Пожалуйста, введите доступный Вам адрес электронной почты. По окончании процесса покупки Вам будет выслано письмо со ссылкой на книгу.

Выберите способ оплаты
Некоторые из выбранных Вами книг были заказаны ранее. Вы уверены, что хотите купить их повторно?
Некоторые из выбранных Вами книг были заказаны ранее. Вы можете просмотреть ваш предыдущий заказ после авторизации на сайте или оформить новый заказ.
В Вашу корзину были добавлены книги, не предназначенные для продажи или уже купленные Вами. Эти книги были удалены из заказа. Вы можете просмотреть отредактированный заказ или продолжить покупку.

Список удаленных книг:

В Вашу корзину были добавлены книги, не предназначенные для продажи или уже купленные Вами. Эти книги были удалены из заказа. Вы можете авторизоваться на сайте и просмотреть список доступных книг или продолжить покупку

Список удаленных книг:

Купить Редактировать корзину Логин
Поиск
Расширенный поиск Простой поиск
«+» - книги обязательно содержат данное слово (например, +Пушкин - все книги о Пушкине).
«-» - исключает книги, содержащие данное слово (например, -Лермонтов - в книгах нет упоминания Лермонтова).
«&&» - книги обязательно содержат оба слова (например, Пушкин && Лермонтов - в каждой книге упоминается и Пушкин, и Лермонтов).
«OR» - любое из слов (или оба) должны присутствовать в книге (например, Пушкин OR Лермонтов - в книгах упоминается либо Пушкин, либо Лермонтов, либо оба).
«*» - поиск по части слова (например, Пушк* - показаны все книги, в которых есть слова, начинающиеся на «пушк»).
«""» - определяет точный порядок слов в результатах поиска (например, "Александр Пушкин" - показаны все книги с таким словосочетанием).
«~6» - число слов между словами запроса в результатах поиска не превышает указанного (например, "Пушкин Лермонтов"~6 - в книгах не более 6 слов между словами Пушкин и Лермонтов)
 
 
Страница

Страница недоступна для просмотра

OK Cancel
E ssays on the E -S volutionary ynthetic T L heory of anguage ALEXEY KOSHELEV ESSAYS ON THE EVOLUTIONARY-SYNTHETIC THEORY OF LANGUAGE On the Crisis in Theoretical Linguistics Basic Meaning and the Language of Thought The Unity of Lexical and Grammatical Polysemy Translated by Alexander Kravchenko in collaboration with Jillian Smith Moscow & Boston 2019 Library of Congress Cataloging-in-Publication Data: Names: Koshelev, A. D. (Alekseæi Dmitrievich), author. | Kravchenko, A. V. (Aleksandr Vladimirovich), translator. | Smith, Jillian, translator. Title: Essays on the evolutionary-synthetic theory of language / Alexey Koshelev ; translated by Alexander Kravchenko with Jillian Smith. Other titles: Ocherki çevolëiìuëtìsionno-sinteticheskoæi teorii ëiìazyka. English Description: Boston : Academic Studies Press, [2019] | Includes bibliographical references and index. | Translated from Russian. Identifiers: LCCN 2018057558 (print) | LCCN 2018058089 (ebook) | ISBN 9781644690031 (ebook) | ISBN 9781644690024 (hardcover) Subjects: LCSH: Language and languages--Philosophy. | Language and languages--Origin. | Cognitive grammar. Classification: LCC P107 (ebook) | LCC P107 .K72613 2019 (print) | DDC 401--dc23 LC record available at https://lccn.loc.gov/2018057558 Copyright © 2019 LRC Publishing House Copyright © 2019 А. D. Koshelev Copyright © 2019 A. V. Kravchenko & А. D. Koshelev, translation into English All rights reserved ISBN 978-1-64469-002-4 (hardback) ISBN 978-1-64469-003-1 (electronic) Typeset by LRC Publishing House Russia, Moscow, Bolshaya Lubyanka, 13/16 http://www.lrc-press.ru/?lang=en Published by Academic Studies Press in 2019 28 Montfern Avenue Brighton, MA 02135, USA press@academicstudiespress.com www.academicstudiespress.com Contents Acknowledgements .................................................................................................. ix Preface. . ........................................................................................................................x 1. The place of the evolutionary-synthetic theory of language among the ranks of linguistic theories...................................................................x 2. Peer reviews of the evolutionary-synthetic theory. . ............................................... xii 3. About the book contents. . ...................................................................................... xiii Chapter 1 On the contradictory nature of contemporary linguistic theories. and how to change it for the better. . ..........................................................................1 § 1. Introduction............................................................................................................1 § 2. A compendium of incompatible linguistic frameworks ........................................3 § 3. Contrastive analysis of some mutually contradicting linguistic theories...............7 1. Introduction.......................................................................................................7 2. On the role of language in thought...................................................................8 3. Is language an autonomous module (“cognitive organ”)? .............................12 4. Other opposing assumptions in the theories under discussion. . ......................13 § 4. Contradictory descriptions of particular linguistic problems...............................15 1. The incompatibility of theories of lexical polysemy......................................15 2. Contradictory descriptions of the syntax-lexicon interface............................16 3. Theories of language origin and evolution.....................................................16 § 5. The futility of attempts to reach a consensus ......................................................17 1. Unproductiveness of scholarly polemics........................................................17 2. Inefficiency of interdisciplinary approaches...................................................18 § 6. Is linguistics a natural science?............................................................................19 § 7. The crisis and its causes.......................................................................................22 1. Is “multiparadigmality” typical of linguistics?...............................................22 2. The status of linguistics as a science. . .............................................................22 3. The root of trouble..........................................................................................24 Contents vi § 8. Principles of an evolutionary-synthetic theory of language.................................26 § 9. Crisis in the cognitive sciences ...........................................................................27 1. Contradictory frameworks in cognitive sciences............................................27 2. The cause of the crisis.....................................................................................31 3. Towards a unified cognitive paradigm............................................................33 Chapter 2 A reference-based approach to describing notional words...................................35 § 1. A reference-based approach to lexical semantics.................................................35 1. The dual structure of lexical meaning . ..........................................................35 2. Basic meanings of the words chair and armchair .........................................50 § 2. An analysis of basic meanings of the action verbs udarit’ ‘hit-PF’, tolknut’ ‘push-PF’, padat’ ‘fall-IMP’, brat’ ‘take-IMP’ and vzbirat’sja ‘climb-IMP’ ................................................................................58 1. Introductory remarks . ....................................................................................58 udarit’ kosnut’sja 2. Contact verbs: ‘hit-PF’, ‘touch-PF’, tolknut’ ‘push-PF’...........................................................................................60 3. The verb padat’ ‘fall-IMP’ . ...........................................................................67 4. The verb brat’ ‘take-IMP’ / vzjat’ ‘take-PF’...................................................69     5. The verb vzbirat’sja ‘climb’ ...........................................................................78 § 3. Verbs of spatial orientation. : . stojat’ ‘stand’, sidet’ ‘sit’, ležat’ ‘lie’, and viset’ ‘hang’ . ..............................................................................83 1. The verb viset’ ‘hang’. . ....................................................................................83 stojat’ ležat’ sidet’ 2. The verbs stand’, ‘lie’, and ‘sit’.......................................89 idti bežat’ polzti § 4. The motion verbs. . ‘walk’, ‘run’, ‘crawl’, šagat’ ‘step/stride’, prygat’ ‘jump’, and exat’ ‘go/ride/drive’..............................94 1. The motion verbs idti ‘walk’ and bežat’ ‘run’. . ...............................................94 2. A comparative description of the verbs idti ‘walk’, bežat’ ‘run’, polzti ‘crawl’, prygat’ ‘jump’ and šagat’ ‘step/stride’.................................110 3. Basic meaning of the verb exat’ ‘go/ride/drive’ . .........................................112 § 5. The structure of lexical polysemy......................................................................117 1. Reference in lexicology ...............................................................................117 2. Lexical polysemy..........................................................................................119 3. Supplement. Three contemporary approaches to lexical polysemy.............127 § 6. Appendix 1. Excerpts from the email correspondence between A. D. Koshelev and I. A. Mel’čuk (February—March 1995). .............133 § 7. Appendix 2. An infant’s early acquisition of the laws of nature .......................141 1. An infant’s view of the physical world.........................................................141 2. Formation of causal relationships.................................................................143 Contents vii Chapter 3 Basic-level concepts as the neurobiological codes for memory. ..........................146 § 1. Concrete concepts and motor concepts..............................................................146 1. Introduction. Two interpretations of basic-level concrete concepts. . ............146 2. Functional schema of basic-level concepts...................................................147 3. The function of an object and a linguistic explanation of this function. . ......149 4. A linguistic explanation of the concept functional schema...........................151 5. Neurobiological grounds for the basic concept schema...............................152 6. Motor concepts. . ............................................................................................154 7. Concrete concepts tropinka ‘footpath’, dorožka ‘track’, doroga ‘road’............................................................................................ 156 8. Concepts ozero ‘lake’ and reka ‘river’. . ........................................159 9. The concept derevo ‘tree’....................................................................161 10. Appendix. On the dual nature of human categories....................................163 § 2. On the psychophysical state and the neurobiology of human actions . .............164 1. Events and their storage in memory (the neural codes of memory).............164 2. Psychophysical state as a memory code for interaction. . ..............................166 3. Biomechanical models of walking and running. . ..........................................170 4. On recognition of observed actions..............................................................172 5. Mirror neurons and action recognition.........................................................173 6. Canonical neurons and object recognition....................................................176 Chapter 4 Elements of a sensory grammar............................................................................178 § 1. On the basic and derivative meanings. . of nominal genitive constructions in Russian (ručka dveri handle-Nom door-Gen, vetka dereva branch-Nom tree-Gen).................................................................178 1. The ‘part-whole’ relationship and its linguistic representation.....................178 2. Objects and their functional parts.................................................................181 3. The basic meaning of the nominal genitive construction Y X-Gen. ..............182 4. The derivative meanings of the nominal genitive.........................................183 5. Objects with a multilevel partitive structure.................................................183 6. The partitive structure of plants....................................................................184 7. The partitive structure of animals and humans.............................................186 § 2. On structural and genetic similarity. . of lexical and grammatical categories. The meaning of transitivity ...............................................................................189 1. The structure of basic meaning.....................................................................189 2. The structure of polysemy. . ...........................................................................189 3. The meaning of transitivity...........................................................................190 Contents viii § 3. The basic and derivative meanings of voice. The active, passive, and reflexive meanings......................................................194 1. A cognitive approach to the analysis of voice meanings..............................194 2. The active voice............................................................................................198 3. The reflexive voice: myt’sja ʻwash-Reflʼ, kutat’sja ʻmuffle-Refl up [in something]ʼ, brit’sja ʻshave-Reflʼ...........................................................198 4. The passive voice: mašina moetsja ʻthe car is washing-Reflʼ, bel’ё sušitsja dom stroitsja plotnikom ʻthe linen is drying-Reflʼ, ʻthe house is building-Refl by the carpenterʼ...............................................200 Chapter 5 On the single structure of lexical meanings of nouns and verbs. . .......................203 § 1. Object and the system of its parts (the partitive concept)..................................203 1. Object and its parts. . ......................................................................................203 2. The function of an object as the sum of the functions of its parts ...............204 3. The role-based hierarchy of object parts.......................................................205 4. The developed concept. . ................................................................................207 5. How is the main part of an object determined? ...........................................210 6. The functional and physical parts of an object.............................................210 7. On the radial position of the parts of an object ............................................211 8. A developed concept represented as a circle.................................................212 9. The recursiveness of a partitive concept ......................................................213 10. A vocabulary of elementary cognitive units . .............................................214 11. On distinguishing physically connected objects ........................................217 § 2. An object and the system of its properties (the attributive concept). The basic meaning of a concrete noun . ............................................................217 1. An object and its properties. . .........................................................................217 2. Conceptualization of concrete properties. . ....................................................219 3. The partitive-attributive extension of a concrete concept and the basic meaning of a concrete noun . .................................................225 § 3. A motor concept and its two extensions: partitive and attributive. The basic meaning of an action verb .................................................................................228 1. The motor concept sxvatit’ ‘grasp’ (seize an object by the hand).......228 2. The action bežAt’ ‘RUN’...........................................................................230 3. The attributive motor concept.......................................................................231 4. The basic meaning of action verbs . .............................................................231 References................................................................................................................233 Index. ........................................................................................................................249 Acknowledgements This book represents the result of many years of research. Throughout that period I have benefited from the support, advice, recommendations, and critical feedback of a large number of people, from close friends and colleagues to anonymous readers. Unfortunately, in all of my researches I have not been able to find one like-minded associate who shared the fundamental principles of the theory of language presented here. Only recently, when in the final stages of writing this book, I at last found such a like-minded individual. I am very glad that this person proved to be so well-known and authoritative a linguist as Liudmila Zubkova. Her support, advice, and critical comments were extremely important to me, and it is to her that I give my first words of thanks. I would also like to express my most sincere gratitude to my family, friends, and colleagues, who have helped me throughout the whole of this work through their discussions, comments on sections of the manuscript, and more. These are: Grigorii Bondarenko, Sergei Zhigalkin, Viacheslav Ivanov, Aleksei Kas’ian, Andrei Kibrik, Mikhail Kozlov, Sergei Krylov, Leonid Krysin, Iulia Mazurova, Igor’ Mel’čuk, Nikolai Pertsov, Tat’iana Samarina, Tat’iana Skrebtsova, Georgii Starostin, Iakov Testelets and Ekaterina Iakovleva. I would also like to thank the many anonymous reviewers for their comments. I am very grateful to the translators Alexander Kravchenko and Jillian Smith, who helped me to correct a number of inaccuracies, and to the editor Vera Stoliarova, for her tireless work on the text throughout its journey to the press. Preface xi 1.2. Causes of the current crisis and ways to overcome it. Regarding the original cause of this theoretical dead-end (pp. 22–26), it is necessary to point out two fundamental flaws inherent (separately or in combination) to modern linguistic theories. Firstly, the synchronic approach, which still predominates, ignores the processes of macro- and microevolution of language, its genealogy, and development. Secondly, when describing the mechanisms of language complex functions, these theories do not take into account the contribution of the related subsystems of thought, knowledge representations, memory, etc. Overcoming these two flaws is the aim of the Evolutionary-Synthetic Theory of Language, a preliminary version of which is presented in this book. 1.3. General principles underlying the Evolutionary-Synthetic Theory of Language. The theory here set forth harkens back to antiquity, to Aristotle’s conceptualization of the divide between reality, thought, and language, and the role of language as an instrument for describing the mental representations humans have of the world. According to this idea, language does not function in isolation but in close interaction with other subsystems. This principle is implemented within the framework of the synthetic component of this theory by means of a systematic merger of the basic concept of language with the knowledge and findings of certain other theories closely related to linguistics. These are theories of: (a) basic concepts (Rosch et al. 1976; Lakoff 1986 and inter alia), 1987; Mervis and Rosch 1981; Mervis 1987, (b) thought (Sechenov 1978), (c) motor control (Bernstein 1967: 223, 227–228), (d) visual perception and event recognition (Johansson 1973 and 1976; Bingham et al. 1995; Runeson 1977, inter alia), (e) neurobiological theory of memory (Tsien 2008). Selected results from research in other disciplines have also been drawn upon in the formation of the present theory. The second principle—the study of language in macro- and microdiachrony—is implemented through a very broad use of general development theory (pp. 31–33). This theory is involved in the analysis of a wide range of language issues, from lexical representation and grammatical meaning to explanations of glottogony and the overall structure and evolution of language. The evolutionary-synthetic theory of language is conceived as an alternative to the many incompatible theories now existing. Naturally, the question arises: is it possible to develop alternatives to these theories? Our answer is: yes, it is possible. But since such alternatives are also bound to be complex theories, there cannot be more than one or two of them. This situation, however, is characteristic of normally developing sciences. Preface xii 2. Peer reviews of the evolutionary-synthetic theory When a new theory emerges, the first peer reviews from specialists are of particular interest. In Zubkova’s (2017, Moscow) monograph, Language Theory in its Development, chapter 15 contains an analytical review of the theory of language presented in the Russian edition of the A. D. Koshelev’s (2017, Moscow) monograph, Essays on the Evolutionary-Synthetic Theory of Language. I provide one excerpt below: The relationship between the world of external events and the inner world of man, between reality, thought, and language, occupies the central place in this theory of language. However, the fact is that the actual mechanisms connecting these two worlds remained undiscovered until the end of the 20th century. The elucidation of these mechanisms became possible through A. D. Ko shelev’s development of a cognitive theory of semantics which combines a referential approach to language with a concept-based approach. The inspiration for this approach, it appears, is the contrast distinguished in antiquity between two worlds: the “visible domain” (the perceptible world) on the one hand, and the “intelligible domain” (the world of ideas) on the other (Zubkova 2017: 534) [translation J. Smith]. In the preface, while analyzing the contribution of various linguistic concepts in the development of a general theory of language, Zubkova writes: A. D. Koshelev takes the next crucial step, toward a synthesis of cognitive and linguistic abilities in their interrelated development, toward a systematic unity of universal and specific, abstract and concrete in language. He begins with patterns in the cognition of objects, from the most basic and holistic representations of an object to the delineation of its distinct features and, further, to the synthesis of the accumulated knowledge of the various facets of an object into a cohesive system for its representation (Ibid.: 20) [translation J. Smith]. Another analysis of the present theory is given in chapter 9, section one, of Skrebtsova’s (2018) monograph, Cognitive Linguistics: classic theories and current approaches. Preface xiii 3. About the book contents This book is a translation of the first part of the Russian edition of Essays on the Evolutionary-Synthetic Theory of Language (Koshelev 2017) and includes chapters 1–4 and §§ 23–25 from chapter 5. In chapter 1, it is argued that over the past 50 years theoretical linguistics has been in a state of crisis. It has remained a compendium of mutually contradicting theories on multiple levels: the level of general theories of language, the level of its main constituents (the lexicon, syntax, and the syntax-semantics interface that connects them), and the lower levels of specific linguistic problems (such as lexical polysemy, lexical-semantic combinability, grammatical meanings, etc.). The Evolutionary-Synthetic Theory of Language aims to overcome this crisis. It implements an interdisciplinary approach in describing language (a) in its ontogenetic development, and (b) in its close interrelationship with other human subsystems: thought, memory, actions, etc. The chapters that follow deal with the semantic component of the evolutionary-synthetic theory. In chapter 2, a reference-based approach to the analysis of basic meanings of concrete nouns and action verbs (that is, object and motor basic-level concepts) is proposed. Two types of cognitive units (elements of the language of thought) for representation of these meanings are singled out—perceptual, accessible for sensory identification, and functional, reflecting human intentions—along with the relationship of interpretation that connects them. The mechanisms for lexical polysemy are also analyzed. Chapter 3 brings to light the structural unity of artifacts and natural concepts (stul ‘chair’, tropinka ‘footpath’, doroga ‘road’, ozero ‘lake’, reka ‘river’, derevo ‘tree’, etc.); object and motor concepts are defined in terms of the language of thought, and their representation in neurobiological memory codes is discussed. In chapter 4, the mechanisms for grammatical polysemy are analyzed, and the structural unity of lexical and grammatical polysemy is brought to the fore. Finally, in chapter 5, the notions of attributive and partitive concepts (that is, the systems of object properties and parts) are introduced; using concrete examples as illustrations, the hierarchic structure of basic meanings of concrete nouns is shown to arise as a result of their step-by-step development in ontogeny. It is shown that basic-level concepts belong to the first level, and attributive and partitive concepts, which develop from basic-level concepts, to the second level of the hierarchy. Preface xiv All quotations from the Russian sources have been translated by Alexander Kravchenko unless otherwise indicated. The transliteration system ISO 9 – 1995 has been employed throughout the book. The ALA-LC Romanization Tables have been used for transliterating personal names. Chapter 1 On the contradictory nature of contemporary linguistic theories... and how to change it for the better § 1. Introduction Our attention will be focused on an analysis of the inherent contradictoriness of contemporary theoretical linguistics that harbors a number of mutually contradicting theories of language. More specifically, arguments will be given in support of the following claims: 1. Over the past 50 years theoretical linguistics has nearly failed to add anything new to the universally recognized facts about language as a global object of study. Throughout several decades it has remained a compendium of mutually contradicting doctrines on multiple levels: the level of general theories of language, the level of its main constituents (the lexicon, the syntax, and the syntax-lexicon interface that connects them), and the lower levels of specific linguistic problems (such as lexical polysemy, syntax-lexicon combinability, grammatical meanings, etc.). 2. This contradictoriness of contemporary linguistic theories is indicative all of a deep crisis. However, it does not follow that of contemporary linguistics is in a state of crisis. The main issue is that theoretical linguistics does not seem to build on direct intuitions of native speakers. As for the concrete research that takes linguistic intuitions into account, it has been growing in scope and in many different directions. 3. It is impossible to overcome this obvious crisis without working out a unified evolutionary-synthetic theory of language. This task, in turn, cannot be solved without (a) taking into account not only the synchronic, but also the diachronic regularities in linguistic functions, and (b) a comprehensive use of research data on the functional regularities of other subsystems in humans, 1. On the contradictory nature of contemporary linguistic theories... 2 such as thought, knowledge representation, etc., that closely interact with the linguistic subsystem. It should be noted that the parallel existence of mutually contradictory theories in contemporary linguistics has often been pointed out. The phenom 1 itself, however, receives different interpretations. According to some linguists, it is the consequence of linguistic theory sliding into a stagnation phase, the signs of which have recently become obvious (Beaugrande 1991: 2). Others attribute it to the very nature of the object of study (Bahner 1983), while some, following Kuhn (1962), consider this phenomenon as evidence that linguistics is at an early stage of its development (“pre-paradigmatic stage”) and has not yet reached the level of maturity that could be expressed by a single dominant paradigm. Finally, there is yet another point of view: that the apparent contradictoriness of many linguistic theories is not of a fundamental nature and can be overcome as these theories continue to develop and come to a point where they may be united. As an illustration of this quite common view, consider Fitch’s (2010) The Evolution of Language. Fitch gives his view on the problem with the study of language at the very beginning of the book by reiterating an old Indian parable about the blind men and the elephant. Taking this parable, he draws an analogy between linguists and the blind men, who, by feeling the elephant with their hands, try to give a full description of it: The palm of one fell on the trunk. “This creature is like a water-spout,” he said. The hand of another lighted on the elephant’s ear. To him the beast was evidently like a fan. Another rubbed against its leg. “I found the elephant’s shape is like a pillar,” he said. […] A core argument in this book is that each of the scholars has grasped some truth about language, but that none of these truths are complete in themselves. Language, I will argue, requires the convergence and integration of multiple mechanisms, each of them necessary but no one alone sufficient (Fitch 2010: 1, 3). In what follows, an attempt will be made to show that the real situation in contemporary linguistics is better described by a different parable: the blind men are feeling not the same elephant which symbolizes language, but several different animals—an elephant, giraffe, rhinoceros, etc.—which symbolize different models of language. One touches the trunk of an elephant, another the neck of a giraffe, a third the horn of a rhino, and they all believe that 1 Some recent examples in Russian are Zubkova (2015), Kasevich (2009), Koshelev (2013a and 2015b); Kravchenko (2015). § 2. A compendium of incompatible linguistic frameworks 3 this is one and the same animal. Clearly, hopes for integrating the resultant descriptions into a single picture in such a case are vanishingly small. Chapter 1 focuses on an examination of some modern linguistic frameworks, both general (various linguistic theories in § 2, 3) and specific (evolution of language, lexical semantics, etc. in § 4). In § 5 the futility of attempts to reach a consensus via polemics and interdisciplinary contacts is discussed; in § 6 the status of linguistics—whether it is a natural science—is analyzed; in § 7 the causes of the crisis in linguistics are discussed; in § 8 the principles of an evolutionary-synthetic theory of language are formulated, and § 9 offers a discussion of a similar crisis observed in other cognitive sciences, first of all in psychology. It is shown that, just like linguistics, these sciences are ridden with conceptual contradictoriness that impedes further development. Possible ways out of the theoretical dead-end are then discussed. § 2. A compendium of incompatible linguistic frameworks If we take an overview of contemporary linguistic research as a whole, the emerging picture will be somewhat paradoxical. We will not find a single, universally recognized theory of language (linguistic framework, school of thought) or at least two competing theories. What we will find will be a large number of mutually contradictory theories developed by Noam Chomsky, Ray Jackendoff, Igor’ Mel’čuk, George Lakoff, Talmy Givón, Anna Wierz bicka, Ronald Langacker, Charles Fillmore, and others. The reason for this is that a newly developed theory does not replace an already existing one by more adequately representing and explaining reality; rather, it begins to coexist with the theories developed earlier. Of course, many schools of linguistic thought are of interest now only from a historical perspective. It is not often that we find references to the theories developed by Hermann Paul, Karl Bühler, Leonard Bloomfield, and other linguists of the past. However, the theories that go out of use do not lose their topicality because the newly arriving doctrines have somehow proven them to be erroneous or deficient in one way or another. They are simply forgotten, pushed out to the background by the more “modern” approaches.  2 2 Remarkably, not any of the new theories of language has become a successor to the three frameworks mentioned. That is to say, each new framework begins to describe language anew by creating its own linguistic paradigm. In the course of this process, some very important achievements of the older theories are often ignored. For example, some findings of Paul (1970) and Bloomfield (1973), neglected by modern schools of thought, 1. On the contradictory nature of contemporary linguistic theories... 4 Meillet (1928) was perhaps one of the first to point out this feature of 20 th century linguistics: Mais, tandis que la linguistique historique a des méthodes précises, éprouvées par un long usage, la linguistique générale est encore mal assurée dans sa démarche; elle tient toujours un peu de la philosophie: chaque auteur procède à sa manière et il y a, semble-t-il, autant de linguistiques générales que de linguistes. Pour que la linguistique générale progresse, il faut qu’elle devienne plus objective et qu’on en fixe la technique. Pour se bien entendre, on a besoin tout d’abord d’une terminologie où les termes aient pour tout le monde le même sens. Or, il apparaît du premier coup une difficulté fondamentale qui tient à la nature du fait linguistique  3 (p. 29). As is well known, ninety years after it was expressed, the wish of this outstanding linguist has not yet come true. There is no less philosophy in general linguistics, and its methods have not become more objective. Having struck this note, we think it appropriate to recall what Jakobson (1971) thought about it: At first glance, linguistic theory of our time seems to offer a stunning variety of clashing doctrines. […] Yet a careful, unprejudiced examination of all these restricted doctrines and vehement polemics reveals an essentially monolithic whole behind the striking divergences in terms, slogans, and technical contrivances. […] In the same way that a general topology underlies and encompasses a wide range of mathematical approaches, the manimutually approaches to language reflect merely the plurality of its complementary aspects (p. 712; emphasis added.—A. K.). In Jakobson’s opinion, it is “the cardinal principles of a structuralist […] approach to language, common to all directions in this research”, that make are in line with a centuries-long linguistic tradition that goes back to Antiquity, and, in our opinion, continue to be of undiminished importance. 3 “While historical linguistics has exact methods, approved by long use, general linguistics is still quite badly supported by its methodology; linguistics is still inclined to philosophy, as each author chooses a research method of their own, and there are, so it seems, as many kinds of general linguistics as there are linguists. For general linguistics to progress it needs to become more objective and its technique needs to become more permanent. of all, in order to understand each other better, we need to have a terminology in which all the terms have the universal meaning for all of us. At first, however, there seems to be a fundamental difficulty lying in the nature of the linguistic fact” [translation Anna Kosheleva]. § 2. A compendium of incompatible linguistic frameworks 5 a variety of doctrines into a whole. Therefore, “[t]he inquiry into language structure is the undeniable aim of contemporary linguistics in all its varieties” (Jakobson 1971). Today, however, it is hard to imagine that anyone would defend this thesis. Already not quite undisputed at the time, it later lost its appeal completely.  4 As for late 20 century theoretical linguistics, there isn’t even any semth of unity. As observed by Kibrik (1995: 99), […] having disintegrated into a variety of competing theories, general linguistics does not represent a system that has… a universally recognized core as the “dry residue” of all that has been accumulated by the theory and is not subject to revision in the near future. Kasevich (2013/2000) expresses a similar concern: […] while in mathematics or physics different schools of thought usually show the relationship of complementarity (each particular school explores its specific problematics more intensely and thoroughly), in linguistics or psychology more typical are schools that are, in effect, in a relationship of opposition (they explore the same problematics, but use different initial assumptions and/or different methods) (p. 22; original emphasis). Speaking of today, it must be acknowledged that the number of mutually opposing theories of language has only grown. Following Meillet and Jakobson, many researchers believe that theoretical linguistics, just as any natural science (or a discipline close to natural science), will sooner or later develop a single dominant theory of language and, accordingly, a single research paradigm. Such hopes have been voiced both by Kibrik (1995: 100)—“the tendency to specialize will be replaced by the tendency to generalize”—and Kasevich (2013/2000: 27): “reaching a consensus on the foundations of our science is a task for the future, the future one would like to believe in”. However, many linguists do not share such sentiments. For example, Serio (1993) believes that different linguistic paradigms neither replace nor refute one another; rather, one is superimposed on another, and they coexist, ignoring each other at the same time. According to Dem’iankov (2009: 31–32), 4 See also Kubriakova (1995) for an overview of research on this topic, where she undertakes “to give a well-reasoned argument in support of the view that the situation in linguistics today is not only marked by a diversity of opinions… frameworks, hypotheses and theories, but also by some intrinsic unity” (p. 144). § 3. Contrastive analysis of some mutually contradicting linguistic theories 7 tention of theoretical linguists; from the point of view of an individual r esearcher, the current state of affairs doesn’t seem so disheartening. Every researcher sees only “his own piece” of the general picture, and, against  5 the background of impressive achievements of particular linguistic research projects, any “local” contradictions are not seen as symptomatic at all. The focus of our discussion is theoretical linguistics. An attempt will be made to show that, in theoretical linguistics, one cannot see any signs of attempts to elaborate and generalize a single established theory of language, unlike, for example, in physics, where Einstein’s Special Theory of Relativity encompassed Newton’s classical mechanics as a particular case of large masses moving at low speeds. Nor has an old theory been replaced by a new theory of a greater explanatory power, as, for example, in the case of Lamarck’s and Darwin’s theories in evolutionary biology, Ptolemy’s and Copernicus’ theories in astronomy, or the theory of phlogiston and Lavoisier’s oxygen theory of combustion in chemistry. Finally, there is no sharp competition between adverse theories (on account of their proponents’ belief that there may be only one true theory), as, for example, in modern physics with its several coexisting alternative conceptions of the general field theory. Instead, we witness an unconstrained growth of many mutually opposed theories of language—or one of its constituent parts (syntax, lexical semantics, etc.). All that being said, one can conclude that, as modern linguistics continues to develop, our scientific knowledge about language becomes both broader and ever more diverse and at the same time less and less profound and indisputable. § 3. Contrastive analysis of some mutually contradicting linguistic theories 1. Introduction It would be natural to begin arguments in support of the abovegiven view by analyzing Saussure’s theory as the “starting point of a new era in linguistics” 5 Even this, however, is not always the case. A linguist belonging to a particular school of thought tends to view other theories as false rather than alternative (Kasevich 2009: 39–40). Moreover, the mutual opposition of different schools is not always obvious. Therefore, any new theory that is not thematically close to a particular linguist is seen by such a linguist as a theory which extends, and adds to, the common stock of linguistic facts. True, there is also an opposite tendency to make ungrounded conclusions about the incompatibility of different viewpoints—see, for example, an instructive analysis of the opposing propositions of generativism and functionalism given by Newmeyer (1999). 