Grammar

Finding Metaphor in Grammar and Usage: A methodological analysis of theory and research by Gerard Steen (Converging Evidence in Language and Communication Research: John Benjamins Publishing Company) Cognitive linguists have proposed that metaphor is not just a matter of language but of thought, and that metaphorical thought displays a high degree of conventionalization. In order to produce converging evidence for this theory of metaphor, a wide range of data is currently being studied with a large array of methods and techniques. Finding Metaphor in Grammar and Usage aims to map the field of this development in theory and research from a methodological perspective. It raises the question when exactly evidence for metaphor in language and thought can be said to count as converging. It also goes into the various stages of producing such evidence (conceptualization, operationalization, data collection and analysis, and interpretation). The book offers systematic discussion of eight distinct areas of metaphor research that emerge as a result of approaching metaphor as part of grammar or usage, language or thought, and symbolic structure or cognitive process.

Excerpt: When cognitive-linguistically inspired researchers of language investigate metaphor, they typically do so by looking at language as either grammar or usage. Moreover, they have to make a choice in focusing on metaphor in grammar or usage as either language, analyzing linguistic forms, or thought, examining conceptual structures, or both. And finally, they have a further choice in adopting either a sign-oriented, symbolic perspective on metaphor, or a behavior-oriented, social-scientific perspective on the processes and products of metaphor in cognition. When these choices are combined, we end up with a field of research that consists of eight distinct research areas displaying their own object and approach.

I have characterized each of the research areas by means of a question about metaphor identification. As can be seen, the questions differ substantially between the distinct approaches. For research on the linguistic aspect of metaphor approached as symbolic structure, there is a difference between research on grammar and usage which may be captured by the contrast between the following two questions (top row):

Q1: When does a conventionalized linguistic form-meaning pairing count as metaphorical?

Q2: When does any linguistic form-meaning pairing in text and talk count as metaphorical?

The former type of research requires evidence about sufficient degrees of conventionalization of metaphor whereas the latter type of research does not. It should be noted that this does not mean that all metaphor in usage is novel: metaphor in usage is novel, conventional, or obsolete. The difference between the study of conventional metaphor in grammar versus usage will be returned to at various points in the book.

The distinction between conventionalized metaphorical forms versus all metaphorical language forms, whether they are conventionalized or not, applies to all of the following sets of questions about metaphor in grammar versus usage. Natural tools in research about conventionalized meanings in grammar are reference works, such as dictionaries and grammars, and corpora. In usage, we are typically dealing with the analysis of text and talk as verbal documents or transcripts. Metaphors in such usage data, whether conventional or not, may be further determined by their discourse characteristics that have to do with their domain (religion, literature and the arts, science and education, law, government and politics, mass media, business, and so on), their function (information, persuasion, instruction, and so on), their type (narrative, argumentative, expository, descriptive), etcetera.

When attention is shifted from the linguistic aspect to the conceptual aspect of metaphor, questions 3 and 4 capture the distinction between grammar and usage for the semiotic approach to the field:

Q3: When does a conceptual structure related to a conventionalized linguistic form count as metaphorical?

Q4: When does a conceptual structure related to any linguistic form in text and talk count as metaphorical?

This type of research analyzes the conceptual complexities of cross-domain mappings. It takes its cue for such an analysis from language data that express such a metaphorical mapping. The methodological problem here is that more metaphorical mappings in conceptual structure may explain the same language data. This is a problem which may be more difficult to solve in the study of usage than in the study of grammar because of the potentially idiosyncratic nature of any single usage event (think of the mind set of individual poets, politicians, and patients). In this connection it is of methodological importance that researchers decide either that the analysis fixes the language data and then explores which conceptual structures may be related to it (the `semasiological' route), or that it fixes the conceptual metaphors and then looks for potential linguistic expressions (the 'onomasiological' route).

The first two rows deal with semiotic approaches to metaphor in grammar and usage, divided by language and thought. The difference between these two areas of research, on the one hand, and research on metaphor in grammar and usage as behavior, on the other, may now be explained as follows, first of all with reference to its linguistic aspects (third row):

Q5: When does the storing, acquisition or even loss of a conventionalized linguistic form-meaning pairing count as metaphorical?

Q6: When does the production or reception in text or talk of any linguistic form-meaning pairing count as metaphorical?

