In this book, Stroik and Putnam take on Turing's challenge. They argue that the narrow syntax – the lexicon, the Numeration, and the computational system – must reside, for reasons of conceptual necessity, within the performance systems.
AUTHORS: Cann, Ronnie; Kempson, Ruth; Gregoromichelaki, Eleni TITLE: Semantics SUBTITLE: An Introduction to Meaning in Language PUBLISHER: Cambridge University Press YEAR: 2009
Alexandra B. Bagasheva, Department of British and American Studies, Sofia University ''St. Kliment Ohrdiski'', Sofia, Bulgaria
INTRODUCTION
The book ''Semantics: An Introduction to Meaning in Language'' as part of the series ''Cambridge Textbooks in Linguistics'' follows the layout and structure of a classical textbook. It can be used as a core reading for a fairly advanced course in Semantics or for individual study. It is aimed at senior graduate students and postgraduate students of linguistics, cognitive science, semantics, etc., as well as practicing linguists with or without a solid background in formal semantics. The book presents the dynamic developments in the formal paradigm of linguistic research in the past fifty years, as well as innovative current approaches that irrespective of their formal/functional/cognitive affiliations contribute to the understanding and adequate description of the dynamics of meaning creation and interpretation in natural languages. The book contains a preface, eight chapters, a bibliography and an index. Each chapter is supplemented with suggestions for further reading, impressive with their richness in terms of titles and with the specification of details which can be found in each recommended source. Practical tasks and exercises are interspersed into the body of the respective chapter, coming just before the summary reflections or at places where new features of the model or new analytical tools are introduced. If used as a course book it should be supplemented with exercises and problem-solving tasks.
The book contains detailed introductions to various logical formalisms as means for describing meaning in natural languages. Its overall theoretical and methodological make-up renders it specifically designed for readers with an interest in formal semantics. Its unique feature is that the introduction to logic is written by linguists, i.e. by the authors themselves, as they are convinced that only outsiders to the highly technical field of logic would know what exactly linguists need to know. The authors formulate ''a model from which to start to explore a formal account of the basis of meaning for natural language'' (p. 7). They justify their choice of method of inquiry by stating that ''if we want an explicit characterization of the nature of language…, we cannot fail to take up the challenge of constructing formal models to reflect the insights about language that we want to express'' (p. 7).
Each chapter is subdivided into relevant subchapters and subparts of the subchapters. The first three chapters guide the reader from general considerations of model building through the principles and fundamentals of propositional logic and predicate logic, model theoretic semantics and type logics to re-evaluation of central concerns in linguistic semantics. In the subsequent five chapters the focus is shifted from the nature of formalism onto central problems of semantic analysis in natural languages and how these can be fruitfully treated within the framework of formal semantics supplemented with innovative, dynamic approaches. These include for chapter four quantification analyzed through generalized quantifiers; chapter five - context-sensitive anaphor explicated with the help of Discourse Representation Theory (DRT), capable of modeling the dynamicity of on-line language processing; chapter six - time and tense/aspect accounted for with the help of a sorted ontology recognizing eventualities and their formalization as discourse referents in discourse representation structures (DRS); chapter seven - an innovative treatment of ellipsis as a unitary phenomenon, even if occurring over conversational turns and interlocutors, to be best analyzed with the tools of dynamic syntax (DS) which provides structural representation of what is needed from context for interpreting semantic content; and finally for chapter eight, studying what a word contributes to the compositional interpretation of meaning by positing words as clusters of context-particular construals. In the coda of the last chapter the authors conclude that the exceptional dynamicity of natural languages and the expressly intrinsically underspecified nature of natural language expressions in terms of denotational content require the joint efforts of pragmatic, psycholinguistic and cognitive research paradigms, disregarding contentions between functional and formalist methodologies, for an adequate description of the on-line, incremental composition of meaning generation/interpretation in linguistic interaction.
