Review of Feature Distribution in Swedish Noun Phrases by Kersti Börjars

ISSN: 0726-8602 (Print) 1469-2996 (Online) Journal homepage: https://www.tandfonline.com/loi/cajl20 Book Reviews To cite this article: (2002) Book Reviews, , 22:2, 265-280, DOI: 10.1080/0726860022000021789 To link to this article: https://doi.org/10.1080/0726860022000021789 Published online: 09 Jun 2010. Submit your article to this journal Article views: 60 Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=cajl20 Australian Journal of Linguistics, Vol. 22, No. 2, 2002 Book Reviews Fixed expressions and idioms in English: a corpus-based approach ROSAMUND MOON Oxford: Clarendon Press, 1998, 338pp. Reviewed by KOENRAAD KUIPER In her modest conclusion in Chapter Eleven, Moon claims that she has established that: more corpus based studies are needed in order to make the picture of FEIs (Žxed expressions and idioms) more accurate and complete. (p. 309) existing models and descriptions need to be revised in the light of emerging corpus evidence, in particular with respect to form and variation. (p. 309) the signiŽcance of the roles of FEIs in discourse should not be underestimated. (p. 309) The preceding ten chapters have certainly done that but they have done a great deal more. Moon’s investigation of what she calls ‘Žxed expressions and idioms’ was performed primarily on an 18 million word corpus. Work in phraseology has not often turned to corpora in the past, but partly for lexicographic reasons it increasingly does so (Aimer 1996; Altenberg 1998). Moon’s view is that ‘[c]orpus data provides the opportunity to corroborate or modify theoretical models.’ (p. 1) To that end she provides an excellent literature review to indicate just what theoretical models there are. And they are many. Her Žrst task is to survey the terminology in the Želd. That is not easy since it is a mess. Many terms are used for the same theoretical units; sometimes the same term is used differently by different analysts; sometimes a term is used ambiguously by the same scholar. For example, the term idiom may be used to indicate that a unit is non-compositional in meaning. It may also be used to designate any lexicalised phrase, compositional or not. Idiomatic can be used to indicate non-compositionality or what Pawley and Syder (1983) term ‘native-like’ performance. Moon’s deŽnition of lexicalised phrases or phrasal lexemes is interesting. She views such units as having three central properties: they are institutionalised, lexico-grammatically Žxed, and non-compositional. She sees these as being ‘the criteria by which the holism of a string can be assessed.’ (p. 6) The Žrst of these conditions has to do with the licencing by a speech community of a lexical unit and thus the criterion of institutionalization generalises across to single word lexemes. Non-compositionality also generalises across all lexical items since many morphologically complex words are non-compositional. Lexicogrammatical Žxedness is a problem area. In fact most Žxed ISSN 0726-8602 print/ISSN 1469-2996 online/02/020265-16 Ó 2002 The Australian Linguistic Society DOI: 10.1080/0726860022000021789 266 Book Reviews expressions are not all that Žxed, as Moon points out in Chapters Žve and six. But all of them are Žxed in some way or other. But these ways also do not generalise across all phrasal lexemes. Perhaps what typiŽes phrasal lexemes is that they have idiosyncrasies of one sort or another. That is enough, according to traditions as old as BloomŽeld (1933), to have them entered in the lexicon, i.e. they have to be learned. What is left, then, is that phrasal lexemes are phrases entered into the lexicon because they are known by native speakers, that being the minimal idiosyncrasy required to enter the item. Other idiosyncrasies might be in any aspect of their representation. They might be non-compositional, syntactically ill-formed, collocationally restricted etc. Phrasal lexemes have been a happy hunting ground for taxonomists. Almost everyone who has looked at the Želd has tried to make distinctions between different types of phrasal lexemes on a variety of criteria. Often they have not noticed that the criteria for making distinctions cross-classify. Which criteria seem important differ from analyst to analyst. For some, the distinction between compositional and non-compositional phrasal lexemes is signiŽcant; for others their syntactic category whether phrasal or clausal. Still others try and distinguish different conditions of use, for example gambits, which are a variety of discourse marker (Schiffrin 1987), from non-gambits (Keller 1981). Moon does a valuable service in accounting for much of this taxonomic labour and then she creates her own taxonomy (pp. 19–25), a thoughful bestiary of the genus phrasal lexeme. However there are two issues with such an endeavour. If one is concerned solely with utility, then a taxonomy is good if it serves the ends of the study. But for realist approaches to linguistics there is a further desideratum. A taxonomy should arise out of a theory which explains the nature of the phenomena. Phrasal lexemes are lexical items. They should thus have the properties predicted of lexical items. Unlike single word lexical items they have the structure of phrases. Their additional properties should arise out of the syntactic properties of phrases and the way these might be represented in a model of the lexicon. These latter concerns arise from the fact that phrasal lexemes constitute part of the competence of a native speaker, i.e. they are part of what a speaker knows when they know a language as a native speaker. There are also performance properties having to do with the use speakers make of phrasal lexical items. These are of two clear kinds. Some usage conditions of phrasal lexemes are part of what the speaker knows about the item. For example, a native speaker knows that the lexeme shit can be used as an exclamation of annoyance. In the case of phrasal lexemes such conditions of use can often be detailed (Kuiper and Tan 1989). Speakers who work at fast food outlets know that Have a nice day is not a greeting but a farewell. A second set of performance properties have to do with what speakers actually do with their phrasal lexemes in producing speech and writing. Some of what they do will be governed by the conditions of use stored as part of the lexical entry for the items. Other aspects will not. For example, phrasal