1. On the contradictory nature of contemporary linguistic theories... 8 (Jakobson 1971: 717). We will also use this theory as base point in our analyses of the subsequent linguistic theories developed by Noam Chomsky, Ray Jackendoff, Igor’ Mel’čuk, Anna Wierzbicka and George Lakoff. It should be noted from the outset that mutual contradictoriness between such broad areas of linguistic research as generativism and functionalism (along with cognitivism, which is close to functionalism) is well known and does not require any special comments (in spite of the fact that some features converge in certain approaches; see, for example, an overview in Mustaioki (2010: 107–108)). However, it will be shown that mutual opposition is also inherent in some closely related theories, for example in Chomsky’s and Jackendoff’s theories (generativism), Mel’čuk’s and Wierzbicka’s theories (semantic atomism, i.e. reduction of linguistic meaning to a set of semantic primes), etc. As will become clear, none of the above-mentioned theories is in a relationship of mutual complementarity to the other theories, as they are all based on different, often incompatible, initial assumptions. And this is despite the fact that Chomsky’s and Jackendoff’s theories represent generative linguistics while Mel’čuk’s theory does not, though it is quite close in spirit and structure to the theories of Chomsky and Jackendoff. At the same time, Mel’čuk’s theory and the version of it developed by the Moscow School of Semantics (Apresian 2009) use methods of semantic description of the lexicon that closely tie them to Wierzbicka’s theory, while a cognitive approach to semantics brings the latter close to Lakoff’s (1987) theory. 2. On the role of language in thought The issue of the relationship between language and thought is of paramount importance to theoretical linguistics. Can the core functions of language be reduced to purely communicative tasks or do they include thought production as well? Is it in the nature of language to fully participate in the formation of thought expressed in a linguistic utterance (phrase) or does it serve only as a tool for the transfer of thought? The aforementioned theories answer this question differently. For Saussure, language is an instrument of thought and there may be no thought outside of language. For Chomsky, language is a tool for the formation of the general structure of an utterance while the meaningful content of the utterance itself belongs to the mechanisms of thought that are outside of language. And for Jackendoff and Mel’čuk, language is exclusively a means of communication seen as a translation of thought into linguistic form. However, between the latter two theories there is also a substantial difference. § 3. Contrastive analysis of some mutually contradicting linguistic theories 9 Let us dwell on this issue in a little more detail. According to Saussure (1966) language is the main instrument of thought, which is totally and completely shaped by language as there is no thought without language. Indeed, Saussure’s theory builds on a hypothesis of the primacy of the speaker’s native tongue over the realm of concepts—units of thought—which is secondary to, and derivative from, language. To Saussure, thought abstracted from its expression in words is something amorphous and undifferentiated. Without the help of linguistic signs we would not be able to strictly distinguish one concept from another. Language, according to Saussure, is special in that it serves as a connecting link between thought and sound. Thanks to language, thought, which is chaotic by nature, is broken into parts and made more precise. Thus, the primary categorization of the world into objects, events, and relations is determined neither genetically nor cognitively. It is the result of language acquisition, a projection of the native speaker’s system of concepts on the initially perceived amorphous world. As Whorf put it, it is “a kaleidoscopic flux of impressions which has to be organized… largely by the linguistic systems in our minds” (Whorf 1940/2010: 6). Clearly, such an interpretation of the function of language dramatically diminishes the role of non-linguistic thought (as an independent process), because both the concepts that constitute thought and conceptual structures are reduced to linguistic meanings and the syntactic structures of language. This model simply has no room for cognitive concepts that are outside of language. Let us now consider Chomsky’s approach. According to his Theory of Principles and Parameters, language (“internal language”, also referred to as “I-language”, or grammar) is a “computational system of the mind/brain that generates an infinite array of hierarchically structured expressions” composed of abstract words of a specialized vocabulary, or “conceptual/lexical ‘atoms’” (Chomsky 2010: 45, 52). Abstract “words” are “bundles” of properties of three types: phonological, syntactic, and semantic. Syntactic expressions made up of these bundles are used for “thought, understanding and organizing activity” (Ibid.: 46) through a system of “semantic-pragmatic” interface. This system converts each syntactic expression into Logical Form, a more semanticized expression of the same type in which the syntactic relations between the constituents fit the semantic relations more closely and which is suitable for semantic-pragmatic interpretations by the human systems of thought.  6 Simultaneously, at the sensorimotor interface, this expression is converted 6 According to Chomsky (2010), generation of syntactic expressions is determined by intralinguistic causes and is not aimed at performing any functions. How such expressions are used is not an issue for linguistic theory; it belongs to the realm of linguistic performance. 1. On the contradictory nature of contemporary linguistic theories... 10 into Phonological Form and bound to the acoustic and articulatory mechanisms. However, the interface systems that bind the generated expressions to these mechanisms are not linguistic. Therefore, according to Chomsky, language is “an internal system that links sound and meaning in a particular way, by means of generated expressions” (Chomsky 2010: 46). Clearly, in such an approach to language its role in the system of thought, though not as all-encompassing as Saussure would have it, is still very important, as the combinatorial forms of thought (conceptual structures) are determined by abstract syntactic expressions. As for the role of non-linguistic  7 thought, it consists in making these abstract forms concrete by filling them with concrete meanings. In other words, while for Saussure thought is, on the whole, determined by language, for Chomsky it is a partial relationship: language is autonomous and is used as an instrument of thought only in combination with strictly cognitive mechanisms. Staying within the generativist framework, Jackendoff (2010: 67) defines language similarly to Chomsky, as a system generating expressions based on the very same abstract “words” (see above). An expression formed by this system, however, comprises three levels instead of one. It is a “parallel architecture” comprising three structures: phonological, syntactic, and “semantic/ conceptual”, each formed by its own generative module, with the connection between them “established by the interfaces” (Jackendoff 2007: 49). As Jackendoff emphasizes, the semantic/conceptual component, being independent from the syntactic component, belongs to both language and thought. It follows that a thought (an “algebraic encoding”, a “conceptual structure”) is wholly and completely formed by the cognitive systems, while language fulfills only a communicative function, embodying this (already prepared) thought in a linguistic utterance: “According to the mentalist stance, the basic function of language is to convert thoughts into communicable form” (Ibid.: 69). Thus, language is assigned a purely communicative, “conversion” function: first, the cognitive system forms a non-verbal conceptual thought and only then does language come in to build a three-level (phonological, syntactic, and semantic) phrase that expresses this thought. It should be noted that Chomsky specifically warns against a communicative interpretation of the function of language: “it is wrong to think of human use of language as characteristically informative, in fact or in intention” 7 As claimed by Hauser, Chomsky and Fitch (2002: 1573), “the computational mechanism of recursion […] is uniquely human”, while the recursive syntactic structure expressing the corresponding structure of the thought is characteristic of human language. § 3. Contrastive analysis of some mutually contradicting linguistic theories 11 (Chomsky 1968/2006: 61). Later, this thought was reiterated in Chomsky 1980: 230. Disagreeing with Chomsky, Jackendoff (2007) argues: Chomsky sets himself apart from common sense here in his oft-repeated claim that language is not “for” communication […] For reasons unclear to me, he has always seemed to believe that language came into existence primarily as an aid to thought […] In at least one recent work (Chomsky 2002) and in a discussion at a conference in Spring 2002, he has justified this stance on the grounds that most use of language is for inner speech. Surprisingly, then, he has fallen into the trap […] of believing that inner is speech thought, rather than (as I will argue) the phonological structure corresponding to the thought (Jackendoff 2007: 70, footnote 13; original emphasis). Jackendoff (2010: 68) sets his theory in opposition to Chomsky’s again when he discusses language evolution. This issue will be briefly touched upon in § 4, section 3. In the theoretical framework developed by Mel’čuk (2012a), language is described by the “Meaning-Text functional model”. This model is based on the assumption that “language is a finite system of rules” for mapping a set of linguistic meanings utterances in a given language. on a set of In other words, language is described as the correspondence “{linguistic meanings} ↔ {utterances in a language}”. This correspondence shows, on the one hand, which phrases express a given linguistic meaning (ideally, the model should be able to yield all synonymic utterances that express a given meaning), and, on the other, which linguistic meaning (or meanings, should there be more than one) is expressed by a given utterance. The Meaning ↔ Text Model is precisely the tool for describing language as such a correspondence (Ibid.: 21–22). Mel’čuk’s model differs from Chomsky’s model in that it is not generative, but transformational (a “conversion” model). Chomsky’s computational model does not use any informational input for the production of expressions. By contrast, the Meaning ↔ Text Model converts the linguistic meaning input from the Concept—Linguistic Meaning interface into a phrase (or a set of all synonymic phrases) that embodies this meaning. While structurally similar to Jackendoff’s model, Mel’čuk’s model includes a level to account for the specifics of a given language. While Jackendoff’s model—through its semantic/ conceptual component—is directly connected to the conceptual level, the Meaning ↔ Text Model is strictly separated from this level, as well as from 1. On the contradictory nature of contemporary linguistic theories... 12 the Concept—Linguistic Meaning interface. Mel’čuk believes that the essence of language lies in the production of phrases that adequately express not universal meanings (non-verbal conceptual structures), but purely linguistic, or language-specific, meaning (Mel’čuk 2012a: 27, 29, 39). Therefore, at the beginning (in the preliminary stage of phrase formation) the interface converts non-verbal conceptual structure (thought) into linguistic meaning that reflects the specific features of a given language, and after that language comes into play, that is, the Meaning ↔ Text Model; it is this model that converts the resulting linguistic meaning into a phrase of a given language. Let us stress again that, unlike in Chomsky’s and Jackendoff’s models, the output of the Meaning ↔ Text Model is not an abstract linguistic expression but a concrete phrase (or a set of synonymic phrases) of a given language. Thus, Mel’čuk’s model stands even farther from cognitive mechanisms than Jackendoff’s model. As can be seen, all three linguistic frameworks are in opposition to Saussure’s conception and to one another in their understanding of the role of language in the system of thought. These models are based on different principles, work in substantially different ways, and have outputs that are typologically different linguistic “expressions”. Chomsky’s model produces abstract syntactic expressions devoid of concrete meanings. In Jackendoff’s model, expressions are filled with concrete but at the same time universal meanings (conceptual structures). In Mel’čuk’s model, abstract expressions are replaced by phrases in a given language along with their language-specific meanings. Note that the abovementioned linguistic frameworks are, in turn, opposed by functional frameworks such as construction grammar, for example Fillmore et al. (1988), Goldberg (1995 and 2006). The basic element of construction grammar is a linguistic construction viewed as an integral unit whose formal and substantive properties are not reduced to the properties of its constituents (e.g. conditional, comparative, adversative, and other types of constructions). 3. Is language an autonomous module (“cognitive organ”)? This question divides linguistic theories into two opposing camps. It is answered in the positive by the generative theories, as well as by Mel’čuk’s theory. It receives a negative answer from the functional theories (Givόn, A. Kibrik, Fillmore, Goldberg, and many others), which are based on an assumption that the fundamental properties of language are determined by its functions (mental, communicative, etc.), and from the cognitive theories based on an assumption that language uses conceptual knowledge shared by the sys- § 3. Contrastive analysis of some mutually contradicting linguistic theories 13 tem of thought and other cognitive systems, i.e. basic concepts, universal conceptual primes, etc. Lakoff (1987), contrasting his cognitive approach with the linguistic theories that treat language as a “modular” system, wrote: One of the principal claims of this book is that language makes use of our general cognitive apparatus [i.e. that which performs non-linguistic functions as well.—A. K.]. […] In fact, the most widely accepted views of language within both linguistics and the philosophy of language are based on inthe opposite assumption: that language is a separate “modular” system dependent of the rest of cognition. The independence of grammar from the rest of cognition is perhaps the most fundamental assumption on which Noam Chomsky’s theory of language rests (Lakoff 1987; original em phasis). As a matter of fact, opponents of the modular view of language (in which language is viewed as a mental organ) are many and include representatives from various fields of research (cf. Bates 1994). Some other oppositions between the “basic assumptions” of cognitive linguistics (the frameworks developed by George Lakoff, Ronald Langacker, and Ray Jackendoff) and traditional reificatory semantics are discussed by Chenki (1996). 4. Other opposing assumptions in the theories under discussion There are substantial contradictions between these theories regarding other important issues. Mel’čuk’s Meaning ↔ Text Model, while subscribing to the idea of language modularity as part of Chomsky’s generative model, radically differs from the latter on some fundamental solutions. In describing the syntax of a phrase, Chomsky makes use of the constituent structure while Mel’čuk uses the dependency structure; according to Mel’čuk, this is a decisive difference (cf. Mel’čuk 2012a: 128; 2012b). It should be noted, however, that this contradiction loses its edge in later versions of generative theory that make use of the concept of government. Moreover, in contrast with Chomsky’s syntax-centered model, the central place in the Meaning ↔ Text Model is given to the lexicon, and the semantic description of concrete lexemes plays a very important role. Wierzbicka’s (1996) semantic theory (interpretation of linguistic meanings with the help of a “semantic metalanguage”: a list of semantic primes and a set of simple syntactic constructions) is fairly close to Mel’čuk’s approach to semantic definitions. Although in the course of its development Wierzbicka’s theory was closely related to an earlier version of the theory developed by 1. On the contradictory nature of contemporary linguistic theories... 16 2. Contradictory descriptions of the syntax-lexicon interface It is a well-known fact that different syntactic theories are often opposed to one another and we are not going to dwell on that. Of a much greater interest is the division of semantic-syntactic information between the lexicon and the grammar. And the picture here does not seem to be different, as a variety of incompatible theories continue to coexist. As noted in a review of two volumes published in the series Oxford Studies in Theoretical Linguistics (Erteschik-Shir and Rapoport 2005; Alexiadou et al. 2004), the “division of responsibilities” between grammar and the lexicon is made by different researchers in quite different ways (Arkad’ev 2008). Commenting on the articles included in the first volume, Arkad’ev observes: [T]he authors do not seem to have a consensus on a number of very important issues; in particular, the syntactic procedures for describing the nonaccusative forms given in various articles seem to be quite convincing in the framework of a given article, but on comparison across the various articles they appear to be hardly reconcilable, at least, if not mutually contradicting (Arkad’ev 2008: 124). A similar conclusion is made about the other volume, which includes the “studies carried out within different theoretical frameworks which sometimes quite obviously contradict one another” (Ibid.: 132) 3. Theories of language origin and evolution The huge gap between animal systems of communication and human language gives rise to wild fantasies. Speaking about the multifarious theories of language origin and evolution, Bickerton (2007: 524) stresses “the striking lack of consensus and the incompatibility of different approaches […] in the field of language evolution” (emphasis added.—A. K.; see also Hewes 1977; Fitch 2010: 401–507). Here we will only touch upon the issue of gradual vs. abrupt language development. Humboldt (1820) and Müller (1885) believed that the abovementioned gap could be overcome “instantaneously”, in one leap. This conviction is also shared by Chomsky (2010), who argues that the recursive syntactic structure characteristic of human language could not have evolved in several steps; therefore, the theory of natural selection is not applicable to the process of formation of this “cognitive organ”. As will be shown, others (e.g. Bickerton 2009 and 2010; Jackendoff 2010) believe that, on the contrary, such a gap could be overcome only in stages by means of protolanguage—a symbolic § 5. The futility of attempts to reach a consensus 17 communicative system that spanned the gap between animal signaling systems and human language. A pioneer in the study of the problem of protolanguage, Bickerton (1990 and 2009) argues that human language evolved in two stages. The first stage was marked by an emergence of asyntactic “lexical” protolanguage; its “words” were “put together like beads on a string, A + B + C” (Bickerton 2009: 187) without forming a syntactic structure, and later this protolanguage transformed into full human language in which words form a hierarchical, tree-like structure. Jackendoff (2010), arguing for a three-level structure of language, suggests that one could imagine “various scenarios in which the language capacity evolves in stages, each adding an increment to the system’s communicative efficiency and flexibility” (p. 71). Indeed, since all three levels of phrase structure (phonology, syntax, and semantics) are independent, it is quite possible to hypothesize that initially only two of them emerged: the phonological and the semantic/conceptual levels. Then the scenario suggested by Bickerton (2009) becomes plausible: at first, asyntactic protolanguage emerges which later acquires syntax and transforms into full human language. Moreover, because these initial levels might gradually become more complex, the resultant picture of language evolution does not contradict Darwin’s theory. As may be seen, the antagonism of theoretical linguistic frameworks leads to incompatibility of the theories of language evolution that stem from these frameworks: while for Chomsky it was an abrupt emergence, Bickerton posits a two-step development via asyntactic protolanguage, and Jackendoff argues for a gradual evolution of the initial protolanguage. § 5. The futility of attempts to reach a consensus 1. Unproductiveness of scholarly polemics It must be admitted that the polemics often entered by the representatives of different schools of linguistic thought turn out to be unproductive. Often linguists do not seem to understand one another—and sometimes they are simply unwilling to understand, holding other theories to be false. However, this is only one of the reasons why linguists from different schools are unable to overcome the observed contradictions or at least help draw the different schools nearer. Another and more important reason is a lack of universally recognized criteria for establishing the validity of a linguistic thesis. 1. On the contradictory nature of contemporary linguistic theories... 18 The futility of linguistic disputes may be illustrated by the well-known Behavioral and discussion of Evans and Levinson’s (2009) target article in Brain Sciences, in which a focused attempt was made to show that practically all language universals posited by generative grammar, Greenberg’s typological school, etc., are illusory (unverifiable or trivial). The discussion included over twenty commentaries by representatives of various schools and traditions as well as the authors’ responses; however, it didn’t seem to draw the opposing sides any nearer. A similar example is provided by another famous discussion started by Hauser, Chomsky and Fitch (2002), which attracted a lot of attention (cf. Jackendoff and Pinker 2005; Pinker and Jackendoff 2005; Bickerton 2007, 2009 and 2010, inter alia) and was followed by another article by the same authors (Fitch et al. 2005). No common ground seems to have been found in this discussion, either. As far as one can judge, a similar outcome marked the discussion started at the time by Newmeyer (1999). 2. Inefficiency of interdisciplinary approaches Some linguists have great hopes for an interdisciplinary approach. Alas, such hopes also remain unrealized because, as will be shown later, psychology and other disciplines are not free from contradictoriness between different theoretical frameworks. It should be borne in mind, however, that researchers looking for solutions to particular problems may be unaware of the hidden dangers of interdisciplinarity. Thus, at a recent conference in Moscow (Prospects for Cognitive Research: Interdisciplinarity and Integrationism, 3 December, 2014, Moscow), the author witnessed the following exchange when a well-known cognitive linguist called for an interdisciplinary dialogue with psychologists. When he was asked, “How do you imagine a dialogue between linguistics and psychology in view of the fact that theoretical linguistics today is a compendium of ten mutually contradicting theories, while psychology—just in the same fashion—is not a unified science, but a multitude of independent ‘psychologies’?”, the answer was: “It’s very simple. Developing my own approach to language, I turn to psychology and borrow from it what is needed.” Obviously, this linguist sees contemporary psychology as a large (and continuously growing) stock of consistent facts that can be used as needed; this may result in borrowing some facts from one branch of psychology and then some facts from another branch which is in opposition to the first one. § 6. Is linguistics a natural science? 19 § 6. Is linguistics a natural science? Can one hope that in its future development theoretical linguistics will meet the expectations of Meillet and other linguists and become a unified theory, or does the nature of language as an object of study leave no room for such hopes? Or, in other words, is the current situation caused by the very nature of language or is it rooted in the specific knowledge about language that we possess at this stage? This question is closely connected with another question: is linguistics a natural science? If the answer is “yes”, then a unified theory of language will be worked out sooner or later. If the answer is “no”, the status of linguistics remains an open issue, because to ascribe it exclusively to the humanities would be, in our view, an erroneous decision. The idea that linguistics could be justifiably recognized as a natural science or, at least, an exact science has been expressed long ago: “In the final analysis, at the turn of the centuries modern linguistics was able to proudly show a number of achievements that confirmed its right to be called a natural science” (Shor 1926: 33). As has been previously mentioned, linguists have not always been inclined to treat linguistics as a natural science, which prompted the following remark from Baudouin de Courtenay (Boduen de Kurtene 1963: 76, footnote 48): “Perhaps, in a short time I will have an opportunity to examine […] the prejudice of scholars that the science of language is a natural science (in the sense of botany and zoology), that it is completely different from philology…” However, the focus of our attention will not be on the arguments given by the proponents of the view that linguistics is a natural science. It will be on the arguments that bear on (a) the nature of our object of study (just as in the case of objects of other sciences, language is a physical object), and (b) the method of research (it is the methodology of physical sciences that is used in the study of language). Among the arguments of the first kind, the most convincing seem to be those given by Müller (1885) in his Lectures on the Science of Language, delivered at the Royal Institution of Great Britain in 1861 (10 years prior to de Courtenay’s article). The first lecture was entitled “The Science of Language One of the Physical Science”. First of all, one should note Müller’s definition of the dichotomy “natural sciences” vs. “mental sciences” and “historical sciences”. As he put it, the object of study of a natural science is a natural object, or “the works of God” (the physical world for physics, nature for biology, etc.), while the “historical” sciences (or, in modern terminology, the humanities) deal with “the works of man” (Ibid.: 22). Secondly, his arguments for viewing human language as a natural object seem quite convincing: according to Müller, 1. On the contradictory nature of contemporary linguistic theories... 20 linguistic change is of a purely endogenous nature and neither an individual nor a collective can change it, arbitrarily or by convention: Although there is a continuous change in language, it is not in the power of man either to produce or to prevent it. […] [But if that] change takes place, will of any individual, mutual agreement it will not be by the nor by the of any large number of men […] the first impulse to a new formation in language, though given by an individual, is mostly, if not always, given without premeditation, nay, unconsciously. […] and the results apparently produced by him depend on laws beyond his control... (Müller 1885: 38–41; original emphasis). It is significant that Müller’s idea about the unpremeditated and unconscious character of linguistic change is shared by various linguistic schools (Saussure 1966; Paul 1970, etc.), while his dichotomy of natural sciences vs. the humanities has been accepted by many linguists. Cf.: Researchers in the humanities are interested not in what is natural, physical in humans, but in the artificial; […] how to consider […] language: is it a phenomenon of the natural or artificial order? […] in linguistics the borderline between the natural-scientific and humanistic knowledge coincides, to some extent, with the borderline between language and speech (Shapir 2005: 45). Language is a natural phenomenon. Can you create language? (Plungian 2009: 5). Arguments of the second kind—the use of natural-scientific methodology— were analyzed by Chomsky (1995), who came to the following conclusion: any science that successfully uses a natural-scientific methodology is a natural science and there is no other criterion to distinguish the natural from the nonnatural. similar view of linguistics as a science has been expressed by Kasevich (2013/2000) in his examination of the stance taken by Lazard (1999). Elaborating on the philosophical ideas of Granger (Granger 1960 and 1979), Lazard claims that theoretical linguistics, abundant in various theories of language, in its current state belongs to the proto-sciences as it lacks a “categorical definition” of its object, explicit definitions of key concepts, elaborate formalizations, etc. Taking up this thought, Kasevich observes: Slightly editing and reformulating such claims, one could say that in a “true science” a researcher replaces his object of study with its formalized abstract analogue (model) with which he then continues to work, comparing § 6. Is linguistics a natural science? 21 the results he gets with the results of observation and experimentation (Kasevich 2013/2000: 20). However, Granger (1979) and Lazard (1999) believe that further development must make linguistics a “true” science with a unified research paradigm. Two comments seem to be in order here. 1. Sharing the conviction that the use of natural-scientific methodology in the study of language is important, we would like to stress that what is crucial here is that the abstract model of language used in such a methodological framework be adequate to the object modeled. Take, for example, Chomsky’s approach to such modeling: “Under familiar and appropriate idealizations, a person’s language is a system of the mind/brain that generates an infinite array of hierarchically structured expressions” (Chomsky 2010: 45; emphasis added.—A. K.). The question is, should the linguistic community accept such an idealization? Or would it irreplaceably reduce the real object of study? Mel’čuk, who strictly separates purely linguistic meaning (which is formed by a separate model lying outside of language) from the Meaning ↔ Text Model (i.e. from the model of language which receives this meaning as input and works with it), emphasizes that “such an interpretation does not pretend […] to correspond to the psychological reality” (Mel’čuk 1974/1999: 24). It is important here to understand to what extent such an interpretation “does not correspond” to reality, how strongly such modularity of the model distorts the object modeled. At the same time, the abovementioned properties of language—resistance to arbitrary changes and regularity of linguistic behavior of its speakers—give serious reasons to believe in the possibility of an adequate model of language. 2. The level of formalization in modern linguistics is quite high. There are a great number of theories that use an abstract (or close to an abstract) model of language and natural-scientific methodology (the frameworks developed by Chomsky, Jackendoff, Mel’čuk, and others). Therefore, to treat modern linguistics as a proto-science would be at least questionable. Yet, despite the fact that formal models and natural-scientific methodology exist, new linguistic theories continue to appear. However, this circumstance does not seem to characterize linguistics as a proto-science either. As has already been mentioned, in modern physics there exist several alternative theories of field, including the gravitational, electromagnetic, and nuclear (weak and strong) fields. It would, therefore, be reasonable to assume that coexistence of mutually opposed theories is a sign of a particular stage in the development of knowledge (in regard to both science and proto-science) when a certain novel aspect of the object of study comes into focus. Chapter 2 A reference-based approach to describing notional words Chapter 2 offers an exposition of a reference-based approach to describing basic meanings of sensory lexical items, that is, nouns and verbs that denote “visible” referents (things and physical actions). The central goal of the reference-based approach is to give a rigorous definition of the basic meaning of a word and, accordingly, of the category of its direct referents. § 1. A reference-based approach to lexical semantics 1. The dual structure of lexical meaning 1.1. The problem with reference (designation). A native speaker uses words—with much ease and without giving it a second thought—to name various fragments of the world around him such as things, actions, states, qualities, situations, etc. And he does it, as a rule, in full accordance with other native speakers. For example, seeing coffee beginning to froth in a coffee pot, a native speaker of Russian will exclaim, Kofe bežit!, lit. ‘The coffee is running!’ (‘The coffee is overflowing!’), and we will understand him clearly, experiencing no difficulty in understanding the referent of this utterance. In a similar fashion and to the same effect, the speaker may say, Moloko bežit! lit. ‘The milk is running!’ However, if it is water, rather than milk, beginning to boil and overflow, he will not say, *Voda bežit! ‘The water is running!’ because, just like any other native speaker of Russian, he knows that you can’t say that about water. This naturally begs the question, “Why?”, as outwardly the referential situations in these cases are practically indistinguishable. One might argue that in the former case the liquid froths while in the latter it does 2. A reference-based approach to describing notional words 36 not. However, soup also froths when beginning to boil, yet an utterance such as Sup bežit! ‘Soup is running!’, though admissible, nevertheless sounds odd. ? Explanatory dictionaries don’t help here either, as may be seen from a definition offered in Aktivnyj Slovar’ (2014, 1: 183): bežat’ 4.4 Kofe bežit. [‘(be) run(ning)’], rare or obs. [‘The coffee is running’] meaning. А1 bežit ‘Liquid substance, А1, is flowing over the edge of the vessel, А2, which contains it, because it is boiling’ [usually about a currently observed process]. This definition does not impose any constraints on the utterance *Voda bežit (‘Water is flowing over the edge of the vessel because it is boiling’).  1 It is worth noting that neither explanatory dictionaries nor lexicographic research provide strict definitions of lexical meanings that would explain in which case an observed phenomenon may be designated by a given word, in which case such designation would appear less acceptable, and in which case it would be simply impossible. This is true not only for derivative meanings of the type bežat’ 4.4, but also for the basic meanings of words—not just referigra versatile words like ‘game’, but even very simple words such as stul ‘chair’, kreslo ‘armchair’, idti ‘go/walk’, bežat’ ‘run’, staryj ‘old’, and the like. The traditional account of a word’s basic meaning. Let us touch very briefly upon the main propositions of the traditional approach, exemplified by the seminal works of John Lyons and Dmitrii Shmelёv. We will start with a quotation from Shmelёv (1977): In lexicology […] words are studied, first and foremost, as units of denotation, i.e., as linguistic units that serve the purpose of naming things and phenomena in the real world, the purpose of singling them out and forming respective notions (p. 3). The meaning of a word is the reflection in it of one or another real world phenomenon (a thing, a quality, a relation, an action, a process). notion [...] The meaning of every notional word is based on a particular which contains some general essential features of a particular fragment of the real world, that is, features that make it possible to group single things and phenomena into definite classes […] It is a notion that lies at the basis of a word’s lexical meaning (pp. 58; 60; original emphasis). 