Wallis Reid's Verb and Noun Number in English is the first comprehensive study of its kind, based on a quantitative methodology. The author develops a pragmatically-oriented treatment of subject-verb agreement for 3rd-person subjects. His analysis intends to disprove the value of the distinction between so-called 'determined' elements of a language such as the -s ending of the English verb and 'freely-chosen', lexical elements. By turning to a sign-based explanatory framework rather than an explanation which has the sentence for a starting-point, his account seeks to show that the principles that govern morphological form are identical to those that govern lexical choice; that is, the communicative goals of the speaker. In other words, in Reid's view the syntagmatic structure of a language is essentially determined by its function as a communicative tool. So-called grammatical 'irregularities' are therefore germane to his theory rather than peripheral.

Reid's argument proceeds from the basic opposition between a sign-based theory and a sentence-based theory. He adopts the first, drawing on the framework developed by William Diver at Columbia University, a framework also known as Form-Content Analysis. This framework, he says, implies a return to the Saussurean conception of a language as an inventory of signs. Such a conception explains the co-occurrence of signs - lexical and grammatical - not in terms of syntactic rules, but as "mutual contribution to a common communicative goal" (preface, x). In terms of subject-verb agreement Reid claims that neither the grammatical number of the verb nor that of its subject determines the other; each is chosen for its semantic value and contributes independently to the communication of the message.

In both spoken and written language use, the verb frequently fails to 'agree' with the number of its grammatical subject, e.g.

(1) This afternoon our panel are three male singers. (Edward Downes, Texaco Metropolitan Opera broadcast) (p.193)

Formal rules can never, according to Reid, successfully describe what count as singular and plural subjects for the purposes of verb agreement, nor can they offer any explanation why there should be agreement. Reid's sign-based theory tackles these shortcomings by abandoning the conception of language as a representational system and adopting instead the "radical functionalism" of Diver's Columbia School. This implies that no formal mapping is posited between linguistic signs and 'the message' communicated. Language users are assumed to operate in the same 'goal-directed' communicative way on the micro-level of grammatical and lexical signs as on the discourse level.

Reid concurs with the remark of generative criticism that functional explanations of linguistic form often amount to little more than loose correlations of forms with some discourse function. He admits that, indeed, "you cannot get from syntactic description to functional explanation" (p.39). Yet to resolve such dichotomy, he says, a functional perspective should be adopted from the outset of description, instead of being added to an initial formal structure, set down independently. By taking the linguistic sign as the basic structural unit of language, the conception of linguistic structure as having a purely formal side is implicitly rejected: the linguistic sign unites 'form' and 'content' in a way that makes an independent analysis of linguistic 'form' impossible. Functionally-motivated connections do not obtain, therefore, between form and concept - their link is considered arbitrary -, but between "the inherent conceptual substance of linguistic signs and people's larger communicative goals" (p.34).

Having developed his framework of analysis in chapter 1, Reid's presentation starts with three chapters (2-4) on the principles of the number system of English nouns, before treating his main topic, 3rd person verb number, in chapters 5-7. Chapter 8 develops a broader conception of functional explanation that shows affinity with the functional theory of Halliday and Hassan (Cohesion in English, 1976), chapter 9 addresses the reasons why the book's functional account of verb number cannot be synthesized with a syntactic account, and, finally, in chapter 10 Reid examines the ultimate phenomena which sign-based linguistic theory attempts to explain.

The analysis in chapters 2-4 puts the plural -s suffix on a par with a lexical stem, i.e. "a signal paired with a meaning" (p.45). The plural 'noun' a category which Reid refrains from using is hence analysed and interpreted as a semantic synthesis of two meaning-bearing units. The unit indicating number is called the Entity Number Ø/-s opposition and its meaning is defined as the relational opposition ONE vs MORE THAN ONE. Since such different semantic values do not reflect the objective structure of reality in Reid's framework, but rather the exigencies of communication, those different values must find confirmation in the ways people employ words. Methodologically, this can only be examined by a careful examination of actual language use in spoken and written texts, not via introspective examination.

Interpretations that lead a speaker to use singular or plural verb number in combination with collective nouns like people or family are characteristics of the message or "the intended and understood import of a communication" (p.95), not of the meaning or semantic category of a lexical item. Meaning is a theoretical construct and notionally constant. Reid thus argues in favour of lexical monosemy: rather than explain the dual usage of some nouns as inherent lexical polysemy, e.g. the meanings 'common domestic fowl' for chicken in I saw a chicken and 'meat from a common domestic fowl' in I ordered chicken, Reid argues that the information supplied by the associated grammatical morphology realises the different construal of the lexical stem chicken and its interpretation. In other words, everything happens at the level of the message communicated, not at a prior stage of conceptualisation.

In one instance Reid's communicative and therefore hearer-oriented perspective lapses into an ad hoc motivation of number choice, in my view, which weakens his argument. In a sentence like

(2) The acoustics influence design considerations.

the stem acoustic is construed as MORE THAN ONE, according to Reid, "not because of the greater semantic suitability of the meaning ..., but because of the greater perceptual salience of its signal" (p.81-82), viz. to solve the problem the hearer may have in determining whether acoustic is applied in an adjectival or nominal capacity. Not only is this explanation invalid for nouns without corresponding s-less adjectives like politics and physics, it also overlooks the role of the conceptualisation that the speaker has formed of this entity through experience. Reid's approach seems to return to a communicative situation ex nihilo, in which signs are moulded into suitable communicative signals, yet carry no representational meaning besides the sum of particular contexts of usage. Yet, hearers are also speakers, and hence a theory cannot ignore the lexical knowledge that results from those particular contexts, which is shared by speaker and hearer.

