if it occurs frequently. In language use, when a construction is repeatedly used, the semantic force and specificities that it carries gradually become weakened. In the case of grammaticalization, when a sequence of words is used together frequently, they become automated as a single processing unit. As the sequence is stored and used as a single unit, it emerges as a distinct construction whose internal structure is obliterated and whose lexical specifics are bleached out. Subsequently, this construction can gain a pragmatic inference from the context in which it is used; over time and use, the pragmatic inference becomes semanticized as a coded reference of the construction.

With the above account in mind, this study will closely investigate diachronic changes in frequency. Statistical information from corpus data can empirically test the role of frequency in the emergence of grammatical markers. In addition, statistical information can serve as an explanatory tool that can illustrate various aspects of grammatical change. For instance, skewed collocation patterns at the initial stage of grammaticalization can be seen to index the particular local context which nurtured the grammaticalization. Thus, in this study, frequency will be discussed not only as a contributor to grammaticalization, but also as an explanatory method for its study.

Lehmann (2004) also introduced the concept of “analogically-oriented grammaticalization” which is a form of grammaticalization whose course of development is directed by analogy. As an example, he cites the grammaticalization of coverbs into prepositions in Modern Mandarin. As in that case, when similar grammaticalization patterns


(from coverbs to prepositions) are repeated, the presence of the precedent models plays a part in further prompting subsequent grammaticalization. For this reason, although the coverbs have individually undergone grammaticalization, the subsequent grammaticalization cases are in fact examples of analogically-oriented grammaticalization. As shown in Table 1.2, in Korean, the same verb is(i)- participated repeatedly in grammaticalization of multiple postpositions and connectives. As can be presumed, analogy played an active role in their grammaticalization, and the term “analogically-oriented grammaticalization” will serve as an useful tool.

1.3 Grammaticalization of is(i)- in Three Functional Domains Table 1.2 shows that the Korean existential verb is(i)- (the Middle Korean form of the Present Day Korean iss-) has engendered multiple grammatical markers. As a verb denoting existence, is(i)- collocated with and eventually grammaticalized with diverse linguistic constructions. In Present Day Korean, traces of the existential verb are found in three grammatical domains: 1) postpositions marking the location (-eyse, -eykeyse, and -hantheyse) and subject (-kkeyse) of the sentence; 2) clausal connectives (-ese, -myense, and -kose); and 3) aspect markers (-e iss-, -ko iss-, and -ess-). In addition, although they are not grammatical markers, locative adverbs, such as etise “where” and melise “at a distance” also have emerged making use of the existential verb.

In the postpositions, clausal connectives, and adverbs that emerged containing the existential verb, the trace of the verb is realized as the bound morpheme se. In the 15th century, when the grammaticalization of is(i)- was actively taking place, the connective e “and” was one of the most frequently used connectives. The combination of the verb is(i)- and the connective -e was always used in its contracted form isye (<isi-e) in the 15th century and eventually became se in Present Day Korean. In the postpositional domain, the verbal construction isye (<isi-e) was combined with the existent postpositions and their combination, postposition + isye morphologized together leading to a new postposition with se. In the connective domain as well, isye (< isi-e) was added to the existent connectives engendering new connectives with se. Locative adverbs followed the same pattern.

On the other hand, in the creation of the tense-aspect markers, the verb stem itself was incorporated. In the 15th century, three allomorphs of the existential verb, is-, isi-, and si- were in use. However, over time, is- unified the other forms and in the process, underwent tensification, leading to the current form iss-. In the tense-aspect markers, the traces of the existential verb are realized as either iss- (when the verb itself is used) or ss (when morphologized with another component in the source construction): the progressive -ko iss-, resultative -e iss-, and anterior/past -ess-.

The grammaticalization of the Korean is(i)- is observed in three functional domains of postpositions, clausal connectives, and tense-aspect markers. This book will be organized on the basis of these three functional divisions.


1.4 Previous Studies Previous studies of each grammatical marker under question will be discussed in the corresponding chapters. This section will focus only on those studies that pay attention to the bound morpheme se which appears across the postpositional domain and the connective domain.