1 A more accurate explanation, showing why utterances such as Voda/sup bežit ‘The water/soup is running’ are odd or incorrect, will be given in § 4, section 1.7. § 1. A reference-based approach to lexical semantics 43 1.5. The dual structure of basic meaning. All of the above allows us to hypothesize that the basic meaning of an action verb has a dual structure analogous to that of a noun, as in (3). As will be illustrated by further examples, the basic meaning of a sensory word may be approximated as follows: (7) Basic meaning of a word = Prototype → Function “Prototype”, here, is the visual feature (shape) of the typical referents (objects and actions), and “Function” the “invisible” work attributed to all direct referents on the basis of their shape. For example, the visible shape of a runlosing contact ning person (‘periodically foot with the surface’) is attributed an invisible function, ‘periodically pushes off with his feet from the ground the goal of moving with to the target place’. The arrow, “→”, sets a relationship of interpretation, indicating that the prototype (shape) of a typical referent predetermines its function. As an illustration (7), consider the following two  4 examples: chair 1 (basic meaning)  5 P rototypical chair → Function → designed for sitting on by a single person in a halfsteady (semi-relaxed) posture tree 1 (basic meaning) P rototypical tree → Function (= ‘the original work that it does’) → grows from the ground by itself, blossoms, and bears fruit 4 We may assume that “Prototype” and “Function” are independent cognitive units of different natures stored in separate cells of the lexicon—the domain of long-term memory which contains the lexical data of a language. It is also a storage place for the relationship of interpretation which connects them and is effected as a stable associative tie between them. For neurophysiological data in support of this hypothesis see section 2.3. 5 The definitions of meanings that we give are not explanatory interpretations since they often use words and pictures as terms designating cognitive concepts; that is why the quote-unquote marks are not used. 2. A reference-based approach to describing notional words 44 1.6. Basic meaning of the verb udarit’ ‘hit’. As another illustration of the dual structure, (7), let us take a brief look at the meaning of the verb udarit’ ‘hit’. We’ll begin by analyzing the following well-known definition: А udarjaet Y-a X-оm ~ ‘A hits Y with X’ ≈ ‘А forcefully and briefly brings a compact object X into contact with object Y’ (Apresian 1974/1995: 108). The basic meaning of an English correlate is explained in a very similar way: hit 1. TOUCH SB/STH HARD; to touch someone or something quickly and hard with your hand, stick etc: He raised the hammer and hit the bell (Longman 2009: 832). Obviously, both of these explanations are essentially imprecise. They describe only a visual prototype of the action “X hit Y”, namely, “X came into contact with Y”, while the function, as its second and primary component characterizing the aftereffect of X’s contact with Y, is missing: ‘Y experienced a momentary hard push’. causal relationship Also missing is the (THEREFORE) which connects the prototype with the function. Here is a short version of a more precise description of the basic meaning of the verb udarit’ noga udarila po mjaču ‘hit’, found in such uses as ‘the foot kicked the ball’, palka udarila po zaboru ‘the stick hit the fence’: Object X hit Y (basic meaning) = Prototype: Compact object X forcefully and briefly came into contact with object Y → Function: THEREFORE Y experienced a momentary hard push and got into a state of shock (and pain, if Y is a living being). This meaning is described in detail in § 2, p. 65. 1.7. The role of ‘prototype’ in describing a word’s meaning. One might be puzzled by the fact that such an experienced lexicographer as Iurii Apresian, who offered a scientific explanation, and such an authoritative dictionary as Longman, where a commonly used definition is provided, describe only the outward feature (prototype) of a hit, leaving aside its substantive characteristic: ‘a push, a shock to Y’. Yet, as has been shown above in the analyses of the nouns chair and armchair, and the verbs walk and run, this is a typical situation. As will be shown later in § 2, section 3, explanatory interpretations of the verbs padat’ ‘fall’, fall etc. are no exception. The paradox may be explained as follows. As has already been mentioned, in a native speaker’s memory the prototypical component of a meaning is closely (by association) related to its functional component (explanatory inter- § 1. A reference-based approach to lexical semantics 45 pretation). That is why perceiving or imagining a description of the prototype, the native speaker unconsciously “recalls” the function that is associated with it. Because of this, the speaker receives the impression that the prototype represents the whole meaning. Only a focused referential analysis allows us to detect the deficiency of prototype-based definitions. 1.8. Pictures of prototypes in dictionary entries. The above reasoning explains the common lexicographic practice of illustrating some definitions by pictures of typical referents—things and actions. A picture is the most direct way to illustrate the meaning of a word (first to its prototype, and then to the function). As an example, consider the definition of the word banana given in the Longman dictionary (Longman 2009: 114), accompanied by the image of a half-peeled yellow banana (Ibid.: 705): banana 1. a long curved tropical fruit with a yellow skin, see picture… In the definitions given by Apresian and in the Comprehensive Dictionary (BTS, cf. section 1.1), a verbal description of a typical banana is used instead of a picture. However, it would be much more informative to replace such verbal description by three pictures showing: (1) a tropical plant with a bunch of ripe bananas, (2) the process of peeling a banana, when it is held in one hand and peeled with the other, (3) a man nibbling on the peeled part while holding the banana at the unpeeled part. Of course, to make the description complete, a purely functional feature would have to be added: ‘the flesh of a ripe banana has a pleasant sweetish taste’. Another illustration of the promoted thesis is provided by the following fragment of a “pictorial” dictionary of English action verbs (Fig. 2). Here, each action is represented by a single, most typical frame; this appears to be enough for a quick and reliable identification of the action which the frame symbolizes. It should be stressed that we are able to identify not only the action’s kinematics (we know how it is going to develop), but its dynamics as well (which efforts on the part of an individual give rise to the action’s kinematics). In other words, using a typical frame of an action, we reconstruct its prototype (its kinematics), and with the help of the prototype, the function (the dynamics, or supporting force) of the action. Thus, a definition (a verbal description of the word illustrated by the picture) is here completely unnecessary. This accounts for the universality of the given dictionary fragment—its ability to be comprehended by anybody, of any ethnic background. 2. A reference-based approach to describing notional words 46 Fig. 2. Pictures of prototypes of human actions that help us instantly identify  6 these actions. There is no need for any additional definitions to explain the meanings of the words which name these pictures In dictionaries, such pictures are often used as visual supports for definitions, when subtle distinctions between look-alike referents, such as animal species, similar actions, etc., need to be explained. The important role of pictures in describing concrete lexical items has been stressed by Apresian (2010a: 22): The dictionary [Longman] contains hundreds of color pictures, drawings, schemas, and photographs of various material objects supplied with the names of their parts, components, and types. In other words, the principle of ostensive definition of lexical items has been realized here. Moreover, many physical actions are also defined ostensively, because it is easier to show subtle distinctions between them with the help of a clear picture, rather than with a verbal description. Examples of such pictures are the pictures showing a jump, a hop, a leap and a skip. 6 Adapted from the dictionary English Verbs of Motion (URL: http://www.fluentland.com/ groups/learn-english/forum/topic/english-verbs-of-body-movement/. Accessed on 24.03.2018) 2. A reference-based approach to describing notional words 52 Now, we can formulate the definitions of basic meanings of the words chair armchair. chair. and Let us begin with In contrast to definitions (1a)– (3a), in which the visual and functional features of a chair are syncretized, we will set them strictly apart: the Prototype shall represent exclusively the outward appearance of a chair, and the Function—exclusively its anthropocentric (functional) features. (8a) chair 1 (basic meaning) = Prototypical chair → Function halfdesigned for sitting on by one person in a steady (semi-relaxed) posture convenient for different kinds of work that involve the use of → hands, table/desk; usually at a the body has one main support for the buttocks and two extra supports—for the back and feet Presented in this way, basic meaning (8a) sets the category “chairs” on two planes: the Prototype sets it as a fuzzy category (degree of similarity to Prototype), and the Function—as a strict Aristotelian category. Now, let us turn to armchairs: (8b) armchair 1 (basic meaning) = Prototypical armchair: “image” → Function: designed for sitting on in a quite steady, secure (almost relaxed) comfortable for a brief rest, posture usually in public (with the outer clothes on); the body has one main support for the buttocks and two extra supports—for the back and the arms. a fully recumbent A bed, in contrast to an armchair, allows one to take posture providing for full relaxation of the body, which has supports for all of its main parts: the head, the back, the buttocks, and the legs. A bed is meant for a prolonged night rest without outer clothes, on bed sheets with a blanket. Note. In light of what has been said above, let us analyze the uses of the word such as in kreslo voditelja ‘driver’s seat’ (lit. ‘driver’s armchair’). In our view, a driver’s seat possesses most of the features characteristic of an armchair. Indeed, the driver must sit, slightly leaning on the back of the seat, in a sufficiently steady posture; his feet and hands must be free from the function of supporting and keeping balance, available for work: stepping on the clutch, accelerator, and brake pedals for the feet, and steering the wheel and working the stick-shift (in the case of a mechanical transmission) for the hands. A driver’s seat often has armrests (for enhanced stability). § 1. A reference-based approach to lexical semantics 53 At the same time, a driver’s seat must provide for sufficient body mobility, so the driver could, when necessary, turn left or right or look back—the degrees of freedom allowed by a chair. As can be seen, in sum, the features of an armchair prevail in a driver’s seat. Let us remember our foreigner now. Having at his disposal the functions from (8a) and (8b), he will be able to use correct designations chair and armchair for perceived objects designed for sitting on. Therefore, these definitions explain (a) how the listener understands these words in discourse or a text, and (b) how the speaker chooses a suitable word to designate an object for sitting. The arguments given above allow for the following hypothesis: a native speaker of Russian, before naming a perceived article of furniture with a seat for one person either a chair or an armchair, first decides, judging by its outward appearance, what its corresponding Function is, i.e., what posture will be taken by a sitting person: half-steady (for work) or almost steady (for a rest or a friendly conversation). And in order to decide, one must mentally “sit” on this object. Depending on the result of such a virtual experiment, the perceived object will be named either a chair or an armchair. It follows that the visual features (presence of armrests, incline of the back, height of the seat, etc.) are used not to directly identify the perceived object (i.e. to decide whether it is closer to the prototype of a chair or an armchair), but to evaluate its functional potential from the point of view of a sitting person. Consequently, the category “chairs” includes not only typical chairs but also non-typical “designer” chairs (Fig. 4) which are nevertheless capable of functioning more or less as typical chairs. Fig. 4. A prototypical chair (left) and non-typical (designer) chairs (right). The unusual shapes of designer chairs do not prevent their inclusion in the category “chairs”, designating them as chairs 2.2. What makes chairs, armchairs, and stools distinct? Now, let us take a look at a stool. It allows one to sit on it in an unsteady posture, demanding full control over the body’s vertical alignment. As a result, the sitting person’s body acquires a higher degree of mobility: his back not leaning on anything, 2. A reference-based approach to describing notional words 54 it is easier for him to stand up, turn his body to the left or to the right, etc. And here certain problems with reference may also arise, because there are articles of furniture with legs, a seat, and a very low back (e.g., American barstools). The height of the back, as a criterion in including such an object into the class of chairs, is also assessed functionally by its ability to serve as a full-fledged support for the back of the sitting person.  7 As a result, this is what we come to: (8c) stool 1 (basic meaning) = Prototype: “image” → Function: designed for sitting on in an unsteady (non-relaxed) posture that provides for maximum body mobility (such as standing up, turning around, etc.) Summing up the arguments given above, the following important concludiscreetness may be drawn: the of the categories ‘kresla’ (armchairs), ‘stul’ja’ (chairs) and ‘taburety’ (stools) is caused by the differentiation of the three postures (states) of a human body when sitting, and by the three kinds of human sedentary activity characteristic of these states. Based on this assertion, it seems reasonable to believe that the suggested functional (causal) differentiation of armchairs, chairs, and stools is cross-cultural by nature; it is characteristic of those nations in which these kinds of sedentary activity are very common and, therefore, clearly distinct. In particular, it is true about the Russian and English lingua-cultures, which is circumstantially attested by the prototypical chair given in Longman (2009: 263): it does not feature armrests and fully coincides with the Russian prototype given in (8a). It means that the categories set by the Russian words kreslo and stul and their English correlates armchair chair, and are strictly separate, despite the fact that in the Russian and English definitions, (2a) and (2b), an armchair is defined as a subclass of chairs. 7 The aforementioned functional differences characteristic of stools, chairs, and armchairs, account for the localization of their “habitats” in the sphere of human activity connected with sedentary position. A stool is used to sit on for a brief period of time, with an option to quickly stand up, turn around, sit down again to face the person sitting next to you, etc. (that is why there is no use for armrests or a back in this case)—in bistros, bars, an eye doctor’s office, where you must sit straight while your eyesight is being tested, etc. A chair is for a more prolonged sitting, with a possibility to do some work using one’s hands (that is why armrests are not needed)—at a desk or a table, etc.; and an armchair is for rest, allowing for a relaxed posture—e.g., in hotels, on trains, on board of airplanes and the like. § 1. A reference-based approach to lexical semantics 55 2.3. Strict segregation of Prototype and Function. There are several important reasons to represent the basic meaning of a word in the form of a pair “Prototype → Function”, where the visual and the functional features of the referents are not mingled (syncretized) but strictly segregated. We will note just two of them. (1) Prototype and Function often take separate parts in generating meanings derived from the basic meaning via metaphor. Consider, as an example, the mammoth stone chair in Rene Magritte’s painting (Fig. 5): metaphorically, it may well be designated by the word chair, although it resembles a chair solely by its outward appearance and its function is quite different. Fig. 5. The Legend Fig. 6. Fig. 7. Double Bench, of the Centuries, Electric chair by Antoni Gaudi (c. 1907) by Rene Magritte (c. 1950) Another example is an electric chair (Fig. 6), which looks like a typical chair. However, its functional (causal) characteristic is totally different and, judged by this component, it does not satisfy the function of meaning in (8a), cf.: the arms of the man sitting in it are strapped to the arm-rests, and he cannot electric (nor is expected to) engage in any activity. The metaphorical expression chair is motivated solely by its outward resemblance with a prototypical chair. Finally, one more example—a piece of furniture made by Antonio Gaudi (Fig. 7). This object may be metaphorically called either a chair (cf. its Russian Stul dlja dvoix name ‘Chair for two’), or an armchair. Since it is an object of  8 art and it wasn’t Gaudi’s intention to make it comfortable for sitting, it combines, in equal measure, the features of both a chair and an armchair (low seat, but straight back etc.). 8 Its name in Spanish, “Banco Doble Casa Batlló” (Barcelona, Gaudi, 1907) translates as ‘Bench for two from Batlló’s Estate’. Note that the word banco translates as ‘bench’ or ‘seat’, not as ‘chair’ or ‘armchair’. § 2. An analysis of basic meanings of the action verbs... 63 3a) force interaction occurs: X transfers to Y all or most of its force of motion THEREFORE 4a) Y suffers a sharp shock (= is shaken hard); X also suffers a sharp shock and is shaken hard. Note. The notions used in (**) are simple and clear even to a child from his personal experience of playing with toys, without any knowledge of physics or, to be more precise, before any such knowledge. A child understands very well that a moving object (e.g. a ball) possesses some force of motion and that to stop it some force should be exerted to block this motion, and that when hitting (e.g. a ball), the force of motion (of a foot or a hand) is transferred to the stationary object and it starts moving (if the force is sufficient).  9 These notions belong to the initial notions that a child forms early on in his active experience of the world. Through these and other similar notions a child interprets the surrounding world (for more details on the development of such notions in children see chap. 2, § 7). Later, these initial notions add up to form basic meanings of the words of the mother tongue acquired by a child. Let us continue the referential analysis of the verb udarit’ ‘hit-PF’, taking the formula in (**) as a working definition. First, X does not necessarily lose its force of motion (as in hitting a ball with a foot) nor Y its force of steady state (as in a stick hitting a fence). Second, the ‘shock’ suffered by Y must be substantial—cf. the following utterance, questionable in a typical situation: *Snežinki udarjali v lobovoe steklo mašiny ‘Snowflakes kept hitting the car’s windshield’. However, the following is Krupnye kapli doždja udarjali v lobovoe steklo mašiny quite normal: ‘Large raindrops kept hitting the car’s windshield’. On the other hand, the force of Y’s steady state (resistance) must also be considerable—cf. the questionability of the phrases with the past tense form udarila ‘hit’ in the following examples: *Pulja udarila v steklo ‘A bullet hit the glass’, when the bullet easily 9 For example, 7-months-old infants understand the action ‘prjamoj tolčok’ [‘direct push’]. Thus, if a child sees object A moving toward object B and then coming into contact with it, after which object B immediately starts moving, the child believes that the cause of B’s motion is A exerting force on B by contact. It has been shown experimentally (Subbotskii 2007: 176–178) that a child doesn’t establish such causal relationship in the case of B beginning to move if A stops without coming into contact with B (a space is left between A and B), or when B begins to move not immediately after the impact, but after a noticeable delay. 2. A reference-based approach to describing notional words 64 pierced ordinary glass leaving a little hole, and *Vzryvnaja volna udarila v stenu doma ‘A shockwave hit the wall of the house’, when the house in question, along with the neighboring houses, was demolished by the shockwave. Of course, being hit with or by X may lead to destruction of Y, cf.: Stupka udarila po čaške i razbila eё ‘The mortar hit a cup and broke it’. What is important is that Y’s resistance be commensurate with X’s sufficiently strong force of motion and able to neutralize or considerably diminish it. Lastly, the requirement that transfer of the force of motion from X to Y be instantaneous is characteristic precisely of a hitting event. In a pushing or throwing event transfer of the force of motion does not occur so quickly. Using the definition of a hitting event in (**), we come to the following description of the function of the verb udarit’ ‘hit-PF’: X udaril po Y-u t (5) ‘X hit Y’ (Function) = At a moment (1а) between a moving object, X, possessing a considerable ‘force of motion’, and a stationary object, Y, possessing a considerable ‘force of steady state’ (2а) a momentary (3a) force interaction occurs: X transfers to Y all or most of its force of motion THEREFORE (4a) Y suffers a sharp shock, is shaken hard; if Y is a living being, it experiences a feeling of pain. 2.3. Basic meaning. Now we have all the necessary information to organize a udarit’ description of the basic meaning of the verb as a dual structure shown in (1): X udaril po Y-u (6) ‘X hit Y’ (basic meaning) = Prototype (4) → Function (5). When the word udaril ‘hit-PF Past Tense’ is used in speech, each of its meaning components (the function and the prototype) performs its own semantic function. Imagine that the speaker observes a collision of a moving object, X, with a stationary or moving object, Y. The identification procedure for the perceived action allows the native speaker to instantaneously (and regardless of his own will) decide which function is satisfied by the action. Let us suppose that, by its causal features, this action satisfies the function in (5). udarit’ In that case the verb is instantly foregrounded in the speaker’s mind and the speaker can use it to designate the perceived action. Let us now assume that the hearer perceives not the action itself but the phrase Molotok udaril po gvozdju ‘The hammer hit the nail’. The interpreta- 2. A reference-based approach to describing notional words 68 by the hill. A fall begins only on a sheer descent when the support practically ceases to affect the speed of the wheel. Similarly, a hewn tree falls because its support does not affect the process. Consider the following questionable utterance: Pylinki medlenno padali na stol ‘Specks of dust were slowly falling on ? padali opuskalis’ the table’ (instead of ‘were falling’ it is better to say ‘were descending’). In this example the downward motion of the specks of dust is impeded by the resistance of the air. 2. A landing may cause damage to X. Let us ask ourselves: why cannot one say about a skydiver moving rapidly toward the ground that he is falling, while it is possible to say so about a balloonist whose balloon begins to descend, slowly but uncontrollably? An answer such as “a skydiver can control his motion and the balloonist cannot” may not be accepted, because, instead of a man with a parachute, it can be a dog or even an inanimate object—for example, a tank. Moreover, one can hardly say about a person diving from a diving board or jumping to the ground from a truck that he is falling, although his motion toward the ground depends only on the force of gravity and is not impeded by any auxiliary means. An answer to the question above may be formulated as follows: the utterance X padajet ‘X is falling’ is correct if X’s downward motion ends in an uncontrolled landing, i.e. collision with the ground wherein X receives/may receive a very strong shock. Indeed, in a parachute descent the impact is weakened and doesn’t lead to deadly consequences, as is also the case with pole vaults, jumping from a diving board or from a truck. On the contrary, when descending with an umbrella instead of a parachute, or jumping off a 10-storey building, a person is unquestionably falling because the impact with the ground is not weakened. If a person jumps from a high-rise onto a stretched canvass below, which cushions the impact, to say that on padaet ‘he is falling’ would not be quite correct. Clearly, the observer’s viewpoint, his knowledge about the perceived descent of an object, and his interpretation of the consequences of the object’s contact with the ground, play an important role. For example, it is incorrect to say about a dive-bomber (and its pilot) On padaet ‘It (he) is falling’ because the pilot controls the dive and can stop it at any time. However, the utterance becomes correct if control over the bomber is lost. By the same token, this utterance can be used about a skydiver in free fall (he intentionally delays opening the parachute), but only metaphorically. At the same time, a person unfamiliar with parachutes and their use may well say about a descending skydiver: On padaet s kakim-to gribovidnym predmetom, lit. ‘He is falling § 2. An analysis of basic meanings of the action verbs... 69 with some mushroom-like object’, because the speed of the skydiver’s descent is fast enough to allow for such an interpretation. Now, a definition of the basic meaning of padat’ ‘fall’ can be formulated: X padaet na zemlju (7) ‘X is falling to the ground’(basic meaning) = Prototype: X is moving rapidly downward → Function: 1) X is under the influence of its own force of gravity (the main cause of its downward motion) and no other forces are speeding up or slowing down its downward motion; and 2) (prediction) this downward motion will result in X’s collision with the ground, wherein it will receive a strong impact leading to X’s damage or destruction. Note. The utterances Pizanskaja bašnja padaet, lit. ‘The Tower of Pisa is Sputnik padaet na zemlju uže vtoroj falling’ (The Tower of Pisa is leaning), god ‘The satellite has been falling on Earth for a second year’ possess metaphoric meanings. The Tower of Pisa is kept from falling by the forces which hold it, and the satellite is influenced by the unusually weak (for earthly conditions) force of its weight. Such uses are also specific in that they realize not the current (referring to what is being observed) but the background meaning of the verb padat’ related to very slow and long-lasting processes (about this meaning see p. 59). Nevertheless, both situations feature important properties which meet the definition in (7) and this is what motivates the metaphors. Indeed, the motion of the Leaning Tower of Pisa or of the satellite is such that in the future they (a) will reach the ground under the force of gravity, and (b) will be destroyed (such is our prediction). 4. The verb brat’ ‘take-IMP’ / vzjat’ ‘take-PF’     4.1. Definitions. Just as in the previously discussed cases, definitions of the verb brat’ (both dictionary and lexicographic) give a good description only of its typical action referents. In a manual on writing dictionary entries for Aktivnyj slovar’ russkogo jazyka Dictionary of the Russian Language), (Active Apresian (2010b) gives a definition (synopsis) of the basic meaning of the brat’: Brat’ 1.1 knigu v ruki verb ‘to begin to hold’ (brat’ ‘to take a book in one’s hands’) (p. 83). In an earlier work by Dobrushina and Paiar (2001) a similar definition is given: “to begin to hold Y [e.g. a book] with the fingers, or the hand, or the hands” (p. 125). V. Apresian’s definition is based on the very same feature: 2. A reference-based approach to describing notional words 74 4.4. Basic meaning of the verb brat’ ‘take-IMP’ / vzjat’ ‘take-PF’. Let us continue our analysis. Rozina (2003), in the interpretation of the phrase in (8), uses the following characteristic: “Y is in X’s hand, and he can do with Y anything he wants” (p. 233). This characteristic needs to be made more precise because, for example, pulling on a branch of lilacs and holding it in his hand a person can, in fact, do anything he wants with it: smell it, break it, wave with it, etc. However, as mentioned earlier, it is not correct to say *On vzjal vetku sireni ‘He took a branch of lilacs’. Rozina points out that the verb vzjat’ ‘take-PF’ “in its basic meaning belongs to the class of verbs of motion, more precisely—to the subclass of movement of an Object (examples of other verbs from this class are sxvatit’ stjanut’ ukrast’ utaščit’ unesti ‘grab’, ‘pull off/away’, ‘steal’, ‘drag away’, ‘carry away’” (p. 231). This claim reflects a typical visual feature. However, Zaxar it is not obligatory; this is proven by a later example given by Rozina: ostanovil na nёm krovavyj, tjažolyj vzgljad, potom, ni slova ne govorja, vzjal butyl’ za gorlo (Bunin), lit. ‘Zakhar laid a heavy look of his bloodshot eyes on him, then, without saying a word, took the large bottle by the neck’. It is clear that the bottle may have been tilted, but not necessarily shifted. As for the main (functional) feature, Zakhar now has the ability to move the bottle in space. According to Seliverstova (2004: 276–277), uses such as Petja vzjal knigu ‘Pete took the book’ are characterized by the feature “X becomes a spatial support for Y”. It is the absence of this feature that explains the impossibility sxvatit’ vzjat’ of changing the verb ‘grab’ to the verb ‘take-PF’ in utterances such as On sxvatil (*vzjal) eё v svoi ob”jatija, lit. ‘He grabbed her in his embrace’ (here X doesn’t become a support for Y). “Certain exceptions” are also mentioned for which the absence of this feature doesn’t bar the use of the verb vzjat’ ‘take-PF’, for example: Ona vzjala ego za pleči i načala trjasti, lit. ‘She took him by the shoulders and started to shake him’; On vzjal eё za ruku i potjanul za soboj, lit. ‘He took her by the hand and pulled her after him’. The characteristic offered by Seliverstova is, on the whole, correct (and functional!), but too strong. If X clasps his fingers on the neck of a bottle and tilts it slightly, the bottle still continues to be partially supported by the surface Ona vzjala ego za but may be moved by X as needed. Therefore, the phrase pleči i načala trjasti ‘She took him by the shoulders and started to shake him’ is quite correct; compare with the similar phrase On vzjal kadku za uški i stal trjasti, lit. ‘He took the tub by the tabs and started to shake it’, which doesn’t mean that the tub was lifted. The phrase On vzjal eё za ruku i potjanul za soboj ‘He took her by the hand and pulled her after him’ is a different matter. If it is § 2. An analysis of basic meanings of the action verbs... 75 about a woman following the agent of her own will and using her own energy, On vzjal lošad’ pod uzdcy i povёl it is metaphorical—compare with the phrase eё, lit. ‘He took the horse by the bridle and led it after him’. If, however, the situation is such that the insensible body of a woman is pulled on its arm, then the meaning is literal—compare with the phrase On vzjal mešok za uško i po taščil za soboj, lit. ‘He took the sack by the tab and dragged it’. Here the sack is only partially supported by the agent. As can be seen, Y may be a living being but only on condition that it is profiled in the situation as a physical body. Joining the presupposition and the assertion gives the following resultant definition: (11) Čelovek X berёt/vzjal ob”ekt Y ‘Person X takes/took object Y (basic meaning) = Presupposition (9) + Assertion (10). To give this definition the heretofore accepted form “Prototype → Function”, the visual and functional features will be combined separately, yielding the following: Čelovek X berёt/vzjal ob”ekt Y (11а) ‘A person X takes/took an object Y (basic meaning) = Prototype: (presupposition) object Y, physically integral and unattached to its surroundings, is in a steady state on a surface + (assertion) p erson X grips the object Y and shifts it → Function: (presupposition) object Y possesses the force of steady state due to its gravity + (assertion) person X, having a goal to control the movement of Y, overcomes its force of steady state (caused exclusively by its gravity) and can begin to move it in any direction. 4.5. Explanation of incorrect uses. The definition in (11a) makes an explanaPetja of the incorrect examples given in section 4.1 possible. In the phrase *vzjal jabloko s dereva ‘Pete took an apple from the tree’ Y (the apple) is not detached from the branch and, therefore, possesses an additional force of steady state, while in the phrase Petja *vzjal beguščuju košku ‘Pete took the running cat’ Y (the cat) doesn’t possess a force of steady state at all. The two remaining phrases, Petja *vzjal šljapu s krjučka (correct verb: snjal ‘took off’) ‘Pete took his hat off the peg’ and Petja *vzjal zapisku s pola (correct verb: podnjal ‘picked up’) ‘Pete took the note from the floor’, are incorrect vzjat’ because the verb ‘take-PF’ in these contexts designates not only the action of “taking”, but also the actions that follow. In the first example, Pete took hold of the hat and raised it a little to get it off the peg, while in the second example he bent over, took hold of the note, and then resumed the initial posture. § 2. An analysis of basic meanings of the action verbs... 79 Analyzing prototypical interpretations suggested by Fillmore and Jackendoff (the latter made Fillmore’s interpretation more precise), Wierzbicka concludes: … this analysis is unsatisfactory, too, because it fails to predict, for example, that if a train went quickly up a hill it couldn’t be described as ‘climbing’. I would propose the following […]: X climbed . . . [a mountain, a tree] = sometimes in some places if people want to move upwards they have to move both their legs and their arms X moved like people move at those times in such places (p. 166ff). As before, the meaning of the verb vzbirat’sja will be described as a dual structure “Prototype → Function”. A crucial distinction between the Function and the Invariant must be emphasized. The Function does not describe a complete set of meanings (uses) of a word; it describes only the basic meaning, the direct referents and therefore of a word. Along with the basic vzbirat’sja meaning, the word has extended, metaphoric meanings. For example, the following uses are unquestionably metaphoric, which is why no attempt needs to be made to include them either in the Prototype or the Function: The train climbed the mountain, The temperature climbed to 102 degrees, Bill climbed down the ladder, Traktor vzbiraetsja na goru ‘The tractor is climbing the mountain’, Ulitka vzbiraetsja po stvolu dereva ‘The snail is climbing up the tree-trunk’. Anna Wierzbicka and her opponents do undertake to include such examples in their definitions of Prototype and Functions, in fact making no distinction between the basic meaning and an extended meaning. But they offer different definitions. Ray Jackendoff speaks of the possible suppression of now one, now another component of the prototype. Wierzbicka’s invariant contains (a) a functional part which defines the prototypical motion ‘a person climbing’, and (b) a relationship of similarity for other “climbing” agents (trains, temperatures, etc.)—“X is moving in a way people move in such places”. Thus metaphorization appears to be implicitly built in the invariant: such occasional uses are explained on the grounds of their similarity to prototypical uses. In the Longman Dictionary (2009) some of the abovementioned uses, on the contrary, are explained separately (see below). vzbirat’sja 5.2. Basic meaning of the verb ‘climb’. Dictionary definitions of this word usually contain both visual (“move upward”) and functional (“with difficulty”) features, cf.: 2. A reference-based approach to describing notional words 80 vzbirat’sja 1. S trudom vzlezat’, podnimat’sja vverxh na čto-nibud’ ‘To climb up with difficulty, to rise to the top of something’ (Ushakov, I: col. 271). climb 1. MOVE UP/DOWN—to move up, down, or across something using your feet and hands, especially when this is difficult to do: Harry climbed the stairs. 2. TEMPERATURE/PRICES ETC. To increase in number, amount or level (Longman 2009: 300). vzobrat’sja, Vzobrat’sja na gorku ‘Climb-PF up a hill’; Malyš vzobralsja na stul Oni vzobralis’ na čerdak po ‘The kid climbed up on the chair’; derevjannoj lestnice ‘They climbed to the attic by the wooden ladder’. MEANING. A1 vzobralsja s A2 na A3 [‘A1 climbed-PF from A2 to A3’] ‘A living being, A1, usually making a big effort and using all the limbs, moved from a surface or from a place, A3, to a surface or a place, A2, higher than A3, often by using a contraption or an object, A4’ (Aktivnyj Slovar’ 2014, II: 106–107). literal use It will be assumed that the action referent of this verb in its can be performed only by a living being, A, which has limbs (a human, an animal, an insect). To begin with, consider the examples of motions of A which satisfy the component “move up using both the feet and the hands” (this component is used both by Wierzbicka and her opponents) but which are not referents of the phrase A vzbiraetsja ‘A is climbing’: 1) A diver is coming back up to the surface of the water. 2) A house-painter is going up in a bosun chair on a cable, rotating the pedals of the lifting mechanism with his hands and feet (cf. the definition component “often by using a contraption or an object, A4” from Aktivnyj Slovar’). 3) A young man is going up a convenient spiral staircase (or is boarding a bus), holding the handrails. Now, let A be an animal or an insect moving, as Wierzbicka (1996) would put it, “in a way people move in such places”: 4) A cat or a leopard is climbing [lezet] ( vzbiraetsja is questionable) up ?  12 the tree. 5) A cockroach is crawling [polzёt] (*vzbiraetsja is incorrect) up the wall. 12 The Russian verbs lezt’ and vzbirat’sja are both translated as climb in English, the difference being that vzbirat’sja implies ‘with difficulty’.—Translator’s note. § 4. The motion verbs... 