Question 5 addresses the cognitive psychological aspect of metaphor in grammar by focusing on the long-term processes of language storing, acquisition, and attrition, in order to be able to tap the cognitive processes and products of sufficiently conventionalized language in individual language users. Question 6 tackles the psychological aspect of metaphor in usage by focusing on the short-term processes of language production and understanding (e.g. lexical access) in order to tap the cognitive experience of individual language users of specific usage events. In both areas of research, attention may be devoted to the behavioral processes themselves or to their products, that is, long-term, conventionalized, or short-term, specific mental representations.

The condensed forms of expression of these questions intentionally focus on the metaphorical nature of storing, acquiring, losing, producing, and receiving a metaphorical expression: in cognitive-linguistic terms, these cognitive processes can only be said to be metaphorical if they display the activation of two conceptual domains or spaces, one of which is mapped onto the other. The data of this type of research are speech and listening behavior, or reading and writing behavior, which, for grammar, ideally should stretch out over longer periods of time (longitudinal studies), while for usage, again, they may be highly specific in relation to the genre of discourse that they occur in. Stretches of discourse divorced from their users do not count as such, for they may be compatible with a range of processes, including those that involve the activation of only one cognitive domain because the relevant conceptual mapping has taken place in the history of the language instead of being reproduced by the individual on every occasion of use. This is clearly one area where cognitive linguists have to look at work in psycholinguistics and interactional sociolinguistics to find the relevant data and evidence for their theses.

The areas of research in Q5 and Q6 are characterized by their focus on metaphor approached as language, not thought. That is, both grammar and usage are defined by their attention to the cognitive processing of linguistic forms without many assumptions about the cognitive identity and content of related conceptual structures. Questions 7 and 8, by contrast, do precisely that:

Q7: When does the storing, acquisition, or even loss of a conceptual structure related to a conventionalized linguistic form count as metaphorical?

Q8: When does the production or reception in text or talk of a conceptual structure related to any linguistic form count as metaphorical?

Researchers working in these two areas investigate the cognitive processing of either conventionalized or all mappings between conceptual domains as part of grammar or usage. This may again happen with special attention to the processes themselves, or to their products. The crucial characteristic of this type of research is the concern with people and their behavior, not signs. This, too, is an area which is less typical of cognitive linguistics than of psycholinguistics and interactional sociolinguistics.

The questions in the lower half of Fig. 1.1 all pertain to metaphor in behavior. They are concerned with cognitive processes and products, which are not necessarily identical with the symbolic structures and their uses that are addressed in the upper half of Fig. 1.1. This is one fundamental divide between areas of research on metaphor in grammar and usage. This is compounded by the fact that, within each of these more encompassing fields, there are distinct areas of research to be discerned for metaphor in grammar versus usage, further divided by investigations of linguistic form versus conceptual structure.

To illustrate these distinctions by means of one example, reconsider the opening quotation from George Bush:

See, without the tax relief package, there would have been a deficit, but there wouldn't have been the commiserate — not 'commiserate' — the kick to our economy that occurred as a result of the tax relief.

Since this is one specific usage event, let us first examine questions 2, 4, 6, and 8. I am only going to point out a number of obviously different phenomena that are typically picked out for research in each of the four areas.

For metaphor in usage, question 2 asks when any linguistic form-meaning pairing, including our noun kick, counts as metaphorical. In this book I will suggest that this is typically the case when it displays an indirect contextual meaning which may be contrasted and understood by comparison with some direct, more basic meaning. In this case, the more basic meaning would be a concrete movement by the foot against another concrete object, whereas the contextual meaning is probably something like an abstract stimulus (in this case a tax relief package) that affects another abstract entity (in this case our economy). Question 4 can then be read as asking what the conceptual structure related to this linguistic form could be and whether it is metaphorical, too. It might be claimed that there is a force-dynamic mapping between the conceptual domains of physical movement and abstract financial processes which includes all sorts of elements in those two domains and projects specific relations between some

of these elements from one domain to another. Question 6 highlights the psycholinguistic process that went on in Bush's head as he was producing the linguistic forms in the utterance. It would ideally hope to examine the way in which the lexical unit kick was accessed, retrieved, inserted into the formulation to be constructed, and finally articulated, in order to, for instance, test whether two senses were activated during production, one pertaining to physical movement and another pertaining to financial effect. Question 8 also highlights the psycholinguistic production processes, but with special attention to the two knowledge domains involved; it broadens into more general cognitive psychological processes of human performance, and could include the issue of whether Bush has a mental image of the economy indeed getting kicked —whatever such an image may look like.

