This book "asserts that the origin and spread of languages must be examined primarily through the time-tested techniques of linguistic analysis, rather than those of evolutionary biology" and "defends traditional practices in historical linguistics while remaining open to new techniques, including computational methods" and "will appeal to readers interested in world history and world geography."
AUTHOR: Andrej A. Kibrik TITLE: Reference in Discourse SERIES TITLE: Oxford Studies in Typology and Linguistic Theory PUBLISHER: Oxford University Press YEAR: 2011
Olga Lovick, Interdisciplinary Studies, First Nations University of Canada
SUMMARY In this extensive monograph, Kibrik tackles discourse reference from the typological as well as from the cognitive perspective. His aim is to understand the ‘basic referential choice’ between full referential devices (i.e. noun phrases) and reduced referential devices (i.e. free pronouns, bound pronouns, and zero), looking at their cross-linguistic variation and at the cognitive foundations of reference.
Part I is introductory. In Chapter 1, Kibrik lays out definitions of referent, reference, referring/referential expression, referential devices. Most crucially, he defines reference as a discourse-oriented phenomenon, not a syntactic one (p. 10). He formulates his goal to unite discourse analysis, typology, and cognitive linguistics in order to arrive at a “realistic and comprehensive theory of human language” (p. 22). Chapter 2, Kibrik defines his goal as the investigation of the basic referential choice between full and reduced referential devices, and limits the study to anaphoric specific definite reference of non-locutor pronouns (p. 34). He briefly summarizes the proposed Cognitive Multi-Factorial approach and closes with a distinction between referential devices (whose primary function is to refer) and referential aids (elements with a different primary function which can be used to resolve referential conflict).
Part II contains five typological chapters on referential devices. Throughout, Kibrik discusses particular patterns and offers a typological perspective by drawing on studies in the ''World Atlas of Language Structures'' (Haspelmath et al. 2005, 2008). Chapter 3 introduces major types of reduced referential devices, viz. bound and free pronouns and zeroes, and shows that languages are not always internally consistent in their referential choice. Kibrik also introduces the important distinction between tenacious pronouns (which co-occur with co-referential NPs) and alternating pronouns (which do not co-occur with co-referential NPs). The last portion of this chapter is concerned with zero reference, which appears to be independent of activation (but see Givón 1983 for another view), syntactic position or semantic role. Chapter 4 is dedicated to an in-depth discussion of pronouns and related devices (demonstratives, classifiers, and social status nouns). He shows that the latter are particularly common in languages with zero reference, suggesting that the lack of pronouns leads to drafting of other categories into referential service. Kibrik also discusses pronouns with double reference and pronouns marked for clausal categories such as TMA or polarity. He proposes a three-way typology with respect to boundness; the three parameters are the expression of 1. the lexical meaning of the verb, 2. clausal categories, and 3. reference. The last subsection is concerned with strong, or accented, pronouns. Chapter 5 looks at languages that do not consistently choose the same reduced referential device. Kibrik argues against the view (held e.g. by Givón 2001) that this choice is always due to different levels of activation and claims instead that some languages are sensitive to coordination, definiteness, or other factors. Chapter 6 is a detailed treatment of bound pronouns, particularly the parameter of tenacity (the ability to occur with a co-referential NP) and its independence from boundness. Kibrik argues against the view that bound tenacious pronouns should be considered agreement markers and discusses in detail the Germanic system with both verbal agreement markers and free weak pronouns. The last subsection deals with bound zeroes. Chapter 7 focuses on the diachronic development of bound tenacious pronouns in three language families: Athabascan, Slavic, and Romance, showing that different branches of a language family can undergo different developments.