The impact that the grammaticalization of the existential verb construction isi-e has made in shaping Present Day Korean is quite extensive and its traces are mostly reserved now as the bound morpheme se. Nevertheless, because isi-e has gone through the grammaticalization process in association with many different linguistic structures and for a long time period (since around the 15th century), native speakers of today cannot easily recognize the shared origin of se in different grammatical constructions. For example, the connection between se in locative postposition -eyse and se in clausal connective -myense is not readily recognizable to the Korean speakers of today. Furthermore, even when the shared origin of se is recognized as such, the source meaning of existence alone is not enough of a clue to account for all of the functions of se in Present Day Korean, such as in the subject marker -kkeyse and the concessive marker -myense. As a consequence, few researchers outside the field of Korean historical linguistics have recognized the significance of this existential verb construction in the Korean language and linguistics.

In spite of these obstacles impeding integrated research on the bound morpheme se, a number of studies perceptively recognized the shared origin of se in different grammatical markers and attempted to explain this elusive morpheme. The first study that recognized the independent meaning of se is Ramstedt (1939). Ramstedt observed the recurrent use of se in postpositions (eg, -eyse “from” and -eykeyse “from”) and defined se as denoting “from” which, he postulated, originated from the existential verb is(i)-. This view was shared by subsequent studies including Martin (1992). Then, a breakthrough was made by S N Lee 1976:. By investigating the historical development of se, S N Lee showed for the first time that se in two remote functional domains (the locative particle -eyse and the clausal connective -ese) in fact share the same lexical origin of the existential verb. He suggested that the primary meaning of se in -ese is durative aspect - a view shared by other major studies on se, namely those of W Huh (1975) and T Y Lee 1988:.

Diachronic studies such as S N Yi 1976, W Huh (1975), and T Y Lee 1988: will be discussed in more detail in later chapters. There were also studies that approached the meaning of se from a more synchronic point of view. These studies generally seem to agree on attributing the meaning of se to its original lexical meaning of existence (Sung 1979, I T Kim 1984) or something related to it, such as the claim of existence (C M Suh 1984), the precedence of existence (T S Kim 1986), or the completion of existence (T L Suh 1988).

There are a few studies that deserve closer attention. First, An (1985), who takes a synchronic approach, can be distinguished from the others in that he attempted to encompass the full range of uses of se. He suggested the original lexical meaning of se


to be “given” or “presupposed,” and applied this definition to the explanation of various grammatical markers that contain se. Next, the most detailed historical investigation of the grammaticalization of the existential verb is(i)- and of the honorific existential verb kyesi- would be T Y Lee 1988. While he provided a detailed analysis of the historical development of various postpositions with se, his discussion does not go beyond particles and the connective -ese. Finally, Strauss (1997) who took a synchronic approach presented the first article-long study of se in English. She analysed se as an inclusive marker and with that, attempted to explain the full range of postpositional and connective uses of se.

Taking these pioneering studies as a point of departure, this study will continue to explore the meaning of se. Considering the long and complicated paths of its grammaticalization, diachronic investigation is essential in understanding the multifunctionality of se in Present Day Korean. This study will primarily take a diachronic approach but it will also examine synchronic data whenever it is relevant. In examining each of these grammaticalization processes of the Korean existential verb is(i)-, this study will: 1) delineate each path of change that this verb has undergone and compare it with the universal paths suggested by other languages; 2) identify the local discourse context and discourse management strategies that have contributed to the emergence and the spread of the new grammatical form; and 3) seek to understand diachronic changes in frequency and how they relate to mechanisms of grammaticalization.

1.5 Methodology The present volume will investigate the grammaticalization processes of the Korean existential verb with special reference to the roles of discourse context and frequency. For this purpose, the methodology of this study is comprised of two main components. The principal method that will be employed is corpus linguistics. The corpus used in this study contains 3.6 million words covering both diachronic and synchronic data. In addition, the study will rely on comparative analysis. As mentioned earlier, some of the grammaticalizational pathways that will be discussed in this study are also commonly observed in other languages. Therefore, comparisons with these languages will be made whenever appropriate.

1.5.1 Corpus LinguisticsA corpus is a collection of linguistic data that is either compiled as a written text or as a transcription of recorded speech (Crystal 1992). Corpus linguistics studies corpus data for the purpose of linguistic research; and for the last three decades, corpus linguistics has proven to be a powerful method. These days, corpora are mostly computerized text files; with the help of computer programs, researchers can process large amounts of corpus data very thoroughly.