95 bežat’ 1. By intensely speeded motion, and quickly shifting one’s feet up and down, to move in some direction (MAS, I: 68). walk 1. to move forward by putting one foot in front of the other (Longman 2009: 1966); run 1. MOVE QUICKLY USING YOUR LEGS; to move very quickly, by moving your legs more quickly than when you walk: He was running towards the door (Ibid.: 1531). Definitions from lexicographic research: idti 1. (1) To move, shifting one’s feet up and down, at a normal speed. bežat’ 1. To move fast, shifting one’s feet up and down (Gak 1977: 28). (2) Čelovek Х idёt iz Y-а v Z [lit., ‘A person, X, is walking from Y to Z”] ≅ ‘Person Х moves over a surface from Y to Z, shifting his feet up and down and never completely losing contact with the surface crossed’ (compare, by contrast, with bežat’—‘periodically losing contact with the surface’) (Apresian 1974/1995: 108). (2а) bežat’ 1.1. Ot lesa napererez banditu bežal policejskij [‘From the wood, a cop was running to intercept the bandit’]; Oleni begut na vodopoj [‘The deer are running to a watering place’]. MEANING: А1 runs to А2 from А3 over А4 to (do) А5 ‘A living being, А1, moves at a run to place А2 from place А3 over surface А4 in order to А5’. begóm Deti begom brosilis’ iz klassa ‘at a run; running’, ADV. (lit. ‘The children, running, dashed from the classroom’). MEANING. ‘Evenly moving by alternate pushes of feet from a surface so that, for a moment, the feet completely lose contact with it’ (Aktivnyj Slovar’ 2014, I: 177, 181). At first sight, all these definitions appear to be quite adequate. They d escribe the typical visual appearances (kinematics) of walking and running, and they use visual features as the main distinctive features: either the speed of motion, which is higher for running and lower for walking (dictionaries, Gak), or “periodic loss/no loss of contact with the surface of motion”, see (2). 1.2. How do the human walk and run differ? The seeming adequacy of the definitions in question may be put down to the fact that we have a very good idea of the visual prototypes of walking and running which are easily recog 15 (cf. Figures 1 and 2). 15 I got these diagrams by fixating, frame by frame, the linear version of the point-light displays of the human walk and run available on the website of the Bio Motion Lab at 2. A reference-based approach to describing notional words 96 Prototypical walk Prototypical run Fig. 1 Fig. 2 However, we shouldn’t forget that, as visual prototypes, they reflect only the subclasses of typical movements. As a matter of fact, movements belonging to one category—for example, versions of a human run—may differ considerably in appearance, while movements belonging to different categories (such as running and walking), on the contrary, may be very similar. Yet a native speaker can always identify them easily and unambiguously. On the one hand, it doesn't matter to him “whether it is Carl Lewis circling a track or Grandma running to the telephone” (Golinkoff et al. 2002: 604). Both kinematic images are identified as running, regardless of how different they are. On the other hand, a native speaker doesn’t have a problem distinguishing between “Grandma’s” run and fast walk, although in either case her feet don’t lose contact with the floor, and her run doesn’t have any “flying” phases, which are typical features for distinguishing a walk from a run. In addition, there isn’t much difference in speed between her fast walk and her slow run. The speaker usually knows (“sees”) when a person is running and when walking. In real life there aren’t any “in-between” cases, when human locomotion could be called either running or walking at the same time. This begs the question, “What features does the speaker use to distinguish between a run and a walk?” As we have seen, the feature “high speed” doesn’t serve the purpose (one person’s walk can be faster than another’s run), just as the feature “loss of contact with the surface” doesn’t. Queen’s University (Kingston, Ontario), headed by Nikolaus Troje (cf. Troje 2002; http:// www.biomotionlab.ca/Demos/BMLwalker.html; and http://www.biomotionlab.ca/Demos/ BMLrunner.html, access 05.04.2018). The site shows not only typified animations of the human walk and run (both point-light and linear), but also more specific patterns for such locomotion of humans, determined by the sex, speed, etc. The animations demonstrate the human ability to instantly recognize not only the type of locomotion as such, but also its variable attributes: whether it is a man or a woman, stout or skinny, happy or sad, nervous or relaxed. § 4. The motion verbs... 97 1.3. An analysis of prototypical definitions. Let us focus on definition (2). To show that it cannot perform the role of function and differentiate between a walk and a run, we will use a technique typical for the reference-based approach. We will think up several exotic, borderline examples of human locomotion that will fit the definition (2) Čelovek idёt ‘The person is walking’ while referring to other types of motion: ‘The person is running/crawling/ riding’. (a) Čelovek bežit ‘The person is running’. An elderly person can run without losing contact with the ground. Pushing off from the ground, he maintains a contact with it which is of the touching type (a shuffling run). No native speaker of Russian would call this kind of motion a walk, in spite of the fact that all the conditions from the visual description (2) are satisfied. Čelovek polzёt (b) ‘A person is crawling’. Imagine the following situation: a captive is lying on his back with his bound arms on his chest. If he started moving, each foot in turn pushing off from the ground, such motion would also satisfy definition (2). However, in reality, the man would be crawling, not walking. edet (c) Čelovek ‘A person is riding’. Imagine an adult sitting on a child’s bicycle and propelling himself by pushing off with his feet from the ground; or, similarly, imagine a boy on a sled moving along an even, snow-covered road by pushing off with his feet. In either case the motion satisfies definition (2), but a native speaker will say Čelovek/mal’čik edet ‘The person/boy is riding’, not Čelovek (mal’čik) idёt ‘The person/boy is walking’. That a person uses an artifact as a means of locomotion is not an issue here; an old person may move his feet while using a special prop on wheels for support (a walker), but he would certainly be walking, not riding. Of course, one could argue that (a)–(c) are rare, specially devised examples that may be ignored. However, such an argument cannot be accepted if our all goal is to define the characteristic feature of direct referents. Let us stress again: the sought feature cannot be singled out by ‘adjusting’ definition (2) and excluding examples (a)–(b). True, example (c) may be excluded if the component ‘X is relatively tall’ is added to definition (2). Yet this component doesn’t help to solve the problem. Imagine a metamorphosis such as a fly growing the size of an elephant. In such a case it couldn’t be said about the fly that it crawls. If, reversely, an elephant becomes the size of a fly, we still cannot say that it crawls (the number of legs doesn’t matter—a turtle crawls, even though it has four paws). 2. A reference-based approach to describing notional words 98 1.4. Basic meaning of the verbs idti ‘walk’ and bežat’ ‘run’. Within the framework of a referential approach the meaning of a sensory (concrete) verb is represented by the pair ‘Prototype → Function’ (cf. definition (7) in § 1). Let us explore the functional components of the verbs idti and bežat’, taking definition (2) as a prototype. It should be reminded that the semantic distinction between these verbs consists in either retaining (idti) or losing (bežat’) contact with the surface. A detailed referential analysis by Koshelev (1989) has shown that, in order to come up with adequate definitions for these verbs, causal features must be used that are not perceptually given: ‘push’, ‘support’, ‘unsteadiness’, etc. The resulting descriptions of a walk and a run are, essentially, as follows. propels himself A walking person, A, over a surface (using his inner pushes off from the surface with his forces). While moving, he alternately supporting foot—the one that bears the weight of his entire body—and gradually, without losing support, transfers the weight of his body to the other foot, which now becomes a support, and so on. At the same time, and throughout the motion, A keeps his balance, being in an unsteady position (he may fall down on the surface over which he is moving), unlike in the case of steady polzёt motion by pushes described by the verb ‘is c rawling’. Čelovek A bežit A phrase such as ‘Person A is running’ designates a similar repeatedly loses support type of locomotion with one difference: the person when moving his body weight from one limb to the other. These observations allow us to formulate the following functions: (3а) Čelovek A idёt po poverxnosti v punkt Z ‘Person A is walking over a surface to point Z’ (Function) = (i) Person A, pursuing his spatial goal to get to point Z, propels h imself; (ii) he alternately rests his feet on and pushes off from the surface, each time moving his body weight from one foot to the other, (iii) not at any time losing the support of the surface; (iv) at any given moment the position of A is unsteady. (3b) Čelovek A bežit po poverxnosti v punkt Z ‘Person A is running over a surface to point Z’ (Function) = (i) Person A, pursuing his spatial goal to quickly get to point Z, propels himself; (ii) he alternately rests his feet on and pushes off forcefully from the surface, each time moving his body weight from one foot to the o ther, 2. A reference-based approach to describing notional words 102 Function”, offered in a number of publications (Koshelev 1989, 1990 and 1996), is close to the dynamics approach to action recognition (Runeson 1977; Runeson and Frykholm 1983; Bingham et al. 1995; Shipley 2003; Bingham and Wickelgren 2008). According to this approach, the specifics of an action’s kinematics are determined by the action’s dynamic features, in particular by its force-dynamic features (such as gravitation, shock, etc.) that are hidden from direct perception. However, humans identify perceived actions precisely by these dynamic, but not outward features (kinematics). Without going into more detail, it is sufficient to mention a curious fact from the history of cinematography. A good idea of the principles of perception of locomotion events is given by Hollywood’s attempts to produce natural movements of fictitious creatures. For example, in the first versions of the Godzilla movie the monster, despite all the efforts of the animators, moved just like a mechanical toy. Only after Hollywood had learned to use good dynamic models to produce appropriate movements did the moveJurassic Park—start ments of such creatures—for instance, the dinosaurs in to look realistic (Bingham and Wickelgren 2008: 276). An observation made by Johansson (1976) directly bears on this fact: in the process of recognition of point-light displays of human motion it is not just the structure of connected segments that is perceived; it is something more, as the observers in the experiment distinguished between point-light displays of humans and similar displays of segmentally connected dolls. 1.5. An animal’s run (quadrupedal locomotion). The function of a run (3b), formulated, for clarity, only for humans, remains in effect for any bipedal agent such as an ostrich, a chicken, a penguin, etc. For quadrupedal locomotion (a cat/dog/horse is running) the meaning of the verb bežat’ ‘run’ has a different prototype but the same function (3b). For evidence, let us consider the run of a horse as an example. For this purpose, we will use the concept of paired take-off limbs. If a horse is running at a trot, it simultaneously pushes off now with the first and the third legs, now with the second and the fourth legs, that is, alternately with one or the other paired take-off limb, each time losing contact with the surface and the support thereof. If a horse is running at a gallop, it alternately pushes off from the ground now with forelegs and now with hind legs (the front and the rear paired limbs) and the support schema of its motion is the same. As a result, the basic meaning of the verb bežat’ ‘run’ has two prototypes: Prototype 1 for humans and bipedal creatures (two single take-off limbs, vertical position § 4. The motion verbs... 103 of the body) and Prototype 2 for quadrupeds (two paired take-off limbs, horizontal position of the body). Both prototypes have the same function for interpreting the run of both bipeds and quadrupeds. The resulting definition can be stated as follows: (5) Čelovek/lošad’ bežit ‘A person/horse is running (basic meaning) ≈ (Prototype 1: A person is moving fast, shifting his feet up and down losing and periodically contact with the surface; synchronously shifting pairs of Prototype 2: A horse is moving fast, legs up and down and periodically losing contact with the surface) → Function: A person/horse, pursuing the goal to quickly get to a certain point in space, alternately rests his/its single/paired limbs on and pushes off forcefully from the surface (with one foot/leg or, synchronously, with a pair of legs), each time moving his/its body weight losing the support from one limb to the other and momentarily of the surface while keeping unsteady balance. Thus, the basic meaning (5) specifies a single category ‘X bežit’, where X is a bipedal or quadrupedal individual. This category is defined by one  16 function and has two prototypes. Note. In the case of a running quadruped, loss of contact with the surface does not always happen. For example, if a dog is running at a gallop, the contact with the surface and the support thereof are lost; and if it is running at a trot or simply moving its legs chaotically, there may be no loss of contact while the support may be lost or greatly weakened (just like in the case of a shuffling human run). However, this not so obvious borderline between the walk and the run of a dog or cat does not make these categories fuzzy. different For a dog or cat, just like for a human, walking and running are biomechanical systems of locomotion, and transition from one to the other is abrupt. Outwardly it is manifested in the absence of what might be called a dog’s or cat’s fast walk. Unlike humans, a dog can’t incrementally increase the speed of its walk; it either walks, moving its paws not fast, 16 Such an integral interpretation is found in many explanatory dictionaries. Definition (5) simply substantiates it—compare the examples in the definitions of the verb bežat’ ‘run’ above: Ot lesa napererez banditu bežal poliсejskij ‘From the wood, a cop was running to intercept the bandit’; Oleni begut na vodopoj ‘The deer are running to a watering bežat’ 1. place’ (Aktivnyj Slovar’ 2014, I). Cf. also To move fast, by forceful pushes of the legs disengaging with the ground. Lošad’ bežala rysju ‘The horse was running at a trot’ (Ushakov, I: 102). 2. A reference-based approach to describing notional words 104 or runs, rapidly shifting its paws. The run itself in this case may be either fast (at a gallop) or not (at a trot). Of course, a well-trained horse can shift between different speeds and types of walk in response to cues from its rider. idti The verb ‘walk’ in its basic meaning can be used when speaking about not only bipedal, but also quadrupedal locomotion if it is not fast: Koška/lošad’ idёt ‘A cat/horse is walking’. However, some languages do not allow for such extension of the basic meaning, cf.: In Modern Chinese […] the verbs of autonomous locomotion (without the instrument of transport) remain exclusively the prerogative of mankind. Only a bipedal being with a vertically aligned body may become zou an agent of the verb ‘walk’. Species from the animal world can only pa ‘crawl’, pao ‘run’, tiao ‘jump’, you ‘swim’ or fei ‘fly’ (Aoshuan 2004/2012: 20, 22). 1.6. The different goals of walking and running. Now the necessity to introduce the notion of A’s spatial goal—‘to move in space to some point Z’—will be illustrated. First of all, it should be kept in mind that, as far as walk run the basic meanings of and go,—‘to move in space to some point Z’ the main goal. is For example, an athlete circling a track at the stadium in preparation for the contest has a different main goal, which is not to move to a point in space but to exercise, grow some muscle, test his stamina, etc. Therefore, in our opinion, this kind of locomotion is named run only metaphorically, not in its basic meaning. If, however, a runner is taking part in a contest, a phrase such as On bežit ‘He is running’ is used in its direct meaning since the runner’s main goal is to move across the finish line as fast as he can. One of the reasons to distinguish between the basic and metaphoric uses of the verbs idti ‘walk’ and bežat’ ‘run’ is that their primary meanings have to do with the goal of changing one’s location in space. Under natural conditions a dog or an ape uses the locomotion technique of walking or running only for its intended purpose. This purpose remains the one and only for animals. A dog wouldn’t run to exercise or show off before its master (as a young athlete might do in front of some girls to attract their attention). It may run around in excitement, though, greeting its master. As for humans, a great many derivative goals have emerged in the course of cultural evolution, and walking and running are used as a means to reach these goals. As an illustration, consider some of these derivative goals and the actions caused by them. Imagine a gardener who has made a new path in the garden, § 4. The motion verbs... 105 covered it with a layer of sand, and is now trampling on it. While doing this, he is moving down the path in full accordance with points (2a)–(4a) from the function (3a), pushing off from the surface of the path with his feet, etc. However, a native speaker would not say in this case that The gardener is walking down the path (the manner of his locomotion would be somewhat different: he stamps his feet on the path, etc.). Something different would be said: The gardener is trampling the path. Similarly, a ballerina may run across the stage while doing some dancing pas, yet it would be inappropriate to say The ballerina is running. The in this case that The phrase to be used would be ballerina is dancing. As can be seen, the manners of walking and running (as types of motion, cf. Talmy 1975) may be used by humans not only for locomotion, which is the primary goal, but also for other, non-spatial goals. Therefore, in description (3a) the spatial goal of a human using the manner of walking—to move to some point Z—must be explicit. Note. There may be a hierarchy of goals for a running person. For example, if a ballerina runs across the stage while doing some dancing pas she may be pursuing a spatial goal as well: she knows the direction and the place to which she must run. This goal, however, is subordinate to another, main goal: to express by means of a run some aesthetic idea or a feeling. It is for this reason that the locomotion of a ballerina using the manner of running may be called run only metaphorically. The same is true of the example with the gardener. Although he makes steps (shifts his feet) along the path, that is, uses the manner of walking, his main goal is not spatial and his motion, therefore, would not be walking. It should be noted that as soon as the main goal of locomotion becomes a non-spatial goal it radically changes the shape of the walk/run, including the pattern of steps, the position of the body, the movement of arms, etc. The ballerina does not think about how to best make a step or a jump but about how, by means of a step or jump, to express the grace and lightness of the movements, the aesthetic essence of her dance. That is why, from a trivial point of view, a ballerina’s walk or run on the stage appear quite artificial. For example, she doesn’t step on her heels but on the tips of her toes, turning her feet sideways, etc. Similarly, the actual movements of the gardener’s feet change noticeably when making a step. He tries to raise his feet higher, stamp harder on the path, etc. It is these special features that help us identify the trampling of a path in the gardener’s walk and the dance in the ballerina’s run. It should be stressed that a walk or a run may be accompanied by other simultaneously pursued goals. For example, a 2. A reference-based approach to describing notional words 112 plete motion. In jumping they are independent and separate (they may be long composite or short, continuous or interrupted by pauses), i.e. they constitute a idti šagat’ motion. This is also true for the verbs ‘walk’ and ‘step/stride’. A polzёt/ Let us draw a brief summary. The basic meanings of the phrases idёt/bežit/prygaet/šagaet ‘A is crawling/walking/running/jumping/stepping’ define the classification of locomotion events (A’s impact-generated motion) shown in the following table. Type of impact-generated motion Motion set by the functional meaning of the verb verb Degree of A’s steadiness Complete/composite motion when moving А polzёt steady integral А idёt locally unsteady integral А bežit globally unsteady integral А prygaet globally unsteady composite А šagaet locally unsteady composite Comparing this table with the table of types of X’s motionless positions set by the verbs stojat’ ‘stand’, sidet’ ‘sit’, ležat’ ‘lie’ and viset’ ‘hang’ given in § 3, section 2.4, we can see that the most important classification feature in both cases is the degree of A/X’s steadiness. It may be hypothesized that this fact universal human nation-specific reflects the (species-specific), rather than the (Russian), classification of human locomotion and position in space. 3. Basic meaning of the verb exat’ ‘go/ride/drive’ 3.1. Definitions of the verb exat’. Let us turn to the dictionary and scientific exat’ definitions of the verb to see what kind of information they provide. (7) exat’ 1. To move in a certain direction using some means of transportation. Exat’ na paroxode, na poezde, na velosipede, na lošadjax, v sanjax ‘To go by boat, by train, on a bicycle, by horse, in a sleigh… (Ozhegov: 182). (8) X edet iz Y v Z na W ‘X is going from Y to Z by W’ = ‘X is moving from Y to Z because X is on W which is moving from Y to Z, and movement from Y to Z is among X’s goals’ (Apresian 1974/1995: 108). The picture we see here is already familiar: both definitions describe the prototype (typical action referents) of this verb. There isn’t a close lexical § 4. The motion verbs... 113 correlate to the verb exat’ in English, which is why an Englishman learning Russian will have difficulty in acquiring this lexical item. Nevertheless, the above definitions will help him to quickly understand the meaning of the verb in a particular (con)text. 3.2. The Function of the verb exat’. Imagine now an opposite situation. An Englishman sees a person riding a horse at a gallop. Guided by definitions (7) and (8), he will designate the observed motion with the phrase Čelovek edet na lošadi, lit. ‘A person is riding on a horse’. However, to a native speaker of Russian this phrase would sound at least odd, while if the horse is walking the phrase becomes correct. As can be seen, to establish the reference of the phrase Čelovek edet (in its actual meaning) a native speaker of Russian uses some other property (the function) that characterizes all direct referents of the verb exat’ rather than just its typical referents. What features constitute this function? The example of a galloping horse suggests the first feature—“support of the surface for the moving object must be continuous”. If a horse is walking or Čelovek edet na lošadi running at a trot one can say because the rider, through the intermediacy of the horse, has the continuous support of the road. If, however, the horse is running at a gallop (when the support of the road is periodically lost), to a native speaker of Russian such a phrase appears questionable, skačet the more so the longer the horse’s strides. In such a case the verb ‘is galloping’ should be used. over a surface” A second feature that deserves attention is “motion (not above or under a surface), hence the incorrectness of the utterance *X edet v podvodnoj lodke, lit. ‘X is going in a submarine’. A third feature is ‘for X, the surface over which X moves must be solid’. Indeed, it is not good to say about a person moving on a light raft down a river, on [‘he’] edet. Note that the expression ‘solid surface’ is used terminologically, meaning ‘the surface that serves as a support for X’s steady motion, preventing X from falling through’. Thus, an utterance such as My edem na paroxode, lit. ‘We are going on a ship’, is acceptable (cf. this example in definition (7) above) even though the motion is on the water and not on the ground. Compare also the incorrect phrases *X edet na lёgkom parusnike / v lodke, is going on a light sailing ship in a boat’ lit. ‘X / (here the surface of the water is inevitably interpreted as not solid) and the correct phrases X edet na voennom korable na vodnyx lyžax na vodnom velosipede, / / lit. ‘X is going on a man-of-war / on water skis / on a paddle boat’; here the surface may 2. A reference-based approach to describing notional words 118 Brugman and Lakoff 1988; Norvig and Lakoff 1987), Evans (2009 and 2015), and others. Iu. Apresian, explaining the specificity of The Active Dictionary of Modern Russian (Aktivnyj Slovar’ 2014, I), compiled by a group of linguists under his leadership, points out that “the purpose of active dictionaries is to meet the needs of spoken communication” (p. 6). This contrasts with the purpose of passive dictionaries, which is to provide information about a word necessary for understanding this word in any randomly chosen text. At the same time, there is no mention of the information necessary for the correct reference of a word in each of its meanings. Yet, as has been shown above (§ 4, section 1.7), absence of such information may lead to incorrect use of a word. To describe the semantics of motion verbs (their basic meanings), Talmy (1985: 57, 61) used the following elements: ʻMotionʼ (change of location), ʻFigureʼ (the moving object or agent), ʻGroundʼ (the reference-object), ʻPathʼ (the course followed by the Figure object), ʻMannerʼ (the mode of motion: shifting one’s feet, flapping one’s wings, etc.), and ʻCauseʼ (the cause of m otion). But all these elements are visual characteristics of a motion event; and as we have seen more than once, visual characteristics cannot serve as a basis for strict differentiation of the various types of motion events described by motion verbs. These elements are also too general. For example, the difference between walking and running lies in a rather subtle distinction between the manners of these motion events and requires a completely separate a nalysis, see § 4. In Lakoff’s (1987) theory of radial categories, the structure of a category includes a central subcategory and noncentral subcategories that are predicted from the central subcategory—the analogues of basic and derived (= figurative) meanings. At the same time, “there are no general principles that can predict the noncentral cases [subcategories.—A. K.] from the central case” (Lakoff 1987: 379). Consequently, the issue of correctness in predicting noncentral categories is ignored. This issue is also present with regard to Evans’s (2009 and 2015) framework, which ignores the issue of a word’s reference when it is used in its basic and figurative meanings (“sense extensions”). As noted by Murphy (2011) in his review of Evans (2009), the following “intriguing problem” is not discussed by Evans: […] some sense extensions are possible but others are not. […] It is common to use an author’s name to refer to his or her work but not the publisher’s name (I have been reading Dickens/*Knopf). In a book-length treatment of how words take on different meanings in different contexts, one would expect some mention of the limits on this phenomenon. […] e.g. § 5. The structure of lexical polysemy 119 newspaper can be used to refer to the company that publishes it, but book cannot […] (Murphy 2011: 393, 394). An answer to Murphy’s question may be found in section 4.2 below. A more detailed discussion of the approaches used by Apresian, Lakoff, and Evans will be offered in section 4. 2. Lexical polysemy 2.1. The unbounded character of lexical polysemy. Words are not just polysemous in the sense of having several customary meanings that are well known and given in dictionaries. The essential feature of lexical polysemy is its poim- unboundedness. In communication speakers often use words promptu, in new senses which they haven’t known or come across previously (cf. Leshchёva 2014: 37, 192). However, despite the novelty of such occasional word use, the listeners do not, as a rule, experience problems in understanding the speaker’s meaning. For example, a boy may run up to his mother, almost out of breath, breathing rapidly, and hear her say: Ax ty, moja sobačka ʻOh, my little doggieʼ. The mother in this case likens her son to their little dog whose breath is also rapid and shallow. Note that this is not a feature typical of dogs. It is typical of a particular dog that both the mother and the son know well. Therefore, there may be as many such occasional uses of the word sobaka ʻdogʼ as the number of individual features identified by the dog’s owners, which is infinitely many. sobaka A minute after the occasional use of the word described in the example above it will be forgotten, and no one will remember about it afterwards. It will not be used repeatedly in other situations and will not be committed to memory by the speakers; that is, the word will not acquire a new customary meaning. Maslov (1987: 104) called such word uses “transient”. Sometimes,  19 however, such uses may become frequent and “take root” in the long-term memory of the speakers, thus becoming a new customary meaning. And vice versa, if a customary meaning stops being used regularly, it becomes forgotten and falls out of use. This begs the question: why does language need such a unique feature that sets it apart from all other sign systems—the employment of its units in occasional uses? The answer is that language needs this feature because it uses a 19 In the English literature the terms “irregular” and “regular polysemy” are used (for a detailed discussion see: Pinker 2007: 114). 2. A reference-based approach to describing notional words 126 Consider the interpretation of the verb sorvat’ ʻpluckʼ given by Kustova: When a person plucks a flower or a fruit, he does not just detach it from the root or the branch; he irreparably SEVERS the organic ties between the object and the point of its attachment, bringing the natural growth of the object to a stop. This causes DAMAGE to the object […] it is this implied sorvat’ damage on which a wide range of meanings of the verb is based kožu na pal’ce rez’bu (sorvat’ ʻ[lacerate] the skin on a fingerʼ / ʻ[strip] the threadʼ / golos ʻ[strain] one’s voiceʼ / urok ʻ[disrupt] a lessonʼ / peregovory ʻ[derail] the talksʼ); these meanings absolutely cannot be derived from the source meaning ʻby pulling and detaching from the root, to begin to have (in one’s hand)ʼ (Kustova 2004: 39; original emphasis). However, severing an organic tie does not always cause damage, cf.: sorval speloe jabloko ʻ[he] plucked a ripe appleʼ. Therefore, the component ʻcause of damageʼ, which is the result of an interpretation of the outcome of sorvat’ the action “sorvat”, is not part of the basic meaning of the verb ʻpluckʼ. This meaning may be specified by using a reference-based approach. Let us ask the question: to what parts of a plant can the action sorvat’ ʻpluckʼ be applied? The expressions sorvat’ plod/cvetok/list ʻto pluck a fruit/flower/leafʼ are quite correct; by contrast, the expressions *sorvat’ sučok / makušku / torčaščij koren’ ʻto pluck a twig / the top / the protruding rootʼ are not. It may be hysorvat’ that the action designated by the verb is applicable only to an integral part of the plant. Yet, although a branch is an integral part of a plant, the expression *sorvat’ vetku ʻpluck a branchʼ is incorrect. It may be inferred that the object must be not only an integral, but also a renewable, reproducible part of the plant. As a result, we arrive at the following interpretation: (3) Čelovek sorval Y (cvetok, jabloko, list) ʻA person plucked Y (a flower/ apple/leaf)ʼ (basic meaning) ≈ ‘a person has a goal to be able to freely handle Y, a reproducible part of the plant (≈ to begin to hold Y in his hand) THEREFORE he takes this part in his hand and by a sudden strong movement detaches it from the rest of the plantʼ. The use sorvat’ kožu na pal’ce ʻlacerate the skin on a fingerʼ mentioned in the quote above is motivated by the same feature as in (3): ʻby an abrupt movement a person detached a reproducible part (piece of skin) from the rest of his bodyʼ. For the other figurative uses—sorval rez’bu ʻstripped the threadʼ / golos urok peregovory ʻ[strained] one’s voiceʼ / ʻ[disrupted] a lessonʼ / ʻ[derailed] the talksʼ—the characteristic feature is not the detachment of Y, but its destruction: there is no thread, the voice is lost, the lesson is interrupted. § 5. The structure of lexical polysemy 127 These metaphors are motivated by another feature in (3): ʻthe plant loses its part Yʼ. Now, consider the metaphor Kartočnyj igrok sorval bank ʻThe gambler broke the bankʼ. It is motivated by a third feature of the basic meaning in (3): ʻby an unexpected (≈ sudden) move the gambler comes in possession of the entire prize sum (≈ gets the ability to manipulate it)ʼ. While in the previous cases the object that lost its part, Y, was damaged, in this case it is the opposite: the gambler, breaking the bank, came into possession of it, profiting from it. Veter sorval šljapu s golovy proxožego, Consider another metaphor: lit. ʻThe wind plucked the hat from the passerby’s headʼ. Here, the passerby and the hat are interpreted by the speaker as a whole, and the hat as an integral part of this whole. To blow the hat off the head, some force must be applied; this makes this case similar to (3). As soon as this similarity disappears, the metaphor becomes incorrect, cf.: Veter *sorval šljapu so stola, lit. ʻThe wind plucked the hat from the tableʼ. As can be seen, the various motivating features singled out in the basic meaning of the verb sorval allow the speaker to use this verb to designate new actions that are not covered by its basic meaning. While agreeing with many of Leshchёva’s (2014) propositions, we cannot subscribe to her general thesis about what word meanings are represented in the speaker’s lexicon: “The word is in the mind with its entire system of meanings, customary and potential, ready ʻto surface on the first occasionʼ (Vinogradov 1977: 17)” (p. 194). As we have tried to show, potential meanings are not in the mind of the speaker such that they can ʻsurfaceʼ on the first occasion. Instead, in each use special mechanisms generate them from a word’s basic meaning. The structure of grammatical polysemy is discussed in chap. 4. 3. Supplement. Three contemporary approaches to lexical polysemy In a brief survey of the three approaches used by Apresian (2005, 2010b and 2014), Lakoff (1986 and 1987), and Evans (2009 and 2015) given below, we will focus solely on an analysis of the principles of describing lexical polysemy. This will allow the reader to compare—on this parameter—these frameworks with the approach described above. 3.1. Iurii Apresian and the Moscow Semantic School’s (MSS) framework. There is a tradition in lexicography to give a definition of the basic (≈ most concrete) meaning of the word in a word entry first, followed by definitions of other meanings. In some contemporary dictionaries—for example, Longman 2. A reference-based approach to describing notional words 128 Dictionary (2009)—a quite different, synchronic principle is used to describe lexical polysemy. Basic meaning is not identified as primary, from which other meanings are derived; all meanings of a word are treated as typologically similar. The most frequent meaning is given first, followed by all other meanings in the order of their frequency, from high to low. Similarly, in a textbook by Shaikevich (2005: 141–154) one finds, along with the traditional approach, a modern understanding of basic meaning as the “most frequent” meaning. The MSS has been developing a similar approach, treating all meanings of a word as typologically homogeneous and genetically indistinguishable. The list of a word’s meanings defined in a dictionary starts with the meaning that is most “topical and elaborated in a given language”, cf.: In most Russian explanatory dictionaries, the word CEL’ has, as its first meaning, ‘target, that which a person intends to hit using a weapon’ [basic meaning.—A. K.], while the second meaning is ‘that which a person intends to achieve’ [figurative meaning.—A. K.]. However, […] in modern Russian the second meaning has an indisputable advantage over the first [it is more “topical”.—A. K.], and it is this meaning that is given preference in the AD [Active Dictionary.—A. K.].  24 The main lexeme is followed by other word lexemes [its other meanings.—A. K.] depending on their semantic closeness to the main lexeme and to each other (Apresian 2010b: 72). Cel’ Thus, the dictionary entry for begins with the most elaborated in the [Russian] language (but also the most abstract) meaning: ‘that which a person intends to achieve’. It is followed by the semantically close meaning ‘ target, that which a person intends to hit using a weapon’ and, finally, by a third, semantically remote meaning, ‘front sight—a small protrusion at the end of the barrel of a gun, used for aiming’. Recall that in the framework of the reference-based approach, which leans on the lexicographic tradition, it is the “visual” meaning ʻtargetʼ that is interpreted as basic. And it is this meaning that is the source of two figurative meanings: ʻfront sightʼ (metonymically linked with the target), and ‘that which a person intends to achieve’ (a metaphoric extension of a shooting action to a whole range of resultative actions; for more details, see section 2). Thereby 24 target Similarly, in Longman Dictionary (2009: 1803) in the word entry for the meaning ʻsomething that you practice shooting atʼ, which is basic and should come first, comes only fourth according to its frequency. § 5. The structure of lexical polysemy 129 the phenomenon of lexical polysemy is explained in a natural way, both as a synchronic state of the system of customary meanings of a word, and a process of diachronic change of this system, i.e. the possibility for a word to develop new meanings (to form occasional metaphors and metonymies). Moreover, this explanation makes clear why figurative meanings, which may be very far from the basic meaning, nevertheless fall within its scope—why, for example, the meaning ʻfront sightʼ belongs to the scope of meanings of the cel’ word ʻtargetʼ. It also makes clear the origin of such an abstract meaning as ‘that which a person intends to achieve’. The MSS lexicographic framework does not have answers to such questions. Rather, the MSS approach is oriented only towards representation of the stock” of the current range synchronic state of lexical polysemy—“taking of customary word meanings that exist at the time of their description. It is appropriate to mention here Potebnia’s (1976) words: Usually, we study words in the very form in which they appear in dictionaries. It is tantamount to studying a plant as it appears in a herbarium—that is, not as a real living plant, but as something artificially prepared for the purposes of a study (pp. 465–466). 