As for metaphor in grammar, questions 1, 3, 5, and 7 could take the Bush example as one usage case out of many, which might or might not add up to a sufficiently conventionalized grammatical phenomenon. For question 1, the question would not only be whether there is a metaphorical meaning in this use of kick, but it would also ask whether such a use can be found in a sufficient number of different cases, which could suggest that the use is conventional and therefore grammatical. For question 3, the same question about conventionality would be asked for the conceptual structures that also play a role in question 4; attention might be paid here to related linguistic manifestations of the postulated conceptual structures in other linguistic forms that can be associated with kick, including its use as a verb, and contextually near-synonyms such as hit or jolt. Question 5 would have to look at the knowledge that the individual speaker Bush has of the English language, in order to ascertain whether the metaphorical use established for kick is a sense that is part of his mental dictionary, so that it is an established metaphorical lexical unit in his personal mental grammar. And question 7 should turn to the related knowledge structures in Bush's mental encyclopedia, to test whether he has an entrenched conceptual mapping available from the source domain to the target domain.

Again, these are just tentative and partial suggestions, meant to illustrate some of the fundamental differences between the eight areas. They are not complete. And they are partly programmatic. The differences between the eight areas of investigation will be further detailed in the rest of this book. But the point is that each of the eight areas displays a distinct set of phenomena that can be researched in various ways. It is therefore important to achieve some sort of understanding of the relations between these diverging types of research.

Methodologically, symbolic approaches predominate over behavioral ones in cognitive linguistics and in most other schools of mainstream linguistics. This is reflected in Gibbs's (1994, 1999b) concern with a number of possible alternative interpretations of the cognitive-linguistic view of a role for conceptual metaphor in language. I will now argue that all of his alternative hypotheses concern the relation between metaphor in language and thought as symbolic structures, on the one hand, and metaphor in language and thought as verbal behavior, on the other. This is of course precisely what might be expected from a contribution by a psycholinguist to the cognitive-linguistic endeavor. If this analysis is accepted, it shows the virtues of adopting an encompassing and systematic approach to the various dimensions of metaphor research since it can explain Gibbs's alternative hypotheses as relating to various perspectives on metaphor in language and thought.

Gibbs's first hypothesis runs as follows:

(1) Metaphoric thought might function automatically and interactively in people's on-line use and understanding of linguistic meaning. (Gibbs 1999b:43)

This is a hypothesis which narrows the cognitive-linguistic claim about metaphoric thought in language down to the research area designated by question 8. It assumes the potential validity of the symbolic analyses of metaphor in language and thought, and projects them onto issues of verbal behavior. The relation of this claim to other behavioral aspects of metaphor in language, addressed by questions 7, 6, and 5, remains implicit.

Gibbs's second alternative reads as follows:

(2) Metaphoric thought might motivate individual speakers' use and understanding of why various words and expressions mean what they do, but does not play any role in people's ordinary on-line production or comprehension of everyday language. (Gibbs 1999b:43)

This hypothesis pertains to people's off-line, long-term understanding of grammar, not to its short-term usage. This suggests that this hypothesis makes a positive claim about the research area characterized by question 7, and by implication a negative one about research area 8. Again, the hypothesis does this by assuming that the symbolic analyses of linguistic and conceptual metaphor as signs have some validity.

Hypothesis 3 cuts the connection with behavior, and looks at conceptual metaphor in language as a sign system:

(3) Metaphoric thought might motivate the linguistic meanings that have currency within linguistic communities, or may have some role in an idealised speaker/hearer's understanding of language. But metaphoric thought does not play any part in individual speaker's ability to make sense of, or process, language. (Gibbs 1999b:42)

I understand this hypothesis as making a positive claim about the research areas designated by questions 1 through 4, and a negative claim about all research areas addressing behavior. The same probably holds for its historical variant, hypothesis 4:

(4) Metaphoric thought might play some role in changing the meanings of words and expressions over time, but does not motivate contemporary speakers' use and understanding of language. (Gibbs 1999b:42)

It turns out, then, that the interaction between the dimensions of the field which I have distinguished creates a grid which offers a natural home to the alternative interpretations of the cognitive-linguistic position on conceptual metaphor distinguished by Gibbs (1999b). They all pertain to the role of conceptual metaphor in grammar and usage, and make a distinction between grammar and usage as behavior (hypotheses 1 and 2) or as a sign system (hypotheses 3 and 4). Each alternative hypothesis highlights a different area of research, which stands for a different aspect of metaphor in grammar and usage.