The corpus used in the present study contains both diachronic and synchronic data. The diachronic component includes various kinds of historical texts (eg, religious texts, letters, and novels) from the first Korean alphabet documents in the 15th century


to the 19th century. The synchronic component includes various genres of Present Day Korean written texts as well as transcriptions of naturally occurring conversational data. While diachronic corpus data are written texts, for Present Day Korean, I have included both written and conversational data. The reason why conversational data are included is that natural conversations are often the locus in which language change can be most vividly observed. The proposal that the written register lags behind the spoken register in language change is well attested, and is supported in this study as well. For instance, historically, the frequency of the connective -ese has increased over time; in Present Day Korean, its frequency is much higher in the spoken register than in the written register. On the other hand, the frequency of -mye has been declining since the 16th century; its frequency is much lower in the spoken register than in the written register. In this way, when used in combination with diachronic data and contemporary written data, spoken data can help us to see the trajectory of changes and even to predict the possible direction of change in frequency of a given construction

The diachronic and synchronic corpus data used in this study were processed using a Korean concordance program, “Uniconc.” The occurrences of each searched word were displayed within its surrounding context (ie, KWIC--Key Word in Context-- displays). In addition, linked with “EmEditor,” this program also allowed easy access to the original text beyond the display of KWIC; whenever necessary, extended contexts were also examined in the original texts.

In this study, computerized texts were classified according to their year of publication and grouped together to create a separate corpus for each century. As will be shown later, the size of the corpus varies for each century. To compare the diachronic changes in frequency across different centuries, it was necessary to make the raw numbers of tokens comparable, independent of their corpus sizes. This process is called “normalization.”

Normalization is used to adjust raw frequency counts from texts of different lengths in order to compare them accurately regardless of the difference in length (Biber et al 1998: p263). Specifically, raw frequency counts are divided by the number of words in the corpus, then multiplied by a chosen basis which is decided on in relation to the corpus size (eg, per thousand words or per million words). In this study, based on the size of the corpus for each century, the basis is mostly per 100,000 words. In the following chapters, at the end of each table that compares frequencies from different centuries, if the numbers are normalized, the basis for normalization will be given (eg, occurrences per 100,000 words). On the other hand, no notes will be added when the numbers in the table represent raw frequencies.

Lastly, I would like to note that this book is based on my doctoral dissertation filed in 2003. When I was writing the dissertation, available corpus data were smaller in size and the only concordance program that could process Middle Korean fonts was HGREP97 which worked in DOS environment. Although I used different data sets for the dissertation (total 2,538,066 words) and for this book (total 3,491,907 words), the same


major statistical findings were consistently observed in both data sets. Any discrepancies in specific numbers (eg, slightly lower or higher frequency rates per 100,000 words) between my dissertation (2003) and this book are derived from the differences in the corpus data sets that were used for each study.

1.5.2 Comparative AnalysisThroughout this paper, cross-linguistic comparisons will be made whenever relevant. In addition to Korean, across many different languages, existential verbs have been shown to be important sources for enriching grammatical resources (Heine et al 1991, Hopper and Traugott 1993, Lord 1993, Bybee et al 1994). As shown in Tables 1.1 and 1.2, existential verbs of different languages show comparable paths of change. Furthermore, as the study unfolds, the discussions will involve other cases of common grammaticalizational pathways, such as the development of the instrumental marker from the verb “take” and the development of the quotative marker from the “say” verb construction. Thus, comparisons with other languages will provide valuable insights for this study, both in understanding the Korean phenomena and in understanding the mechanisms behind these common paths of language change.

1.6 Data Description1.6.1 Middle Korean Corpus DataIn 1998, the Korean government started a ten-year project in memory of King Sejong, under whose sponsorship the Korean alphabet was created. As part of this commemoration project, the National Academy of the Korean Language digitized most Middle Korean and Early Modern Korean documents. This massive database covers about six hundred years, beginning with the first Korean alphabet documents in the 15th century and continuing until the early 20th century. It contains more than two hundred texts. For this study, each historical document was classified as belonging to a period of 100 years — from the 15th to the 19th century. Table 1.3 below lists texts from the 15th and 16th centuries, the time period on which the study will focus the most. The list of texts from the 17th to the early 20th century can be found in the Appendix.