SUMMARY
Chapter 1: Preliminaries for model building After demonstrating the vastness of meaning in language and illustrating the unsuitability of an inductive approach to the nature of meaning in language, the authors define the requirements that should be met by a theory of semantics which has to provide an explication of interpretation for natural language expressions. The theory should in their view be able to account for the general phenomenon of ''context-dependence, which is in part conventionalized within a language'' (p. 2). The authors provide ample linguistic examples to illustrate context sensitivity in various areas of language and focus the reader's attention on the nature of compositionality which is complicated by mode of combination, variability in word meaning, immediate context, etc. Their contention is that an explicit characterization of the nature of language is best accomplished by the construction of a model which will provide a formal account of the basis of meaning for natural language. In order to explain its workings we need to account for entailment relations among expressions, which guarantees inferential reasoning. In view of language being just one possible input system, the authors posit as a desideratum for such a model the consideration of ''how good an interface it allows with general cognitive processing'' (p. 11). The model should also be equipped with appropriate tools to explicate the ''language-world relation as part of the grammar of natural languages'' (p. 11). Thus they subtly introduce the contention between representationalist and denotationalist accounts of natural language content. Then they establish the relation between truth-conditionality and inference in utterance exchanges in language. In looking for a way of correlating ''the syntax of expressions with their interpretation in a principled and formal way'' (p. 18), the authors resort to the postulates of the post-Gricean theory of Relevance. The authors acknowledge the universal applicability of the cognitive principle of relevance which underlies inferential communicative behaviour of humans. An adequate model of natural language semantics in their view should incorporate all these complex features without abandoning its formal nature which warrants the model's satisfactory level of explicitness and scientific rigor.
Chapter 2: The syntax of logical inference The chapter introduces the reader to the principles of propositional and predicate logic as the rules for inference at the level of sentences in the case of the former and at a level below the sentence in the case of the latter. Human beings are defined as ''inference engines'' (p. 27). Natural deduction is presented as the proof system that ''is closest to on-line human reasoning'' (p. 27). A formal model of patterns of inference is introduced, defining the connectives and quantifiers necessary for the construction of proofs. Unlike other languages, natural human language displays high ''sensitivity to the process by which information unfolds'' (p. 33) which should be adequately reflected in the model of analysis. This requirement on the model leads to the conclusion that ''natural-deduction proof systems, being explicitly defined to bring out the fine structure of how some valid inference is step-wise defined, give us the closest congener to the use of natural languages for human reasoning'' (p. 33). The ultimate utility of formal semantic models the authors see in their ability to express all major semantic relations in proof-theoretic, syntactic terms. Illustrating this, the authors define entailment as ''a relation between assumptions in a proof, and the conclusion deducible from that set'' (p. 62) and synonymy as proof-theoretic equivalence, the ''strict syntactic analogue of two formulae having identical truth conditions'' (p. 62). By defining these two important semantic relations in fully representationalist terms, i.e. system internally, the authors conclude that such a powerful formal pattern for scientific modeling can identify what ''we should take inference to be grounded in'' (p. 66).
Siding with Fodor, they contend that ''syntax is what is in the head, semantics is the relation between what's in the head and the world around that is outside it'' (p. 28). In view of the fact that they have already provided a syntactic account of valid inference, the three scholars go on to evaluate the hypothesis of the possibility of ''a proof-theoretic characterization of inference defined over those formulae to determine the necessary entailment relations displayed in natural language'' (p. 66).
To pursue their goal they turn in the next chapter to the study of the semantics of the formal system, which implies a movement towards a denotational account of the mapping from strings defined by the language onto an interpretation.
Chapter 3: The semantics of logical inference: models and semantic types In this chapter the authors introduce the basic concepts of model-theoretic semantics. They present model theories for both propositional and predicate logic, and redefine inferential relations in semantic terms. They proceed with a critical evaluation of syntactic and semantic characterizations of inference. After establishing the definitional co-extensiveness of structural and denotational characterizations of inference in classical logics, the three semanticians address a central and controversial question of the nature characterizing inferential relations in natural language. The dilemma encompasses a denotational possibility where inference in natural language is accounted for on the basis of the denotational content of entailing and entailed sentences (a semantic approach) and a representationalist possibility in terms of ''the manipulation of cognitive representations mind-internally' (p. 84) (an essentially syntactic account). The authors identify two fundamental questions which further complicate the characterization of inference in natural languages: a) the need to account for context sensitivity and the issue of how this could be formally modeled and b) the ''lack of any notion of concept in formal-language semantics with which to get any handle on what the intrinsic content of a word might be'' (p. 84). In order to address these questions the scholars try to establish mechanisms for improving the match between predicate-logic types and natural-language categories. To this end they utilize the lambda operator to make compatible the syntax of functional application with that of predicate logic. The application of Lambda Conversion yields a fully compositional account of the semantic representations of phrases and sentences. The authors introduce the Curry-Howard isomorphism as the underlying principle of typing natural-language expressions. With this correspondence recognized as a core property of the syntax-semantics interface, Cann, Kempson and Gregoromichelaki offer means for the manipulation of typed expressions in order to derive representations of the semantic content of natural-language expressions in a compositional manner. Despite this obvious success of their enterprise at the end of the chapter the authors recognize yet another caveat - the logical type to be assigned to quantifying expressions in natural language, as they recognize a major structural difference between quantification in predicate logic where quantifiers are propositional operators binding open propositional formulae and that in natural-language expressions which are subparts of noun phrase expression. This becomes the focus of their next chapter.