lexical items lend themselves to being used for verbal humour, as Moon demonstrates in Chapter ten. That is not part of native speaker competence in the Chomskian sense. Nevertheless, Moon’s taxonomy and its supporting literature are a very useful contribution to phraseology. Chapter two deals with collocation. Collocation is a contested term like so many others in phraseology. Moon produces a useful taxonomy in this area. She says Book Reviews 267 ‘Collocations are surface, lexical evidence that words do not combine randomly but follow rules, principles, and real world motivations’ (p. 26), and then divides collocations into three varieties: collocations created by selectional restrictions, lexicogrammatical plus selectional restrictions such as rancid collocating with butter and the third type arising through subcategorization properties of phrasal heads in the way they select for certain complement types. It is clear that, deŽned in this way, collocation is an epiphenomenon arising from properties of the grammar that speakers have acquired and the lexical properties of words as they project grammatical properties. Moon also recognises anomalous collocations of various kinds. These are not epiphenomenal since they have to be learned by native speakers and are thus part of their lexical knowledge. Native speakers know that one gets on the bus not into the bus. Both are grammatical but one is institutionalized and known by native speakers. The other is not. Chapter two also contains a short section on the psycholinguistics of speech production and perception and the role phrasal lexemes play in both, and a section on lexicalization, the processes by which phrasal lexemes gain the property of being institutionalized. It is a thought-provoking question as to whether particular phrases are, or are not institutionalized. For example, in her discussion of prepositional phrases headed by under (p. 38), it is an interesting question whether under orders is or is not known by native speakers. Moon points out that the preposition does not have its literal meaning in such an expression but is still compositional on a metaphorical reading. Intuition is very clear here. The expression under orders is either known or not known by native speakers regardless of its compositionality. For a corpus linguist who does not wish to use intuitive data, the problem of how to establish institutionalized status remains. In the last section of the chapter Moon briey looks at diachronic properties of phrasal lexemes. Since they are lexical items, they both exist and persist (Aronoff 1976). That being so they have a history like other lexemes. They are coined, gain currency and in some cases change and become archaisms or vanish from current use. Clearly some are loan translations such as long time no see which is a Chinese four character idiom greeting. Moon points out that some biblical sayings were originally in a language other than English and came into English through translation. Many enter the language from various text forms such as cinema and television, e.g. Go ahead, make my day. Chapter three deals with the way in which corpus data and computation manipulation have yielded the data and results of the succeeding chapters. Since Moon was one of the researchers who produced the Collins Cobuild English Language Dictionary (CCELD) she uses phrasal lexemes from that work selecting just under 7,000 for study. This is interesting since she says ‘the complete set of (phrasal lexemes) of English … is uncharted, unquantiŽed, and indeterminate’ (p. 44). All three assertions are correct. However guesstimates of the number of phrasal lexemes a native speaker knows are possible. Pawley and Syder (1983) guess over 200,000. If that is anywhere near accurate then 7,000 is a relatively small subset. But one has to start somewhere and 7,000 is not a bad number. The items were annotated in 17 Želds for various properties. The text data which was interrogated for the appearance of the 268 Book Reviews target lexemes was the Hector Pilot Corpus of which Moon says ‘[t]here was far too little spoken data and far too great a proportion of journalism.’ (p. 48) Also Moon recognises that many phrasal lexemes ‘have different distributions in different genres’ (pp. 48–9). So this raises the obvious point as to what would constitute a balanced corpus. That depends entirely on what one wants to Žnd out. There are occupational varieties of a language in which virtually everything that is said is formulaic. If one wanted to Žnd out what phrasal lexemes constituted native speaker competence in that variety then a database consisting solely of texts from that variety would be a good start. For example, if one selected no spoken weather forecasts for one’s database then one would be unlikely to Žnd such expressions as a deep depression is centred over REGION X. One also wouldn’t be able to Žnd out whether it is possible for a forecaster to say over REGION X a deep depression is centred. Since the selection of all corpora is arbitrary, even balanced ones, it remains an interesting question what one means by a representative or balanced corpus. Moon concludes about the Cobuild dictionary source of her set of phrasal lexemes that ‘[of] the dictionaries available at the time only the CCELD (1987) set out to analyse from Žrst principles all (or virtually all) of the tokens in a corpus of current and general English; the FEIs (phrasal lexemes) it included may be taken as a reasonable indication of which FEIs were actually in use in the 1980s’ (p. 45). This is clearly a questionable conclusion. I am not trying to make light here of the problems facing corpus linguists. They are real and serious. But they do raise crucial issues relating to the relationship between native speaker competence and texts produced by speakers in performance. For example, lexicographers who use corpus-based tools are attempting to provide on paper an account of what a native speaker of a language knows when he or she knows the words contained in a dictionary. What comes out of the mouths of people and the keyboards of writers is not a perfect reection of this knowledge. Exactly the same issues occur in discourse analysis where the analyst’s selection of texts raises the same issues and makes the conclusions of the analysis similarly problematic. Moon outlines the computational issues of working in the way she did. This includes a discussion of the size and complexity of available corpora and the methods of tagging them for various kinds of information. Moon suggests that a lookup list is valuable to check to see if various phrasal lexemes are contained in the corpus. But she believes that intuitional data have no part to play in the process. ‘The starting point for all such work must be detailed corpus-based, text-based description.’ (p. 56) This seems unnecessarily severe. Any native speaker of a language will have useful intuitions about which phrases they know to be phrasal lexemes. I know that one gets on the bus and not into the bus. That intuition might be useful if one wanted to Žnd restricted collocations related to travelling on public transport. Chapter four deals with frequency details of the various searches conducted on the corpus since ‘[o]ne of my (Moon’s) core aims in examining the evidence in OHCP was to gather data concerning the frequency of FEIs.’ (p. 57) The general result was that only about 5% of the corpus text consisted of phrasal lexemes. This seems very low given, for example, Altenberg’s results (Altenberg 1998) which suggest well over 50% of texts in a corpus are phrasal lexemes. There are various possible reasons for the disparity. First Moon requires a signiŽcance threshold of 5. In other words any Book Reviews 269 collocated text occurring fewer times than that in the corpus was not counted. That may be Žne for statistical reasons but the banning of intuitional data makes this necessary. There are no doubt corpora in which the word atlas does not occur more than four times. But it doesn’t follow that atlas isn’t a word of English. Moon recognises this by noting that many phrasal lexemes, such as hang Žre and out of practice do not occur at all in the corpus. (p. 60) Other factors include deciding what counts as a phrasal lexeme. Altenberg is more catholic in his criteria. The bulk of the chapter is devoted to providing frequency data for the various types of phrasal lexemes in Moon’s typology. In line with the comments made above about specialist varieties of a language Moon states that ‘[i]t is clear that the exact densities and proportions of different kinds of items in different genres will only emerge after exhaustive, corpus-based and text-based studies.’ (p. 74) Chapter Žve looks at the lexical and grammatical form of the data. An interesting observation is that, in her data, many normal single word lexical items do not appear in any phrasal lexemes. That may be a function of the set of phrasal lexemes selected to interrogate the corpus. It would seem unlikely that any verb would not be a constituent of a phrasal lexeme. One has only to think of a verb at random to be able to think also of a number of phrasal lexemes consisting of that verb and a lexicalised complement. For example dry has dry NP’s tears, dry NP’s eyes and a number of phrasal verbs such as dry up based around it. The crucial point here is the size of the set of phrasal lexemes with which the corpus is being interrogated. The chapter also contains a useful set of bound lexemes, single word lexemes which are bound within a phrasal lexeme. They include cases such as loggerheads, and umbrage. It also contains a number of syntactically ill-formed phrasal lexemes such as more fool you. Then Moon surveys the frequency with which various phrasal syntactic categories are represented in the corpus and here there are problems occasioned by the choice of syntactic theory Moon has adopted. The data has been annotated in accordance with a Hallidayan functional grammar. This results in there being no VP constituent and thus any phrasal lexeme containing a verb must have a clause as its dominating node. The result is that a phrasal lexeme such as admit defeat (p. 86) is assumed to have a subject slot in its lexical entry. However an X bar analysis would say that this is a V bar constituent and that the subject was not necessarily a slot in its lexical entry. The appearance of subjects is thus a syntactic not a lexical requirement. The majority of Moon’s data appear to be V bar constituents. That being the case tense and aspect are freely selectable and inect in the syntactically predictable ways since they are not constituents of the phrasal lexeme. However within a phrasal lexeme inectional properties appear to be unpredictable. Moon points out that parts of the body in phrasal lexemes sometimes inect and sometimes they don’t. For example there is only one ear bent in bend NP’s ear and only one leg pulled in pull NP’s leg. Slots often occur in V bar phrasal lexemes in complement position such as in the above examples. At other times they may have more than one slot as in leave NP to NP’s tender mercies. Some of these slots are determined by the theta role assignment of the verb (although this is not how Moon would describe this phenomenon). Others are speciŽc to the phrasal lexeme. For example in driving a wedge between NP and NP, the 270 Book Reviews two NPs are human rather than in the literal reading of the phrase which is not lexicalised where the slots are to be Žlled by inanimate NPs. The section on slots is again useful in surveying what kinds of slots there are and where they occur. Moon also looks in this chapter at the degree of exibility which her lexemes have under movement. Many of the transformations which she examines would no longer be regarded as such in generative theory. For example, she has negatives as transformational variants. Her account is thus close to that of Fraser (1970). It shows that some phrasal lexemes are relatively exible under movement while others are not. There seem to be no overarching constraints on this. Chapter six carries this forward by looking in general at surface variability. Moon suggests that the amount and kinds of variability ‘calls into question the whole notion of Žxedness, shedding doubt on the viability of the notion of the canonical form’ (p. 121). In discussing lexical variation Moon raises the issue of words which alternate in phrasal lexemes such as Žll/Žt the bill. These are termed ‘selection sets’ by Kuiper and Everaert (2000). Crucially it is impossible to say which one is basic to the phrase. The same thing happens with nouns as in tempt fate/providence. It is signiŽcant, however, that only one’s intuition can determine whether these alternations constitute alternations of a word in one and the same phrasal lexeme or whether they are two constituents of different