In order to argue for the validity of one or another of each of these alternative hypotheses, evidence would have to be collected within the boundaries of the relevant field of research, with the appropriate methods, in order to be able to make a connection between that area and the cognitive-linguistic research on metaphor in thought (conceptual metaphor). This assumes that researchers respect the various distinctions between the areas of research, since they involve different phenomena. This is not only advocated by Gibbs but also by other psycholinguists who have taken the proposals of cognitive linguistics seriously. And this position has also been adopted in the more general discussion about the relation between linguistic and psychological analysis of language in cognitive linguistics.

A map such as the one provided by Fig. 1.1 may be helpful in defining the conceptual framework within which such distinctions can be defined and evaluated. This may also throw light on the increasing interest in cognitive linguistics and, elsewhere, in the role of converging evidence, as we shall now see. In the next section I will also present my main methodological claims for this book.

Converging evidence and the role of methods: Main claims of this book

Converging evidence is particularly important for the debate over conceptual metaphor. This is especially relevant for the question whether evidence about one particular phenomenon in one area of research, metaphor in language, allows researchers to draw conclusions about other phenomena in other areas of research, metaphor in thought. That is, if cognitive linguists have found metaphor in language in various ways, can they then also claim that they have found metaphor in thought? And can they then continue by utilizing these conceptual metaphors in deductive fashion to go back to language and find other linguistic manifestations of the same conceptual metaphors without becoming vulnerable to the accusation of circularity?

Here is the objection as phrased by Keysar, Shen, Glucksberg, and Horton (2000:577):

How do we know that people think of happy and sad in terms of up and down? Because people talk about happy and sad using words such as up and down. Why do people use expressions such as his spirits rose? Because people think of happy in terms of UP.

But cognitive linguists claim that there is a lot of converging evidence for the existence of a conventional conceptual basis for metaphor which can get expressed in language and in other codes, such as visuals, gesture, and signing. Cognitive linguists also hold that this provides an independent basis for subsequent searches for metaphor in language.

Lakoff and Johnson (1980, 1999a) argue that the ubiquitous and systematic patterns of metaphor in language can be seen as various types of evidence for the fact that there is an underlying pattern of conceptual structure which is metaphorical itself. They claim that there is converging evidence for conceptual metaphors in many areas of language, in the form of for instance polysemy generalizations, novel case generalizations, and inference generalizations. They have gone so far as to suggest that these conceptual metaphors are not cultural patterns of thought, as has been weighed as an option by for instance Gibbs (1999c) and Tomasello (1999). Instead, Lakoff and Johnson hold that conceptual metaphors are cognitively real in individual people's minds (1980, 1999a); what is more, they even claim that they are neurally encoded in every individual's brain (1999a). Such conceptual metaphors are held to explain the various types of data outside thought, including language.

However, the existence of conceptual metaphors as part of the individual mind (let alone the individual brain) has been challenged by a number of researchers outside the cognitive-linguistic school. Even though they highly appreciate the cognitive-linguistic proposals, they contend that the various types of linguistic evidence for conceptual metaphor marshaled by Lakoff and Johnson and other cognitive linguists can be explained alternatively. Vervaeke and Kennedy (1996:283), for instance, issue the following warning:

..., the idea that metaphors govern thought needs to be firmly restricted. Rather, a metaphor is often chosen from a set of alternative metaphors with widely differing implications to express an idea that is literal.

These critics even sometimes suggest that the evidence itself, whether linguistic or non-linguistic, is not reliable. Thus Vervaeke and Kennedy (1996) emphasize that the conceptual metaphors proposed by Lakoff and Johnson and their followers are quite dependent on the linguistic examples selected for their support: if the examples are changed, the conceptual groupings are modified, also allowing for lower or higher levels of generality at which a conceptual metaphor may be postulated. This raises questions about both the testability as well as the validity of the theory. These are the most urgent reasons why a methodological assessment is in order of the ways in which metaphor can be found in language and, for that matter, in thought.

