This book "supplies a vocabulary of English words and idiomatic phrases 'arranged … according to the ideas which they express'. The thesaurus, continually expanded and updated, has always remained in print, but this reissued first edition shows the impressive breadth of Roget's own knowledge and interests."
Although a number of monographs have been published on specific types of noun classification or categorization, particularly gender (e.g., Corbett 1991), or on specific languages, such as Japanese (Downing 1996), this book, originally published in hard cover in 2000, is the first thorough worldwide survey of such systems. Despite the title, therefore, it is not restricted to what are generally known in the literature as 'classifiers' (usually, 'noun classifiers' and 'numeral classifiers', two subtypes in her usage). In the book the ''term 'classifier system' refers to a grammatical system of noun categorization device(s) in a particular language'' (vii). A's approach is traditional comparative typology, and she has aimed at comprehensive coverage: she includes examples from all relevant languages known to her from the literature or her own fieldwork. The book is intended to be ''an up-to-date introduction to the field'' for linguists, anthropologists, psychologists, and others, and can be used ''both as a sourcebook for future typological studies, and as a textbook'' (vii). As it also includes copious new data and insights from A's extensive fieldwork on American and Australian languages, the book is a major contribution to linguistics in general.
The introductory first chapter, ''Preliminaries,'' gives an overview of the book's coverage and introduces the author's theoretical framework-general linguistic theory--and her approach to the study of noun categorization systems, which will henceforth, following the author's usage, be called 'classifiers'. She explains that the distinctions she makes among subtypes are based largely on ''the morphosyntactic loci (or environments) of classifier morphemes'' ranging from ''the highly grammaticalized gender agreement classes of Indo- European languages'' to ''the lexical numeral classifiers of Southeast Asia'' (2-3). Her classifier subtypes ''correspond to prototypes, or focal instances, which display all the definitional and most of the contingent properties of a type'' (14).
Chapter 2 covers ''noun class and gender systems,'' such as those of Romance or Bantu languages, which ''correlate--at least in part--with certain semantic characteristics (particularly in the domain of human and animate referents)'' and ''are sometimes called concordial classes'' (19). Chapter 3 covers ''noun classifiers,'' which co-occur in the same noun phrase with the nouns they classify semantically. They are best known from American and Australian languages. Chapter 4 covers ''numeral classifiers,'' the most well-known type of classifier and the most widely represented among some of the more populous languages of the world, including Chinese, Japanese, and many Southeast Asian languages. Chapter 5, ''Classifiers in possessive constructions,'' covers a relatively rare type found mainly in Austronesian and American languages. Chapter 6 covers ''verbal classifiers,'' including three subtypes: ''classificatory noun incorporation, whereby a noun is incorporated into a verb to categorize an extra-predicate argument,'' ''verbal classifiers affixed to the verb,'' and ''suppletive 'classificatory' verbs'' (149). Chapter 7 covers ''locative and deictic classifiers.'' A considers this type, which is based almost exclusively on data from Palikur (North Arawak), mainly from her own fieldwork, to be rare and therefore not firmly established (172). Chapter 8 covers the problem of ''different classifier types in one language.'' Interest in this topic was piqued by Dixon's theory of the typological distribution of noun classification types (Dixon 1986), which--though now considered to be disproven (10)-- stimulated a great deal of interesting work on classifiers.
A's proposal to base a new typology ''on the morphosyntactic locus of coding of noun categorization devices (together with their scope of categorization, principles of assignment, and the kind of surface realization)'' is supported in this chapter by her discussion of the distribution of different classifier types within the same language. Among the many interesting examples given in this chapter are some from Minangkabau (Western Austronesian), about which A notes, ''A few morphemes can be used both as numeral classifiers and as noun classifiers, with a difference in meaning. Batang, when used as a noun classifier, means 'trees as a class'. As a numeral classifier, it is used to refer to long vertical objects, e.g. trees. As an independent noun, it means 'tree trunk''' (190). Numeral classifiers and noun classifiers can be used in the same noun phrase, but have different meanings.
