Semantics Research Paper

1. Introduction

Traditionally, the main focus of linguistic semantics has been on word meaning, or lexical semantics. Since classical times writers have commented on the fact, noticed surely by most reflecting individuals, that the meaning of words changes over time. Such observations are the seeds of etymology, the study of the history of words. Over longer stretches of time, such changes become very obvious, especially in literate societies. Words seem to shift around: some narrow in meaning such as English ‘queen,’ which earlier meant ‘woman, wife’ but now means ‘wife of a king.’ Others become more general, while still others shift to take on new meaning or disappear altogether. Words are borrowed from language to language. The study of such processes is now part of historical semantics (Fisiak 1985). Another motivation for the study of word meaning comes from dictionary writers as they try to establish meaning correspondences between words in different languages, or, in monolingual dictionaries, seek to provide definitions for all the words of a language in terms of a simple core vocabulary. In lexicology similarities and differences in word meaning are a central concern.

The principled study of the meaning of phrases and sentences has only become established in linguistics relatively recently. Thus it is still common for descriptive grammars of individual languages to contain no separate section on semantics other than providing a lexicon. Nonetheless it has always been clear that one can identify semantic relations between sentences. Speakers of English know from the semantics of negation that nominal negation has different effects than sentence negation, so that ‘No-one complained’ may aptly be used to answer ‘Who complained?,’ while ‘Someone did not complain’ may not. Two sentences may seem to say essentially the same thing, even be paraphrases of each other, yet one may be more suited to one context than another–like the pair ‘Bandits looted the train’ and ‘The train was looted by bandits.’ A single sentence may be internally inconsistent, such as ‘Today is now tomorrow,’ or seem to be repetitive or redundant in meaning, such as ‘A capital city is a capital city.’ Another feature of sentence meaning is the regularity with which listeners draw inferences from sentences, and often take these to be part of the meaning of what was said. Some inferential links are very strong, such as entailment. Thus we say that ‘Bob drank all of the beer’ entails ‘Bob drank some of the beer’ (assuming the same individual Bob, beer, etc.), because it is hard to think of a situation where acceptance of the second sentence would not follow automatically from acceptance of the first. Other inferential links are weaker and more contextually dependent: from the utterance ‘Bob drank some of the beer’ it might be reasonable to infer ‘Bob didn’t drink all of the beer,’ but it is possible to think of situations where this inference would not hold. We might say that a speaker of the first sentence is implying the second, in a certain context. Speakers of all languages regularly predict and use such inferential behavior to convey their meaning, such that often more meaning seems to be communicated than is explicitly stated. All these aspects of sentence meaning are under study in various semantic frameworks.

Semanticists share with philosophers an interest in key issues in the use of language, notably in reference. We use this term to describe the way in which speakers can pick out, or name, entities in the world by using words as symbols. Many scholars, especially formal semanticists, accept Frege’s distinction between reference (in German, Bedeutung) and sense (Sinn); see Frege (1980). Reference is the act of identifying an entity (the referent) while sense is the means of doing so. Two different linguistic expressions such as ‘the number after nine’ and ‘the number before eleven’ differ in sense but they both share the same referent, ‘ten.’ For semanticists it is particularly interesting to study the various mechanisms that a language offers to speakers for this act of referring. These include names such as ‘Dublin,’ nouns such as ‘cat,’ which can be used to refer to a single individual, ‘your cat,’ or a whole class, ‘Cats are carnivorous,’ quantified nominals such as ‘many cats,’ ‘some cats,’ ‘a few cats,’ etc. Linguists as well as philosophers have to account for language’s ability to allow us to refer to nonexistent and hypothetical referents such as ‘World War Three,’ ‘the still undiscovered cure for cancer,’ ‘the end of the world.’

Semanticists also share interests with psychologists, for if sense is the meaning of an expression, it seems natural to many semanticists to equate it with a conceptual representation. Cognitive semanticists, in particular, (for example Lakoff 1987, Talmy 2000), but also some generative linguists (Jackendoff 1996), seek to explore the relationship between semantic structure and conceptual structure. One axis of the debate is whether words, for example, are simply labels for concepts, or whether there is a need for an independent semantic interface that isolates just grammatically relevant elements of conceptual structure. As Jackendoff (1996) points out, many languages make grammatical distinctions corresponding to the conceptual distinctions of gender and number, but few involve distinctions of colour or between different animal species. If certain aspects of concepts are more relevant to grammatical rules, as is also claimed by Pinker (1989), this may be justification for a semantic interface.

2. Approaches To Meaning

Even in these brief remarks we have had to touch on the crucial relationship between meaning and context. Language of course typically occurs in acts of communication, and linguists have to cope with the fact that utterances of the same words may communicate different meanings to different individuals in different contexts. One response to this problem is to hypothesize that linguistic units such as words, phrases, and sentences have an element of inherent meaning that does not vary across contexts. This is sometimes called inherent, or simply, sentence meaning. Language users, for example speakers and listeners, then enrich this sentence meaning with contextual information to create the particular meaning the speaker means to convey at the specific time, which can then be called speaker meaning. One common way of reflecting this view is to divide the study of meaning into semantics, which becomes the study of sentence meaning, and pragmatics, which is then the study of speaker meaning, or how speakers use language in concrete situations. This is an attempt to deal with the tension between the relative predictability of language between fellow speakers and the great variability of individual interpretations in interactive contexts. One consequence of this approach is the view that the words that a speaker utters underdetermine their intended meaning.