The debate over converging evidence for conceptual metaphor presupposes that the methodological picture about each of the separate types of evidence for metaphor in language and thought is clear. This, however, is a presupposition which is simply not endorsed by all linguists outside the cognitive-linguistic fold. It also presupposes that the connections are clear between the various areas that the converging evidence is derived from. This is a presupposition which is not generally supported either. My concern in this book will therefore be with the relations between evidence for all metaphor within and between these areas of research.

My general thesis will be twofold. First of all, I will argue that caution needs to be observed when evidence from different domains of investigation is crossed. It is problematic to go from conceptual metaphor to linguistic metaphor within symbolic analysis, or to go from either of these phenomena to its correlate in behavior. It is not impossible, but this type of research needs to be aware of a number of issues. I will spell some out in Chapter 2, where I will look at the way in which cognitive linguists and others have employed deductive argumentation to go from conceptual metaphor to other areas of research. For now, I should like to formulate the following general assumption:

"Converging evidence" presented about distinct phenomena, such as metaphor in grammar versus usage, or metaphor in language versus thought, or metaphor in symbols versus behavior, or their more specific configurations, is exciting but problematic, unless clear models of the relation between the distinct phenomena can show how the evidence for one type of phenomenon can be said to point to the same conclusion as the evidence for another type of phenomenon.

The second part of my thesis will be more positive. I will argue that obtaining converging evidence in the field of metaphor research can best be promoted by applying a form of methodological pluralism within each of the various areas of research. Thus, for the identification of metaphor as linguistic form in grammar, different methods and techniques may serve to produce various kinds of evidence that may, or may not, converge to the same conclusion. But the phenomenological condition for this type of success is that it has to be clear from the outset which area of research is being addressed, which is a condition that is not often met. Another condition is methodological, and imposes the same norms of data collection and analysis on the two types of evidence that are said to converge; this condition is not often met either.

"Converging evidence" presented about the same phenomenon, whether metaphor in grammar versus usage, metaphor in language versus thought, metaphor in symbols versus behavior, or their more specific configurations, is exciting and attractive, provided the same norms of data collection and analysis are adhered to in evaluating the evidence obtained with one type of method as with the other type of method.

The reason why these two methodological claims can be made about converging evidence may now be clear, too. In the first case, we are dealing with evidence about different phenomena, deriving from distinct areas of research, such as cognitive processes and symbolic structures, or grammar and usage. The relations between such distinct phenomena constitute different and more complex objects of research. They require separate evidence about each of the two phenomena that are hypothesized to be connected with each other. In the second case, by contrast, we are dealing with comparable or identical phenomena, located within one area of research, such as metaphor in one specific area of grammar or usage. Converging evidence here has a bearing on one sphere of reality, and distinct studies with different methods and techniques can be compared against the same framework of assumptions. This is the reason why a map of the field may be helpful in ordering the issues of methodology that have been placed on the agenda over the past decade.

What is more, the map also helps to understand that converging evidence within each of these areas, or methodological pluralism, is an extremely worthy goal to pursue. The areas characterized by a semiotic approach to language and thought are typically researched by means of introspection and observation, with experimental manipulation trailing far behind. There is moreover a good deal of intuition and qualitative analysis which decides what counts as metaphorical in these areas. This is objected to by researchers who work in the areas characterized by a behavioral approach, who place a lot of emphasis on experimental manipulation and quantitative analysis. Such methods and techniques can also be applied in research on the symbolic structures of metaphor, but not many linguists have adopted them yet.

Furthermore, if linguists aspire to be students of cognition, they also need to address cognition by analyzing behavioral data in ways which can be taken seriously by cognitive and social scientists. Then they need to collect the appropriate behavioral data with adequate methods and techniques, and do the required qualitative and quantitative analyses. Moreover, they have a critical and constructive role to play in this area, as some linguists have objected to the lack of representativeness of some of the experimental materials utilized by psycholinguists doing behavioral work on metaphor processing. This might be improved when the role of observational approaches regarding usage is upgraded in such behavioral research. Experimental effects are rewarding, but their validity across all language materials, language users, and situations of language use may not always be quite clear.