Example (1), quoted from A (190) but slightly modified for clarity here, has both. (N.B.: NUM.CL = numeral classifier; NOUN.CL = noun classifier)
This example is of course nearly identical to one of the famous, supposedly stereotypical examples of classifier phrases invented by John Lyons, *three fruit banana = 'three bananas' and *three tree banana = 'three banana trees' (Lyons 1968: 288). A strong point of this chapter is its discussion of Palikur (North Arawak), which has at least five noun classification systems, including genders and several kinds of classifiers (192-198). Chapter 9, on ''multiple classifier languages,'' covers those in which ''the same, or almost the same, set of morphemes can be used in more than one classifier environment'' (204). The above example from Minangkabau also illustrates this situation, but some of the languages discussed in this chapter, especially Baniwa and Tariana (both North Arawak languages), are extremely complex (230-241). Chapter 10 covers the interaction of classifiers with other grammatical categories, including number (243- 252), person (252-255), grammatical function, i.e., ''nominal case (in dependent marking languages)'' or ''verbal cross- referencing (in head-marking languages)'' (255-257), possession (257-259), politeness, where very brief reference is made to distinctions in first and second person pronouns (260-262), declensional classes (262-263), verbal categories (263-265), deictic categories (266), and derivation and the lexicon (266-268).
Chapter 11, ''Semantics of noun categorization devices,'' and Chapter 12, ''Semantic organization and functions of noun categorization,'' cover one of the main topics of classifier studies in general, well known from the title of George Lakoff's book _Women, fire and other dangerous things_ (1987), which refers to the semantics of one noun class. As A says, ''Semantic features encoded in noun categorization reflect principles of human cognition and world perception. This is why it has often been argued that classifiers offer 'a unique window' into studying how humans construct representations of the world and encode them into their languages'' (307). Chapter 13, on the ''Origin and development of noun categorization devices,'' analyzes the birth, development, and death of classifiers and classifier systems. Chapter 14, ''Noun categorization devices in language acquisition and dissolution,'' focuses on child speakers. Chapter 15, ''Conclusions,'' summarizes the themes and findings of the book. There are three appendices: ''Noun categorization by means other than classifiers''; ''From noun to classifiers: further examples of semantic change'', including (a) body part terms as sources for classifiers and (b) sources for configurational (''shape-based'') numeral classifiers; and a ''Fieldworker's guide to classifier languages.''
Since the publication of _Noun Classes and Categorization_ (Craig 1986), classifiers in usual parlance (i.e., excluding some of the types A includes under the term) have become a fairly well-known type of linguistic category. One of A's stated purposes in the book is to clarify terminological usage in the field of noun classification studies, because ''there exists a pervasive terminological confusion in the literature'' (1). She also states that the terminology she uses ''for each classifier type relies as much as possible on currently accepted terminology'' (13). However, while an umbrella term for all grammatical noun categorization devices might be convenient, her choice of the term 'numeral classifier' for the 'classifier' of usual parlance, and extension of the term 'classifier' to virtually all noun categorization devices--including even measures, which she calls ''mensural'' classifiers, even though she very thoroughly explains why they are _not_ classifiers (114-120)--would seem to increase, rather than decrease, the confusion in the field.
Instances of such confusion actually occur in the book, such as in her discussion of the previous literature, where she says, ''it is not always clear what is a classifier and what is a concordial noun class in each particular case'' (10). Significantly, she uses the terms 'noun class' and 'gender' (rather than 'classifier') throughout chapter 2, which is devoted specifically to the concordial type of class marking. This is a fortunate distinction (19) that I wish she had extended to the introductory material and (even) the title of the book.
The author discusses pronominal classification, such as with the English personal pronouns (which in third person singular distinguish three noun classes) in her discussion of gender in Chapter 2. She divides her discussion of 'noun classes' into two types of system, one which ''is used with personal, demonstrative, and other pronouns, and for verbal cross- referencing; this is called 'pronominal gender/noun class','' and one which ''is used with adjectives (and sometimes other modifiers, such as numerals); this is called 'nominal' gender/noun/class''' (68). There is a rather sharp distinction between these two types and their distribution--for example, gender is usually concordially marked, whereas pronominal classification, as in English, usually is not--as she points out in some detail (68). She also notes the existence of a ''rare'' type of classifier, ''locative and deictic classifiers'' (12), to which she devotes much of Chapter 7. It is difficult to understand why she does not include 'pronoun classifiers' together with locative and deictic classifiers in one category; ''deictic classifiers'' would not then be so rare.