Reid's discussion of noun number yields two principles, crucial to his treatment of verb number (p.118):

1. the relation between a linguistic meaning and a message is one of communicative efficacy

2. the meanings in a grammatical system are exploited in terms of their relational oppositions

(3) Returning to the woody undergrowth, Jake found his traps had caught half a dozen rabbits, enough for his hungry companions back at the house.

The settlers often trapped rabbit that year, which were their only source of fresh meat.

The factor 'degree of interest in the individual' examined for animal reference returns in Reid's theory of verb number, developed from chapter 5 onwards. He calls the number system of English verbs the Focus Number System, consisting of -s and Ø attached to the lexical stem of the 'verb', which - as in the case of the Entity Number System - stand for semantic substances, viz. ONE (-s) and MORE THAN ONE (Ø) 'entity in focus'. The Focus Number System categorises the subject or 'participant' of an utterance in terms of the 'amount of attention' concentrated upon it with respect to the 'occurrence', i.e. Reid's term for the action of the verb. This amount of attention - and hence the reason for number discord and concord - is motivated by contextual features of the predicate. For example,

(4) Those who study dust estimate that 43 million tons of it settle over the U.S. every year. About 31 million tons of this is natural and the other twelve million man-made. (Penny Ward Moser, Discover).

Reid points out that in (3) the first "communicative point" (p.224) is the large amount of dust that settles over the US yearly, and the meaning MORE THAN ONE of settleø "reflects the quantification of the dust" (p.224) into a plurality of tons by the phrase 43 million tons. In the second part of the sentence the point shifts to the origin of the dust, which requires dividing the initial quality according to its source. The counting category shifts from 'tons' to 'natural' (and 'man-made'), and the meaning ONE of is reflects this new grouping.

Deviations from grammatical agreement between 'occurrence' and 'participant in focus' cannot - in a statistically significant way - be assigned to speaker errors. Hence, Reid puts forward the hypothesis that Entity and Focus number are independently chosen, in correlation with various contextual features of the predicate supporting the relational contrast between the meanings ONE and MORE THAN ONE. The corpus analysis to test the Focus Number oppositions draws on naturalistic data from a variety of spoken and written sources (radio and TV shows, magazines, newspapers...). Six example types are examined, viz. the occurrence of concord and discord with three types of subjects: type 1 morphologically singular subject, type 2 morphologically plural subject, type 3 singular conjoined nouns as subject. The statistical results of the analysis show that speakers exploit the relational opposition of Focus Number 100 per cent of the time. This means that choices of concord or discord are motivated each time by the attention focussed on the 'participant' or subject in relation with the contextual elements in an utterance. For example,

(5) With only two hundred bears roaming the park lands, ten bears is a significant number.

In this case, numerical quantification of bears has an effect comparable to that of a lexical collective: it can be categorized as either singular or plural.

Still, the analysis of possible combinations and their communicative motivation does not yet explain which communicative strategies speakers favour and why - in particular, why the one strategy of choosing noun and verb concord occurs most frequently. To motivate this, Reid invokes psycho-linguistic factors. In chapter 8 he introduces the term textual resonance: textual resonance is created when the meanings of various signals jointly contribute to the communication of the same feature of the message. In other words, the idea of linguistic elements being in harmony - echoing the concept of cohesion introduced by Halliday and Hasan (supra) - is extended from discourse to sentence level and supplants the traditional formal treatment of syntax. Reid argues that textual resonance is explained by inherent characteristics of the communicative process: by building in a high degree of resonance or harmony between elements of the message the speaker mitigates the problems of perceptual indeterminacy of linguistic signals and meanings. What cannot be explained at this point is how much textual resonance a language requires and how this functional factor may have motivated linguistic change.

Reid's final theoretical position is that sign-based theory is "empirically grounded" in a way that sentence-based theory is not; in other words, opting for sign-based theory is not a choice motivated by an a-priori commitment to functional explanation, but rather relies on more general principles of scientific inquiry. Sign-based theory sees language as a facilitating communicative instrument rather than as an all-encompassing cognitive system. It does not assume a perfect match between categories of language and categories of thought. According to Reid, therefore, sign-based theory may establish a linguistics that "finds human language of interest in its own right, not just for what it can tell us about the nature of thought, cognition, or the mind" (370).

One of the great merits of Reid's study is its successful application of quantitative data to analyse and interpret phenomena at the centre of grammatical theory. The work shows that an account of actual language use rather than an idealised body of grammatical facts based on intuition and introspection need not run the risk of falling short of consistency or linguistic relevance. Reid states, rather, that within a sign-based conception of linguistic structure the principles responsible for observational diversity are intrinsic to the functioning of the system itself. So, although all speakers share the same grammatical system, the very nature of this system implies that different speakers will exploit the principles differently.

Nevertheless, it remains difficult to agree with Reid's fundamental statement that the supposed difference between a speaker making a lexical choice and the same speaker making a grammatical choice "is simply an artefact of the analyst's perspective" (p.65), that neither is prior to the other, but that they are interrelated in serving a common communicative goal. In the assumed absence of a mapping between concept and formal structure, it becomes difficult to examine syntactic issues, viz. direct links of pragmatic function and syntactic form.

Reid's claim that functional linguistic theory is empirically grounded may basically be right, but for the sake of linguistic analysis, a theoretical framework that is based on communicative goals also has to make a hierarchical distinction between the communicative goals that linguistic elements serve, viz. a distinction between the relative weight of 'lexical' and 'grammatical' elements - in other words, an analysis that seeks to improve our understanding of the workings of language has to rely on analytical categorisation of a certain type. Otherwise it may be difficult to account for historical phenomena like grammaticalisation from a functional perspective.

Lieve Jooken, Leuven