There are hence plenty of opportunities for researchers in each of these areas to benefit from the methodological expertise of researchers in other areas when they wish to increase the quality of their knowledge in any of these areas. Manipulation does not always require sophisticated laboratory techniques, so that it is more open to linguists than they may think. And observation does not necessarily require advanced corpus-linguistic techniques to enable closer inspection of language data. Once this type of methodological imbalance between the two encompassing research areas of semiotic versus behavioral analysis has been redressed by a healthier methodological pluralism, it is possible to forge better and more comparable connections between the more distinct areas included by them so that phenomenological pluralism becomes less problematic.

The differentiation and integration of the three perspectives on metaphor in language have led to a new map of the field for metaphor identification. It provides the overall structure to the methodological analysis of metaphor identification in parts two and three. Part two will address metaphor identification in the four areas focusing on grammar, and part three will tackle metaphor identification in the four areas targeting usage. In each of these areas, I will examine how various methods may be utilized to collect converging evidence within that particular area of research. In that way, I hope to show that methodological pluralism as defined above is possible and exciting.

Finding metaphor in grammar, part two, will first be approached from a semiotic
perspective (Chapters 6 through 8), to be followed by one chapter on the behavioral perspective (Chapter 9). Chapters 6 through 8 will make a distinction between identifying linguistic forms of metaphor in grammar (Chapter 6) and identifying conceptual structures of metaphor in grammar (Chapters 7 and 8). Identifying metaphor in grammar as cognitive process and product will be addressed in Chapter 9, which will treat the linguistic forms and conceptual structures of metaphor in one go.

Finding metaphor in usage will be discussed in part three, in the same order. Chapters 10 and 11 will deal with the semiotic approach to metaphor in usage, approaching metaphor and usage as symbolic structure. Chapter 12 will then continue with a discussion of behavioral approaches to metaphor in usage, dealing with metaphor in usage as cognitive process and product. In these three chapters, linguistic and conceptual aspects of metaphor in usage will be addressed at the same time, albeit in consecutive stages.

Parts two and three will be preceded by a more general part one, which provides an overall impression of the various methodological issues that have to be tackled. Chapter 2 begins by highlighting what is in my opinion the most important difference between cognitive-linguistic approaches and other approaches, the role of conceptual metaphor. Researchers may be looking in deductive fashion across research areas, for instance for linguistic evidence supporting (or falsifying) the presence of a particular

conceptual metaphor, such as TIME IS SPACE or THE MIND IS A COMPUTER. This would be the prime example of phenomenological pluralism. But researchers may also follow other strategies, in which the identification of a conceptual metaphor may be an accidental by-product of the results. This global difference in research strategies can affect the comparability of the results of cognitive-linguistic and other approaches.

With a clear view of this essential characteristic of the cognitive-linguistic approach, we can turn to the three fundamental stages in the methodology of any research project: conceptualization, operationalization, and application. Chapter 3 will deal with the conceptualization of metaphor in language from the general cognitive perspective which has held reign over metaphor studies for the past 25 years, and discuss a number of competing theoretical definitions and models for metaphor as well as a number of fundamental issues that they raise. Chapter 4 will then address the question of how such models can be operationalized for doing empirical research on metaphor in the eight research areas distinguished above; I will look at, for instance, the derivation and ordering of criteria for metaphor that may be inferred from such models. Chapter 5 will finally look at the application of these operational definitions in data collection and analysis, and develop the distinction between introspection, observation and manipulation as well as between qualitative and quantitative analysis.

These chapters lay the foundations for the methodological analysis of metaphor in the various areas of grammar versus usage in parts two and three. Each of the separate chapters in parts two and three will follow the same order as the global discussion in part one: they will begin with deductive approaches setting out from conceptual metaphor, then turn to the most important issues of conceptualization and operationalization, and then to their application in introspection, observation, and manipulation. The concluding chapter will take a step back and look at the map of metaphor in grammar and usage for the future.

Meaning and Universal Grammar: Theory and Empirical FindingsVolume 1 edited by Cliff Goddard, Anna Wierzbicka, Kristi J. O'Neill (Studies in Language Companion Series, 60: John Benjamins) Meaning and Universal Grammar: Theory and Empirical Findings Volume 2 edited by Cliff Goddard, Anna Wierzbicka, Kristi J. O'Neill (Studies in Language Companion Series, 61: John Benjamins) This two-volume set of studies takes as its starting point an old idea: the idea that universal grammar is based on meaning. It seeks to give this idea a solid theoretical foundation, and to explore its viability through detailed empirical studies in a set of typologically divergent languages (Lao, Malay, Mandarin Chinese, Mangaaba‑ Mbula, Polish and Spanish).