Chapter 4: Quantification and plurality In this chapter the reader is presented with a brief introduction to Montague's account of quantification and generalized-quantifier theory. After a critical evaluation of the proposed analyses, the authors face the serious problems that plurality poses to formal semantic theorizing. In discussing the treatment of 'many' as underspecified the authors contend that ''pragmatic choices must be allowed to precede semantic interpretation'' (p.114). They recognize context-dependence as intrinsic to the content of natural-language utterances. They acknowledge that ''we need to allow semantics and pragmatics to interconnect much more thoroughly than is typically assumed in formal theories of semantics'' (p. 114). The next step for them is to seek an answer to the question of how the ''feeding relation between contextual parameters and denotational content'' should be defined. After introducing merological structures in the ontology of the model in order to provide a unified account for cardinal quantifiers, plurals and mass terms without increasing the complexity of semantic types, the authors conclude that besides context sensitivity, the changing human point of view, i.e. changes in the conceptualization of the things denoted should figure prominently in accounts of the semantics of natural languages. Before addressing the latter issue, in the subsequent two chapters, the authors try to propose a systemic explanation of highly context-dependent language expressions and phenomena.
Chapter 5: Anaphor, discourse and context The first such problematic phenomenon identified and discussed in detail in the fifth chapter is anaphora. First problems in the analysis of pronouns of all types (E-type pronouns, indexical uses of pronouns, donkey pronouns, etc.) are discussed. Then a solution is proposed for a possible unified treatment of different types of anaphor. The solution is Discourse Representation Theory which in addition to truth-conditional content, takes into account ''the impact of the context in which processing of discourse occurs'' (p. 147). Inter-sentential anaphora is identified as a crucial testing area for the adequacy of accounts of meaning in natural language. Its adequate explication can only be achieved by an account which distinguishes ''the contribution of distinct ways of presentation'' (p. 149). Central in this chapter is the recognition of the legitimacy of the way humans process language and how this affects the interpretations as an object of enquiry for natural-language semantics. The research programme equipped with tools necessary to incorporate ''changes in the information states of the interpreters'' (p. 167) in the scholars' view is Dynamic Semantics. Thus they are able to provide means for analyzing semantic representations independent of their model-theoretic interpretation. Discourse Representation Structures are recognized as an analytical tool suitable for capturing dynamic change intrinsic in the composition of the grammatical categories of tense and aspect which become the focus of discussion in the next chapter.
Chapter 6: Time, tense and events In this chapter the purely extensional semantic systems offered up to this moment are augmented with a tensed logic which is systematically proven inadequate and insufficient to account for the representation of the intricacies and fine degree of granularity with which natural language interpretation has to be characterized. The authors present a simple tensed logic operating with an interpretation of tense in terms of logical operators over formulae, where the existential quantification over times is achieved at the level of the metalanguage. Parametrising each-truth-conditional definition to a particular time however falls short of providing an adequate formal semantics tool for the treatment of temporal reference in natural language. To provide a better analysis, after introducing a new sorted ontology in the model to accommodate events, the authors present a classification of types of eventualities which are conceptualized in terms of part-whole relations. They emphasize that eventuality distinctions can only be determined by taking into account the particular linguistic and non-linguistic context of predicates. The scholars adopt Reichenbach's interpretation of tenses in terms of three reference times and suggest that temporal intervals be treated as discourse referents within the framework of Discourse Representation Theory where sortal construals of eventualities are directly represented in discourse representation structures. The analysis gets further complicated with the necessity to incorporate accounts of aspectual considerations. This posits further a necessity to account for inferences over the unexpressed semantic properties of the individual words in an utterance and possible different modes of construal. The authors conclude the chapter with the claim that Discourse Representation Theory with its representationalist flavor might provide a suitable analytical tool encompassing semantic explication and the characterization of content and context simultaneously.