lexemes. Moon also describes irreversible binomials (Malkiel 1959) (pp. 152–6) and what she calls idiom schemas (p. 158) which look very like the entities theorized in construction grammar (Goldberg 1995). In all this is a very full chapter with a great deal of interesting and useful observation about the ways in which phrasal lexemes vary in text. Chapter seven examines semantic properties of phrasal lexemes, speciŽcally their ambiguity, polysemy and use of metaphor. Unlike single word lexemes, phrasal lexemes are normally not polysemous, so while a particular phrase in a text may be ambiguous as between a literal reading of a freely generated structure and a non-compositional phrasal lexeme, phrasal lexemes themselves are seldom polysemous. Her data suggest 5%. Examples are expressions like tread water. However in context they are seldom ambiguous since since ‘[p]olysemous FEIs are often associated with different collocations or realizations of subject and object and these effectively disambiguate.’ (p. 189) There are interesting and much researched questions for psycholinguists as to how hearers determine whether to look up the meaning of a phrasal lexeme or parse the structure as a freely generated expression. Moon mentions a number of these and again the discussion is careful and useful. She also notes that ‘[c]orpus investigations show very clearly that literal equivalents to metaphorical FEIs (phrasal lexemes) occur relatively infrequently.’ (p. 180) She also provides a useful taxonomy of the ways by which non-literal meanings of phrasal lexemes are created such as personiŽcation and metonymy. Chapter eight deals with the discoursal functions of phrasal lexemes. It is clear that these are often used for structuring discourse and Moon’s account of this is again carefully handled and useful. She uses Hallidayan taxonomic categories for the various functions and illustrates them. She also shows that a given expression may be ambiguous as between various functions or multifunctional. They not only organise Book Reviews 271 discourse but provide expression for evaluative stances towards what is being said. For anyone engaged in discourse analysis this is a very useful chapter in pointing to discourse markers by which discourses are organised in various ways. It is also interesting that phrasal lexemes provide one of the ways of doing this. Chapter nine looks further at the evaluative and interactional information which phrasal lexemes can provide in discourse. Evaluation may, for example, be personal or, if it is part of the cultural loading of a phrasal lexeme, cultural. As an example, on pp. 250–1 Moon provides a piece of discourse analysis illustrating nicely the evaluative role played by phrasal lexemes in a text. That being the case cultural information in the form of a schema underlying sets of phrasal lexemes may provide information about cultural underpinnings, as suggested by Teliya et al. (1998). Of that possibility Moon says, ‘I am arguing here that FEIs such as metaphors, proverbs, and many other formulae or collocations operate as discourse devices by appealing to socio-culturally conditioned schemas. This then gives access to predetermined evaluation defaults which inform the discoursal position. Such schemas are rhetorically powerful, coercing agreement and pre-empting disagreement.’ (p. 259) This is a clear statement of the hegemonic capacities of the phrasal lexicon in creating socio-cultural solidarity. The same goes for politeness. Going back at least as far as Ferguson (1976), it has been clear that phrasal lexemes play an important role as politeness markers. Again Moon provides a good deal of useful evidence of their role in this area. She also notes the use of speech act formulae, after Austin (1976). Chapter ten deals with the function of phrasal lexemes in creating cohesion in text. Again the model used is Hallidayan and the discussion clear and useful. All the Hallidayan cohesion functions are carefully outlined and illustrated. Moon also makes a start on a discussion of humour resulting from word play with phrasal lexemes in this chapter. For academics the following will provide a nice illustration: ‘I have Žnally realised why universities are known as ivory towers. It must be because we are hanging on by the skin of our teeth’. (p. 289) I have reviewed this book in detail because it seems to be a most important step on the way to giving what the Russian and German scholars term ‘phraseology’ the kind of treatment that it needs. The book is not perfect in that the models of human language on which it draws seem to me to be inadequate, but the wealth of detail and care with which the analysis has been undertaken more than makes up for this. To conclude with the book as book, it is beautifully produced, typographically clear and free of typographical errors. It shows how far one can currently get with corpus linguistics and how rich a Želd the phrasal lexicon is. It is warmly recommended for all university libraries, for teachers of foreign languages, particularly those who have been inuenced by Nattinger and de Carrico’s suggestive but solidly inadequate treatment of the phrasal lexicon for second language teachers (Nattinger & De Carrico 1992). It is also essential reading for phraseologists, even those who do not know that they are. Department of Linguistics University of Canterbury, New Zealand 272 Book Reviews References Aijmer K 1996 Conversational Routines in English Longman London. Altenberg B 1998 On the phraseology of spoken English: The evidence of recurrent word-combinations. In Cowie, AP (ed) Phraseology: Theory, analysis, and applications Clarendon Press Oxford: 101–122. Aronoff M 1976 Word Formation in Generative Grammar MIT Press Cambridge, Mass. Austin JL 1976 How to Do Things with Words Oxford University Press London. BloomŽeld L 1933 Language Allen & Unwin New York. Ferguson C 1976 The structure and use of politeness formulas Language in Society 5:137–151. Fraser B 1970 Idioms within a transformational grammar Foundations of Language 6:22–42. Goldberg A 1995 Constructions: A construction grammar approach to argument structure Chicago University Press Chicago. Keller E 1981 Gambits. Conversation strategy signals. In Coulmas F (ed) Conversational routine Mouton The Hague: 93–113. Kuiper K & Martin E 2000 Constraints on the phrase structural properties of English phrasal lexical items PASE papers in language studies Proceedings