As the twentieth century recedes, linguists seem increasingly to agree that the "anti‑semantic turn" inaugurated by Leonard Bloomfield and continued by Noam Chomsky was a wrong turn. It is now widely believed that the grammatical properties of a word follow, at least in large measure, from its meaning. But it is true to say that in generative linguistics, semantics is still in a theoretically and methodologically underdeveloped state. It would not be going too far to compare much modern semantic work to phonology before the advent of phonemic analysis. There is little agreement on fundamental methodological problems such as how to distinguish semantic invariants from contextual effects, how to distinguish polysemy from semantic generality, and, above all, on the nature of semantic representation itself. All too often, individual researchers cobble together frameworks to suit their immediate needs, with little regard to how these frameworks can be integrated into a comprehensive system.

Viewed against this background, the natural semantic metalanguage (NSM) framework stands out sharply (cf. Wierzbicka 1972, 1980, 1992, 1996; Goddard 1998; Goddard and Wierzbicka Eds 1994; and many other studies listed in this volume, chapter 1). It has been developed and refined over some thirty years. It provides a comprehensive system equally applicable to lexical, grammatical, and illocutionary meaning. It provides a clear and practical methodology for semantic analysis, which has proved itself in literally hundreds of studies in descriptive semantics.

The NSM framework is based on evidence supporting the idea that there is a set of simple, indefinable meanings ‑universal semantic primes ‑ which have concrete linguistic exponents in all the world's languages. By using universal semantic primes as a vocabulary for semantic description we can achieve semantic analyses which are maximally intelligible, testable, and intertranslatable, as well as enabling the maximum possible resolution of semantic analysis.

How does all this connect with universal grammar? The basic idea is a simple one ‑ that universal semantic primes have an inherent grammar which is the same in all languages. Recent work in the NSM framework (see chapter 1) has yielded a very substantial set of hypotheses about this postulated universal grammar. For example, the prime SAY is postulated to allow, universally, valency options of "addressee" and "locutionary topic", so that one can express, in any language, meanings equivalent to `X said something to Y', and `X said something about Z' (notwithstanding that the formal marking of these valency options will vary from language to language). For an example of a different kind, we can take the prime IF. This is postulated to occur universally in a biclausal frame, so that one could express in any language the semantic equivalent of a sentence like `if you do this now, something bad will happen' (notwithstanding that in some languages the exponent of IF coincides with the exponent of WHEN). For a third example, the prime Two is postulated universally to allow a "selective" frame, so that one could express in any language the semantic equivalent of a phrase like `two of these people' (notwithstanding that in some languages the word for Two is verbal in character).

These are highly concrete and testable claims about universal grammar, and it is the goal of this two‑volume collection of studies to test them, and many other equally concrete claims, against data from a set of divergent languages. The first two chapters (this volume) are of a general nature. Chapter One describes the theoretical framework of the natural semantic metalanguage approach, and summarises recent research. Chapter Two lays out in detail a large number of meaning‑based hypotheses about universal grammar. Then follow six studies of individual languages: Malay, Spanish and Mandarin Chinese (this volume) and Mangaaba‑Mbula, Polish and Lao (volume 1I). In each case the author identifies the inventory of semantic primes in that language and then works through the full set of NSM hypotheses about universal grammar. The final chapters (volume II, chapters 4 and 5) explore how the metalanguage of semantic primes can be used as a semantic foundation for grammatical typology, and identify directions for future research.

In a sense, the grammar of semantic primes in any language represents a "core grammar" of that language. The key concern of this set of studies is: Are the essential properties of these core grammars universal, and therefore present in an isomorphic form in all human languages? It is, of course, true that the realisation of the core grammar differs from language to language ‑ in word order, in morphological marking, and in constituent structure. Each of the individual language studies can also be read as a typological profile of that language, based on a principled, semantically‑based sample of questions about argument structure and valency options, complementation,, quantification, specifications of time and location, negation, and so on.

Each core grammar has, then, both a universal aspect and a language‑specific aspect. The approach taken in this set of studies allows us to separate these aspects and to study them both independently and in relation to one another. As ever, the quest for language universals and the study of language diversity and typology go hand in hand.