Chapter 7: Ellipses as a window on context In this chapter Cann, Kempson and Gregoromichelaki recapitulate the enrichment of the flexibility of the formal-semantic methodologies presented so far and postulate the need of at least two levels of representation: a) of structures in syntax and b) of structure in semantics, ''reflecting structural aspects of assigned content'' (p. 209). After presenting current debates on different types of ellipsis, the authors contend that ''the most promising way to get an appropriately integrated account of ellipsis is to give a procedural slant to the concept of context'' (p. 213). For this they turn to Dynamic Syntax as a model of natural language content which postulates ''a level of representation that reflects what is needed for semantic interpretation'' (p. 225). In this model context is construed as encompassing a complex of content, structure and the actions used to build up interpretation. Within the model, structures are built up in a strictly incremental way and the syntax of the grammar is this same process of building up the structures. Thus the articulation of the natural-language syntax is the articulation of constraints on growth of interpretation. The overall conclusion is that interpretations for natural-language expressions need to be in terms of the construction of representations rather than in terms of the content assigned relative to context.
Chapter 8: What a word can mean No matter how far one extends the concept of context-sensitivity and procedural dependences, applying the principle of compositionality only as a guiding principle, the meaning of a sentence, in the view of the authors is ''a (monotonic) function of the meanings of all the words it contains together with the constructions used to put the words in order'' (p. 250). This in turn implies that ''lexical meaning must be seen as forming the basis for all (linguistic) meaning'' (p. 250). After reviewing semantic minimalism and the generative lexicon paradigms in accounting for lexical meaning, the three semanticians guide the reader towards new avenues of research in word meaning taking into account context variability. They define context as ''the cognitive context relative to which content decisions are progressively made'' (p. 259). Then the authors suggest a formalism introduced by Larsson interpreting word meaning as concept-clusters as a viable tool. Seeking to come to grips with the variability of word meaning and grammar, while preserving amenability to formalization, the scholars promote Cooper's Type Theory with Records as a means to capture the dynamics of constructing information types for analyzing word meaning extensibility. The authors go so far as to challenge the teleology of the concept of sentence meaning in determining the systematic compositionality of content as expressed in natural language. Taking into account formal theoretic, pragmatic and syntactic considerations, the three scholars reformulate the research agenda for formal semantics in terms of providing the tools to analyze human language adequately not as corresponding to any formal system of logic, but as ''an intrinsically dynamical system, a system whose parts are available for use in ways that allow open-ended but not unlimited modes of construal'' (p. 272). To pursue this agenda successfully, Cann, Kempson and Gregoromichelaki invite scholars of all persuasions from the whole spectrum of inquiry into language (pragmatics, cognitive science, psychology, etc.) to join forces in meeting the challenge of unraveling the mystery of meaning in human language.
EVALUATION
''…a philosophically interesting story may be told about the relationship between meaning, truth, and the use of language. An important research programme within linguistics and philosophy of language, sometimes called 'natural language semantics' or 'formal semantics', is grounded on a particular notion of how such a story, or at least an important portion of it, is supposed to go'' (Predelli 2005: 1) The story offered in the book reviewed here in addition to the features described provides a unified perspective and an overall dynamic model able to satisfy the requirements to incorporate formal accounts of context-sensitivity and item-concept meaning contributions to compositionality, while at the same time keeping track of the relevance theoretic assumption that ''the cognitive system…is constrained to maximize relevance by engaging in minimum cognitive effort for adequate inferential effects'' (p. 47)
Despite its claimed introductory nature, the book is far from a ''baptismal'' textbook in Cruse's terms (Cruse 2000: x). The book presents a step-by-step approach, leading towards the construction of a system for a formally explicit but cognitively informed description of the nature of meaning in natural language in on-line communication. It is cogently and clearly written with precise cross-referencing and a useful index. Problems are picked up at relevant places and discussed anew in view of the currently elaborated argument.