of the 8th annual conference of the Polish Association for the Study of English, at Wroclaw: 151–170. Kuiper K & Gek Tan Lin D 1989 Cultural congruence and conict in the acquisition of formulae in a second language. In Garcia O & Otheguy R (eds) English across cultures: Cultures across English Mouton de Gruyter Berlin: 281–304. Malkiel Y 1959 Studies in irreversible binomials Lingua 8:113–160. Nattinger JR & De Carrico JS 1992 Lexical phrases and language teaching Oxford University Press Oxford. Pawley A & Syder F 1983 Two puzzles for linguistic theory: nativelike selection and nativelike uency. In Richards J & Schmidt R (eds) Language and Communication Longman London: 191–226. Schiffrin D 1987 Discourse markers Cambridge University Press New York. Teliya V, Bragina N, Oparina E, & Sandomiskaya I 1998 Phraseology as a language of culture: Its role in the representation of a cultural mentality. In Cowie AP (ed) Phraseology: Theory, analysis, and applications Clarendon Press Oxford: 55–75. Syntax: Structure, Meaning and Function ROBERT D VAN VALIN , JR. & RANDY J LAPOLLA Cambridge: Cambridge University Press, 1997, xxviii 1 713 pp. Reviewed by CYNTHIA ALLEN This book is intended as ‘an introduction to syntactic theory and analysis which can be used for both introductory and advanced courses in theoretical syntax’. However, readers might well be surprised to see that the book is intended for introductory courses when they see that such things as a basic knowledge of phrase structure are assumed. Syntax for Dummies would not be an appropriate alternative title for this rather hefty tome. But as the authors clearly state in the ‘Notes for Instructors’, the book presupposes a ‘standard introduction to the basic notions in syntax and morphology’ (p. xxi). This is a bit confusing, because many syntacticians would assume that an introduction to syntactic theory would necessarily introduce the student to the basic notions of syntax. The book is in fact mainly an introduction to a speciŽc syntactic theory Book Reviews 273 which the authors have developed—viz. essentially Role and Reference Grammar, with extensions made by drawing on various other theories where it is helpful. In general, the authors are quite scrupulous in giving credit to those whose insights they have incorporated, which is a nice contrast with many textbooks. The Notes for Instructors section offers suggestions for what sections can be left out in an introductory course and tells the teacher how an instructor’s guide can be obtained. The Žrst chapter of the book is a very helpful discussion of the goals of linguistic theory and overview of current approaches to syntax. The authors make it clear that their own theoretical orientation lies in the ‘communication and cognition’ perspective, but they are very even-handed in their treatment of the ‘syntactocentric’ perspective and do a good job of explaining to the reader the concerns of this general approach.1 The authors conclude this introduction with a comment that they intend both to present a descriptive framework which will be useful to Želd workers and to develop an explanatory theory of syntax which addresses the major issues of contemporary syntactic theory. The theory which Van Valin and LaPolla develop is introduced in the second chapter. The framework assumes a single level of syntactic representation and a single level of semantic representation, with a linking algorithm connecting the two levels. After the introductory discussion of why no more than this single level is necessary, Chapter Two is devoted to the structure of simple clauses and noun phrases and introduces the ‘layered structure of the clause’ (and related layering of the NP), which is argued to be superior to immediate constituent representations for capturing universals. The discussion is generally quite clear, but the beginner is likely to suffer some confusion over a couple of points. For one thing, it is not at Žrst made clear how categories such NP and Det are meant to Žt into this layered structure; such categories show up in the ‘formal’ representations of the layered structure of the clause but no indication is given of how they got there. Things become clearer, however, once we get to the concluding section of the chapter, where the different possibilities for specifying the proposed structures are discussed and we learn about the constructional templates that RRG uses. Similarly, the novice in RRG is likely to be puzzled as to why the phrase what did seems to be dominated by an NP node in Žgure 2.2, the representation of the layered structure of the clause for the sentence Yesterday, what did Robin show to Pat in the library? The position of did in this clause structure becomes clear in the next section, in which operators such as tense are discussed and we learn that they are assumed to have a domain of their own and are represented as a distinct projection of the clause from predicates and arguments. It is a relief to realise that did is not in fact dominated by that NP in Žgure 2.2 nor just suspended there disconnected from anything. In fairness to the authors, it must be recognised that it is extremely difŽcult to present a system like this without some problems of this sort, since it is simply not possible to present all the relevant information at once. However, I think 1 A rather surprising omission from the discussion of Role and Reference Grammar in this chapter is the lack of any mention of Van Valin and Foley (1980), which presented a preliminary sketch of the theory, or Foley and Van Valin (1984), which elaborated the theory. However, Foley and Van Valin (1984) is referred to elsewhere in the book. 274 Book Reviews the exposition could be improved somewhat by brief assurances to the student that the unexplained parts of the diagrams will be explained later. Similar difŽculties with the presentation are found in other chapters and will not be discussed in this review. None of them represent insurmountable difŽculties; they mostly simply require that the reader be willing not to understand everything as they are going. However, it seems to me that there is a more fundamental difŽculty with the discussion in Chapter Six of the role of controllers and pivots in the grammar, speciŽcally in voice discussions. Most of this chapter is easy enough to follow. There is an excellent discussion of how grammatical relations are deŽned and arguments that not all languages have