In keeping, perhaps, with her emphasis on American and Australian languages, her coverage of numeral classifiers-- which are generally considered to be stereotypical 'classifiers' par excellence, and which are the subject of the majority of published typological and theoretical studies of classifiers--is not as thorough as other parts of the book. Though she has cited many studies of this type of classifier, including typological surveys such as that in Pamela Downing's book on Japanese classifiers (1996), she overlooks some important points. For example, in the discussion of numeral classifier semantics in Chapter 12, the fairly well-understood distinctions among configurational (or 'shape') classifiers, heterogeneous classifiers, and taxonomic (or 'kind') classifiers, which are generally described as segments on a continuum of types, are largely ignored. As for the analysis of classifier semantics, she says ''a taxonomic approach to classifiers may be useful, but only in limited circumstances'' (317). However, her primary approach, prototype theory, is not applicable to the vast majority of classifiers in Japanese (Downing 1996), so it is hardly better than the taxonomic approach. Since the same difficulty applies to most other languages with similar numeral classifier systems, it would appear that semantic analysis of classifier 'classes' remains problematic. The chapter also suffers from the nearly total omission of group classifiers, which tend to occur in languages that have few if any other classifiers, such as English (among other languages such as Russian), where they are generally referred to, misleadingly, as 'collective nouns'. She says of them only, ''quantifier constructions in English _three heads of cattle_ are in fact a subtype of genitive constructions'' [sic--CIB] (116), but the example is wrong (read _three head of cattle_; _head_ is well-known as perhaps the only true unit classifier in English), and so is the analysis.
Although A includes some discussion of the parameters of classifier choice, she does not stress the importance of the fact--for typological analysis, at least--that classifier choice per se is possible at all. In concordial gender languages such as Latin, agreement marking is generally obligatory due to the portmanteau nature of the bound fusional morphemes that mark gender. Changing a morpheme to reflect the speaker's views about the noun referent would more often than not also change the case or number marking as well--though it is not true that ''in a noun class system, every noun has to be assigned to a class'' (334), as there are indeclinables in Arabic, Latin, and most, if not all, other 'noun class' or 'gender' languages. In isolating languages, by contrast, classifiers are relatively free morphemes, and speaker choice is inherently possible. It has also been widely noted (but is not mentioned by A) that a speaker's use of different classifiers--or of a wide variety of classifiers--in such languages is generally considered by other speakers to be a sign of linguistic and cultural expertise, and use of too limited a number of classifiers a sign of a lack of such expertise.
Cases of multiple classification systems in individual languages occur in complementary distribution (in different environments), as she shows amply in the book. This is a major contribution to the field. The many arguments in the earlier literature about multiple classification systems somehow disproving the existence of clear typological distinctions in noun classification systems would now seem to be moot. On the other hand, she also gives examples of languages in which classification systems seem to overlap in the same language, so that ''two noun classes can be marked within one morphological word'' (76). One little-known language, Paumar� (Araw�), is given as the only example of conflicting overlap within a language. She devotes considerable discussion to it (71-75), but it is far from clear from her presentation (which in this instance happens to be far from clear) that there is any significant overlap or conflict among the various systems noted or implied by her presentation, including the examples. Since none of the Paumar� examples she gives mark two noun classes within one morphological word, it would appear that even Paumar� is not an example of the putative phenomenon.
In her discussion of multiple classifier languages in Chapter 9 the author mistakenly states that in Mandarin Chinese the so-called ''generic classifier'' _ge_ ''is written as a separate word in [''_s�n ge_ three CL 'three'''], where it is used as a numeral classifier, but in [ ''_n�i-ge_ that-CL 'that'''] it is written as a suffix to the demonstrative. This reflects the intuition of speakers; they have a 'feeling' that a classifier forms a closer unit with a demonstrative than it does with a numeral'' (208). In fact, _ge_ is actually written in _pinyin_ (the official Chinese transcription system used in her examples) as part of the same word with a numeral or a demonstrative, e.g., _s�nge_ and _n�ige_. There are several reasons for this, the most salient one being the fact that when _ge_ is used in the classifier slot of a classifier phrase in normal speech it loses its tone, marking it as a clitic.
The book includes several maps showing distribution of various phenomena, but more care should have been taken with their preparation. In the case of Map 1, ''Distribution of noun classes and genders in the languages of the world'' (78), ''noun classes and genders'' are shown not only in most of the Indian subcontinent west of Bengal but also across Tibet into Eastern Turkistan (Xinjiang). However, Tibetan should have been marked as having ''fossilized gender markers'' (one of the three categories marked on the map), while its Himalayan relative Limbu has functional adjectival gender markers, and by contrast Uighur (the native language of most of East Turkistan) has no grammaticized noun class or gender marking at all. On the other hand, Ket, the last remaining member of a Siberian language family, has three genders, as A notes in the text (77), but it is not marked on the map.