In short, the goal of this set of studies is to establish empirically that there is a universal core of grammar which is based on ‑ indeed, inseparable from meaning, and in this way to lay the foundations for an integrated, semantically-based approach to universal grammar and linguistic typology.

This set of studies is founded on the idea that universal grammar is based on indeed, inseparable from ‑ meaning. The theoretical framework is the natural semantic metalanguage (NSM) approach originated by Anna Wierzbicka over thirty years ago and developed since then in collaboration with Cliff Goddard and other colleagues (for references, see vol.1, chapter 1). The NSM framework is based on evidence supporting the idea that there is a set of simple, indefinable meanings, ‑ universal semantic primes ‑ which have concrete linguistic exponents in all the world's languages.

The NSM system is perhaps best known as the methodology for a large body of descriptive studies in cross‑linguistic semantics and pragmatics, but it also has fundamental implications for the theory of universal grammar. The key idea is that universal semantic primes have an inherent grammar (including combinatorics, valency and complementation options) that is the same in all languages.

Chapters One and Two of Volume I provide theoretical background and propose a substantial set of concrete hypotheses about universal grammar. Subsequent chapters of that volume test the viability of these hypotheses through studies of Malay, Spanish, and Mandarin Chinese.

This volume continues with three further languages: Mangaaba‑Mbula, Polish, and Lao. In each case the author identifies the inventory of semantic primes in that language and then works through the full set of NSM hypotheses about universal grammar. The final chapters explore how the metalanguage of semantic primes can be used as a semantic foundation for grammatical typology, and identify directions for future research.  

Old Church Slavonic Grammar 7^th  edition by Horace G. Lunt (Mouton de Gruyter) This description of the structure of Old Church Slavonic (OCS) is intended to present fully the important data about the language, without citing all the minutiae of attested variant spellings. The facts have been treated from the point of view of structural linguistics, but pedagogical clarity has taken precedence over the conciseness required for elegant formal description.

OCS was used over a period of some two hundred years and in various geographical parts of the Slavic world precisely at the time when the Slavic languages were undergoing rapid, fundamental, divergent changes. Some of these changes are doubtless reflected in the variant spellings in the few texts which have survived from this period, so that while most variations in grammar and vocabulary are the sorts of stylistic and idiosyncratic differences that are found in the standard or literary language of any single epoch, some important variant details result from different regional dialectal history. It has thus been necessary to include occasional references to historical and comparative linguistics in the first half of this book, although in principle these problems do not fall within the scope of a strictly descriptive, synchronic grammar.

It is necessary to normalize forms to present the grammatical structure as a consistent whole, and the normalization inevitably obscures the differences in the language of the various manuscripts. A clear picture of the different combinations of linguistic elements making up each of the texts is not to be achieved by lists of spelling variants or tables of percentages, but it is worthwhile to point out some of the striking variations. First‑hand acquaintance with the texts and constant comparison of variant readings is the only way to arrive at an understanding both of the underlying unity of the texts as a whole and of the major and minor differences between them.

Little mention is made here of another type of comparison‑the relationship of the OCS translated texts to the Greek originals. And yet it is in the Greek and in the translation technique that the explanations of hundreds of tiny problems (especially of syntax) are to be found, and certain major structural problems need to be posed in terms of the influence of Greek on OCS. However, so few students have enough Greek to profit by such comparisons that it did not seem worth the considerable space that would be required. Excellent work in this field is available, though some scholars tend to forget that even a poor translator is governed by the structure of the language into which he is translating. The "Notes on Syntax" in Chapter Six are offered on the premise that something is better than nothing. It is particularly in this area that translation techniques need to be analyzed.

After forty years of teaching OCS and related topics in the history and structure of modern Slavic languages, my views on the nature of language and the models for describing language have evolved away from the Bloomfieldian structuralism of my training. The data of OCS have not changed importantly from the material described by scholars a century ago, although some details from imprecise editions have been discarded and a few new details must be accounted for. I continue to believe that every language is a coherent structure, and that each language can be described in terms of static and dynamic elements and learned by novices who do not have the slightest knowledge of its history. Chapter Six is an entirely new addition to the original text. It contains a sketch of the development from Late Indo-­European to Late Common Slavic and will be of interest not only to Slavists but to linguists in general.

 

Headline 3

insert content here

WT Main | About WT | Review Links | Contact | Review Sources | Search