purely grammatical relations, as distinct from semantic and pragmatic relations. There are purely syntactic grammatical relations in a language only if there is at least one construction in the language with a restricted neutralisation of semantic and pragmatic relations for syntactic purposes. Furthermore, a grammatical relation such as ‘subject’ is not the same in all languages. The traditional notions of grammatical relations based on Indo-European languages are not typologically adequate, and it is proposed that they be replaced with the notions of syntactic controller and syntactic pivot, which involve neutralisations of the semantic metaroles introduced in Chapter Four. Privileged Syntactic Argument (PSA) is a cover term for these controllers and pivots. Pivot refers mainly to the PSA of a linked or dependent core in a complex construction, such as the argument which can be deleted from a subordinate clause under coreference with something in a matrix clause, while controller refers to the argument that is privileged in such constructions as verb agreement, an example of which is the matrix controller in a complex construction. We are repeatedly reminded that controllers and pivots are construction-speciŽc. A language may have a syntactic controller/pivot in one construction, such as verb agreement in English, but a semantic controller or pivot in others, such as in English relative clauses. It is this construction-speciŽc nature of controllers and pivots which allows a good description of a language like Jakaltek, which is very different from English in having different PSAs for different constructions. So far, so good. But it seems to me that there is a fundamental difŽculty with the discussion of voice in section 6.4. In this section, we are told that one part of an English-like passive is that a non-actor is the pivot or controller. The problem which I have here is that being the controller of a passive construction seems to in fact involve being the controller or pivot of a number of constructions; the argument in question is the controller of verb agreement and initial position and both the controller and pivot of coordinate subject deletion, etc. ‘Privileged Syntactic Argument’ with respect to the passive construction seems to be referring to an argument which is privileged in various ways at the clausal level, rather than one which is privileged in a speciŽc construction, but this is not made clear. This causes problems in Chapter Seven when we try to Žgure out exactly how voice Žts into the linking algorithms which related semantics to syntax and vice-versa. It should be mentioned that this is not a problem which is unique to this textbook, but is one that I have had with presentations within Role and Reference Grammar generally; it is simply not clear to me what status statements like ‘non-actor occurs as pivot/controller’ have within this theory, although it is clear enough what is being referred to. Book Reviews 275 This fundamental problem aside, this is an excellent introduction to voice systems. I found the discussion of the correlation between types of reference-tracking devices and voice systems particularly illuminating. The presentation of semantics and pragmatics and the relationship of these two areas of language to syntax is also impressive. As beŽts an introduction to a theory which emphasises the centrality of meaning in language, two entire chapters (Three and Four) are devoted to semantics, and one chapter (Five) is devoted to information structure. Chapter Three is an admirable introduction to argument structure. The approach adopted is based on Vendler’s (1967) four basic Aktionsart classes and employs lexical decomposition. I expect to draw heavily upon this chapter in my presentation of aspect in syntax classes; I think that the discussion of the difference between perfective and perfect aspect (p. 172) will be particularly useful. From the discussion of the lexical representation of predicates in Chapter Three, we move on to the semantic representation of noun phrases, adjuncts and operators and the lexicon generally in Chapter Four. In this theory, the semantic macroroles Agent and Undergoer, which are generalisations across the argument-types found with particular verbs, play a key role in grammatical rules. The lexical entry of a verb has the logical structure of the verb at its heart, and a hierarchy regulating the assignment of macroroles to arguments, coupled with mapping principles which relate these macroroles to syntactic functions, makes it unnecessary to stipulate much syntactic information in the lexical entries of most verbs. Chapter Five deals with information structure and draws heavily on the work of Knud Lambrecht, as the authors freely acknowledge. It offers a concise and easily followed distillation of the central ideas presented in Lambrecht (1994) concerning topic, focus etc., with some additional material based on the authors’ own knowledge of more ‘exotic’ languages and a good explanation of how information structure is integrated into syntactic structure. I have already commented on Chapter Six, and the difŽculty I have with the notion of the PSA as applied to voice constructions. This difŽculty caused me problems in Chapter Seven, which attempts to formalise the linkage between semantics and syntax and vice-versa. As this is not a derivational theory, it is not assumed that syntax is derived from semantics; rather, the two levels are related by linking algorithms. For the reasons which I mentioned above, I had trouble Žguring out exactly how PSAs Žt into these algorithms. On the positive side, this chapter contains an insightful discussion of the linkage between case marking and semantics which goes well beyond the usual sort of presentation of case marking in a text book. The treatment of reexives within this theory is also extensively covered. The Žnal two chapters deal with complex structures. Chapter Eight looks at the layered structure of the clause in complex sentences and the syntactic templates necessary to deal with these complex sentences. Syntactic theories traditionally recognise only coordinate and subordinate relationships between the units involved in complex constructions, but RRG assumes a third relationship of cosubordination. I think that the exposition in this chapter could be improved considerably. We get pretty far into the chapter before it is made clear that the term ‘coordinate’ within RRG has a rather different meaning from the traditional one. While the traditional term assumes 276 Book Reviews a direct correspondence between conjunction and coordination, in RRG we have coordination when there is no subordination and also no obligatory sharing of operators. One cannot help wishing that the traditional terms which are based on ‘formal construction types’ had been replaced with other terms in RRG, because it is startling to a reader with any grounding in traditional grammar to Žnd sentences like John must tell Bill to wash the car treated as coordination without sufŽcient warning. The warnings are given, but it would be better to highlight this difference from the traditional notion of coordination earlier, rather than burying it at the end of the discussion of cosubordination in non-Indo-European languages. The presentation of cosubordination is itself confusing; in section 8.3, where the notion is introduced, we are told that the traditional contrast between subordination and coordination seems to be very clear cut for languages like English and are given the impression that cosubordination is found only in languages outside the Indo-European family, but in the next section we are suddenly given examples of cosubordination in English. This is not the only place in this book where I longed to have more explicit information about the connection between the ‘formal construction types’ and the ‘abstract syntactic relations’ of RRG. But there is much worthwhile material in this chapter, which also has detailed discussions of types of relative clauses and the status of elements such as to and for in complex constructions, as well as an interesting and useful discussion of the connection between semantics and the choice of juncture-nexus type. The Žnal chapter deals with the question of how the linking algorithms of Chapter Seven apply to complex sentences and need to be modiŽed to deal with constructions which involve the obligatory sharing of core arguments. As the authors note, the question of how to deal with this obligatory sharing of core arguments has played a prominent role in theoretical debates within contemporary syntactic theories. The authors do an excellent job of presenting ‘control’ constructions and in pointing out the weaknesses of purely syntactic approaches to control. I think they are rather less successful in their explanation of what they call ‘matrix coding constructions’, using the theory-neutral terminology of Frajzynger (1955) for what were treated as ‘raising’ constructions in classical transformation grammar. In my opinion, the exposition would have beneŽted from a clear statement at the outset that what the two types of matrix-coding constructions have in common is that a noun phrase which is not an argument of a given core nevertheless has syntactic and morphological properties which suggest that it belongs to that core. More references to the discussion of lexical entries in Section 4.3 would also be helpful here. And after the attention which has been paid in this book to pointing out the deŽciencies of arbitrarily marking particular lexical items for particular syntactic properties when similar lexical items can be shown to have similar syntactic properties, it comes as a disappointment to Žnd that the ‘non PSA matrix-coding construction’ (‘object to subject raising’) has to be treated as an exception to their 9.60b, one of the principles for template selection. It seems to me that this is an area where the authors might indicate to the student that there is a need for further research. However, the amount of detail devoted to these constructions is welcome, even if it is somewhat difŽcult to work through. The remainder of the Žnal chapter is devoted to extending the theories of case marking and reexivization developed in Chapter Seven to complex sentences and to Book Reviews 277 revising the linking algorithms to deal with the new material. The chapter ends with an interesting discussion of how the interaction of syntactic structure and focus structure results in the well-known restrictions of wh-questions and related constructions. These restrictions are argued to be neither purely syntactic nor purely pragmatic. I am in complete agreement with the approach of the book: to introduce students to syntactic theory by giving them a good grounding in one particular theory. I don’t think that textbooks which try to give the student an overview of, say, three different syntactic theories are terribly successful. Students are really able to compare syntactic theories only when they have acquired the ability to apply one theory to the analysis of data. I also applaud the liberal use of examples and analysis from non-Indo-European languages. However, I think that this book is too large and makes too many demands on the reader to be a textbook which I would expect to be able to use successfully in my own classes. On the other hand, I expect to draw on this book a good deal in my teaching, and I think it would be especially useful for postgraduate students interested in syntactic typology. I am certain that most syntacticians will Žnd it useful as a reference, as in many places it goes well beyond the restatement of well-known facts and widely accepted notions which is more typically found in textbooks. It should be a welcome addition to the bookshelf of anyone interested in syntactic theory and/or analysis. Department of Linguistics The Australian National University References Foley WA & Van Valin RD 1984 Functional Syntax and Universal Grammar Cambridge University Press Cambridge. Frajzynger Z 1995 A functional theory of matrix coding Paper presented at the Conference on Functional Approaches to Functional Grammar Albuquerque, NM. Lambrecht K 1994 Information Structure and Sentence Form Cambridge University Press Cambridge. Van Valin RD and Foley WA 1980 Role and reference grammar. In Moravcsik E & Wirth J (eds) Current Approaches to Syntax Academic Press New York: 329–352. Vendler Z 1957[1967] Linguistics in Philosophy Cornell University Press Ithaca, NY. Feature Distribution in Swedish Noun Phrases KERSTI BÖRJARS Oxford/Boston: Blackwell Publishers, 1998, 290 1 xiv pp. Reviewed by ALAN R. LIBERT This book deals with issues involving noun phrases in Swedish, some of which turn out to be more complex than one might expect. After a short introduction, in the second chapter, ‘Determiners and ModiŽers’, Börjars discusses the