Table 1.3 MidK Documents from the 15th and 16th C in the Corpus Data15th Century

Yongbiechenka (1445) Hwunmincengum (1446)Welin chenkang cikok (1447) Sekpo sangcel (1447)Welin sekpo (1459) Nungemkyeng enhay (1462)Pephwakyeng enhay (1463) Amitakyeng enhay (1464)Sencong yengkacip enhay ha (1464) Panyasimkyeng enhay (1464)Kumkangkyeng enhay (1464?) Sangwensa kwensenmwun (1464)Wenkakkyeng enhay (1465) Kwukuppang enhay (1466)Mokwucaswusimkyel (1467) Mongsan pepe (1472)


Nayhwun (1475) Chokan twusi enhay (1481) Samkang hayngsilto chokan (1481) Kumkangkyeng samkahay (1482)Nammyengcip enhay (1482) Kwanumkyeng (1485) Kwukupkanipang (1489) Yukcopeppotankyengenhay (1492)Cinenkwenkong (1496) Yukcotankyeng (1496)Sinsenthayulcakumtan (1497) Kaykan pephwakyeng (1500)

16th Century

Sok samkang hayngsilto (1514) Penyek pakthongsa (1517)Penyek nokeltay (1517) Mongsan pepe-kowunsa (1517)Cengsok enhay (1518) Penyek sohak (1518)Ilyun hayngsilto (1518) Pepcippyelhaynglok (1522)Kani onpang (1525) Mongsan pepe-simwensa (1525)Mongsan pepe-pingpalam (1535) Wumayangceyem eykpyeng chilyopang (1541)Pwunmwunonyekihaypang (1542) Kwuhwang chwalyo (1554)Wukca senceng (1560) Pwumouncwungkyeng (1563)Mongsanhwasangyuktoposel (1567) Chiltay manpep (1569)Senkakwikam (1569) Yessi hyangyakenhay-hwasan (156?)Yessihyangyakenhay-ilsek (1574) Payklyenchohay (1576)Anlakkwukthaycacen (1576) Mongsanpepe-songkwangsa (1577)Kyeychosimhakinmwun (1577) Senkakkwikam enhay (1579)Cwungkan kyenminpyen (1579) Sok sangkang hayngsilto-cwungkan (1581)Iungthaymyo enkan (1586) Sohak enhay (1586)Cwungyong enhay (1590) Cwuyek enhay (1590)Mayngca enhay (1590) None enhay (1590)Tayhak enhay (1590) Hyokyeng enhay (1590)Senco kwukmwunkyose (1593) Mwuyeyceypo (159?)Swunchenkimssimyo kanchal (?) Samkang hayngsilto-tongkyeng (?)Cangswumyelcoyhoceytongcatalanikyeng (?)

The years of creation assigned to each document here follow those listed by the National Academy of the Korean Language. The (?) mark indicates that the text was created during the listed century but it is not possible to pinpoint the exact year. Some of the texts from the National Academy of the Korean Language database are not included, either because they have a dictionary format or because the texts are repetitions of similar sentence styles (eg, government’s records). Table 1-4 shows the corpus data used for this study.


Table 1.4 Size of Corpus for Each Century

Number of words15th C 741,518 16th C 242,820 17th C 415,936 18th C 523,877 19th C 707,936 Present Day Korean (written) 439,365 Present Day Korean (spoken) 420,455 Total 3,491,907

In the 15th century, with the first proclamation of the Korean alphabet by King Sejong, the government led various publishing projects to promote the newly invented alphabet including translations of Chinese Buddhist texts. This is how the 15th century corpus data came to be so large.

Among the historical documents that were used for this study, three groups of texts deserve special attention. The first group is a series of translations of language textbooks, namely nokeltay, pakthongsa, and chephay sine. Among the three, the most important is nokeltay. One Chinese text of nokeltay has been translated into Korean many times spanning nearly three hundred years. These translations allow us to track historical changes in the Korean language. Furthermore, as they were intended as a language textbook for government translators, these texts are unlike other historical documents in that they reflect the spoken language of their respective periods. Among various translations of nokeltay, this study will examine the four versions that are most commonly compared by Middle Korean specialists.

Table 1.5 Publication Years for Different Nokeltay TextsText Year Text YearPenyek nokeltay 1517 Nokeltay enhae 1670Chenge nokeltay 1765 Monge nokeltay 1790

Two other language textbooks, Pakthongsa and Chephay sine, are also very valuable in that they too reflect the spoken form and have been translated multiple times. Therefore, they both are capable of illustrating changes in the Korean language over time. Paktongsa was a Chinese textbook; its three different versions, published in 1517, 1677, and 1765, are often compared. On the other hand, Chephay sine is a Japanese textbook; its three different versions from 1676, 1748, and 1781 can also be compared.