3.2. George Lakoff’s theory. Unlike the MSS framework, Lakoff’s radial semantic categories framework does not ignore the genetic relationships between word meanings. Lakoff distinguishes a central subcategory (an analogue of basic meaning) from noncentral subcategories derived from it. At the same time, he claims that “there are no general principles that can predict the noncentral cases from the central case” (Lakoff 1987: 379). An example of a radial category is provided by the category Mother: “there is a central subcategory [ʻmotherʼ as such.—A. K.] […]; in addition, there are noncentral extensions […] (adoptive mother, birth mother, foster mother, surrogate mother, etc.). These variants are not generated from the central model by general rules; instead, they are extended by convention and must be learned one by one” (p. 91; original emphasis). To describe the relationships between the central and peripheral subcategotheory of motivation ries, what is needed is, firstly, a (which has not been worked out) and, secondly, “cognitive models of various sorts: propositional, metaphoric, metonymic, and image-schematic” (p. 153). As can be seen, along with the metaphoric and metonymic models, two more general models are involved: propositional and image-schematic. It is these models that complicate the entire picture. § 5. The structure of lexical polysemy 131 LONG, THIN OBJECT SCHEMA; “[t]his image-schema transformation is one of the many kinds of cognitive relationships that can form a basis for the extension of a category” (p. 106). Yet this interpretation doesn’t explain why, for example, the phrase *Uzkaja proseka bežala k reke ʻThe narrow glade ran towards the riverʼ is incorrect— considering that the aforementioned schema covers the glade as well. We see the cause of such incorrectness in that a glade doesn’t have the function of a road, because a glade isn’t made for running on; therefore, the verb bežit doesn’t allow for a metonymic transfer to the glade. This property of metonymy also explains the examples given by Murphy (2011) (see section 1 above), who observed that, for some reason, (1) an author’s name can be used to refer to his work while the publisher’s name cannot been reading Dickens/*Knopf), newspaper (I have and (2) can be used to refer to the company that publishes it, but book cannot. This is because an author and his book form a functional whole, while a publishing company and a book published by it don’t. A book is created by its author, but to the publisher it is an external, alien object, accidental to a certain degree. In the case of a newspaper the situation is different: the publisher of the newspaper is also its creator, therefore the metonymy is correct. 3.3. Vyvyan Evans’ theory. According to Evans (2009; 2015), his theory of lexical concepts and cognitive models (LCCM) can easily explain generation of lexical meanings with the help of such notions as “lexical concept” and “cognitive model”. A lexical concept is associated with a word (a word form) and is a component of linguistic knowledge, the semantic pole of a symbolic unit that encodes a range of various types of schematic linguistic content (Evans 2015: 359). Lexical concepts provide direct access to various cognitive models. A cognitive model is a complex of multimodal knowledge derived from experience, interoceptive systems, and propositional information (achieved via language, cultural learning, and so forth). There are two kinds of cognitive models in the LCCM framework: primary (directly accessed via the lexical concept) and secondary (sub-structures of primary models, indirectly accessed via the lexical concept). The formation of lexical meanings is illustrated by the following examples containing the word France: (3) a. France is a country of outstanding natural beauty. b. France is one of the leading nations in the European Union. c. France beat New Zealand in the 2007 Rugby world cup. 2. A reference-based approach to describing notional words 132 As pointed out by Evans, the meanings of the word France vary across these sentences. In (3a) it is ʻa geographical landmassʼ, in (3b) ʻa political entity, a nation stateʼ, and in (3c) ʻthe 15 players who make up the French Rugby teamʼ. Such variation of meaning is explained, briefly, as follows. The word (form) France is associated with the lexical concept [FRANCE]. This concept is directly related to the first-level cognitive models GEOGRAPHICAL LANDMASS and NATION STATE. In addition, the latter is related to the second-level cognitive models NATIONAL SPORTS and POLITICAL SYSTEM. The differentiated meanings of the word France arise as follows. In example (3a), the lexical concept [FRANCE] is related, due to the context, to the first-level cognitive model GEOGRAPHICAL LANDMASS, which is interpreted as a geographical region. In (3c), again because of the context, another first-level cognitive model is activated, NATION STATE, which is related to the second-level cognitive model NATIONAL SPORTS; and the latter model is interpreted as a team of French players who represent the French nation on the rugby field. Here is one of the concluding propositions made by Evans: One reason for the distinction in literal versus figurative interpretations is a consequence of the cognitive model profile, and a distinction, therefore, in terms of the range of analogue concepts directly and indirectly accessed by the lexical concept. Literal interpretations involve activation of a primary cognitive model […] while figurative interpretations involve activation of secondary cognitive models. […] While there is unlikely to be a neat distinction between primary and secondary cognitive models, and while the distinction is likely to vary from individual to individual, and indeed across discourse communities, there appears to be a sound basis for making a qualitative distinction of this sort (Evans 2015: 366–367). In keeping with this proposition, the meaning of the word-form France in (3a), ʻa geographical landmassʼ, may be interpreted as basic, and the meaning in (3c), ʻthe 15 players who make up the French Rugby teamʼ, as figurative (metonymic). As Evans pointed out, because there isn’t an explicit rule for the isolation of cognitive models and their hierarchical distribution, there is hardly a distinct division between primary and secondary models and, as a result, between literal and figurative word uses. This may easily be shown with some examples. In the sentence France is a state with a democratic form of government the word France is used in its basic meaning. However, a secondary cognitive § 6. Appendix 1 133 model is activated in this sentence: the lexical concept [FRANCE] → NATION STATE → POLITICAL SYSTEM. Thus, in the framework proposed by Evans, the word France—in this case—must have a figurative meaning. § 6. Appendix 1. Excerpts from the email correspondence between A. D. Koshelev and I. A. Mel’čuk (February—March 1995) This Appendix is primarily of historical interest, and there are two reasons for this. First, it shows that the reference-based approach was formed about 20 Moskovskij lingvističeskij al’manax years ago. In 1996, (The Moscow Linguistic Almanac) published my article “Referentsial’nyj podxod k analizu jazykovyx značenij” (“A reference-based approach to analysis of linguistic meanings”, p. 82–185), in which a detailed analysis of many typologically different linguistic meanings was offered, both lexical (concrete and abstract) and grammatical (aspect and tense). Second, the polemics in the correspondence clearly outline the antagonistic positions taken by each side and the impossibility to somehow reconcile one with the other. As may be seen from the contents of сhapter 1, the situation has not changed in the years that followed.  25 Letter № 1 From A. Koshelev to I. Mel’čuk 27 February, 1995 Dear Igor’! [...] I am glad that you asked me to read the volume and give some comments, because I was going to do it anyway. I doubt, however, that my comments will be very much to the point. You know, my interests lie in the field of semantics, which, I imagine, is hardly touched upon in your volume. Speaking of semantics. Here (at a workshop at RGGU [Russian State University for the Humanities]) we’ve been engaged in semantic battles around semantic primes, definitions, and the like. At the latest workshop meeting (where Iu. D. Apresian’s recent 1994 article “On the language of definitions and semantic primes” was discussed) I quoted your book “Opyt Teorii…”, p. 22: “[…] the meaning [now I call it conceptual representation.—I. M.] 25 Other excerpts from this correspondence are given in § 6 of сhapter 2, where the concept živoj—živoe [animate masc.—animate neut.] is analyzed. 2. A reference-based approach to describing notional words 134 of utterances […] lies outside the model ‘Meaning ⇔ Text’ and […] is formed by a separate mechanism (some other model) in the motion from meaning to text. If, for example, we speak about a verbal description of a particular situation, the procedure for correct translation of such a situation (which is the ‘meaning content’, or meaning, of the corresponding texts) must be supported by a special model ‘World ⇔ Meaning’.” And, finally, the crucial point (p. 12): “[…] At the same time, semantic representation must be a natural outcome of the function of the model ‘World ⇔ Meaning’ when the motion is reversed— when the given automaton somehow formalizes particular visual and other auditory ‘impressions’ from the world […]. Therefore, a semantic repre sentation is also determined by the efficacy of the models ‘World ⇔ Meaning’ that use it.” I used these quotations as an additional argument in support of my main thesis: A DEFINITION OF A NOTIONAL WORD MUST ADEQUA TELY DESCRIBE A SET OF ITS DIRECT REFERENTS. [By no means!— I. M.] In other words, a human-like robot (one that has a body and a sensory apparatus similar to those of a human [along with the similar knowledge of referents.—A. K.]) should be capable of correctly relating a word to a fragment of the world (the referent) by using such a definition. An implementation of this thesis is the first step to the model ‘World ⇔ Meaning’ from the side of language (semantics). Arguments in brief: 1. A child learns the semantics of notional words mainly through references (made by those around him along with his own), not through definitions. [Right, but that is HOW he learns, not WHAT he learns!—I. M.] 2. Throughout his adult life his referential capability is sustained and doesnʼt change [I’m totally unconvinced! I don’t think this is the case!—I. M.] (therefore, the referential component of a wordʼs meaning, which plays, in this respect, the role of the meaning nucleus (function), doesnʼt change, either). [It is the function that doesn’t change, because it is ‘recorded’ in the language of ‘human’ interpretations of reality (predetermined by the abilities and needs of the human body and its psyche and thus independent of the knowledge a human has). And this is true not only of concrete lexical items dom stol lošad’ bežat’ čitat’ (derevo ‘tree’, ‘house’, ‘table’, ‘horse’, ‘run’, ‘read’), but of quite abstract items as well (obman ‘deceit’, igra ‘game’, xoxma izučat’ vospityvat’ ‘gag’, ‘study’, ‘bring up’), not to mention grammatical meanings. It is sets of wordsʼ referents that change (are extended), because the correctness of relating a function to a referent depends on knowledge about the referent.—A. K.] § 6. Appendix 1 135 3. It becomes possible to verify, in a relatively objective way, the accuracy of a definition: it must cover all direct referents, and only those referents (whether an object or process belongs to the set of referents of a noun or a verb does not, as a rule, cause much controversy). [Linguists donʼt need such verification—this is not language.—I. M.] [Such verification makes it possible to study linguistic meanings by studying the linguistic categorization of the world these meanings induce.—A. K.] 4. In this regard, well-known definitions are substantially inadequate; there are, however, examples of adequate definitions (they have been discussed in detail at the workshop). 5. It becomes possible (a) to go over and beyond language boundaries when defining the meanings of words [And this is something that should NOT be done (by linguists!).—I. M.], and (b) to understand what meanings should be considered as primes: namely, those that would be elementary to a humanlike robot (and not those that are most frequent in the definitions of linguistic meanings, according to Apresian). Yours, A. K. Letter № 2 From I. Mel’čuk to A. Koshelev 2 March, 1995 Dear Alexey, [...] In your debate with Apresian, I’m 100% on Apresian’s side. It all depends on WHAT semantic primes youʼre looking for. If those are mental or perceptual primes, then, of course, they should be sought in the world around us [in our mental representation of the world, to be more precise.—I. M.]. If, however, they are primes used to define the word MEANINGS (and NOT to define things!), they exist only in language. All smart non-linguists make the mistake of confusing LINGUISTIC meaning with PERCEPTION RESULTS. Funny that you should quote me in the opposite sense! I’m convinced (have always been, and am quite sure now) that the definition of a word should NOT describe a set of its referents! To me, this is the central belief of a linguist. To be able to identify a horse, I must define the horse, and NOT the word horse. I think this is precisely what I tried to say in 1974. I canʼt agree with your propositions—alas!—either globally or minutely. People identify things and actions very precisely, often being unable to name them even in their native tongue. Try and describe a complete stranger to me 2. A reference-based approach to describing notional words 136 so I could identify him! Or express verbally how Georgians differ outwardly from Armenians or Jewish brunettes—even though the difference is quite obvious. No, language is a very singular matter. Yours, I. M. Letter № 3 From A. Koshelev to I. Mel’čuk 3 March, 1995 Dear Igor’, I failed to make myself clear. The point in question is not at all the definition of the perceptual image of a set of referents. And not because it is hard to define verbally, but because IT IS NOT THE IMAGE that is the characteristic feature of a referent. Making reference to something, the speaker ATTRIBUTES a characteristic feature [= generalized interpretation.—A. K.] to an object/ process based on its perceptual properties, observed behavior, and what not. There are Georgians who donʼt look like typical Georgians at all (for example, they may be the spitting images of Armenians or Jews), and there are Jews that look like Arians (fair-haired and blue-eyed). [But then they wonʼt be identified as such by their looks!—I. M.] [Of course! Georgian ≈ ‘1) a typical appearance (prototype) and 2) a specific (to Georgians) system of behavioral stereotypes (function)’.—A. K.] It is not the appearance that defines a ‘set of referents’ for a nation. Let’s consider an example. Take a typical definition of the verb dogonjat’ ‘overtake’: (1) A dogonjaet [‘is overtaking’] B = ‘(1)  A and B are moving in the same direction (2) A is behind B; (3) the distance between A and B is shortening’. Doesnʼt it seem like a good definition? Letʼs begin its referential analysis by identifying the process-referents this definition doesnʼt cover. [I can’t see anything referential here; you keep speaking about meaning!—I. M.] 1. On some sections of the path, A and B may be moving in different, even opposite, directions (parallel streets racing), but the correctness of reference may still hold. Therefore, property (1) is not obligatory. 2. Athletes on the running track at a stadium. B may be ahead of A (locally, visually) but at the same time overtaking him. Property (2) is not obligatory, either (if we assume the usual meaning of the word pozadi ‘behind’). § 6. Appendix 1 137 3. The distance between A and B may grow on some sections of the path without affecting the correctness of reference (take the very same parallel streets racing). Property (3) is not obligatory. Thus, what we have is a definition of the prototypical referent, or prototypical meaning: A and B are moving over a straight section of a highway (and this definition is similar to the definition of a Georgian by their typical appearance); but it is not a definition of the characteristic property of the referent (= description of the meaning sought for). [I absolutely donʼt understand the relevance of this!—I. M.] Counter-example referents prompt the definition in (2) below. Let us begin by illustrating some novel concepts: let A be moving along a sinusoid toward a greater value of X, while B moves along axis X in the same direction. In this case, the sinusoid will be the path of A, axis X the path of B, and the points of intersection of the sinusoid and axis X will constitute the shared (congruent) path of A and B. Thus we have: (2) A dogonjaet [‘is overtaking’] B = ‘1)  A and B are moving along a SHARED PATH; [The concept (meaning) of ‘shared path’ should be defined linguistically!—I. M.] [But of course! Sinusoid and axis X are used for brevity.—A. K.] 2)  at every point on this path B arrives LATER than A; 3) Bʼs delay time is shortening, and it is very likely that A and B will reach some farther point at the same time’. I believe that a 5-year-old child watching cartoons and establishing the dogonjat’ reference of the verb ‘overtake’ makes use of a schema close to (2), and not to (1). [I don’t know! I doubt it.—I. M.] And he ATTRIBUTES this schema to the perceived dynamical image, thus making it a referent. Yours, A. K. Letter № 4 From I. Mel’čuk to A. Koshelev 6 March 1995 Alexey, Your arguments are of much interest to me; but, alas, not all may be made clear in an email. The point is that youʼve formulated a good [and quite linguistic.—I. M.] criticism of definition (1). But: if I can say about something, X is overtaking Y, 2. A reference-based approach to describing notional words 138 I can ALWAYS say that X is behind Y, and that the distance between them is shortening; this is inferred from language and has nothing to do with something in the real world. BESIDES, people have a way to ‘pull’ clear and (relatively) simple linguistic meanings onto wild referents: thatʼs a totally different tune! No, I still insist—just as in 1974—that referents should not be taken into account (not generally, of course, but in LINGUISTIC semantics; then rules for matching meanings with referents will be needed, and this is where your considerations would become applicable). [The following are what I would like to be distinguished. There are referents in the world: external reality; there are mental representations of these referents: conceptual representations; and there are linguistic representations of conceptual representations: linguistic meanings. The roles of referents and knowledge about referents may be different, depending on lexemes. For example, interpretations of lošad’ ‘horse’ or koška ‘cat’ are based on knowledge about the referents, whereas interpretation of podtalkivat’ (kogo k čemu) ‘urge (sb. to do sth.)’ isnʼt based on it at all!—I. M.] [I would define the external world (referents) as a two-layered structure: (1) a layer of perceptual (available for perception) images, and (2) a layer of interpretations of such images—hypotheses about the images based on one’s knowledge about them. Linguistic meaning has a similarly binary structure: prototype (generalized image)—function (generalized interpretation). Refer ence is a correct mapping of the function onto a suitable INTERPRETA T ION of the referent.—A. K.] Yours, I. M. Letter № 5 From A. Koshelev to I. Mel’čuk 9 March, 1995 Igor’, [...] Turning to the topic of semantics, I’d like to emphasize that, knowing how busy you must be, I donʼt NECESSARILY expect answers to the questions raised. Let me give a response to your answer first. I agree that “if I can say about something, X is overtaking Y, I can ALWAYS say that X is behind Y, and that the distance between them is shortening; this is inferred from language and has nothing to do with something in the real world”. And hereʼs why: § 6. Appendix 1 139 The meaning of a notional word is made up of two components: the FUNCTIONAL component that defines the set of ALL direct referents (meaning (2) A dogonjaet ‘is overtaking’ B), and the PROTOTYPICAL component that defines the image of a typical referent (meaning (1)). Prototypical meaning (1) is the result of numerous performed or observed direct references (mappings of meaning (2) onto external world) crystallized in the linguistic consciousness of an individual. Thus, (1) is the superficial and CONSCIOUS component of meaning. Therefore, in a sentence (in the absence of a referent [to the hearer, who simply canʼt see this referent.—A. K.]), prototypical component (1) represents typified external reality (the typical referent). [Alas, I donʼt understand any of this.—I. M.] As for your second proposition, “people have a way to ‘pull’ clear and (relatively) simple linguistic meanings onto wild referents: thatʼs a totally different tune!”, it isn’t quite clear to me. Prototypical meaning (1) cannot be “pulled” onto “wild” referents [Hwhiiiyy cain’t et? ‘The distance is shortening, but in just such a wild sense’.—I. M.], while functional meaning (2) covers all of them. It seems reasonable to assume that this component belongs to the realm of the subconscious knowledge of a word’s meaning and reveals itself only when “wild” referents are analyzed. [I don’t understand.—I. M.] Yours, A. K. Letter № 6 From I. Mel’čuk to A. Koshelev 17 March, 1995 Alexey, [...] I have a difficulty understanding you but am convinced that linguistic meaning should be defined WITHOUT any recourse to referents (although it is, of course, formed [in the speakerʼs brain.—I. M.] on the basis of referents both direct and indirect, just as you claim). What you call “the subconscious knowledge of a word’s meaning” is, to me, NOT a word’s meaning (even a subconscious one); rather, it is a mechanism for mapping [the meaning of a word onto a concept, and then.—I. M.] onto a referent. Yours, I. M. 2. A reference-based approach to describing notional words 140 Letter № 7 From A. Koshelev to I. Mel’čuk 19 March, 1995 Igor’, [...] About meanings. I agree that the referential component of a wordʼs meaning should not be viewed as PURELY linguistic. It may be viewed as referential meaning, let us say. [Concept!—I. M.] But its direct relationship to language is beyond any doubt to me. [Surely concepts relate to language (directly!).—I. M.] [In the chain, Word—Meaning—Concept—Referent that you suggest, the status of Concept (‘mental representation of referents’) is not clear to me. Why, not being a component of meaning, does it directly relate to language? If it is directly related to Meaning, where is the border between them (the boundary defined by the function of Concept)?—A. K.] And hereʼs why. Along with the function of communication, language also has another function which is no less important: to describe the perceived external world (a witnessʼs tale, a reporter’s account, and the like); this function cannot be performed without referential meanings [True!—I. M.]. It is these meanings (and not PURELY linguistic meanings) that lie at the basis of linguistic categorization of the world [True!—I. M.]. Hence the inevitability of going over and beyond the boundaries of language IN A STUDY OF LANGUAGE (analyses of referents, human perception specifics, etc.) [But this is not true. In such a case the object of study is not language, but linguistic behavior.—I. M.] Letter № 8 From I. Mel’čuk to A. Koshelev 20 March, 1995 Alexey, [...] I absolutely agree that language doesnʼt function separately from the referents. I also agree that the function of description/analysis of the world is no less important for language than the function of communication. I would only like to SEPARATE (into levels) meanings and their relationship with words on the one side [This, and only this, is LANGUAGE.—I. M.], and referents [Via concepts; this is not language.—I. M.] on the other [Of course, it doesnʼt really matter WHAT is included WHERE and WHAT is named HOW. But the tech- § 7. Appendix 2 141 niques of description in language (discrete, like algebra) and outside of language (continuous, like analysis) are QUITE DIFFERENT.—I. M.]. [I call the Function “meaning”, because, just like the Prototype(s), it is DIRECTLY connected with Word: Function—Word—Prototype(s). With its functional meaning Word is oriented to the external world (referential function), while with its prototypical meanings it is oriented to text (communicative function). Moreover, Function, in fact, made concrete in Prototypes, plays the role of MEANING INVARIANT.—A. K.] It would be fascinating to talk to each other, wouldnʼt it? It seems the time has come for me to start planning a trip to Moscow again. Yours, I. M. § 7. Appendix 2. An infant’s early acquisition of the laws of nature 1. An infant’s view of the physical world Early on, infants begin to understand and use verbs of physical action and udarit’ tolknut’ spatial orientation (posture verbs) such as ‘hit-PF’, ‘push-PF’, padat’ brat’ stojat’ viset’ bežat’ ‘fall-IMP’, ‘take-IMP’, ‘stand’, ‘hang’, ‘run’, etc. Meanwhile, the meanings of these verbs contain non-trivial conceptual components. For example, the meaning of the verb udarit’ ‘hit-PF’ contains invisible force-dynamic characteristics, such as ‘one object transfers the force of motion to another object’, and a causal relationship THEREFORE that connects the effects of these invisible forces: ‘THEREFORE the other object suffers a shock’. Similarly, to understand the verbs padat’ ‘fall-IMP’ or stojat’ ‘stand’, infants must have some ideas of the invisible force of gravity, the support a motionless object must have, the locus of an objectʼs center of gravity, conditions under which an object falls, etc. It is only natural to ask: how and when does an infant acquire such knowledge? The data from numerous experiments show that these and many other conceptions of the laws of the physical world are formed by the age of 13–14 months—much earlier than one would think. Thanks to such formed conceptions, infants can already at an early age understand force-dynamic interactions and cause-effect relationships between objects in the surrounding visible world. It must be emphasized that the fact that infants can grasp such interac- 2. A reference-based approach to describing notional words 142 tions and relationships between objects is determined exclusively by infants’ cognitive development. Acquisition of the lexicon of the child’s mother tongue does not substantially affect this process. Supportive evidence can be found in how infants view the world in general. As recent cognitive research has shown, already during the first year an infant acquires diverse and detailed knowledge about objects, living beings, and interactions between them (Spelke et al. 1992 and 1995; Baillargeon 1994, 1999 and 2002; Carey 2009; Murphy 2002; Sergienko 2006; Subbotskii 2007, inter alia). Thus, at 6 months, infants begin to distinguish between inanimate objects and living beings. They already understand that, unlike humans and animals, inanimate objects are incapable of self-propulsion (for example, they cannot begin to move without some external influence from another object, or come to an abrupt stop in the absence of an obstacle), nor can they move purposefully: circumvent an obstacle, halt before it and move in reverse, etc. For example, in an experiment by the psychologist Elizabeth Spelke, a baby is shown a ball rolling behind a screen and another ball emerging from the other side, over and over again to the point of boredom. If the screen is removed and the infant sees the expected hidden event, one ball hitting the other and launching it on its way, the baby’s interest is only momentarily revived; presumably this is what the baby had been imagining all along. But if the screen is removed and the baby sees the magical event of one object stopping dead in its tracks without reaching the second ball, and the second ball taking off mysteriously on its own, the baby stares for much longer. Crucially, infants expect inanimate balls and animate people  26 to move according to different laws. In another scenario, people, not balls, disappeared and appeared from behind the screen. After the screen was removed, the infants showed little surprise when they saw one person stop short and the other up and move; they were more surprised by a collision (Pinker 1994: 423ff.). Quite remarkably, infants begin to acquire such knowledge very early, not later than 2 months of age. From that moment on, their knowledge expands and grows in depth extremely quickly, following a number of steps in infants’ cognitive development. It involves, among other things, the dynamics of acquisition of the laws of the physical world and, in particular, the nature of 26 See (Spelke et al. 1992 and 1995). On how infants, at 12 months, interpret perceived motion as purposeful and optimal, see (Csibra et al. 2003; Gergely et al. 1995).— Author’s note. 2. A reference-based approach to describing notional words 144 In a series of experiments conducted by Leslie (1984), infants at 2.5 to 3 months of age watched as one object (a red block) moved up to another object (a green block) and stopped, while the green block slowly came into motion (“direct impact”); this did not cause any surprise in infants (they ceased to pay attention to the event rather quickly). However, if the red block stopped just a little short of the green block and the latter, nevertheless, started to move (“impact without collision”), infants were surprised by such a development. After 3 months of age, infants were also surprised to see one object (a red block) come up to another object (a green block) and stop, while the latter started to move not at once, but with a delay (“deferred impact”). Infants showed the same reaction when they watched a similar motion of squares of light instead of blocks (which is understandable because, at this age, infants do not make much of the shape, size, and color of the object observed). These and many other similar experiments allowed the researchers to conclude that a necessary condition for perceiving a direct impact as the bearer of causal relationship between the subsequent motions of the blocks is the spatio-temporal continuity of their motions. It has also been shown, in another series of experiments, that at 7 months infants distinguish a ‘direct impact’ event from the continuous motion of a block that changes its color from red to green in the course of motion. Taken perceive two subsetogether, these data indicate that, from 7 months, infants quent motions of two colliding discrete objects as ‘direct impact’, as i.e. two motion events that are in a causal relationship to each other (Leslie 1984; Subbotskii 2007: 177–178). Note. The data described above allow us to conclude that infants develop an ability to spontaneously make causal “subconscious inferences” as  27 follows: if infants perceive two visual images whose subsequent motions satisfy the condition for “direct impact” (spatio-temporal continuity of “motion transfer”), they automatically attribute a causal relationship to this motion. Importantly, rational reasoning that may show this subconscious inference to be false in a particular case cannot cancel or repudiate this relationship. Michotte (1962), in his pioneer study (based on similar experiments) of how adults perceive direct impact and interpret the inherent causal relationship, produced screen projections of direct impact, not of colored blocks, but of squares of light. The adults in the experiment, who understood very well the true mechanism of the squares’ motion, could not 27 This is the term used by H. Helmholtz; for its explanation see, for example, (Sechenov 1952: 356ff). § 7. Appendix 2 145 get rid of the impression that one square gave a “slight push” to the other. The impression that there was a causal relationship between these motions was, apparently, false, yet it persisted (for a more detailed discussion of this property of subconscious inferences see Koshelev (2015a: 226–231)). Note also that, according to Leslie (1984), who elaborates on Michotte’s (1962) ideas (see also Carey 2009: 217–221), and some other researchers (see references in Subbotskii (2007)), the direct mechanical impact discussed above encapsulates a causal relationship and is, to the child, the source of a more general understanding of the cause-effect relationships between objects. Cf.: … at 2 to 3 years of age, children recognize, as a cause of a phenomenon, only such an event which is materially connected to the phenomenon one way or another (that is, it sort of “passes on” its “energy”); thus, if, in the presence of two working fans, a burning candle goes out, to identify the cause children point to the fan blowing air in the direction of the candle. In other words, this module [the direct impact.—A. K.] brings the child to the idea of the necessity of spatio-temporal continuity between cause and effect (Subbotskii 2007: 180) For a detailed discussion of this view see Carey (2009: 221–246). Notice the remarkable subtlety with which infants classified different types of interactions between blocks. Not only did they single out direct impacts as a special, central class, they also identified three other independent classes of “impacts”: “impact without collision”, “deferred impact” (infants did distinguish between them), and “deferred impact without collision”, which integrated the two deviations from “direct impact”. If infants saw that the red block stopped short of the green block (first deviation from direct impact) and the latter started to move only after several moments have passed (second deviation), their surprise (an eagerness to see such an event again) was greater than in the case of only a single deviation (Subbotskii 2007: 177–179; Leslie 1982 and 1984). Chapter 3 Basic-level concepts as the neurobiological codes for memory § 1. Concrete concepts and motor concepts 1. Introduction. Two interpretations of basic-level concrete concepts In the area of cognitive research associated with the names of R. Brown, E. Rosch, C. Mervis, B. Berlin, and G. Lakoff, among others, an important discovery has been made concerning how children acquire generic categories. By making use of contrasting categories, it was shown that a child acquires generic categories first, and that these categories are maximally distinct (cf. Lakoff 1987: 58–67). Therefore concepts—cognitive structures that shape generic categories—are called basic, or basic-level, concepts. As has been found by Rosch and her colleagues (Rosch et al. 1976), the basic level is, among other things, the highest level at which category members have similarly perceived overall shapes and at which a person uses similar motor actions for interacting with category members. It is also the first level named and understood by children (cf. Lakoff 1987: 46). Leaning on previous research by his predecessors, Lakoff (1987: 36) identifies two interconnected features that form a basic concept: overall shape and typical motor interactions, “the possibilities for which are also determined by overall shape”. Thus, a typical interaction with chairs is to sit on them, with flowers to smell them, with cats to pet them, with a ball to play with it in a certain way. In other words, according to Lakoff, a basic concept is constituted as follows: (1) Concept = overall Shape + physical Interaction. Note. A similar schema is also given in Gallese and Lakoff (2005: 466). They extend the use of this schema, as described in the following excerpt: § 1. Concrete concepts and motor concepts 147 We believe that the same basic structures—schemas structuring sensory-motor parameterisations—can be used to characterise all concrete concepts. Take, for example, the basic-level concepts […] chair, car, etc. As we saw, basic-level concepts are defined by the convergence of (1) gestalt object perception (observed or imaged) and (2) motor programmes that define the prototypical interaction with the object (again, performed or imaged). Thus, a chair looks a certain way (and we can imagine how it looks) and it is used for sitting in a certain position (and we can imagine sitting that way) (Gallese and Lakoff 2005: 469). According to this definition, the basic-level concept stul ‘chair’ is constituted as follows: (2) concept stul (basic meaning of the word stul ‘chair’) = Shape & Human interaction with it Video clip Video clip & As can be seen, schemas (1) and (2) provide visual identification of the basic-level concept; both of its components are easily stored in a native speaker’s memory. According to another, similar view (Nelson 1973; Mervis 1987; Rakison 2000), basic-level concepts have binary structure: (3) Shape — Function. Here, the overall shape of an object is matched with its general function rather than interaction. 2. Functional schema of basic-level concepts It seems that the formula for a basic concept given in (2) should include two more functional features. An object’s shape must be complemented by its function (the pair “Shape → Function”; see the definition of the Function on pp. 41 and 43), and visually perceived human interaction must be complemented by the psychophysical state of the interacting person (the pair “human 3. Basic-level concepts as the neurobiological codes for memory 148 Interaction ← psychophysical State”; for a definition of psychophysical state see § 2 section 2). Let us use the symbol “&” to designate the relationship of association between these two pairs. Then we get the following: Object Interaction (4) Basic-level concept = (Shape → Function) & typical (human Interaction → psychophysical State). Or, in a shorter version: (4′) Basic-level concept = Object (Shape → Function) & Typical motor concept. arrow (→) designates a relation of interpretation: the visual property (Shape) is ascribed a functional property (the initial job done by the shape), which is a feature of a different nature. The word ‘typical’ indicates that the associated interaction is probable and potential, but not actual; it will often be omitted in the subsequent discussion. The above definition may be clarified using the basic-level concept stul (‘chair’) as an example. The general function of a chair (the function of its shape) is to be sat on by a person in a certain posture: the person leans on the back of the chair, his buttocks rest on the seat of the chair, and his feet are supported by the floor. The state of a person sitting on a chair (‘half-steady position of the body’) is a range of his typical sensations caused by his interaction with the chair. Clearly, the function of a chair and the state of a person sitting on this chair are two interrelated but completely different features. The first function characterizes the chair, while the second function characterizes the person. As a result, we come to the following: stul (5) concept stul (basic meaning of the word ‘chair’) = (Shape → Function) & (Interaction Allows A person’s back a person & and buttocks to sit are supported like this: by the chair: → → Psychophysical state) 1) With the objective to do some work at a desk/table, a person → 2) takes this body posture and 3) is in a half-steady (semi-relaxed) state § 1. Concrete concepts and motor concepts 149 Some clarification is in order here. The definition features the same “picture” twice. However, in the first case the picture explains the function of a chair—to allow a person to take a certain posture, while in the second case it explains the psychophysical state of a person taking such a posture. In the first component of the concept invisible properties—durability, stability, ability to support a human body—are attributed (→) to a visually perceived shape. If, for example, we take a chair made of papier-mâché, we will see that these properties do not always come with the shape of chair. In the second component, it is the human activity of “sitting on a chair” to which a ‘semi-relaxed state’ is attributed, caused by the interaction. That such an interpretation is not obligatory may be illustrated by the following circus act: a clown is sitting on a chair in the middle of the arena in the posture of Rodin’s “Thinker”. The clown’s mate sneaks up from behind and pulls the chair from under him. But the clown continues to “sit” in the Thinker’s posture (to the delight of the laughing audience). The sitting posture is now practically the same (except there is no support from the back of the chair) but is not associated with any, even partial, relaxation of the clown’s body. It is crucial that the component ‘psychophysical state’ be included in the definition of basic concept. As will be shown in § 2 section 2, it is precisely the different psychophysical states of a sitting person that distinguish the categories ‘Stul’ja’ (‘Chairs’), ‘Kresla’ (‘Armchairs’), and ‘Taburety’ (‘Stools’) as strictly separate from one another. At the same time, neither the difference in shape between chairs and armchairs nor the different sitting postures associated with them allow for a strict differentiation. 