The uniqueness of the perspective taken in the book is the attempt to adapt and evolve formal tools/methodologies so that they can capture new insights stemming from different modes of studying meaning in natural languages. One significant feature of the book is the smooth, natural reconciliation of denotationalist and representationalist stances in semantics. They have put into practice Pietroski's recommendation that ''one shouldn't try to define the scope of semantic theorizing in advance of investigation... One cannot determine a priori which facts a semantic theory should explain. In any domain, what a theory SHOULD explain depends on what gets discovered'' (Pietroski 2005: 26-27).
The presentation of the dynamic developments in the formal semantic paradigm is neither panoramic, nor exhaustive. It is extremely valuable as it is tailored so as to make prominent the unique perspective from which the authors discuss hotly debated issues in semantics, offering a unified formal but dynamic approach. Most valuable is the recognition of the context sensitivity of language and the conception of language as a formal abstraction on the world, as underspecified prompts for interpretation construction as central research issues in a workable agenda for formal semantics.
On the cover of the book one reads that the exposition assumes only a basic knowledge of concepts in semantics and pragmatics but at many places where caveats are offered only to be discarded later, analyses are presented, and the discussion presupposes profound knowledge of syntax and syntactic terminology. The elaboration of the argument also presupposes familiarity with disputes over the syntax-semantic bootstrapping debates in linguistics and fine-grained knowledge of the properties of the syntax and/or semantics of formal systems, as well as the nature of intricate correspondences between an object language and a metalanguage. The argumentation presupposes solid knowledge of natural language syntax - various theory-specific terms are freely used without being defined. For the full appreciation of the textbook prior knowledge of basic logic is necessary. For people of non-formalist persuasion the book is a demanding but stimulating read. For people coming from different linguistic backgrounds, the book lacks sufficient contextualization of the place of formal theorising in contemporary debates in semantics - the cognitive paradigm, structural semantics, etc.
The textbook might have benefited from a glossary section at the end clarifying the most focal terms.
''Formal semantics is based on an even more disastrous equivocation of form and notation, as well as on a complete misunderstanding of the relationship between form and meaning. It takes meanings and 'formalizes' them, i.e. expresses them in a notation. In doing so it is aware that form and meaning are inextricably linked, but it believes that by expressing meanings in a notation it has captured those meanings formally and thus explained them scientifically. It is, of course, sheer folly. Notations are not forms, they are mere notations; and the relationship between form and meaning in natural language is that form creates meaning - natural language is itself a formalization of the universe. The meanings we perceive are determined by the forms through which we see them. And by scrutinising those forms carefully and scientifically we can gain greater insight into the meanings which they create/express. But to translate meanings into a notation is a trivial and senseless activity.'' (Beedham 2005: 93) What ''Semantics: An Introduction to Meaning in Language'' achieves is a powerful and convincing refutation of Beedham's views, revitalizing formal semantics with a new challenging agenda for research. The authors offer a solid formalist account of how and what natural language as a formalization of the universe offers to reasoning agents in their interactive composition and interpretation of meaning with a full consideration of the cognitively relevant contributions of both content and context irrespective of their nature (visual, linguistic, etc.). Reading this book is a profitable intellectual investment.
REFERENCES
Beedham, Christopher (2005) Language and Meaning: The Structural Creation of Reality, John Benjamins Publishing Company, Amsterdam/Philadelphia. Cruse, Alan (2000) Meaning in Language: An Introduction to Semantics and Pragmatics, Oxford University Press. Pietroski, P. (2005) Events and semantic architecture, Oxford University Press. Predelli, Stefano (2005) Contexts: Meaning, Truth and the Use of Language, Oxford University Press.
ABOUT THE REVIEWER
ABOUT THE REVIEWER:
Alexandra Bagasheva teaches General Linguistics and English Syntax at the
Department of British and American Studies at Sofia University, Bulgaria.
Her main interests broadly lie in the areas of cognitive and functional
linguistics, typology, linguistic anthropology, the semantics of compounds
as the result of a complex double blending process, and the
semantics-cognition-pragmatics interface.