categories to which prenominal members of the noun phrase 278 Book Reviews belong. The basic question here is whether certain lexemes are determiners, adjectives, or neither. The lexemes under examination include maÊ ngen ‘many a’, alla ‘all’, annan ‘another’, faÊ ‘few’, sista ‘last’, and dylika ‘such’. Börjars uses a set of criteria for status as determiners and adjectives (e.g. adjectives can be modiŽed, while determiners cannot), and Žnds that there is a single lexeme which all of these indicate is a determiner, namely maÊ ngen, while none of the lexemes possesses all the properties of adjectives. However, she says (p. 39) that ‘there are few “real” adjectives which have all of these characteristics’, and since ‘each element in the position classes under discussion shows some characteristics of adjectives’ (ibid.), they should be classiŽed as adjectives, with the apparent exception of maÊ ngen. Given the complex facts, some readers might be tempted to take the view that determiners and adjectives are not discrete categories, but that there is a kind of ‘squish’ of the sort described by Ross (1973) for ‘nouniness’. Swedish has a sufŽx which marks deŽniteness, -en (and its variations, depending on number, gender, and whether it is attached to a consonant or vowel). The question dealt with in the third chapter, ‘DeŽnite Feature or DeŽnite Article’, is whether this element is a separate syntactic category from the noun. Börjars believes that it is not; according to the group of tests she applies it is not a clitic, and it ‘behaves differently from the syntactic determiners’ (p. 60). Chapter 4, ‘The Head of the Noun Phrase’, is concerned with whether (Swedish) noun phrases are headed by Ds or by Ns, that is, whether they are NPs or DPs. There does not seem to be a straightforward answer, since the criteria which Börjars uses give conicting results. She says (p. 130), ‘Considering how important the notion of headedness is to the type of structure I have adopted in this book, this is a very disappointing result.’ However, I think it preferable to admit to such an unclear conclusion, as she does, than to try to force a conclusion which is clear-cut, but unwarranted from the data examined. Börjars later (p. 134) states, ‘Until further evidence is found, I believe that the choice [of the head] can only depend on one’s speciŽc theoretical assumptions and how they, in effect, render irrelevant a subset of the headedness criteria.’ She opts to consider N the head, and says, ‘I will claim that there are reasons to prefer an NP analysis t[o] one in which D is claimed to be the head’ (ibid.). In Chapter 5, ‘Features in DeŽnite Noun Phrases’, Head-driven Phrase Structure Grammar is introduced. Börjars chose this framework because in it ‘an element is not required to be the head of the phrase in order to select its sister’ (p. 136). This is important given the facts of Swedish and Börjars’ choice of N as the head of the noun phrase. The framework is applied to several questions about Swedish deŽnite noun phrases, the ‘most obvious’ (p. 140) of which is the ability of singular nouns to stand alone as noun phrases, if they bear the deŽnite sufŽx, illustrated in (1): (1) Mus-en aÊ t ost-en mouse-DEF ate cheese-DEF ‘The mouse ate the cheese.’ (ibid.) Book Reviews 279 The matter is signiŽcant for Börjars, since she does not consider the sufŽx a ‘syntactic determiner’; as she notes, if it is taken to be one, there is then the question of how to account for the fact that it can be found with another determiner, as in (2): (2) den mus-en that mouse-DEF ‘that mouse’ (p. 141) An intriguing additional wrinkle is the fact that when a noun is preceded and modiŽed by most adjectives, the deŽnite sufŽx alone is not sufŽcient for a well-formed noun phrase: *Glad-a mus-en aÊ t ost-en happy-DEF mouse-DEF ate cheese-DEF b. Den glad-a mus-en aÊ t ost-en the happy-DEF mouse-DEF ate cheese-DEF ‘The happy mouse ate the cheese.’ (ibid.) (3) a. A laudable feature of this book is that for each of the situations described above, Börjars discusses earlier accounts as well presenting her own ideas on solutions; the reader can thus get a good idea of the not inconsiderable work done in this area. In her proposal, nouns with deŽnite sufŽxes bear a feature value stating that they may optionally have a determiner as a speciŽer. This will account for the sort of data exempliŽed in (1) and (2). With respect to the situation in (3), an important fact is that the adjectives occurring in such contexts are weak. (Swedish has a distinction between weak and strong adjectives.) Among the features posited for weak adjectives is one which states that ‘they combine … with an N9 which obligatorily selects a deŽnite determiner’ (p. 189). The sixth and last major chapter is ‘Features in IndeŽnite Noun Phrases’. The facts to be dealt with here are somewhat different. Count nouns which are meant to be indeŽnite, and hence are without the deŽnite sufŽx, generally cannot occur in argument positions in the singular; that is, they require an indeŽnite article. In this chapter Börjars again presents other accounts as well as her proposals, this time for various types of indeŽnite noun phrases including those involving ‘Nouns with determiners’ (section 6.2) and ‘Noun phrases containing adjectives’ (section 6.3). She supposes that nouns which lack the deŽnite sufŽx have the feature value [DEF u], rather than [DEF -], that is, they are ‘unspeciŽed for deŽniteness’ (p. 223). Because they bear neither the feature value [DEF 1 ] nor the value [DEF -], they cannot ‘trigger the semantic interpretation of a determiner’ (ibid.), and so require a ‘syntactic determiner’. The necessity for an indeŽnite article in some kinds of indeŽnite noun phrases (even predicative ones) containing an adjective is accounted for in a similar way as in the discussion of deŽnite noun phrases. Even those who do not agree with the choice of framework or the accounts put forth may Žnd this book to be a valuable resource if they are concerned with noun phrases in Swedish or noun phrases in general. Not all possible elements of Swedish noun 280 Book Reviews phrases are dealt with; for example, post-modiŽers are not treated. It would be interesting to see how these would be handled, and to see the approach extended to some other languages. University of Newcastle Reference Ross, J R 1973 ‘Nouniness’ in O Fujimura (ed.) Three Dimensions of Linguistic Theory TEC Company Tokyo.