Second, personal letters are also important data since their styles offer the closest resemblance to spoken language. There are two groups of letters in the corpus: the Swunchen Kim Family Letters (192 letters, 16th century) and the Cincwu Ha Family Letters (about 130 letters, 17th century)3. These letters were exhumed from the tombs


of the Swunchen Kim and Cincwu Ha Families. As can be inferred from the numbers of letters, the letters are written by various family members of the clans.

The last group of texts to note is the ethics texts Penyek sohak (1518), Sohak enhay (1587), and different versions of Samkang hayngsilto. Penyek sohak (1518) and Sohak enhay (1587) are translations of Chinese sohak, which means “children’s learning.” As reflected in the title, these texts, including Samkang hayngsilto, are designed to teach morals through stories of model figures. Hence, the texts are composed with the clear intention of impressing young students with heart-felt moral behaviour; thus, they tend to overtly underscore those sections of the stories that will move readers. Due to this strategy in storytelling, these texts tend to be more animated and to adopt new emphatic expressions. Consequently, they serve as invaluable data that show how the grammaticalization of connectives with se has been influenced by discourse management strategies.

Table 1.6 Publication Years for Important Ethics TextsText YearSamkang hayngsilto chokan 1481Sok samkang hayngsilto 1514Penyek sohak 1518Sohak enhay 1587

In citing sources of examples, I follow the conventions in Korean studies. Each example from historical documents will be followed by a mark as in (3).

(3) [Sekpo sangcel 1447, 11:25b]

Sekpo sangcel is the title of the cited text; the number 1447 represents the year of publication; 11 represents the volume number; and 25b indicates the page number. The historical classification of the Korean language is presented in Table 1.7 below.

Table 1.7 Historical Classification of the Korean Language

Classification YearsOld Korean 7th C - 9th CMiddle Korean Former Middle Korean 10th C - 14th C

Latter Middle Korean (MidK) 15th C - 16th CEarly Modern Korean 17th C - 19th CModern Korean 20th CPresent Day Korean (PDK) Contemporary Korean

This classification of time periods follows K M Lee (1963 [1998]). In the seventh century, the Korean peninsula was first unified as one country and the language used in this unified country Silla served as the foremost source for Present Day Korean. Former


Middle Korean (from the 10th to the 14th century) corresponds to the period of the Korye Dynasty; Latter Middle Korean (15th and 16th century) corresponds to the first 200 years of the Chosun Dynasty. The Korean alphabet was first created in 1443. Before that, the Korean language was recorded by using Chinese letters in different systems including itwu and kwukyel. Latter Middle Korean is the first period in which we can clearly see the history of the Korean language. From Chapter 2 of this volume, Latter Middle Korean will be abbreviated as MidK; most cases of grammaticalization that will be discussed in this study date back to this period.

The end of the seven-year war with Japan (the Hideyoshi Invasions) in 1598 demarcated the start of Early Modern Korean (17th through 19th centuries) which was still part of the Chosun Dynasty. The long war destabilized the social order; this period is characterized by more pragmatic thinking and a focus on basic industries such as agriculture and trading. Along with that emphasis, this period sees the emergence of people’s literature such as the consumption of Korean novels by ordinary people.

The early 20th century witnessed an influx of western culture which prompted western-style novels (called sin sosel “new novels”) and poetry; this is the period of Modern Korean. The diachronic corpus data used for this study range from the 15th to the 19th centuries. However, since the new novels are also digitized and available, whenever it is necessary to examine early 20th century data, I will also make use of them. Lastly, Present Day Korean is the contemporary Korean language; from Chapter 2 of this volume, Present Day Korean will be abbreviated as PDK.

1.6.2 Present Day Korean Data In addition to historical texts, the National Academy of the Korean Language also digitized written and spoken data of Present Day Korean. This study makes use of these data. In selecting different parts of the written corpus for this study, I sought to balance various genres (short novels, short essays, and explanatory textbooks) and also to include many different writers.

The Present Day Korean spoken data used in this study are transcriptions of naturally occurring conversations. The spoken data for this study consist of three different data collections: the first is my own data collection (35,430 words); the second is a data collection borrowed from my colleague, Mee-Jeong Park at the university of Hawaii (26,209 words); and the third is a data collection made by the National Academy of the Korean Language (358,816 words).