3. The function of an object and a linguistic explanation of this function The function of an object (the initial job it does) is its most important taxonomic characteristic. It is closely connected with the object’s shape but dominates it in the following sense: the shape of objects belonging to the same category may vary to a certain extent, while their function always remains the same. That an object has a certain function may easily be determined, since it is the function that defines the parts into which the object can be divided. For example, the function of a chair—‘make it possible for a person to sit in a certain manner’—defines its division into three parts: the back of the chair, with the particular function ‘support for the back of the sitting person’, the seat, with the function ‘support for the person’s buttocks’, and the legs of the chair, whose function is to ensure ‘the position of a person’s legs and the sup- 3. Basic-level concepts as the neurobiological codes for memory 150 The sum of these three particular functions port of the floor for the feet’.  1 constitutes the general function of a chair. An armchair features additional parts: armrests as supports that provide for a much steadier state of the body of a sitting person. Because of an object’s general function, a native speaker always knows which physical part of the object is its functional part proper and which part is not. For example, the function of a knife ‘≈ to cut something holding a knife in a hand’ determines its division into two constitutive parts, the blade (with the particular function ‘to cut’) and the handle (with the function ‘for holding in a hand’). These particular functions together define the general function of a knife. Importantly, an object’s parsing structure is verified linguistically. In Russian, the relationship ‘Y is a part of X’ is usually expressed by a noun Y Х-а phrase (‘Y of X’), which consists of two nouns; one of these nouns is in the Genitive case, e.g.: lezvie (Y) noža (X-a) ‘the blade of a knife-Gen’, ručka kastrjuli spinka kresla (Y) (X-a) ‘the handle of a pan-Gen’, (Y) (X-a) ‘the back of an armchair-Gen’, etc. (for more details see chap. 4, § 1). Notably, noun Y functionrefers not to just any physical part of object X, but specifically to its ally relevant part, that is, the part whose function directly contributes to the object’s general function. Therefore, expressions such as lezvie noža ‘the blade of a knife’, ručka noža ‘the handle of a knife’ are correct. However, a corkscrew or fork in a Swiss Army knife, being its physical parts, are not its functional parts as they don’t contribute to the knife’s general function (‘to cut’). Consequently, expressions such as *štopor noža ‘the corkscrew of a knife’, *vilka noža ‘the vilka u fork of a knife’ are incorrect (in Russian the correct phrasing would be noža ‘the knife’s fork’). Similarly, the general function of a door is ‘to open and close the doorway’, therefore a handle is its functional part. But an eyehole in a door does not contribute to this function (the function of an eyehole is ‘to show who is behind the door’); therefore, it is quite correct to say ručka dveri ‘the handle of a door’, but one cannot say *glazok dveri ‘the peephole of a door’ (it should be dvernoj glazok ‘the door-ADJ peephole’). The correctness of the expressions spinka/siden’e/nožki stula (‘the back/ seat/legs of a chair’) is explained in a similar way: each of these parts contrib1 One of the most felicitous definitions of an object’s shape has been offered by Pereira and Smith (2009: 68): “objects are perceptually parsed, represented, and stored as configurations of geometric volumes (‘geons’). Within this account, object shape is defined by two to four geometric volumes in the proper spatial arrangement, an idea supported by the fact that adults need only two to four major parts to recognize instances of common categories […]”. § 1. Concrete concepts and motor concepts 151 utes to the general function of a chair. But imagine that a head-rest has been attached to the back of a chair, and a foot-rest to its legs. The incorrectness of the expressions *podgolovnik stula ‘the head-rest of a chair’ and *podnožnik stula ‘the foot-rest of a chair’ shows that, becoming physical parts of a chair, they don’t become its functional parts (they don’t contribute to the general function of chair but rather limit the required degree of mobility of the sitting person). For an armchair, however, these parts are natural and functionally significant because they increase the steadiness of the person sitting in it. Therefore, the podgolovnik kresla similar expressions ‘the head-rest of an armchair’ and podnožnik kresla ‘the foot-rest of an armchair’ are quite correct. These, as well as other examples (see below), show that a native speaker is well aware of the functions of objects. Only by taking these functions into account can he easily “compute” the functional parts of objects. This enables him to identify the correct and incorrect uses of expressions that designate parts of objects. 4. A linguistic explanation of the concept functional schema As shown in (4) and (5), a basic concept includes four interrelated components: [Shape → Function] & [Interaction → State]. There is evidence of various kinds that each of these components is differentiated and possesses an independent status. As an illustration, consider Schema (5). 1. Shape. In Rene Magritte’s painting “The Legend of the Centuries” (see Fig. 5 in chap. 2, § 1, section 2.3), a gigantic stone object is depicted that resembles a chair by shape. This resemblance is quite sufficient for naming the object metaphorically: This is a chair. 2. Function. Consider a chair (Fig. 1) whose shape is not typical (it does not This is a chair fit the prototype). If it were made of papier-mâché, the phrase would also be metaphorical, but for a different reason: this object would not possess the function of a chair as it would be crumpled should someone sit on it. Fig. 1 3. Interaction. One could say, metaphorically, about the clown who continues to keep the posture of the thinker after the chair has been pulled away 3. Basic-level concepts as the neurobiological codes for memory 152 from under him, that he is sitting. The action ‘to sit on a chair’ (its manner, outward appearance) is, therefore, quite independent and separate from the other components of the schema in (5). It is this action that motivates the metaphor The clown is sitting, even though the clown’s state is obviously not semi-relaxed.  2 4. State. One could say about a stack of bricks propped against a wall in This is a chair. such a way that one could sit on it, leaning on the wall: This metaphor, by contrast with the previous ones, is motivated solely by the ability of the construction described to produce the required (semi-relaxed) state in a person sitting on it. The concept schema introduced in (4) and (5) defines the algorithm for recognizing (identifying) an object perceived by a person as a chair. The object’s shape prompts an initial hypothesis: this might be a chair. Then the possibility of a specific interaction with this shape (‘to sit’) is tested. If there is such a possibility, the person performs such an action in his mind to check if the object can keep him in a sitting position (the function of a chair) and whether the person sitting on it finds himself in the required psychophysical state. If that is the case, then the perceived object is a chair. If, however, a similar but different psychophysical state is obtained—‘almost completely relaxed position of the body’—then it is an armchair. The proposed four-component schema of a basic-level concept was recently corroborated by unexpected, even though circumstantial, evidence: it seems that similar schemas form the basis for object taxonomies of animals, briefly discussed in the following section. 5. Neurobiological grounds for the basic concept schema Joe Tsien’s (2008) innovative research on human memory studies gives reasons to believe that the components of the concept stul (‘chair’) in (5), which define the category “Stul’ja” (‘Chairs’), are directly encoded in human memory as an ensemble of specific groups of neurons (neural cliques). Neural 2 Consider an example of a child’s metaphor motivated solely by the form of interaction with an object. Jacqueline (aged 2 years and 3 months), holding a comb over her head, said: This is an umbrella (Eliseeva 2008: 86). The correctness of such an occasional use of the word umbrella is due exclusively to the fact that the interaction with the comb is the kind of interaction typical for an umbrella. Had the girl held the comb in a different manner (not in the manner an umbrella is held), the phrase would have been incorrect. Therefore, in a child’s mental schema of the meaning of the word umbrella, the component ‘interaction with an umbrella’ is featured as quite independent. § 1. Concrete concepts and motor concepts 155 The above is also true for motor concepts: čelovek sidit (“a person is sitting”) is a basic-level motor concept. Its prototypical interaction—“a relatively vertical position of a person’s body whose buttocks are in contact with a horizontal seat”—is the most generic, as is the function “a person’s body has a main support at the bottom part and is in a partially steady position” (cf. the basic meaning of sidet’ ‘sit’ in chap. 2, § 3, section 2.3). Similarly, the motor concepts čelovek stoit/ležit (‘a person is standing/lying’), čelovek idёt/bežit (‘a person is walking/running’) are also basic-level concepts. The concepts čelovek sidit na stule / v kresle / na polu (‘a person is sitting on a chair / in an armchair / on the floor’) are subordinate. Consider, as an example, the motor concept given earlier in (5): (7) Motor concept čelovek sidit na stule (the meaning of the phrase Čelovek sidit na stule ‘A person is sitting on the chair’) = Human Interaction → Psychophysical State A person’s back and 1) (goal) buttocks are in contact With the goal to do some work at a with the chair: desk/table, a person → 2) (dynamics) takes this body posture and is in a halfsteady (semi-relaxed) state It must be noted that a human motor concept possesses an independent status that is not determined by the object with which a person interacts. Thus, in the human basic-level concept čelovek sidit both the shape and the state are strictly the features of a human interaction; they do not include the features of the object sat on. The same is true for the concept in (7). Here the left-side component—physical interaction—is, rather, of an illustrative, explanatory nature because the posture of the sitting person registered by the receptors in the human body (proprioceptors) is stored in the psychophysical state (in the coded memory of a typified interaction “sit on a chair”, cf. §1 section 2). The chair itself is not part of this state. Now consider the basic-level concept čelovek idёt ‘a person is walking’. The object with which a person interacts is a hard level surface (a road, a trail). Taking into account the meaning of the phrase Čelovek idёt (cf. chap. 2, § 4, section 1.4), we get the following: 3. Basic-level concepts as the neurobiological codes for memory 156 (8) Motor concept čelovek idёt ‘a person is walking’— the meaning of the phrase Čelovek idёt = Human Interaction → Psychophysical State A person is moving Person A, down the path, but not 1) (goal) pursuing his spatial goal to get to a quickly different location, 2) (dynamics) alternately rests each foot on and pushes → off from the surface, each time moving his body weight from one foot to the other, not at any time losing the support of the surface; at any given moment the position of A is unsteady Here the psychophysical state (memory code) of the walking person is likewise an internal feature of the person’s psyche which does not include the features of the surface with which the person interacts. The kinematics of walking is also stored in this memory code (in the proprioceptive data). As for the left-side component (visual prototype), it is included in the definition of the concept (and basic meaning) to clearly show this component of the state. Along with the concepts that describe human actions there are motor concepts that describe actions of objects: kamen’ letit (‘the stone is flying’), voda tečët (‘the water is flowing’), etc. They possess a similar structure except that instead of a psychophysical state an affective constituent (a person’s interest in the action, expected outcome (negative or positive) from it, etc.) is ascribed to the shape of the action. 7. Concrete concepts tropinka ‘footpath’, dorožka ‘track’, doroga ‘road’ Following the approach to the definitions of chair and armchair, we can now define some types of surface over which motion takes place—a footpath, track, or road—guided by their functions and the psychophysical state of the walking/running/moving persons, respectively. We will begin with the concept tropinka ‘footpath’ and define it in a way similar to that in which we defined the concept stul ‘chair’. While a chair is a man-made object used by a person for sitting on, tropinka ‘footpath’ is an entity that comes into being spontaneously and is used for the passage of a single person. § 1. Concrete concepts and motor concepts 157 In explanatory dictionaries the word tropinka ‘footpath’ is defined with the dorožka help of the word ‘track’, cf.: Tropinka [‘footpath’] 1. Uzkaja pešexodnaja dorožka, protoptannaja lud’mi ili životnymi v lesu, v pole, po snegu i t. p. [‘A narrow pedestrian track beaten by people or animals in a forest, field, snow, etc.’] (Ushakov, IV: col. 808). This definition is in need of specification. First, unlike a track, a footpath is made exclusively by humans; compare the correctness and incorrectness of the expressions zverinaja tropa ‘wild animal track’ and *zverinaja tropinka ‘wild animal footpath’, respectively. Second, in its basic meaning the word tropinka ‘footpath’ refers to a strip of ground beaten in a forest, field or highland, but not in snow—consider the questionability of the phrase tropinka v snežnom pole ? ‘a footpath in the snow-covered field’. Third, a footpath is a very narrow track, just dorožka tropinka wide enough for one person. And, finally, the words ‘track’ and ‘footpath’ are semantically very close and one feels reluctant to explain one word with the help of the other. The following definition will be accepted here: (9) Tropinka [‘footpath’] 1 a) is a narrow strip of beaten ground used by people in an area inconvenient for walking; b) it is hard and even, convenient for walking; c) it is used by people for walking alone or in single file. As can be seen, the entity ‘tropinka’ [footpath] is defined through (a) its prototypical shape (“narrow strip of ground surface”) and (b) its function (“hard and plane, convenient for walking”), as well as (c) the motor concept čelovek idët ‘a person is walking’ (“used by people for walking alone or in single file”). This is quite consonant with the definition of a concrete concept presented in the schemas in (4) and (5). As a result, only one potential motor concept is associated (&) with the pair “Prototype → Function” as one of an object’s components: (10) The concrete concept tropinka ‘footpath’ (basic meaning of the tropinka) word = (Prototype → Function) & Motor concept Narrow strip of ground surface, hard → and plane enough for & čelovek idёt (8) a person to walk conveniently 3. Basic-level concepts as the neurobiological codes for memory 158 The function of the footpath defines its parts: the middle, which can be walked on, and the edges, which separate it from the adjoining surface (which is unfit for walking); hence the correctness of the expressions seredina/kraja tropinki ‘the middle/edges of the footpath’. A rut or a bump on a footpath, for instance, are not parts of it as they do not contribute to its function, hence the incorrectness of the expressions *rytvina/*bugor tropinki ‘the rut/bump of a footpath’. Dorožka ‘track’, which is already an artifact, has a man-made hard surface and is usually designed not just for walking, but also for running and cycling. Therefore, the concrete component of the concept is associated with a set of three potential motor concepts: čelovek idёt (‘a person is walking), čelovek bežit (‘a person is running’), and čelovek edet na velosipede (‘a person is cycling’). As a result, we get the following: (11) The concrete concept dorožka ‘track’ (basic meaning of the word dorožka) = (Prototype → Function) & A set of motor concepts Narrow strip of čelovek idёt, ground with a hard čelovek bežit, → even surface made for & čelovek edet convenient travel for na velosipede one or two persons A still greater number of motor concepts is associated with the concept doroga ‘road’. On a road people can walk, run, drive (a car, tractor, etc.) and, at the same time, transport something in either direction. Note that a crosswalk marked by white stripes is not part of the road (it impedes traffic), while traffic lane markings (which facilitate traffic) are, cf.: *perexod dorogi ‘crosswalk of a road’ vs. razmetka dorogi, lit. ‘(lane) markings of the oad’. In our further discussion the second component of the concrete concept shown in (4’)—the typical motor concept (or a set of such concepts)—will often be omitted for the sake of simplicity, and a simplified formula will be given: (12) Basic-level concept = Shape of object → its Function. § 1. Concrete concepts and motor concepts 159 8. Concepts ozero ‘lake’ and reka ‘river’ A similar approach will now be used with regard to a lake and a river. Their prototypes are evident, and their general function may be formulated as follows: ‘a continuously sustained large mass of fresh water which naturally appears on the surface of the ground, completely covering the area’. The distinctive features are, briefly, these: a river is ‘a mass of fresh water moving continuously and unidirectionally’, and a lake is ‘a mass of still fresh water’. This allows for the following representation of the concept ozero ‘lake’: (13) Concept lake = Prototype (Shape) of a lake → its Function Constantly keeps from spreading a large → mass of still fresh water A few words must be said about the second component of the concept ozero, omitted in (13) but associated with the first component, namely, a set of potential motor concepts. Among other things, they determine the human scale of a lake relative to a sea on the one hand, and to a pool of water on the other. A lake is relatively small and can be swept by a single glance together with its shores. It is a water barrier for passage to the other shore, and it is deep enough for a person to drown while trying to swim it. Its width is relatively small, such that a person can swim across it at some points or easily reach the opposite shore in a boat. One can swim in a lake, relax on its shore, drink its water, go fishing, wash oneself, wash and rinse linen, etc. By contrast, a sea is so large that a person cannot take it in at a glance, see the opposite shore, or reach the opposite shore by swimming or in a boat. A person sees only a part of the seashore on which he is standing, and the adjoining part of the sea. Drawing an analogy, one could compare the difference between the motor concepts plyt’ po ozeru ‘swim/sail over a lake’ and plyt’ po morju ‘swim/sail over a sea’ with the difference between the motor concepts sidet’ v kresle ‘to sit in an armchair’ and sidet’ na stule ‘to sit on a chair’. Sea water is salty and cannot be used to quench thirst, etc. A pool is a small, temporarily appearing mass of fresh still water. It can be stepped over or crossed at a few steps. One cannot swim in it, etc. § 1. Concrete concepts and motor concepts 161 9. The concept derevo ‘tree’ Heretofore we have discussed only categories of inanimate objects. In the case does not change. of an animate object the concept schema given in (12) Consider the concept derevo ‘tree’ as an example; this is what it looks like as a first approximation (the prototype of a tree is shown as extended over a time period): (14) Concept derevo = Prototype of a tree → its Function (‘grows from the ground by itself, blossoms, and bears fruit’) Bears fruit periodically, i.e. grows fruits and seeds. This is done as follows: from the roots in the ground, which provide nourishment, a hard thick sprout (trunk) grows; from the trunk hard thinner The trunk of the tree stands vertically and → sprouts (branches) grow in motionlessly on the ground; the roots are in different directions; the ground. The tree gradually grows in periodically (in spring) buds size, its trunk gets taller and thicker, and appear on the branches; from the branches get longer and thicker. Every the buds come blossoms, and spring new branches shoot forth from the from the blossoms come fruit trunk and the older branches; blossoms (seeds) which ripen (in appear on the old branches to be followed autumn) and fall from the tree by fruit, which falls to the ground in autumn The prototype of a tree is shown as three chronologically ordered “profiles” of a tree: a sapling, a young tree, and a grown tree. The text below these profiles describes the typical visual characteristics of growing, blossoming, and fruit bearing, while the function explains these observable characteristics, i.e. our intuition about the tree as a living organism whose growth and development depend on its internal activity and interaction with the ground. One might think that this function, which gives an idea of the vital functions of a tree, reflects not the trivial knowledge of a native speaker but a truncated projection of our knowledge gained from a school course in botany. To show that this is not the case, we turn to the functional parts of a tree; as can easily be seen, their composition and structure are defined by the function of a tree as formulated above. 3. Basic-level concepts as the neurobiological codes for memory 162 Indeed, just as parts of a chair—the seat, back, and legs—conjointly define its function, the roots, trunk and branches define the function of a tree. The roots procure nourishment from the ground; the trunk supports the branches, and the procured nourishment is delivered via the trunk and branches to the buds and fruits. The functional relevance of this parsed structure is manifested in the correctness of the expressions korni/stvol/vetki dereva ‘the roots/trunk/branches of a tree’ in contrast with the incorrectness of similar expressions referring to alien objects physically connected with a tree, cf.: *gnezdo dereva ‘the nest of a tree’ or *skvorečnik dereva ‘the starling-house of a tree’ (of a nest built on a branch of the tree or a starling-house nailed to the trunk). Quite a few human motor concepts are associated with trees: sažat’/ polivat’/udobrjat’ derevo ‘to plant/water/fertilize a tree’, srezat’ vetki / sryvat’ plody / lazit’ na derevo ‘to cut off branches / pick fruits / climb a tree’, sidet’ v teni dereva ‘to sit in the shade of a tree’, pilit’ derevo ‘to saw a tree’, etc. These constitute a second, motor component of the concept derevo. A shrub has a different parsing structure. It has multiple trunks which are low and thin, and therefore the scope of potential human interactions with it is different: one doesn’t have to climb the shrub to pick its fruit; it will not provide shade to hide from the sun, etc. Let us summarize what we have looked at in this chapter. Lakoff (1987: 31–32), following Brown (1965: 318–319), believes that, just like a man-made object, a living object is defined by a single characteristic interaction: with flowers—to smell them, with cats—to pet them, etc. However, there are reasons to believe that the category of living objects is defined by two sets of interactions: the object’s interaction with its surroundings and the interactions of humans with the object. For example, the following actions are characteristic of a cat: to climb trees, to walk on fences and roofs, to stalk prey and then catch it with its paw. Quite different actions are characteristic of a dog. A dog joyfully greets its master coming home, it likes to be taken out for walks, it can guard the house—all of these are quite alien to interactions between a human and a cat. Another example would be that of a feral cat and a lynx, which show similar parsing structures and habits; however, a cat is not a danger to man, while a lynx is. § 1. Concrete concepts and motor concepts 163 10. Appendix. On the dual nature of human categories The formula for a concrete concept in (12) gives grounds to believe that the semantic category of noun, i.e. the category of referents defined by the basic meaning of a noun (concrete concept), is simultaneously defined by two criteria of different nature: the prototypical shape of an object, and the object’s function. The prototype criterion assigns an underspecified fuzzy category of referents, while the function criterion sets a rigid category. Specifically, the function criterion, which appears at a later stage, does not cancel the prototype criterion, but it has a higher priority. Thus, the principal controversy between Lakoff and Wierzbicka is resolved in a natural way. Since the 1970s, research done by Eleanor Rosch, George Lakoff and other cognitive scientists has led to the development of a new theory of natural categories which are defined, for example, by notional words such as bird, boat, cup, bachelor, mother, game, healthy, climb, etc. Leaning on works by game Wittgenstein (1953: 66–71) (his analysis of the word and his notion of “family resemblance”), Austin (1961: 71) (who discussed the fuzzy semantics of the word healthy), and using the results of some psycholinguistic experiments, Lakoff (1987) proposed a new approach to human categorization. The core claim of his proposal was that human categories are defined on the basis of “embodied” basic-level concepts which reflect a child’s imaginative perception and motoric conceptualization of the world. Therefore, human categories are prototypical by nature and, as a consequence, essentially fuzzy, as they are based on prototypes defined by cognitive models. Thus, the classical Aristotelian view of category as a set defined by the properties shared by all its members could no longer be considered adequate. Wierzbicka (1990) disagrees with Lakoff and claims the opposite: semantic categories of the very same notional words (bird, boat, cup, bachelor, mother, game, healthy, climb) are quite strict. The definition of a concrete concept given in (12) resolves this contradiction. It allows us to claim that natural human categories are at least dual, i.e. simultaneously they are defined both as prototypical and fuzzy (based on prototypes) and as classical and strict (based on functional features). 3. Basic-level concepts as the neurobiological codes for memory 164 § 2. On the psychophysical state and the neurobiology of human actions 1. Events and their storage in memory (the neural codes of memory) Our definition of the term ‘psychophysical state’ (of a human) will lean heavily on the research on human and animal memory structure conducted by Joe Tsien and his colleagues (Tsien 2008; Tsien et al. 2013). A brief overview of some of their findings is given in Tsien (2007), which discusses the results of experimental investigation on how memories of staged dramatic events are encoded in the episodic memory (in the hippocampus) of a mouse. The tests consisted in the following. Using a technique for simultaneously recording the activity of 260 neurons in the hippocampus (in the CA1 region, which is important to forming memories in both people and animals), the experimenters subjected mice to seven stressful episodes separated by periods of rest for several hours. Among such events were (1) a lab version of an earthquake, induced by shaking a small container holding a mouse, and (2) a brief vertical free fall inside a small “elevator” (provided by a cookie jar). The researchers reasoned that such dramatic events should produce memories in the mouse by forming new patterns of neuronal activity (firing) in the hippocampus, which would later manifest themselves during perception or recall of such events. And indeed, new patterns of activity in the CA 1 neural ensembles were found within a computer-monitored population of 260 neurons. As it turned out, the basic units of memory (functional coding units) are the so-called neural cliques—ensembles of neurons in the neural population of the CA 1 region—which show similar features and selective response reactions. In other words, a clique is a group of neurons that respond similarly to an event and thus operate collectively as a robust coding unit. For example, in the course of the experiments a “general startle” neural clique was activated in a mouse’s hippocampus which responds to all types of startling stimuli, including the elevator drop, earthquake, and air-blow (Tsien 2008: 407). Furthermore, cliques were activated that encode more specific types of startling events. Thus, a subgeneral startle clique responds to a combination of only two types of startling events: free fall and earthquake. Finally, highly specific cliques encode various features of an event: one encodes shaking, another location of the shake, etc. According to Tsien and his colleagues, their findings show that a perceived event is encoded in the memory of a mouse by a whole assembly of cliques 3. Basic-level concepts as the neurobiological codes for memory 166 2. Psychophysical state as a memory code for interaction Going back to the notion of “psychophysical state” and its interpretation, a psychophysical state is a neural code of long-term memory (an ensemble of cliques) that encodes a typified immediate interaction of a human, that is, a motor concept. In other words, a psychophysical state is a complex of polytypic data generated by various regions of the human brain in the course of performing a concrete physical action. The following are examples of various kinds of such interactions (motor concepts): sit on a chair, sit in an armchair (these interactions cause different psychophysical states of the sitting person), drink water, drink brandy, shave with an electric razor, shave with a razor, run, walk, ride a bicycle, ride on a bus, cut bread with a knife, cut cardboard with scissors, etc. A psychophysical state integrates input from different subsystems of the nervous system (limbic, vestibular, somatic, the cerebellum, etc.) which register human motives and goals as well as various kinds of sensations (physical and emotional) that accompany a particular physical interaction. Each subsystem contributes to the psychophysical state, activating its neural clique: the limbic system is responsible for the emotional-motivational aspect of the interaction, the vestibular system deals with the information about balance (sustainment of vertical orientation of the body), acceleration or deceleration of motion, and the like, the cerebellum coordinates the motions that require sequential contractions of multiple muscles, controls the balance of the body, forms the current plan of interactions and adjusts its implementation. Let us focus on the somatic system, which is comprised by the afferent (sensory) and efferent (motor) neurons. The afferent constituent (proprioceptors) provides for the muscle-joint feeling which helps to control the position of the body and its interrelated parts in space, sustain awareness of the direction and speed of motion, and define the muscular strength required for a given motion or for keeping the joint in a particular position. The efferent constituent (the motor efferent neurons found in the spine and brain), processing the input from proprioceptors, sight, and hearing, provides for control over the body and its locomotion by causing contraction and relaxation of the skeletal musculature. As we already know, an immediate human interaction has a three-component structure: (1) the motive (goal), (2) the dynamics (force-dynamic schemata), and (3) kinematics (changes in the spatial orientation of the body and § 2. On the psychophysical state and the neurobiology of human actions 167 its parts). The goal of an interaction is represented in the psychophysical state by a clique that receives input from the limbic system. The interactional dynamics are represented by the input from the efferent constituent of the somatic subsystem which, among other things, controls implementation of the force-dynamic schema of motion. Of special interest to us is the motion kinematics input. The memory code receives this input from the afferent constituent of the somatic system—a system of proprioceptors located in the muscles, ligaments, and joint capsules of a human body. Thus. speaking of locomotions such as walking, running, and so forth, Bernstein noted: The main afferents on this level […] are proprioceptions of joint-angular and geometric velocities and positions, to which a vast complex of general exteroceptive sensations of pressure reception, deep sense of touch and friction is added […], with exact ‘local marks’ characteristic of such receptions (Bernshtein 1990: 71; original emphasis). Thanks to this input humans are constantly aware of the position of their limbs and joints, whether they are in motion or at rest; they possess a feeling of body posture (position of all their limbs in space), a feeling of locomotion (awareness of the direction and speed of their joints in action), and a feeling of force (ability to assess muscular strength required for moving or keeping a joint in a particular position). Therefore, it is input from the afferent system of the somatic subsystem that serves to register a human’s current activity (or inactivity). an illustration of what has been said above, compare two motor concepts and their respective psychophysical states (these concepts have been discussed in detail in § 1, section 6). (1) Motor concept čelovek sidit na stule ‘a person is sitting on a chair’ (the meaning of the phrase Čelovek sidit na stule) = A Person’s Interaction → Psychophysical State A person’s back and buttocks 1) (goal) are in contact with the chair: With the goal to do some work at a (kinematics): desk/table, a person → 2) (dynamics) takes this body posture and is in a half-steady (semi-relaxed) state 3. Basic-level concepts as the neurobiological codes for memory 168 (2) motor concept čelovek idёt (po doroge) ‘a person is walking (on the road)’ (the meaning of the phrase Čelovek idёt) = A Person’s Interaction → Psychophysical State A person is moving down the 1) (goal) road, putting one foot in front A person, pursuing his spatial goal to of the other, but not fast get to a different location, (kinematics) 2) (dynamics) alternately rests his feet on and pushes → off from the surface, each time moving his body weight from one foot to the other, not at any time losing the support of the surface; at any given moment his position is unsteady Note the specific distinctions between the memory codes for motor concepts (1) a person is sitting on a chair and (2) a person is walking. In the latter case, there are full-fledged interactional dynamics that result in the locomotion of the body, its acceleration and deceleration. Therefore, there is, along with other cliques, a clique in the memory code that stores input from the vestibular system. In the former case, the role of proprioceptors is also important; they register the position of the motionless body, that is, the specific location of its parts such as the back, buttocks, and legs/ feet with their specific supports. There are also cliques that store input from the cerebellum (the state of partial stability of a motionless body) and the limbic subsystem (motivation for the interaction). At the same time, the vestibular subsystem does not contribute to the psychophysical state ‘sit on a chair’ because its receptors are not activated if the body is motionless. Note 1. All that being said, it becomes clear how visual prototypes (kinematics) of a person’s actions (shown in Fig. 1 and 2), as parts of the definitions of the motor concepts (1) A PERSON IS SITTING ON A CHAIR and (2) A PERSON IS WALKING, are connected with the psychophysical state of a person sitting on a chair or a walking person. They are reflected in the proprioceptive input and represented in the memory code of the motor concepts by a corresponding neural clique that stores this input. It has already been discussed above (chap. 2, § 1, section 1.9) why, when explaining the meaning of a word to a native speaker, it is enough just to describe the prototype (outward appearance, or the kinematics) of the word’s referents. It is enough, it has been argued, because the prototype (the visual component § 2. On the psychophysical state and the neurobiology of human actions 169 of basic meaning) triggers activation of the function in the native speaker’s memory—the main component of basic meaning. But what is the actual mechanism of connecting the prototype and the function? Now this mechanism becomes clear: during perception of a motion’s kinematics a corresponding clique in the memory code is activated in which proprioceptive data about these kinematics are stored, and this activation is followed by activation of the other cliques of the memory code (the function). Fig. 1 Fig. 2 Gallese and Lakoff (2005) argue that the “neural substrate” used in imagining a motion is the very same substrate used in its understanding. This Harry picked up the glass: hypothesis is illustrated by the simple sentence If you can’t imagine picking up a glass or seeing someone picking up a glass, then you can’t understand that sentence. Our hypothesis develops understanding is imagination, this fact one step further. It says that and that what you understand of a sentence in a context is the meaning of that sentence in that context (Gallese and Lakoff 2005: 456; original emphasis). In other words, a claim is made that the meaning of the given sentence is a “neural substrate” that provides only the visual component (kinematics) of the basic meaning of the verb pick (up), that is, only the proprioceptive component of the memory code of the action “pick up”. Thus, the second component of the meaning—the function—is overlooked. Moreover, as has been shown, a single kinematic component is not enough to understand an action (its goal), because there may be several memory codes with similar kinematics but different dynamics or goals. The notion of psychophysical state may be explained with the help of a simple example. It should be remembered that the basic meaning of a sensory verb is its current meaning (see the definition in chap. 2, § 2). Imagine a marathon runner approaching the finish line. He can be described by a phrase such Čelovek bežit k finišu, as lit. ‘The man is running to the finish line’, in its current (basic) meaning. This phrase expresses the psychophysical state of running. However, if the runner has stopped for a moment to grab a bottle of water offered by a spectator and take a few gulps, his psychophysical state will change 3. Basic-level concepts as the neurobiological codes for memory 170 dramatically; it will be expressed with the help of the phrase Čelovek p’ёt vodu, lit. ‘The man is drinking water’. In this case, what changes is, in the first place, the runner’s proprioceptive input; it registers that the runner is now standing, holding a bottle at his mouth and swallowing. Inputs from other subsystems also change: the limbic system registers that the thirst is being quenched, and the cerebellum that the body is in a steadier position now, etc. Similar psychophysical states determine the basic meanings of such verbs as udarit’ ‘hit’, padat’ ‘fall’, brat’ ‘take’, vzbirat’sja ‘climb’, risovat’ ‘draw’, s tojat’ ‘stand’, and others. Note 2. In discussions of the essence (or invariant) of verbal meaning, such properties as ‘motion’ and ‘energy expenditure’ are those often considered. However, none of them is a universal property. Thus, a man holding a suitcase may be motionless but expending energy. A man who is lying down is motionless and doesn’t expend energy, yet the verb ležat’ ‘lie’ is no bežat’ less a verb than the verb ‘run’. One might think of ‘time’ as a universal property, but it is rather opaque. We believe that the essence of verbness and of the physiological sense of time is expressed by input from the human proprioceptive subsystem; this continuously updated input signals at every moment about the position of the body and its parts, regardless of whether it is in motion or at rest. By reducing verbal meaning to a psychotekuščee state that includes proprioceptive input, the expression vremja ‘current time’ acquires an explicit cognitive property; that is, it becomes a term of the cognitive metalanguage of linguistics. 