With regard to Map 2, showing world distribution of ''noun classifiers'', despite the author's statement in the text that Taic languages and Tibetan have noun classifiers (82, 97), only the area of Thailand is so marked on the map. Similarly, on Map 3, showing world distribution of ''numeral classifiers'' (122), Japanese is unmarked for such classifiers, despite extensive discussion of them throughout the book. Also, the text refers (121) to the classifiers of Hungarian (Uralic) and ''many Turkic languages'' (though so far only Uzbek has been noted to have a grammaticized noun classification system), but these languages are not marked on the map as ''isolated instances of numeral classifiers.'' Map 5, showing distribution of ''verbal classifiers,'' does not include Ojibway (Algonquian) and the Tibeto-Burman languages discussed in the text (154-156, 297); the text accompanying the map unfortunately reinforces these two oversights with the omission of Algonquian (169) and with the flat statement, ''There are no verbal classifiers in the languages of Africa or Eurasia or in the Austronesian family.''
Generally speaking, due to the choices of things to mark on them, the maps give the false impression that most languages of the world have no noun classification systems, though the text, by contrast, correctly notes that very few languages lack noun classification. Historical implications are also made, probably unconsciously, by the maps. For example, Russian, a gender language frequently mentioned in the book, was spread in all directions by late medieval Muscovy, so that Russian is now the dominant or sole language spoken in most of the vast Russian Republic. Yet Map 1 shows noun classes and genders extending northward from Africa through the area of Turkey (which has none), western Russia, and Finland (though colloquial Finnish has no noun classification whatsoever, as the author notes in the text), to the Arctic Ocean; the majority of Russia is blank. By contrast, the area of Manchuria is included in the ''numeral classifier'' language area of East Asia on Map 3, but it has been Chinese speaking for only about a century. Since graphic devices are especially vivid communicators of information, I hope that if this edition of the book is reprinted the maps will be revised to at least reflect the data in the text.
The volume has been attractively produced and is relatively well edited. I recommend A's book to linguists in general, and especially anyone interested in language typology or cognition.
Beckwith, Christopher I. 1997. Toward a comprehensive theory of noun categorization, with special attention to Thai. In U. Warotamasikkhadit, ed. _Proceedings of the fourth annual meeting of the Southeast Asian Linguistics Society (1994)_. Tempe: Arizona State University, Program for Southeast Asian Studies, 73-82.
Corbett, Greville 1991. _Gender_. Cambridge: Cambridge University Press.
Craig, Colette G., ed. 1986. _Noun classes and categorization_. Amsterdam: John Benjamins.
Dixon, R.M.W. 1986. _Noun classes and noun classification in typological perspective_. In Craig 1986: 105-112.
Downing, Pamela 1996. _Numeral classifier systems: The case of Japanese_. Amsterdam: John Benjamins.
Lakoff, George 1987. _Women, fire and other dangerous things: What classifier categories reveal about the mind_. Chicago: University of Chicago Press.
Lyons, John 1968. _Introduction to theoretical linguistics_. Cambridge: Cambridge University Press.
ABOUT THE REVIEWER
Christopher I. Beckwith is a professor in the Department of Central Eurasian Studies (formerly the Department of Uralic and Altaic Studies) and the Department of Near Eastern Languages and Cultures at Indiana University. His interests are in classifiers, phonology, and comparative-historical linguistics. Most recently he has worked on Japanese classifier partitions, the extinct Koguryo language and its relationship to Japanese (Japanese-Koguryoic), Old Tibetan syllable margins, and the reconstruction of Early Old Chinese.</PRE>
ABOUT THE REVIEWER:
ABOUT THE REVIEWER Christopher I. Beckwith is a professor in the Department of Central Eurasian Studies (formerly the Department of Uralic and Altaic Studies) and the Department of Near Eastern Languages and Cultures at Indiana University. His interests are in classifiers, phonology, and comparative-historical linguistics. Most recently he has worked on Japanese classifier partitions, the extinct Koguryo language and its relationship to Japanese (Japanese-Koguryoic), Old Tibetan syllable margins, and the reconstruction of Early Old Chinese.