Part III is dedicated to the discussion of referential aids. In Chapter 8, Kibrik distinguishes ad hoc devices such as semantic incompatibility, from conventional referential aids, which sort activated referents based on a distinctive feature (p. 294). He discusses in turn absolute stable sortings (taxonomies, noun classes), relative stable sortings (hierarchies), and broad (4th person in Navajo, proximate/obviative systems, and inversion) and narrow (logophoricity, topicality/subjecthood, and activation status) current sortings, considering for each of these where they are found and the locus of expression (on free or bound pronouns or as part of the verb stem). The chapter closes with a classification of referential aids and introduces the locus of its expression as a typological parameter. Chapter 9 considers the functional load of referential aids across languages. Kibrik finds considerable cross-linguistic variation with respect to both the number of referential aids used by a given language, and their respective importance. He then pursues the question of whether the avoidance of referential conflict is the primary function of conventional referential aids. Two case studies from Pulaar and Sereer, two related Senegalese languages, show that, while the languages have comparable noun classification systems, only Pulaar uses the system as referential aid.
Part IV presents the Cognitive Multi-Factorial approach (CMF). Chapter 10 begins with a discussion of attention and working memory and their application to discourse study. Kibrik summarizes the CMF: various discourse factors (e.g. referential distance, episode boundary, a referent’s centrality to the discourse and others) contribute to a referent’s activation. Each factor has a numerical value; the sum of values for each factor is numerical expression of a referent’s activation score. Three thresholds of activation can be identified: Above the first one, only reduced referential devices are possible; between the first and second, reduced devices are preferred; between the second and third, full NPs are preferred; below the third, only full NPs are possible. This numerical approach, according to Kibrik, solves the problem of circularity inherent in many approaches to discourse reference. The CMF is applied to a written Russian text in Chapter 11 and to a written English text in Chapter 12. In both chapters, Kibrik discusses the impact of various activation factors in each language. He also introduces two filters which operate independently from the other activation factors: The “world boundary filter” and the “referential conflict filter”. In the conclusion to Chapter 12, Kibrik observes that “[t]he general model of referential choice is supposed to be universal but the set of activation factors, their relative numerical weights, and thresholds for the activation status range are language-specific” (p. 444). Chapter 13 is concerned with the CMF’s significance for cognitive linguistics in general. Kibrik demonstrates for instance that the sum of activation scores for all active referents within an elementary discourse unit is rarely greater than 4, which is also approximately the number of items that can be held in working memory. He also considers the question of activation decay (= forgetting from working memory) and how discourse linguistics can further contribute to cognitive science. In Chapter 14, Kibrik points out several problems with the method presented in Chapters 11 and 12. To address these issues, he employs an artificial neural network, using the activation factors identified earlier as weighted nodes. Kibrik then considers several other corpus studies with similar results, and finally reports on the application of the CMF to Japanese, a zero reference language.
In Part V (Chapter 15), the focus is widened to include visual aspects of reference. Kibrik looks at the related phenomena of deixis and exophora with an emphasis on their link to first-mention and anaphora. He then discusses reference in Russian Sign Language and virtual pointing in Russian speech.
In the final chapter, Chapter 16, Kibrik summarizes his findings, briefly discusses aspects of reference that were not included in this study, and offers a preview of future research.
The book also includes an appendix for fieldworkers that may be used as a guide to the collection structured data on reference.
EVALUATION The monograph’s first goal was to provide an overview over reduced referential devices and referential aids in the world’s languages. Using discourse examples from over 200 languages, Kibrik amply demonstrates the variety and identifies typological parameters. He makes extensive use of the studies included in the ''World Atlas of Language Structures'' (Haspelmath et al. 2005, 2008) to identify correlations between these parameters and to set up typologies based on them. He also investigates the frequency of particular structures.
Throughout this study, Kibrik discusses a number of very important points that are highlighted here. First of these is his view of reference as a discourse phenomenon rather than a syntactic one. This is not to say that he disagrees with the fact that constructions such as zero reference in coordination structures in English (shown in (1a)) or the use of reflexive pronouns in the case of coreferentiality of subject and object (shown in (1b)) have a syntactic motivation.
(1a) Jill-i went home and Ø/*she-i/*Jill-i went to bed. (1b) Jack-i looked at himself/*him-i/*Jack-i in the mirror.