3. Biomechanical models of walking and running A rather detailed description of particular dynamic components of human locomotion has been given above, as well as their cyclic occurrence (pushing, losing support, transfer of support from one foot to the other, etc.). However, to describe locomotion in full it is necessary to view it as an entire process, i.e. as an integrated biomechanical model. This notion may be discussed using walking and running as examples. As has been noted earlier, an increase in the speed of a walk does not turn it into a run. Indeed, a person may be walking very fast (faster that a person running slowly), yet his manner of locomotion will not qualify as a run. For example, playing a film of a walking man in ‘fast forward’ mode does not turn the man’s motion into a run. Bernstein viewed locomotion as a live morphological object: […] The idea that motion is, in many respects, not unlike an organ (that exists in the coordinates of x, y, z, t, just like anatomical organs) seems to § 2. On the psychophysical state and the neurobiology of human actions 171 be very fruitful, especially when such a durable and universal kind of moissue. as locomotion is at […] As studies of walking have shown, motion displays very selective reactions. To changes in one component it reacts by changes in a number of other components, which are sometimes quite far from the former in space and time; at the same time, those components which, normally, are right next to the initially changed one, almost merging with it, remain unaffected. Therefore, locomotion is not a chain of components, but a structure differentiated into components; an integral structure characterized by high differentiation of its components and various-selective forms of the relationships among them. […] A run as a biomechanical structure is, by and large, the opposite of a walk […]; walking and running solve the mechanical task of moving a body in space in almost opposite ways (Bernshtein 1990: 336, 337, 353; original emphasis). This insightful claim may be elucidated by a simple experiment the reader can easily perform himself. If you begin to walk at a normal pace and then increase the speed of motion incrementally, you won’t feel any substantial changes in the motion’s biomechanics except greater pushing efforts. But the moment you start running, even if it is a very slow jogging run, the biomechanics of your locomotion change abruptly. Now your feet push off from the ground not only in a forward direction, but in an upward direction as well to make you ‘fly’ over the ground just a little bit. Your arms will bend at the elbows and start moving as counterbalances not only back and forth, helping to keep the already unsteady balance of the body, but also up and down to counter the pushes of your feet, to ease takeoff and soften landing. In sync with the pushes, your shoulders and torso will turn now left and now right, facilitating your motion forward; landing after a push will shake your entire body, and you will need special shock absorption provided by the legs (the leg, taking the role of support, bending slightly at the knee), something completely absent during a walk; your body will involuntarily lean forward, facilitating motion, etc. Thus, during walking and during running the human locomotive system works in two different modes, using two different biomechanical models of motion. It doesn’t matter who the runner is—a sprinter, a child, or an old lady. Of course, in the case of an old lady all these changes in motion will be much less noticeable; however, the totality of her running movements will readjust in a systemic way just as in the case of a sprinter. Her arms may not be bent at the elbows, but to keep better balance they (especially her elbows) will be more prominently drawn apart than in a walk, moving more actively in sync with the legs; ground support will be periodically lost or considerably weakened, and so on. 3. Basic-level concepts as the neurobiological codes for memory 172 It may be hypothesized that the biomechanical models of walking and running are stored in human psychophysical states. Therefore, the psychophysical states of walking and running are strictly different and discrete, which means that the categories “A person is running” and “A person is sitting”, set by these states, are likewise discrete. 4. On recognition of observed actions The arguments given above allow for a step-by-step description of the process of recognition of an observed action. As has been repeatedly mentioned, an action has a three-level structure: goal => dynamics (force-dynamic schema) => kinematics (changes in the position of the body and its parts in space, reflected by proprioceptors and the visual image of the motion). It has also been shown that both the dynamics and the goal are reflected in the kinematics, which contain their input (imprints). An action cannot be recognized without taking into consideration these inputs. For example, it has been shown that it is very hard to distinguish between a walk and an old person’s shuffling run by kinematics alone. However, if the dynamics input is taken into account, such differentiation becomes easier. For example, during a shuffling run a person periodically loses ground support, and to keep balance he often draws his elbows apart. In a similar fashion, the effect of the motive of locomotion on its kinematics should also be taken into account. Thus, if a gardener is trampling a newly laid path, he raises his feet higher and brings them down with a greater force than when he is simply walking on it. Going back to the psychophysical state of a human, i.e. the memory code of an action, it can be said in somewhat general terms that this code includes three components (three neural cliques). One component lays down the kinematics of the action (input from proprioceptors that reflects changes in the position of the body and its parts in space throughout the action), another the dynamics (input from the efferent system that controls motor neurons), and the third component stores the data about the motive (goal) of the action (input from the limbic system). It may be hypothesized that during perception of an action the recognition procedure involves three phases. First, using kinematics as a template, all memory codes with a similar kinematic component are found. Then, if there is more than one memory code, the dynamic aspect of the kinematics is assessed (contribution of dynamics to kinematics). If after that there are still several memory codes left, the action’s goal component is checked. If only one memory code is left in the end, it means that the perceived action has been recog- 3. Basic-level concepts as the neurobiological codes for memory 174 lar—are instantly recognized by a single static image (see Fig. 3a that shows a fragment of a painting on an ancient vase; the image of the runner is slightly skewed because of the shape of the vase). Fig. 3a Fig. 3b Fig. 3c Fig. 3d Fig. 3e Moreover, we instantly recognize single “linear frames” of walking and running (see Fig. 3b and 3c). But this is probably the limit of what humans can do, as we cannot recognize running by a separate point-light frame (see Fig. 3d and 3e). Mirror neurons also respond to static images (fragments) of actions, for example, to a single ‘frame’ of the grasping movement of a hand (Kourtzi and Kanwisher 2000; Urgesi et al. 2006). Thus, it appears that mirror neurons respond not only to observed actions but also to point-light animation displays ) and to (which look like apparently chaotic motions of a set of point-lights  5 single static images of actions. It is not quite clear how recognition, guided by “ such poor (and, therefore, ambiguous) visual stimuli, can be based exclusively on the vocabulary of acts and the motor knowledge”, since even fullblown images of some actions are often ambiguous. That being said, a claim may be made that the explanation of how actions are recognized, offered by the mirror neuron theory, is not quite adequate. It seems that the issues that emerge here go back to the reduced two-level representation of a physical action (kinematics—goal). As a matter of fact, such oversimplification involves at least two aspects of an action. First, the intermediate level of action dynamics, which connects the kinematics and the goal, is not taken into account. Second, it is ignored that, unlike for macaque primary monkeys, for humans a single action may have not only a (main) goal but also a secondary goal. For example, one can eat to satiate hunger 5 See, for example, the following links: http://www.biomotionlab.ca/Demos/BMLwalker. html, and http://www.biomotionlab.ca/Demos/BMLrunner.html. § 2. On the psychophysical state and the neurobiology of human actions 175 (primary goal) or to oblige one’s hostess (secondary goal); see chap. 2, § 4, section 1.6. The above said can be illustrated with the help of an example of how the type of human locomotion is recognized. Imagine a ballerina moving on the stage as if she were gliding, her feet keeping contact with the floor. Is her motion a fast walk or a dance? If we exclude the context (which is absent in Fig. 3c), much will depend on the reconstructed dynamic (support—take off) schema. If, after the next push of her take-off foot, the ballerina’s body loses support (for a fleeting moment her feet only touch the floor), then she is running. If, however, the support is not lost, then it is a fast walk. Suppose we have identified it as a run—to be more precise, while moving on the stage the manner ballerina uses the of running. Now we must understand the goal of using this manner, whether it is primary or secondary. If the ballerina is trying to use the manner of running as a means to express some emotional state (secondary goal), then she is dancing; it would be odd to call it running. However, if the ballerina’s main goal is a primary goal—simply to move to a different spot on the stage—then she is really running. And this goal is “computed” by taking the dynamics into account. Building on the arguments above, it should not be difficult to describe the process of recognition of a single (but typical) frame of an action, i.e. to explain how it happens to be so informative as to define the action as a whole, including its kinematics, dynamics, and goal. Consider again the image in Fig. 3c. First of all, we automatically build its dynamic (support—take off) schema. In this case it is clear that the person lands on his left foot after a rather strong take-off push and a short “flight”. The arms bent at the elbows, the right foot raised high, and the torso leaning forward are all typical signs of landing after a short jump. Clearly, it is not a walking step (compare to Fig. 3b); neither is it the final phase of stepping over a puddle; in that case the motion dynamics and, consequently, kinematics would be different. In particular, the arms would be in quite a different position. Obviously, the main goal of this jump is just to move forward. For example, the light jump of a ballerina in a dancing run would have different dynamics and kinematics. It is most likely not the final phase of landing in a single long jump; therefore, it should be a fragment of some cyclical motion. But what kind of motion? Were the person hopping on one (left) foot, then, in accordance with the dynamics of this motion, the right foot would also be in a different position. There is no choice but to assume that this is just a sequential mini-jump with the goal of moving forward, and, by inference, that it will be instantly followed by just another such mini-jump. Therefore, what 3. Basic-level concepts as the neurobiological codes for memory 176 integrated we see is running. A sequence of its separate frames is so tightly systemi c ally—is an “integral structure”, according to Bernstein (see the quote above in section 3)—that a single frame defines this structure as a whole. Thus, a single frame (Fig. 3c) allows us to recognize a person running, i.e. to reconstruct the three-level structure of locomotion: kinematics—dynamics—goal. Every component of the structure, including kinematics, is chosen from several alternatives in the course of cyclical comparison of the intermediate results of the recognition process. In the mirror neurons approach, recognition of a perceived action is seen as a result of mirror neurons firing during perception of this action (“action simulation”). Based on our conclusions, it may be hypothesized that the mirror neurons’ response is collateral to the independent process of recognition of an observed action. Indeed, during perception of an action’s kinematics a clique that stores proprioceptive input is activated in the memory code for the action. This is followed by activation of other cliques in the memory code, including the clique that stores efferent input from the motor neurons that fire while the action is being performed. It is the activation of this clique that produces the mirror neurons effect. 6. Canonical neurons and object recognition The recognition procedure for an observed object—its inclusion into one or another category of objects—is based on formula (4) given in § 1, which defines a concrete object as a system of visual and functional components. The process has been briefly described in section 4 of the same paragraph. It will be recalled here by considering perception of a chair with armrests as an example. We need to decide whether we are looking at a chair or an armchair. Guided by the object shape, we advance an initial hypothesis: this is a chair. This instantly activates the motor concept associated with the chair—sit on a chair. In our mind, we sit on the perceived “chair” and test whether the psychophysical state ‘semi-steady position of the body’ pertains. If this state is acknowledged (as potentially experienceable) by the sitting person, then the perceived object is a chair. If it cannot be acknowledged, another hypothesis is tested: this is an “armchair”, because the shape of the perceived object resembles that of an armchair. Now a different concept is activated—sit in an armchair. In our mind, we sit down as we would sit in an armchair and test whether the (potential) state of ‘almost fully steady position of the body’ may be acknowledged. If the answer is “yes”, then the perceived object is an armchair. If the answer is “no”, it is neither a chair nor an armchair. § 1. On the basic and derivative meanings... 179 The expression dver’ doma ‘the door of the house’ can refer only to the entrance door, not to the door that leads, for example, from the corridor to the kitchen, in spite of the fact that it is also a part of the house. Gates are zabora part of a fence, yet the expression *vorota ‘the gates of the fence’ is incorrect. Clearly, for the nominal genitive to be used correctly in the construction Y X-Gen, it is not enough for Y to be a physical part of X. As will be illustrated further, for the construction Y X-Gen (where Y and X are concrete nouns) to be used correctly, the following condition must be met: Y, being a physical part of X, is also X’s functional part; that is, the function of Y directly contributes to the general function of X. Thus, the entrance door is a functional part of the house (it opens and closes access to the house), while an interior door is not a functional part of the house as it does not contribute to the general function of the house. Therefore, one can say, for example, dver’ doma ‘the door of the house’ about an entrance door but not about a kitchen door. Note that the expressions dver’ kuxni/komnaty ‘the door of the kitchen/room’ are quite acceptable, while the expression *dver’ koridora ‘the door of the corridor’ is questionable (the correct expression is koridornaja dver’ ‘the corridor-Adj door’) as it is not quite clear how the door contributes to the function of the corridor (after all, there may be corridors without doors). The function of a fence is to prevent trespassing, so it does not have to have a garden gate or gates—as in the case of a fence separating two adjoining zabora gardens. Therefore, the expressions *kalitka/vorota ‘the garden gate/ gates of the fence’ are incorrect. Similarly, a door handle is operationally connected with the function of the door, helping to open and close the door. A peephole in the door, on the other hand, performs a different function which is not directly connected with the function of the door: it allows one to see who is standing outside the door. Therefore, the expression ručka dveri ‘the handle of the door’ is correct, while the expression *glazok dveri ‘the peephole of the door’ is not. For the same reason, speaking about a mailbox attached to the entrance door or gate, one cannot say *počtovyj jaščik dveri/ kalitki ‘the mailbox of the door/gate’. The walls also perform a direct and obvious function of a house (stena doma ‘the wall of the house’), but a window in the wall does not have a function associated with the wall (*okno steny ‘the window of the wall’). Like a wall, a window has its own (and different) function contributing to the function of the house. Speaking about a prop holding up the leaning wall of a barn, one can say podporka steny saraja ‘the prop of the wall of the barn’, thereby acknowledging that it is a functional part of the wall; however, one cannot say 4. Elements of a sensory grammar 180 something similar about a hook for work clothes nailed to the same wall; the expression *krjučok steny saraja ‘the hook of the wall of the barn’ is incorrect, as the hook has nothing to do with the function of the wall. For the same reason, the expressions kryl’co/podval/čerdak doma ‘the porch/basement/attic of *veranda/prixožaja/pogreb/ the house’ are correct, while the expressions balkon doma ‘the veranda/hallway/cellar/balcony of the house’ are not. It is podporki doma doma correct to say ‘the props of the house’ but not *otmostka ‘the perimeter pavement of the house’, just as it is incorrect to say *garaž doma ‘the garage of the house’ about a built-in garage. Consider the functional structure of a living-room. The expressions okna/ dver’/potolok/pol/steny/ugly/ventiljacionnaja rešёtka komnaty ‘the windows/ door/ceiling/floor/walls/corners/ventilator grating of the room’ are correct, but the expressions *balkon/časy/stol/škaf/batarei/ljustra/kovёr komnaty ‘the bal c ony/clock/table/wardrobe/radiators/chandelier/carpet of the room’ are not. It may be concluded that the latter referents of Y (the noun that designates both a physical and functional part of the room such as a balcony, table, chandelier, etc.), in contrast with the former referents (the windows, ceiling, ventilator grating), do not perform any particular functions of the room. The expressions *gazovaja plita /mojka kuxni ‘the gas stove/sink of the kitchen’ do not sound natural in Russian because the function of the kitchen is much broader than cooking and having meals or washing the dishes. Similarly, the function of a bathroom is broader than having a bath in it, therefore the expression *Vanna našej vannoj komnaty udobnee, čem vanna vo fligele ‘The bathtub of our bathroom is more comfortable than the bathtub in the wing’ sounds odd. The examples given above suggest that the incorrect use of the nominal genitive in the construction Y X-Gen—and, correspondingly, the functional dependency of Y on X—may entail one of the following: either Y is functionally connected not with X itself but with some functional part of X (the expressions *fortočka doma ‘the ventilation pane of the house’ or *pol doma ‘the floor of the house’ are incorrect as compared to the correct expressions fortočka okna ‘the ventilation pane of the window’ or pol komnaty ‘the floor of the room’), or Y, while remaining a physical part of X, is functionally independent of X. For example, a balcony or an aerial are functionally independdoma/komnaty of the house. It is incorrect to say *balkon ‘the balcony of the house/room’ or *antenna doma/kryši ‘the aerial of the house/roof’, but the naružnaja antenna televizora expression ‘the outdoor aerial of the TV’ is quite correct. § 1. On the basic and derivative meanings... 183 4. The derivative meanings of the nominal genitive Similarly to words, which may have extended meanings, the nominal genitive construction also has extended (metaphoric) meanings derived from the basic meaning, as shown in (1): a) ‘Y is a portion of substance X’: stakan moloka ‘a glass of milk’, kilogramm mjasa ‘a kilogram of meat’; b) ‘Y belongs to X’, if Y and X are autonomous objects: kniga Ivana ‘Ivan’s mašina načal’nika bookʼ, ‘the boss’s car’; c) ‘Y ‘belongs’ to X’: aktёr kino ‘the actor of the movies’, and the like. For a more detailed discussion of the different meanings of the nominal genitive see Shakhmatov (2001: 314–318). 5. Objects with a multilevel partitive structure Let us consider a bicycle. As a first approximation, its function may be formulated as follows: (2) Bicycle (function) = ‘a wheeled means of transport allowing a person to move while sitting on it; the person turns its wheels by pushing on the pedals with his feet and steers it by turning the handlebars to the left or to the right’ The main parts of a bicycle are the frame, handlebars, saddle, wheels, and pedals (hence the correctness of the phrases rama/rul’/siden’e/kolёsa/pedali velosipeda ‘the frame/handlebars/saddle/wheels/pedals of the bicycle’), but not the spokes, pump, bell, lamp, chain, brake, tires, etc. (hence the incorrectness of the phrases *spicy/nasos/zvonok/fonar’/ cep’/tormoz velosipeda ‘the spokes/pump/bell/lamp/chain/brake of the bicycle’). Meanwhile it is quite tormoz perednego kolesa šina correct to say ‘the brake of the front wheel’, zadnego kolesa ‘the tire of the rear wheel’, velosipednye spicy ‘bicycle-Adj spokes’ or velosipednyj fonar’ ‘bicycle-Adj lamp’. The hierarchical structure of the bicycle The main parts of the bicycle (wheels, saddle, handlebars, etc.) form the first level of functional hierarchy; parts of these parts (the spokes and tire of each wheel, saddle springs, etc.) form the second level, and parts of these (spoke nuts, tire valves) the third level. As mentioned earlier, the incorrect use of the construction Y X-Gen can entail one of two things: either the given part belongs to a lower level of X’s partitive hierarchy, or Y is functionally i ndependent of X altogether, although it is connected with it physically, 4. Elements of a sensory grammar 184 cf.: *nasos velosipeda ‘the pump of the bicycle’ or *fonar’ rulja ‘the lamp of the handlebars’. It may be assumed that a native speaker of Russian can decide whether the expressions given above and many other similar expressions are correct just because, on the one hand, he is familiar with the general function of the bicycle as defined in (2) and with how it works in principle and, on the other hand, he hears such expressions from the people around him. Thanks to this knowledge, he intuitively builds (“computes”) the hierarchical structure of the bicycle and its functional parts as described above. An objection could be raised that the described partitive structure is too complicated for the average native speaker, and that he is unlikely to know and use it. However, were this the case the speaker would most probably not be capable of judging the correctness or incorrectness of many expressions of the kind given above. 6. The partitive structure of plants It may be assumed that the native speaker’s knowledge of the functions of objects is subconscious. Definition (1) allows us to use the genitive construction Y X-Gen to explicate the partitive structure of various objects and their general functions. So far we have discussed the genitive construction only with regard to the non-organic objects that this construction describes. Let us now consider cases where X is a living organism. Take a tree, for example. A native speaker of Russian subconsciously interprets its functional structure as substantially different from the structure of an inanimate object. He assumes that practically all physical parts of the tree—not just the roots, trunk, and branches, but also the leaves, buds, and bark—belong to the tree as a whole. In other words, the tree has a single-level partitive structure. This is demonstrated by the correctness of the expressions vetvi/korni/stvol dereva ‘the branches/roots/trunk of the tree’ and list’ja/počki dereva ‘the leaves/buds of the tree’. It seems that the leaves and buds should be a second level in the hierarchy, as they physically belong to the branches; however, the incorrectness of the expressions *list’ja/počki vetok ‘the leaves/buds of the branches’ shows that there is no second level in the hierarchy. This is also true with regard to stvola the roots, which are part of the tree and not of the trunk, cf.: *korni ‘the stvola roots of the trunk’. Neither can one say *sučok ‘the twig of the trunk’ or *sučok vetki ‘the twig of the branch’. This observation is of a general character: any single part (that is, a part with a function) of a living organism is an immediate functional constituent of § 1. On the basic and derivative meanings... 185 the organism as a whole. This claim will be shown to be true with regard to humans as well (see section 7). Note that the functional subordination of some parts of the tree is twofold: kora stvola kora dereva ser‘the bark of the trunk’ and ‘the bark of the tree’, dcevina stvola ‘the heartwood of the trunk’ and serdcevina dereva ‘the heartwood of the tree’. However, this does not contradict the general conclusion made above: the bark, being subordinate to the tree, is also in local functional subordination. An interesting, but quite explainable, exception to this general rule should be mentioned: at a certain moment in time the speaker begins to view the fruit of the tree as functionally autonomous “alien” parts of the tree. Thus, Zavjazi ètoj jabloni xorošo vzjalis’ the phrase ‘The ovaries of this apple-tree ètoj jabloni uže sozreli have taken on well’ is correct, while *Jabloki ‘The apples of this apple-tree have already ripened’ is not. Meanwhile the phrase Jabloki ètoj jabloni uže v korzinax (prodany) ‘The apples of this apple-tree are already in the baskets (sold)’ is quite correct. The genitive construction here has the metaphoric meaning ‘the relationship of belonging’, because the apples are interpreted as autonomous objects separated from the appletree—compare with a similar expression synov’ja materi ‘the sons of the mother’. This interpretation sometimes escapes the attention of researchers. Thus, Liashevskaia (2004) observes: “The secret of countability of nuts, berries parts and beans lies in the fact that they designate these fruit as of the plant, jagody eževiki boby kanavalii cf.: ‘the berries of the blackberry bush’, ‘the beans of Canavalia [a kind of legume]’…” (p. 237; original emphasis). In view of what has been said above, these phrases do not express a ‘partwhole’ relationship between the berries and the blackberry bush or between the beans and the legume because the fruit has already been picked and separated from the plants. If, however, the fruit is still attached to the plant and ripening, thus remaining, physically, part of the plant, such expressions are incorrect. Any alien entity physically connected with a tree (a bird’s nest, a starlinghouse or a basket ring nailed to the tree, a rope-swing tied to the tree branch, etc.) is interpreted by the speaker as functionally autonomous, hence incorrectness of the expressions *gnezdo/skvorečnik dereva ‘the nest/starling-house of the tree’ or *kačeli dereva ‘the swing of the tree’. Now, is a tree growing from the ground and physically connected with it an independent, functionally autonomous object (organism), or is it part (Y) of the ground (X)? Incorrectness of the expressions *derevo zemli/xolma/veršiny 4. Elements of a sensory grammar 186 utёsa ‘the tree of the ground/hill/top of the cliff’ shows that the speaker views the tree as an autonomous organism. To conclude this section, let us take a quick look at the partitive structure of an apple. The following main functional parts of an apple can be singled out: the skin, the flesh, the pips (seeds), and the pedicel (or fruit stalk ) on  2 which it is suspended while growing and ripening. This is confirmed by corkožura/mjakot’/zёrnyški/plodonožka jabloka rectness of the expressions ‘the skin/flesh/pips/pedicel of the apple’. This set of parts is not accidental; it stems from the speaker’s intuitive understanding of the general function of the apple: ‘to ensure the ripening and subsequent germination of the pips—the apple seeds’. Each part from the set contributes to the realization of this function. The skin serves a protective function, the flesh provides nourishment for the germinating seeds, and the pedicel supplies nourishment necessary for the growth and development of seeds and the apple as a whole. As an illustration, consider the apple in Fig. 1. It is quite corFig. 1 rect to say plodonožka jabloka ‘the pedicel of the apple’, but incorrect to say *listik jabloka ‘the leaf of the apple’ or *listik plodonožki ‘the leaf of the pedicel’. The explanation is simple: the speaker, guided by the function of the apple and its parts, understands at once that, by contrast to the pedicel itself, a leaf that grows from it does not contribute either to the general function of the apple or to the particular function of the pedicel. 7. The partitive structure of animals and humans We will now consider the functional structure of humans and the human body. It will be shown that in terms of the part-whole relationship this structure is similar to the structure of a tree. Rakhilina (2000) gives a detailed analysis of the uses of the genitive conY X-Gen struction with reference to parts of the human body. Briefly, her approach is as follows. Expressions such as pal’cy levoj ruki ‘the fingers of the left hand’, pjatka pravoj nogi ‘the heel of the right foot’ are correct because there is another, alternative whole X to which Y belongs (the left hand for the fingers, the left foot for the heel). If, however, there isn’t such an alternative tela for Y, the use of the genitive becomes incorrect, cf.: *ruki ‘the arms of the 2 Plodonožka [‘pedicel’]—“part of the stalk that connects the fruit with the plant” (Ushakov, III: col. 302). 4. Elements of a sensory grammar 188 sion pal’cy nogi ‘the toes of the foot’. The sole and heel, by contrast, contrib 3 to the function of the foot, making expressions such as pjatka/podošva stupni ‘the heel/sole of the foot’ quite acceptable. It is possible to say koža ruk ‘the skin of the hands’ only on condition that the skin performs some specific function of the hands, for example to show how well the hands of a woman are taken care of, cf.: šelkovistaja koža eё ruk ‘the silky skin of her hands’. If, however, a farmer or a worker is described, the state of his skin will be characterized in a different way, for example, ogrubevšaja koža na ego rukax ‘the callous skin on his hands’ (compare with the questionable expression ogrubevšaja koža ego ruk ‘the callous skin of his hands’). ? In addition, some extrinsic formations may appear on a human body which have nothing to do with the functions of the body itself, cf.: *mozol’ pjatki ‘the corn of the heel’, *mozol’ čeloveka ‘the callousness of the man’, *borodavka nosa ‘the wart of the nose’ (the correct expression would be borodavka na nosu ‘the wart on the nose’). Thus, similarly to a tree, almost all parts of the human body are functionally subordinate to the person and not to his body, cf.: telo/golova/ruki/nogi/ glaza/uši/bok/koža čeloveka ‘the body/head/hands/feet/eyes/ears/side/skin of a person’. Sometimes such functional subordination is twofold: kisti čeloveka ‘the wrists of the person’ and kisti ego ruk ‘the wrists of his arms’, nogti čeloveka ‘the nails of a person’ and nogti ego pal’cev ‘the nails of his fingers’, etc. Just as in the case of the fruit of plants, a human fetus is not viewed as a functional part of the female that bears it; the expression *plod ženščiny ‘the fetus of the female’ is incorrect, and the acceptable expressions are plod ‘the fetus’ or materinskij plod ‘the mother’s fetus’. The situation here is even more obvious. While ovaries are, up to a certain moment, viewed as functional parts of a tree—cf. the acceptable expression zavjazi jabloni ‘the ovaries of the apple-tree’—a human fetus is viewed as functionally autonomous from the start; hence incorrectness of the expression *zarodyš materi ‘the embryo of the mother’ (one should say materinskij zarodyš ‘the mother’s embryo’ or simply zarodyš ‘the embryo’). It is possible to say rebёnok materi ‘the baby of the mother’ only about a baby that has been born and physically separated from the mother. This expression would become anomalous if used to refer to a baby still in the womb. 3 noga In Russian, may refer both to the entire lower limb (leg) and to the foot, while stupnja (also stopa, from the verb stupat’ ‘step’) refers only to the foot. This creates certain difficulties in translating expressions such as pal’cy stupni/nogi.—Translator’s note. § 2. On structural and genetic similarity... 189 § 2. On structural and genetic similarity. . of lexical and grammatical categories. The meaning of transitivity Chapter 2 was devoted to an analysis of lexical meanings of sensory words stol derevo bežit (nouns and verbs), that is, words (such as ʻtableʼ, ʻtreeʼ, ʻis runningʼ, udaril ʻhitʼ, etc.) whose typical referents (objects and actions) are quite explicit: they are identified by their appearance, without the help of any context. We can always distinguish between a typical table and a typical chair, or between a typical running event and a typical walking event, and so forth.  4 As a result of such an analysis, we identified two general structures bearing on the semantics of sensory words. 1. The structure of basic meaning In §§ 1–4, the following structure of the basic meaning of a word was iden tified: (1) Basic meaning = visual Prototype → Function. Here the Prototype defines the typical visual (or perceptual, in a broad sense) feature of word referents, and the Function defines the functional (causal, intentional) feature characteristic not only of typical but of all referents. This structure is a first step towards developing a cognitive language of thought. The components of the structure—the prototype, the function, and the relationship of interpretation—are of a purely cognitive, non-verbal nature. Concrete and motor concepts made up of these components are basic units of this special language, the language of thought. 2. The structure of polysemy In chapter 2, § 5, it was shown with a number of examples that lexical polysemy has, approximately, the following structure. A sensory word, typically, has one basic meaning (it is more concrete and obvious, and is given first in explanatory dictionaries) and a range of derivative meanings that are metaphoric and metonymic extensions from the basic meaning. In other words, the following claim seems justified: 4 It should be noted that not all concrete nouns are sensory in this sense. For example, certain parts of objects such as the leg of a sofa/chair, the shaft of an umbrella or the base of a standard lamp, in order to be recognized, require context—that is, the object, of which they are parts (for more details see § 1). 