My own data contains 270 minutes of conversation recorded both in Korea and Los Angeles and transcribed by myself. It is composed of nine different naturally occurring conversations with various speaker groups from boys and girls in their early teens to speakers in their late forties. Next, Mee-Jeong Park’s data collection contains four sessions of Bible study meetings in Los Angeles. The participants are five male and six female students in their late twenties and early thirties. Their conversations revolve around various informal topics as well as formal discussions about the Bible.4 Lastly, the

2� | GrAmmAtiCAlizAtion in KoreAn: the evolution of the existentiAl verb | minju Kim

data collection made by the National Academy of the Korean Language is composed of 56 naturally occurring conversations. While most of the speakers are in their 20s and 30s, as the conversations were collected by university students, they also include teenagers and older speakers.

1.� Organization of the StudyThe book is organized primarily by the functional domains in which the existential verb is(i)- grammaticalized. Chapter 2 presents a short survey of grammaticalization theory and a more detailed explanation of the source forms. Chapters 3 to 9 cover actual case studies of grammaticalization in Korean. While each chapter from 3 to 9 primarily describes cases of grammaticalization, theoretical topics are also presented for the chapters to move them beyond simple description to explanation and to ensure their relevance to scholars outside of Korean linguistics.

More specifically, Chapter 3 investigates the grammaticalization of particles with se. Under the encompassing term “particle,” the chapter analyzes three different groups of particles: spatial postpositions with se (-eyse, -eykeyse, -lose, and –hantheyse), spatial adverbs with se (eg, melise, kakkaise, and etise), and subject markers with se (-kkeyse and -eyse).

Chapters 4, 5, and 6 will examine the grammaticalization of connectives -ese, -myense, and -kose, respectively. The morpheme se in these connectives functions as a relevance augmentation marker. With the addition of se, the new connectives -ese, -myense, and -kose emphasize more strongly the ground-event relationship between the combined clauses. Based on the iconicity principle, of the two simultaneous markers, -mye and -myense, only -myense acquired a concessive meaning; and of -e and -ese, in Present Day Korean -ese is mostly used in combining clauses while -e is used in combining serialized verbs.

Chapter 7 examines the grammaticalization of several aspectual markers, namely, anterior/past tense -ess, resultative -e iss- and progressive -ko iss-. This chapter demonstrates that contrary to Bybee et al (1994), diachronic crossover between the perfective and imperfective aspectual domains is possible. Historically, -ko-type connectives gradually replaced -e-type connectives, and accordingly, the gradual replacement of e isi- (the erstwhile form of the Present Day Korean resultative marker) by -ko isi- (the erstwhile form of the Present Day Korean progressive marker) is observed in the history of Korean.

Chapter 8 discusses the grammaticalization of the Korean verb kaci- “take.” This verb grammaticalized into the instrumental marker -ul kaciko and the sequential and causal connective -e kaciko. In contrast with previous studies which explained using metaphor the common development of “take” into the instrumental marker and the clausal connective, this chapter will demonstrate that the grammaticalization took place in local contexts of use.

Finally, Chapter 9 examines the grammaticalization of “say” verb constructions into the quotative, and then into hearsay evidential markers. Across different languages,


“say” verb constructions often grammaticalize into quotatative markers. In the history of Korean, this development occurs three times, with the connective -e, then with the connective -ko, and finally with the connectives -mye and -myense. Using Korean diachronic corpus data, this chapter will examine the mechanism behind this common path of grammaticalization.

Notes1 Between the first and second editions of the book, Grammaticalization, this study will

mostly use the 1993 first edition unless it is necessary to use the 2003 second edition. 2 Many studies in grammaticalization and historical linguistics greatly benefit from Heine and

Kuteva (2002). In the present study as well, cross-linguistic examples will be primarily drawn from their study.

3 The Cincwu Ha Family Letters are also called the Hyenphwung Kwak Family Letters; Kwak is the husband’s last name, while Ha is the wife’s last name. This study follows the original labelling in the Sejong Corpus.

4 I would like to thank my colleague, Mee-Jeong Park at the University of Hawaii, for lending me her data. Also, I would like to thank Sun-Young Oh at the Seoul National University for providing tokens of -kose and -konase from her conversational data for this study. As can be seen in Chapter 6, the frequencies of -kose and -konase are very low.