Instead, he argues, the syntactic patterns illustrated in (1) are largely grammaticalizations of discourse patterns; syntax is thus a special case of discourse (p. 36). This point is not only theoretically important, but it also impacts the type of data that can be used for the study of reference. If reference is viewed as a discourse phenomenon, then it can only be studied using discourse data, not sentences uttered in isolation (or constructed by the linguist). This constraint is taken very seriously by Kibrik, and much of the data in this book is taken from natural or naturalistic discourse (wherever constructed data is used, this is indicated in the discussion of the example).
I did however find the PRESENTATION of data in this part of the book difficult to follow. Textual data is presented in a two-line format. The first line contains a transcript of the data indicating morpheme breaks, the second line contains a morpheme gloss. The free translation is given only at the end of the example, which, due to the length of the textual examples, is frequently on the following page. A randomly picked example of this is the second line of (5.8), p. 173. To accommodate the email format, I do not include the original Masa (Chadic) line. The second line of this example is glossed as shown in (2):
The free translation of this line appears on the following page (p. 174): ‘Since he has been killed, we start cooking him’. To someone with no knowledge of Chadic languages, this gloss is not entirely predictable from the morpheme glosses, so in order to follow the points illustrated by this example, I had to flip from data to free translation multiple times, and I had to do this for many, many examples. Presenting discourse data that includes morphemic analysis is notoriously difficult because of the amount of information involved, and I understand the rationale behind presenting the free translation of an example as coherent text at the end of the example. However including free phrase translations as a third line would have facilitated reading this book greatly, and would have been worth the increased page count.
A second important point raised by Kibrik concerns the establishment of equivalence between bound and free pronouns. He shows convincingly that not only do bound pronouns fulfill the same functions as free pronouns in that they are referential and case-marked, but also that the parameter of tenacity (frequently taken as evidence for an interpretation of bound pronouns as agreement markers) operates independently from that of boundness. This leads to interesting typological findings, such as the fact that the ‘exotic’ pattern of bound tenacious pronouns, as present e.g. in Abkhaz, is more common cross-linguistically than the ‘basic’ pattern of free alternating pronouns in the Germanic languages.
Here again I had a small concern. Kibrik (pp. 94-95) strongly argues in favor of verb-marked case, and replaces the common terminology of ‘subject’ and ‘object’ pronouns in Athabascan, for example, consistently by the labels ‘nominative’ and ‘accusative’ pronouns (see e.g. p. 209 for a discussion of Navajo). When discussing Russian on the other hand, he talks about ‘subject’ and ‘object’ pronouns, even though the labels ‘nominative’ and ‘accusative’ pronouns would be perfectly accurate in this situation.
Excepting these few minor issues, I was very impressed by Parts I-III. Both the breadth of this overview -- the assembly of discourse data from that many languages is no mean feat, since grammatical accounts frequently do not include a sufficient number of textual examples -- and its depth -- the organization and analysis of such an amount of textual data is challenging and time-consuming -- are remarkable. Furthermore, much of the data stems from Kibrik’s own fieldwork on a number of unrelated languages, which demonstrates not only an impressive fieldwork record, but also bestows additional authority on this account. Another interesting aspect is the inclusion of the Russian research tradition on discourse; since much of that is produced in Russian, it is not as accessible to many linguists as, say, the American research tradition instigated by Chafe (1980, 1994) or Givón (1983).
Part IV, the discussion of the cognitive foundations of basic referential choice, is more difficult to evaluate. The links between discourse linguistics and cognition drawn in chapters 10 and 13 were very interesting, and Kibrik reports on a number of corpus studies on referential choice in chapter 14. While I am not a cognitive linguist, I found these chapters illuminating and clear. I did however have some serious concerns regarding chapters 11 and 12. In these two chapters, Kibrik presents a way to calculate the activation status for each mention by assigning a numerical weight (positive or negative) to each activation factor and adding them up. In order to figure out whether a particular referential choice is obligatory or optional, Kibrik gives altered variations of the text to research participants and asks them to evaluate whether the substitution of, e.g., a pronoun for a full NP is acceptable or not. This experiment establishes three thresholds for referential choice. This technique is applied to a Russian written text in chapter 11 and to an English written text in chapter 12.