4. Elements of a sensory grammar 190 (2) The structure of meanings of a sensory word is as follows: basic meaning—a variety of meanings derived from the basic meaning (metaphors and metonymies). The structure given in (2) allows us to explain the mechanisms for the emergence of new word meanings and occasional uses. As has been mentioned in § 5, the most important feature of human language is the ability to use its lexical and grammatical items in an occasional way, that is, in novel senses that have not been encountered before. Thereby the already existing linguistic means are used to describe novel real world phenomena. Thus, explaining the mechanisms of occasional word uses is one of the central problems in the theory of cognitive semantics. One possible solution of this problem is provided by the structure in (2). In what follows it will be shown that a great number of grammatical items also include sensory units, and their meanings are structurally and genetically similar to lexical meanings. In particular, the basic meaning of a sensory grammatical unit has the structure shown in (1), while its polysemy is structured as shown in (2). In other words, the lexical and grammatical meanings of sensory linguistic units are structurally and genetically indistinguishable. The validity of this hypothesis for verb transitivity will be shown in the next section, and for voice, in § 3. 3. The meaning of transitivity There is a long-standing tradition to give a semantic interpretation of the syntactic feature of transitivity—the ability of a verb to take a direct object in the accusative (cf. relevant references in Letuchii 2014). Here we will touch upon the direction of research that aims to define the prototypical transitive event, or the transitive scenario, and is represented in the works of Slobin (1982), Givόn (1990: 565), and Wierzbicka (1996). For example, siding with Givόn, Wierzbicka writes: In my terms, the prototypical transitive scenario can be represented as follows: (61) at some time, someone was doing (did) something to something because of this, something happened to this something at the same time this person wanted this (to happen) Of course “transitive sentences” don’t have to meet all the aspects of this scenario, but a departure from any of them is likely to lead to a de- § 2. On structural and genetic similarity... 191 crease in syntactic transitivity (manifested in case assignment, passivizability, and so on) (Wierzbicka 1996: 420). Upon closer scrutiny this otherwise clear formulation gives rise to a number of questions. For example, what is the meaning of the words “something happened”? If someone hit a ball with his foot (udaril po mjaču nogoj), did “something happen” to the ball or did it not? If something did happen, why is the verb udaril ʻhitʼ used intransitively? Or, when a boy is reading a book (mal’čik čitaet knigu), does something happen to the book or not? If nothing happens, why is a transitive verb used? In our definition of transitivity two notions are introduced: Prototype— visual (objective) changes in the object caused by the Agent’s contact with the object—and Function (interpretation); they indicate that the changes in the object, both observable and unobservable, are caused by the Agent, are relevant to the Agent, and are the Agent’s goal. For example, if a farmer measures pole), the field (izmerjaet it changes, from his point of view, as it acquires a new important property—the square area. Therefore, the use of a transitive verb is appropriate here, although the field doesn’t undergo any visible changes. protoThus, our definition of semantic transitivity includes the following: type, which is similar to Wierzbicka’s interpretation and describes typical cases (scenarios) of transitivity, and function, which describes the interpretation of the actions by the Agent. This interpretation covers all (not just typical) cases of transitivity. (T) Verb transitivity (basic meaning) = Prototype: The Agent performs an observable CONTACT action with the direct object (Patient) which simultaneously undergoes visible changes → Function: The Agent performs an action THEREFORE the direct object simultaneously undergoes changes that are relevant to the Agent and are the Agent’s goal. By this definition, verb transitivity is included in the category of sensory units. It should be noted that there are certain ties between the properties of the prototype and the function. Firstly, the relation ʻAgent in CONTACT with direct objectʼ is a manifestation of the causal relation ʻAgent performs an action THEREFORE the direct object undergoes changesʼ. If we recall the experiments conducted by Leslie (see chap. 2, § 7, section 2), they have shown that at 7 months infants interpret the collision (contact) of a moving ball with a stationary ball as the cause of the subsequent motion of the latter (cf. also Pinker 2007: 67). Secondly, the visible change in the direct object is a mani- 4. Elements of a sensory grammar 192 festation of objective relevance of its change, not just from the point of view of the Agent. These two properties are typical of verbs that designate actions of Agent coming into physical contact with objects, these actions causing visible changes in the objects: gruzit’ ʻloadʼ, myt’ ʻwashʼ, krasit’ ʻpaintʼ, rubit’ ʻchopʼ and so forth. Cf.: […] In the three object syntagms given above (vzjal topor ʻtook an axʼ, srubil derevo ʻfelled a treeʼ, and postroil dom ʻbuilt a houseʼ) the roles of arguments […] differ on many features (and this difference may be linguistically substantial), but from the point of view of the Russian language they display more common features than differences: all three roles correspond to the “object” (or “patient”) which is the final point of application of energy on the part of a conscious doer (“agent”) and which, as a result, suffers observable physical changes (Plungian 2011: 114). Some changes in the object (its destruction, change of shape or location, etc.) are objectively relevant to human ethnic communities; therefore, verbs from respective classes (razdavit’ ʻcrush/squashʼ, ubit’ ʻkillʼ, etc.) are transitive in many languages (Letuchii 2014). In this respect, the Russian verb udarit’ ʻhit-PFʼ is of special interest. In the Russian lingua-culture the shaking of an object, caused by something or someone hitting this object, is not interpreted as a relevant change in the object; therefore, in phrases such as Ivan udaril po mjaču nogoj / po zaboru palkoj ʻIvan hit the ball with his foot / the fence with a stickʼ the verb udarit’ is used intransitively. However, when it is a human or an animal that is hit, the verb is used transitively, cf.: Ivan udaril Petra/sobaku ʻIvan hit Peter / the dogʼ. In this case, the changes in the Patient (the suffering of pain) are interpreted as relevant. In some cases, the relevance of changes in the object, caused by the Agent, are of a subjective character. If the speaker interprets such changes as substantial, he uses a transitive verb, whereas if he interprets them as unsubstantial, he uses an intransitive verb. For example, a girl complains about her brother On menja obryzgal vodoj to her mother: ʻHe spattered me with waterʼ (a transitive verb is used because the changes are relevant to the sister). Her brother Da ja bryznul na neё doesn’t see the changes as relevant, and his response is: vsego paru raz, lit. ʻWell, I spattered [water] on her just a couple of timesʼ (an intransitive verb is used; see an exhaustive analysis in Pinker 2007: 42–51). A whole range of transitive verbs designate changes in the object that are subjective by their very nature and are not universally relevant. For example, in the clause Ivan osveščaet dorogu fonarikom, lit. ʻIvan lights the road with a flashlightʼ, the observed action (‘lights’) of the Agent does not, strictly speak- § 2. On structural and genetic similarity... 193 ing, change the road. To Ivan, however, the road changes in an obvious way; therefore, this use satisfies the function from definition (T). Transitivity of the phrase Mal’čik čitaet knigu ʻThe boy is reading a bookʼ is explained in a similar way. The changes that occur here are relevant only to the boy reading the book: to him, the part of the book that has been read differs from the unread part of the same book. Compare also the following uses: Ivan smotrit na knigu ʻIvan is looking at the bookʼ (the direction of his gaze is fixed, the book does not undergo any changes) and Ivan smotrit knigu, lit. ʻIvan is looking [through] the bookʼ (is leafing through the book to learn what it is about—to him, the book undergoes changes). The clauses Vrač smotrit na pacienta ʻThe doctor is looking at the patientʼ and Vrač smotrit pacienta, lit. ʻThe doctor is looking [= examining] the patientʼ are another example of similar uses. The transitive expression razgljadyvat’ kartinu, lit. ʻto scrutinize the pictureʼ, means to shift one’s gaze, to look at it from different angles and learn something new about it that changes the picture in the eyes of the beholder; the analizirovat’ kartinu same is true about the transitive expression ʻto analyze vsmatrivat’sja v kartinu the pictureʼ. By contrast, the intransitive expression ʻto peer at the pictureʼ indicates that the subject is looking intently at the picture; whether he learns something new about it or not remains unspecified. Heretofore we have considered examples of transitive verbs used in their basic meaning; the situations they designated satisfied the function from definition (T). Let us now look at the metaphoric and metonymic uses of transitive verbs. Consider transitive verbs designating states, taking, as an example, the sentence Maša požalela bezdomnogo ʻMasha took pity on the homeless manʼ. This sentence is an example of metaphoric use of transitivity which does not satisfy the function from (T). The referent of the verb in this case is not the action performed by Masha, but the change of her state. Moreover, the semantic roles of the predicate arguments have changed: Masha is not the Agent, she is the Experiencer (she perceives visual and acoustic data), and the homeless man is not the Patient but the Stimulus (the source of data for Masha). At the same time, the homeless man has changed from Masha’s point of view—now he makes her pity him. The metaphor created from the basic meaning is founded just on this similarity. As can be seen, metaphoric use of transitivity results in a loss of important features of the basic meaning.  5 5 This is typical of metaphors. In the metaphoric phrase Ivan bežit na meste ʻIvan is running on the spotʼ the crucial features of running are absent—Ivan’s motion and his goal to reach a certain point in space. The metaphoric use of bežit ʻis runningʼ is motivated solely by the apparent similarity between running on the spot and just running. This ex- § 3. The basic and derivative meanings of voice 195 in the rest of this paragraph, where the meaning of the reflexive voice will be shown to be a metonymy from the active voice, and the meaning of the passive voice a metaphor from the reflexive voice. There is a lot of literature on the meanings of voice (for bibliographic reviews see: Testelets 2001: 435–436; Plungian 2011: 223–227; for a brief case history see: Kulikov 2011: 368–369). Traditionally, voice is defined as the relationship between the structure of semantic roles of verb arguments (Agent, Patient, etc.) and the structure of their syntactic roles (subject, direct object, etc.). For example, in the active voice of a transitive verb Agent is expressed by the subject and Patient by the direct object in the accusative (Maša-Nom pričёsyvaet kuklu-Acc ʻMasha is combing the doll[ʼs hair]ʼ). In the passive voice the same semantic roles are designated differently—Patient by the subpričёsyvaetsja Mašej-Ins). ject and Agent by the indirect object (Kukla-Nom Finally, in the reflexive voice both semantic roles are combined and are designated by the subject (Maša pričёsyvaetsja ʻMasha is combing [her hair]ʼ). A different approach uses the concept of ʻcommunicative rankʼ of the verb argument (cf., e.g., Plungian 2011: 187–188). The approach discussed below is, essentially, as follows. Among many real world phenomena relevant for the vital functions of human beings, changes that occur in the world are, perhaps, the most important. Therefore, in human adaptive interactions with the world, perception and categorization of the observed changes in living beings, objects, dry substances, liquids, scenery, and so forth play a crucial role. We believe that the meanings of the active, passive, and reflexive voice reflect three most general types of change undergone by the Agent. The active voice reflects independent changes in the Agent (Mal’čik moet mašinu ʻThe boy is washing the carʼ), the passive reflects dependent changes (Mašina moetsja ʻThe car is being washedʼ), and the reflexive both independent and dependent simultaneous changes (Mal’čik moetsja ʻThe boy is washing himselfʼ). Let us assume that the dichotomy independent vs. dependent changes in the object is intrinsically primary for humans. We believe that a human individual perceiving a change occurring in an object instantly identifies it, categorizing it either as an independent change (for example, a person is running/ falling/waving a hand) or a dependent change (e.g., a stone is getting warm: the sunrays light the stone THEREFORE it gets warm; a boat is sailing: the wind fills the sail THEREFORE the boat sails). It is likely that the human ability to distinguish between independent and dependent changes is speciesspecific and does not depend on language. This ability becomes manifested in children during the first 12 months and from then on it keeps improving. 4. Elements of a sensory grammar 196 Acquisition of the laws of nature by children and the causal relationships that exist in the world are discussed in more detail in chap. 2, § 7. The formation of the dichotomy “independent vs. dependent changes” is based on the ability of humans to distinguish pairs of observable changes that are linked by a causal relationship: Change 2 occurs BECAUSE Change 1 occurs (e.g. a stationary ball, 1, began to move BECAUSE a rolling ball, 2, collided with it). Change 2 is a dependent change. If, at the same time, there isn’t Change 0 such that it directly causes Change 1, the latter is an independent change. An independent change may be both single (when it doesn’t cause another, dependent change) and dual, causing another, dependent change. In bežit/ language, single changes are described by intransitive verbs (čelovek padaet veter duet ʻa person is running/fallingʼ, ʻthe wind is blowingʼ), and režet xleb vzjal knigu uronil nož dual changes by transitive verbs (čelovek / / ʻthe person is cutting the bread / took the book / dropped the knifeʼ). This claim needs clarification. Consider an action such as “Ivan režet xleb nožom” [“Ivan is cutting/slicing the bread with a knife”]. It consists of two separate actions linked by a causal relationship: (С) Ivan režet xleb nožom ʻIvan is cutting/slicing the bread with a knifeʼ = Bringing the knife into contact with the loaf of bread, Ivan is moving the blade of the knife across the loaf (independent action) THEREFORE the loaf is divided in two parts at the point of contact (dependent a ction). As can be seen, this isn’t a single action; there are two causally linked actions. However, just as in the case of a single action, they are designated by a single verb. This may account for the false impression the speaker gets that (C) is also a single action performed by Ivan, and this action “passes over” onto the loaf. According to the contemporary view, a verbal predicate is the main part of the clause that defines the referent situation. Keeping to an earlier tradition (Testelets 2001: 88–89), we believe that a clause designates a situation that is defined by the predicative nucleus of the clause—a nominal subject and a verbal predicate. In such a case, the voice meaning of the verb indicates the kind of change in the agent, the referent of the clause subject. The phrase Ivan ubiraet urožaj ʻIvan is harvesting the crops-Accʼ describes an independent action of Ivan (the active voice), while the phrase Urožaj ubiraetsja Ivanom ʻThe crops-Nom are harvested by Ivan-Insʼ describes a dependent action (the passive voice). § 3. The basic and derivative meanings of voice 197 There is one more category of changes that refer mostly to humans—selfinduced changes: a person performs an independent action, directing it at himself and causing dependent changes in himself (e.g. a boy runs a knife across his finger THEREFORE his finger starts bleeding). Such changes in the subject are expressed by the verb predicate in the reflexive voice: Mal’čik porezalsja nožom ʻThe boy cut himself with a knifeʼ. Thereby the reflexive meaning defines one more type of change in the subject of the predicate (the participant of higher communicative rank, in another terminology). Thus, there are three types of voice meaning that correspond to the three categories of change introduced above: 1) in the active voice the subject undergoes independent changes: Ivan bežit Reka tečёt Maša pričёsyvaet kuklu ʻIvan is runningʼ, ʻThe river flowsʼ, ʻMasha is combing the doll[ʼs hair])ʼ, Veter sorval kryšu ʻThe wind blew off the roofʼ, Kloun razmaxivaet kartonnym mečom ʻThe clown is wielding a cardboard swordʼ; 2) in the passive voice the subject undergoes dependent changes: Kukla pričёsyvaetsja Mašej ʻThe doll[’s hair] is combed-Refl by Mashaʼ, Dom razrušaetsja, lit. ʻThe house is destroying-Reflʼ (ʻThe house is going to ruinʼ), Voriška nakazyvaetsja plet’mi ʻThe pilferer is punished-Refl with a lashingʼ; 3) in the reflexive voice the subject simultaneously undergoes two types of changes—an independent change and a dependent change caused by the subject; in other words, the subject changes himself or itself: Maša pričёsyvaetsja ʻMasha is combing [her hair]ʼ. The semantic interpretations of voice offered above possess a universal, non-linguistic status as they make use of strictly cognitive concepts—the types of changes in the subject. Because of this, a single semantic basis can be used for cross-linguistic comparisons of voice meanings. It may also be claimed that the active, the passive, and the reflexive voice are all sensory units. Our next goal is to analyze two meanings of the postfix -s’/sja: the reflexive meaning (mal’čik moetsja, lit. ʻthe boy is washing-Reflʼ, čelovek breetsja, stroitsja, lit. ʻa person is shaving-Reflʼ) and the passive meaning (dom lit. ʻthe house is building-Reflʼ, bel’ё sušitsja, lit. ʻthe linen is drying-Reflʼ), and to demonstrate that they are derivative from the active voice (mal’čik moet mašinu ʻthe boy is washing the carʼ). More specifically, one could speak of the following sequence: the active voice—the reflexive voice (a metonymy derived from the active voice)—the passive voice (a metaphor derived from the reflexive voice). 4. Elements of a sensory grammar 198 2. The active voice A discussion of the meanings of the postfix -s’/ sja should start with a definition of basic meaning of a transitive verb, as it is the source of all other meanings. We will give a simplified concrete definition, in which the second interpretation (after the equation symbol “=”) plays a crucial role: it provides an explicit definition of the expression “the Agent’s action passes onto the Patient”. (1) Mal’čik moet mašinu ʻThe boy is washing the carʼ (active voice) = a transitive verb in the active voice designates a situation in which independent the subject in the role of Agent performs an and purposeful action; the action passes onto the direct object in the role of Patient and changes it = the subject in the role of Agent performs an independent and purposeful action; THEREFORE the direct object in the role of Patient simultaneously undergoes important dependent changes. 3. The reflexive voice: myt’sja ʻwash-Reflʼ, kutat’sja ʻmuffle-Refl up [in something]ʼ, brit’sja ʻshave-Reflʼ Most researchers view reflexivity as the expression “of a special type of coreference, namely, a full or partial coincidence of the object (or some other participant) and the subject of an action” (Geniushene and Nedialkov 1991: 246; see also Faltz 1985; Givón 1990). Wierzbicka (1996: 420) gives the following definition of prototypical reflexive meaning: I hypothesize that the prototypical meaning which, on a subconscious level, guides linguists in their actual use of the term “reflexive” can be represented as follows: (R) at some time, someone did something because of this, something happened to the same person at the same time. Our definition, given in (2) below, is similar to the definition in (R). However, the important distinction is that the meaning of the reflexive voice is interpreted as metonymically derived from the meaning of the active voice, (1). (2) Mal’čik moetsja (reflexive voice—metonymic transfer from the active voice) = a verb with the postfix -s’/sja designates a situation in which the subject in the role of Agent performs an independent and purposeful action, which passes onto the subject playing the role of Patient and changes it = § 3. The basic and derivative meanings of voice 199 the subject in the role of Agent performs an independent and purposeful action; THEREFORE the same subject, which also plays the role of Patient, simultaneously undergoes dependent changes. Let us ask ourselves: why do we easily understand that the subject in this case refers not only to the doer (Agent), as in (1), but also to the direct object? The answer “because they are physically combined in one person” is not good Mal’čik moet sebja ʻThe enough. In the clause boy is washing himselfʼ they are also combined, yet we understand the clause in a customary way: the subject (mal’čik) and the direct object (sebja) are independent, separate, and designated by the subject and direct object, respectively. This fact indicates that the difference in participant roles prevails over physical combination. It may Mal’čik moetsja be assumed that our understanding of the clause ʻThe boy is washing-Reflʼ is based in a universal mechanism of metonymic transfer. The subject in this clause metonymically designates two participants which have different roles but closely interact with one another as they are physically combined. It is similar to the case described by the clause Ivan celuju tarelku s”el, lit. ʻIvan ate a whole plate[ful]ʼ, where the form tarelku ʻplate-Accʼ is used metonymically and also refers to two different but closely interacting (because of physical contact) objects—a plate and its content. Note. Metonymy is usually described as the transfer of a name motivated by spatial or temporal contiguity (Reformatskii 1996: 47). This condition, however, is not sufficient. It is necessary that there be, between the source object and the contiguous object, a generally relevant functional interaction that links both objects. For example, one can use the word tarelka ʻplateʼ metonymically in Ja s”el tarelku syra ʻI ate a plate[ful] of cheeseʼ only in the case when the plate is filled with cheese cut into little pieces that fill the plate repeating its shape. If the very same cheese lies on the plate in one or two large chunks, such a metonymic use becomes incorrect. Small pieces of cheese are in a functional interaction with the plate, while a single large piece is not. Another peculiarity of a clause with a reflexive verb is that with the loss of direct object it loses the means to indicate what part of the Agent-Patient is involved in the action. Therefore, understanding the reflexive meaning of a verb depends on the knowledge of the typical part of the Agent involved in the Agent’s interaction with self. If there are several such parts, the priority is given to the largest part by default. The clause Mal’čik moetsja ʻThe boy is washing-Reflʼ indicates, by default, that the ʻentire bodyʼ is being washed, 4. Elements of a sensory grammar 200 without any explicit clarification or special support from the context. Therefore, the clause Mal’čik moet ruki ʻThe boy is washing [his] handsʼ does not have a reflexive meaning. Similarly, the clause Mužčina breetsja ʻThe man is shavingʼ (in a referential use) means, by default, that the man is shaving his face. If the man is bald but growing a beard, periodically shaving only his neck, he can say Ja brejus’ ʻI’m shaving-Reflʼ (= ʻI’m shaving my neckʼ), but only to those close to him, accustomed to such shaving. That being said, it becomes clear why the utterances such as Ženščina breetsja ʻThe woman is shaving-Reflʼ or ? Ivan čistitsja ʻIvan is cleaning-Reflʼ are questionable: it remains unspecified ? to what part of the agent the action is applied. For example, Ivan may be cleaning either his shoes, his suit, or his teeth. The clause Ivan krasitsja, lit ʻIvan paints-Reflʼ, is less opaque: most likely, it means that Ivan dyes his hair. 4. The passive voice: mašina moetsja ʻthe car is washing-Reflʼ, bel’ё sušitsja ʻthe linen is drying-Reflʼ, dom stroitsja plotnikom ʻthe house is buildingRefl by the carpenterʼ Let us first consider the function of the passive voice compared with the active voice. Comparing the following sentences: (a) Bol’šinstvo teoretikov otverglo ètot argument. ʻThe majority of theoreticians have rejected this argumentʼ (b) Ètot argument byl otvergnut bol’šinstvom teoretikov. ʻThis argument has been rejected by the majority of theoreticiansʼ Plungian (2011) argues that the difference between them “consists in the fact that in (a) the speaker, most likely, wants to make some assertion about the theoreticians, while in (b) an assertion is made about the argument. […] This pragmatic distinction has a formal correlate: clauses (a) and (b) have different subjects” (pp. 183, 187). While concurring with this argument, we would like to point out the following. The passive voice is intended to ascertain the state of the Patient (either intermediate or final), namely, that “the Patient is in a state of change caused by the action of some Agent (usually unknown or absent at the moment of observation)”. Let us clarify our claim. Imagine that a person has noticed that a car that was dirty not long ago is now clean. Clearly, the car must have been washed, but the person doesn’t know who did it. The passive voice allows the speaker to report only the observed result (final state of the car): Mašina vymyta ʻThe car [has been] washedʼ, without any indication as to who the doer of the action was. One can also say Mašinu vymyli, lit. ʻThe car [they] washedʼ, but in this case it is not the state (ʻclean carʼ) that is reported, but a 4. Elements of a sensory grammar 202 metaphoric extension of the reflexive) = ‘The subject of the predicate, which plays the role of Patient, is in a state of dependent change caused by an independent action of another participant—Agent or Cause ’ =  7 The subject of the predicate, which plays the role of Patient, is in a state of dependent change BECAUSE another participant (Agent or Cause) performs an independent action. Shakhmatov (2006: 95–96) pointed out a group of verbs that have a passive meaning when used with an animate subject: obvinjaetsja ʻ[is] accusedštrafuetsja fined-Reflʼ, ʻ[is] punished-Reflʼ, etsja ʻ[is] denounced-Reflʼ, vyzyvaetsja ʻ[is] summoned-Reflʼ and so forth; apparently, these are the verbs that have an unspecified collective as the doer of the action, with regard to which a single individual may become an object. Definition (3) still holds for them; it is just that the doer of the action here is a third force—neither Agent nor Cause, but Collective, to which the subject is in some social relationship. 7 Here, Cause is “a participant (typically, inanimate or acting unconsciously) that is the cause of the situation (dožd’ zatopil posevy [ʻthe rains flooded the cropsʼ]; strax gnal ego v put’ [ʻfear drove him onʼ]” (Plungian 2011: 116; original emphasis). Chapter 5 On the single structure of lexical meanings of nouns and verbs § 1. Object and the system of its parts (the partitive concept) 1. Object and its parts As has been noted previously, a native speaker does not experience any difficulty in subdividing objects into parts and naming these distinct parts. The explanation of this human ability is far from trivial. In this paragraph I will  1 reiterate some of the conclusions from chap. 4, § 1 in order to include them in a different system of reasoning. In that paragraph and elsewhere (Koshelev 2006; 2008: 35–38; 2011a: 23–24, 28; 2015a: 25–113) it has been demonstrated, using several examples, that the objects around us, both artifacts (such as chairs, bicycles, etc.) and natural objects (such as lakes, rivers, bananas, trees, etc.) possess specific functions that reflect the human interpretation of the essence of these objects, that is, how such objects “work”: a chair allows a person to rest in a sitting posture, a tree grows and bears fruit, etc. Because of this, the object concept that defines the mental image of an object has a dual structure: “Prototype → Function”. Parts of objects possess similar structures, “Prototypical Physical Part → its particular Function”. It has also been shown 1 It appears that a detailed discussion of the range of issues pertaining to this problem was attempted for the first time in Ivan Sechenov’s seminal work The Elements of Thought, first edition of which was published in 1878 in St. Petersburg in the journal Vestnik Evropy ‘The European Bulletin’; herein it is cited from a later edition (Sechenov 1952). The current approaches to the problem are discussed in a review article by Tversky et al. (2008). For a comparison of my view with the approach of Tversky and co-authors see the end of section 8. 5. On the single structure of lexical meanings of nouns and verbs 204 that objects (concepts) are mentally represented as sets of object parts, that is, in a partitive hierarchy. Consider the following example: Here, the object concept CHAIR is constituted by the pair “Prototypical Chair → its Function”: ʻallows an individual to take a sitting postureʼ; and the parts BACK, SEAT, and LEGS are constituted by similar pairs: “Prototypical Part → the Part’s Function”. In what follows, an abridged notation for the partitive hierarchy will be used: CHAIR ⇒ (BACK, SEAT, LEGS). The arrow ⇒ (or ) indicates the ⇓ “Whole ⇒ Parts” relationship. Here are some examples: LAKE ⇒ (BOTTOM, WATER, SHORE) BANANA ⇒ (SKIN, FLESH, FRUIT STEM) KNIFE ⇒ (HANDLE, BLADE) NUT ⇒ (SHELL, KERNEL) BICYCLE ⇒ (WHEELS, FRAME, HANDLEBARS, SADDLE, PEDALS) RIVER ⇒ (BANKS, WATER FLOW, RIVERBED, MOUTH, SOURCE) TREE ⇒ (ROOTS, TRUNK, BRANCHES) 2. The function of an object as the sum of the functions of its parts It was also shown in chap. 4, § 1, that an object is subdivided into parts according to the object’s general function. The division of an object into functional parts marks the next, and much deeper, level of comprehension of the object. We intuitively understand the role of a separate part of an object and its contribution to the general function of the object. This general function, therefore, appears to be the sum of the particular functions performed by separate parts of an object. For example, the seat of a chair has one particular function (it ‘serves as a support for the buttocks of a sitting person’), the back another (it ‘serves as a support for the back of a sitting person’), and the legs a third (they ‘support the seat of the chair—and, therefore, the sitting person—at a required level above the ground’). Thus, we have three functional parts of § 1. Object and the system of its parts (the partitive concept) 205 Their particular functions colleca chair, the seat, the back, and the legs. tively constitute the general function of a chair. Similarly, the general function of a lake (ʻa continuously sustained large mass of still fresh water covering an area of ground, directly available for human perception and use’) consists of three functional parts: ʻa large mass of still fresh waterʼ, ʻthe shores within which the water is heldʼ, and ʻthe bottom that prevents the water from drainingʼ. Similarly, the general function of a banana (ʻa ripe fruit of a banana tree used as foodʼ) allows for its division into three functional parts: ʻthe flesh of a banana, consumed as foodʼ, ʻthe banana skin that protects the flesh from birds and insectsʼ, and ʻthe fruit stem through which the flesh receives nourishmentʼ. Soon after the age of 24 months, children learn the nominal genitive Y + X-Gen constructions of the type stvol dereva ‘the trunk of the tree’, dver’ komnaty ‘the door of the room’, etc. This type of construction expresses the relationship ‘Y is a functional part of X’; as is observed by Tseitlin (2000: 135), “at the age of 24 to 36 months, children learn […] constructions with the meaning ‘part—whole’.” However, as has been shown in chap. 4, § 1, the noun, Y, in these constructions refers not to just any part, but only to a functionally relevant part of the object, X; this functionally relevant part contributes directly to the general function of the object. For example, in Russian the expression ručka dveri ‘handle door-Gen’ (‘the handle of the door’) is quite correct as the handle performs a specific function attributed to the door—it is used to open or shut the door. By contrast, the expressions *glazok dveri ‘the peephole of the door’ and *krjučok dveri ‘the hook of the door’ (about a hook nailed to the door) are incorrect. One should say dvernoj glazok ‘the door peephole’ / dvernoj krjučok ‘the door hook’, because the peephole and the hook possess their own separate functions not related to the function of the stvol dereva door. Similarly, the expression ‘the trunk of the tree’ is correct, while *gnezdo dereva ‘the nest of the tree’ (about a nest made in the tree) is not; the nest does not take part in the vital functions of the tree. It would appear, then, that the nominal genitive cannot be learned without an understanding of the partitive structure of an object (a door, tree, etc.). Thus, it may be assumed that children understand the functions of objects around them, and the functions of object parts, early on. 3. The role-based hierarchy of object parts Let us now try to answer the following question: is the composition of an object’s functional parts something unique? May there be two or more alterna- 5. On the single structure of lexical meanings of nouns and verbs 210 5. How is the main part of an object determined? Sometimes the question of which functional part of an object is the main part cannot be easily answered. For example, it could be argued that the roots of a tree are more important than the trunk or the seeds of an apple more important than its flesh. There is a simple naming method to determine the main part of an object: only the main part can be referred to by the name of the object while any other part cannot. For example, the word nut can be used to refer either to a whole nut or its kernel (the main part), but not to the nutshell. The blade of a knife without a handle can be called a knife, but a handle without a blade cannot. The roots of a tree cannot be called a tree, but a trunk without roots and branches can. The flesh of an apple can be called “apple”, but its seeds or skin cannot. We can imagine something fantastic, such as a wide horizontal water flow sustained in midair by an invisible gravitational channel; in such a case, we can call this water flow a river even though we cannot see its bottom or bed. It may be presumed that the credibility of the results obtained is increased by the mutual support of these methods from two separate knowledge areas: cognitive rules for singling out the main part of an object and linguistic rules for its denotation. 6. The functional and physical parts of an object The functional parts of an object should be strictly differentiated from its physical parts that are visually distinguished due to their specific shape. For example, a pump which has been attached to the frame of a bicycle, while being physically attached to the bicycle, is not a functional part (it is not directly involved in the process of cycling). A bird’s nest is not a functional part of the tree in which the nest has been made (it does not play any role in the vital functions of the tree). A fish is not a functional part of the lake in which it swims (the fish does not affect preservation of water in the lake). The Russian nominal genitive, discussed above, helps to determine whether a given physical part of an object (a part that is in contact with the object) is a functional part of that object. When it is not, as in the case of the bicycle pump, bird’s nest, fish in a lake, etc., the nominal genitive construction becomes incorrect, cf.: *nasos velosipeda pump bicycle-Gen ‘the pump of the dereva *ryba ozera bicycle’, *gnezdo nest tree-Gen ‘the nest of the tree’, fish lake-Gen ‘the fish of the lake’, etc. Some clarification should be made here. The relationship “Whole → functional Part” is not a transitive one. If Y is a functional part of X and Z a functional part of Y, then Z is not a functional part of X because its function is at- § 1. Object and the system of its parts (the partitive concept) 211 tributed to Y and does not exist outside of Y. In such cases the nominal genitive is incorrect. One cannot say *spicy velosipeda spokes bicycle-Gen ‘the spokes spicy kolesa velosipeda of the bicycle’ (the correct phrase is spokes wheelGen bicycle-Gen ‘the spokes of the bicycle wheel’), because the spokes are a functional part of the wheel. Therefore, the partitive structure of a functional part of an object is independent; it does not depend on the partitive structure of the object as a whole (for more details, see chap. 4, § 1). 7. On the radial position of the parts of an object The layout of the spatial positions of the functional parts of an object (in the case of a simple artifact or natural object) has one noteworthy property. This layout is similar to that of a flower: the main functional part is always at the center, like a pistil, and the supplementary parts surround it and are physically connected to it, like the petals. For example, the water in a lake takes the central position while the bottom and shores surround it, the flesh of a banana is contained within while the skin and fruit stem surround it, and so forth. This radial contact positioning of supplementary parts has a simple explanation. The function of every supplementary part must be added to the function of the main part. And this happens only when there is direct physical contact between the supplementary parts and the main part (on account of the property of intransitivity of the parts’ functions, as discussed above). Other physical parts may have contact with the main part even though their functions do not contribute to the main function (for example, a bicycle pump attached to the frame of a bicycle does not contribute to the function of the frame, neither does a tool bag attached to the saddle). However, if the function of a certain part contributes to the function of the main part of the object, this specific part must have physical contact with the main part. In view of the aforesaid, the spatial positioning of the parts of a partitive –N–> part concept may be described as follows: “main (central) supplemen–N–> (contact) parts”, where stands for the role relationships. Based on this characteristic, a developed object concept has the following structure: (7) Developed concrete CONCEPT = a. Basic-level concept: OBJECT ◊ –N–> b. Partitive concept: (Main part (in the center) Supplementary parts (around and in contact with the main part))