While the discussion of the methodology is for the most part very clear and inclusive, Kibrik presents us with the weights of activation factors in Russian on p. 411 (in chapter 11) and English on p. 436 (in chapter 12), but does not explain how he arrived at these weights until p. 441. As a result, one has to read the whole discussion of Russian without knowing why particular weights have been assigned to various activation factors. I found this a tad frustrating, since it could easily have been avoided by switching the order of chapters 11 and 12.
My second concern regards the methodological issues in the calculation of the activation score, all of which Kibrik lists on pp. 460-461 and addresses by developing a neural network that a) identifies those factors that don’t significantly contribute to activation, b) is not as labor-intensive as the assigning of numerical weights by hand, c) recognizes that there may be non-linear dependencies between activation factors, and d) ensures all activation scores are neatly between 0 and 1. The resulting neural network described on pp. 461-470 seems well-designed and remarkably robust in generating reliably the predicted (i.e. original) referential device, even after substantial pruning. This model however raised a question: If such an advanced technique is available, why then spend two whole chapters (about 80 pages) on a manual, time-consuming, and problematic technique? Given that the results from these studies have been published since the late 1990s (see chapter 11, footnote 1; chapter 12, footnote 1), it would have been possible to briefly summarize the relevant findings, use them as input for the computer model, and then move on to other topics.
Part V comprises only one chapter, where Kibrik includes visual evidence from pointing and also from sign languages. Again, I am not a specialist and cannot comment on the accuracy or innovativeness of the chapter. I was a little surprised about the chapter’s placement; it appears to be almost an afterthought to the book. It would have been more intuitive to me had it immediately followed the typological chapters, since this would have supported Kibrik’s view that “the discussion of reference in human languages would be grossly incomplete if I totally ignored the evidence of sign languages in this book” (p. 499).
To sum up: Kibrik has clearly succeeded in providing the reader with an overview not only over the linguistic variation and typological tendencies of reference, but also over the different research traditions in East and West. The book is extremely comprehensive and very clearly written. I am not sure whether parts of it would be suitable as a “sort of a textbook” (p. 552) even at the graduate level, since the chapters build very strongly upon each other, so that isolating portions for students to read could be difficult. The book could rather be used for self-study both by graduate students and others working in the field. In particular, Kibrik succeeds in pointing out many avenues for future research: by providing a questionnaire for fieldworkers, by identifying typological parameters, and by demonstrating that computer modeling and corpus linguistics both are valuable tools in the investigation of reference. In short, this book is an invaluable and very accessible resource for those who want to read up on discourse reference as quickly as a book of this heft permits.
REFERENCES Chafe, Wallace L. (ed.) 1980. The pear stories: Cognitive, cultural and linguistic aspects of narrative production. Norwood, N.J.: Ablex.
Chafe, Wallace L. 1994. Discourse, consciousness, and time: The flow and displacement of conscious experience in speaking and writing. Chicago: The University of Chicago Press.
Givón, Talmy. 1983. Topic continuity in discourse: A quantitative cross-linguistic study. Amsterdam: John Benjamins.
Givón, Talmy. 2001. Syntax: A functional-typological introduction. Amsterdam: John Benjamins.
Haspelmath, Martin, Matthew S. Dryer, David Gil, and Bernard Comrie. 2005. World Atlas of Language Structures. Oxford: Oxford University Press.
Haspelmath, Martin, Matthew S. Dryer, David Gil, and Bernard Comrie. World Atlas of Language Structures. Max Planck Digital Library. http://wals.info/ 2008.
ABOUT THE REVIEWER
ABOUT THE REVIEWER:
Olga Lovick is an Associate Professor of Linguistics and Dene Language
Studies at the First Nations University of Canada. Her research interests
include discourse linguistics, conversation analysis, and prosody in the
Athabascan languages and beyond.