Semantics as a branch of linguistics is marked by the theoretical fragmentation of the field as a whole. The distinction between formal and functional approaches, for example, is as marked in semantics as elsewhere. This is a large subject to broach here but see Givon (1995) and Newmeyer (1998) for characteristic and somewhat antagonistic views. One important difference is the attitude to the autonomy of levels of analysis. Are semantics and syntax best treated as autonomous areas of study, each with its own characteristic entities and processes? A related question at a more general level is whether linguistic processes can be described independently of general psychological processes or the study of social interaction. Scholars in different theoretical frameworks will give contradic- tory answers to these questions of micro-and macro-autonomy. Autonomy at both levels is characteristic of semantics within generative grammar; see, for example, Chomsky (1995). Functionalists such as Halliday (1996) and Harder (1996) would on the other hand argue against microautonomy, suggesting that grammatical relations and structure cannot be under- stood without reference to semantic function. They also seek motivation for linguistic structure in the dynamics of communicative interaction. A slightly different external mapping is characteristic of cognitive semantics, for example Lakoff (1987) and Langacker (1987), where semantic structures are correlated to conceptual structures.

Another dividing issue in semantics is the value of formal representations. Scholars are divided on whether our knowledge of semantics is sufficiently mature to support attempts at mathematical or other symbolic modeling; indeed, on whether such modeling serves any use in this area. Partee (1996), for example, defends the view of formal semanticists that the application of symbolic logic to natural languages, following in particular the work of Montague (1974), represents a great advance in semantic description. Jackendoff (1990), on the other hand, acknowledges the value of formalism in semantic theory and description but argues that formal logic is too narrow adequately to describe meaning in language. Other scholars, such as Wierzbicka (1992), view the search for formalism as premature and distracting. There has been an explosive increase in the research in formal semantics since Montague’s (1974) proposal that the analysis of formal languages could serve as the basis for the description of natural languages. Montague’s original theory comprised a syntax for the natural language, say English, a syntax for the logical language into which English should be translated (intensional logic), rules for the translation, and rules for the semantic interpretation of the intensional logic. This and subsequent formal approaches are typically referential (or denotational) in that their emphasis is on the connection of language with a set of possible worlds, including the real, external world and the hypothetical worlds set up by speakers. Crucial to this correspondence is the notion of truth, defined at the sentence level. A sentence is true if it correctly describes a situation in some world. In this view, the meaning of a sentence is characterized by describing the conditions which must hold for it to be true. The central task for such approaches is to extend the formal language to cope with the semantic features of natural language while maintaining the rigor and precision of the methodology. See the papers in Lappin (1996) for typical research in this paradigm.

Research in cognitive semantics presents an alternative strategy. Cognitive semanticists reject what they see as the mathematical, antimentalist approach of formal semantics. In their view meaning is described by relating linguistic expressions to mental entities, conventionalized conceptual structures. These semanticists have proposed a number of conceptual structures and processes, many deriving from perception and bodily experience and, in particular, conceptual models of space. Proposals for underlying conceptual structures include image schemas (Johnson 1987), mental spaces (Fauconnier 1994), and conceptual spaces (Gardenfors 1999). Another focus of interest is the processes for extending concepts, and here special attention is given to metaphor. Lakoff (1987) and Johnson (1987) have argued against the classical view of metaphor and metonymy as something outside normal language, added as a kind of stylistic ornament. For these writers metaphor is an essential element in our categorization of the world and our thinking processes. Cognitive semanticists have also investigated the conceptual processes which reveal the importance of the speaker’s perspective and construal of a scene, including viewpoint shifting, figure-ground shifting, and profiling (Langacker 1987).

3. Topics In Sentence Semantics

Many of the semantic systems of language, for example tense, aspect, mood, and negation, are marked grammatically on individual words such as verbs. However, they operate over the whole sentence. This ‘localization’ is the reason that descriptive grammars usually distribute semantic description over their analyses of grammatical forms. Such semantic systems offer the speaker a range of meaning distinctions through which to communicate a message. Theoretical semanticists attempt to characterize each system qua system, as in for example Verkuyl’s (1993) work on aspect and Hornstein’s (1990) work on tense. Typological linguists try to characterize the variation in such systems across the world’s languages, as in the studies of tense and aspect by Comrie (1976, 1985), Binnick (1991), and Bybee et al. (1994). We can sketch some basic features of some of these systems.

3.1 Situation Type And Aspect

Situation type and aspect are terms for a language’s resources that allow a speaker to describe the temporal ‘shape’ of events. The term situation type is used to describe the system encoded in the words of a language, while aspect is used for the grammatical systems which perform a similar role. To take one example, languages typically allow speakers to describe a situation either as static, as in ‘The bananas are ripe,’ or as dynamic, as in ‘The bananas are ripening.’ Here the state is the result of the process but the same situation can be viewed as more static or dynamic, as in ‘The baby is asleep’ and ‘The baby is sleeping.’ As these examples show, this distinction is lexically marked: in English, for example, adjectives are typically used for states, and verbs for dynamic situations. There are, however, a group of stative verbs, such as ‘know,’ ‘understand,’ ‘love,’ ‘hate,’ which describe static situation types. There are a number of semantic distinctions typically found amongst dynamic verbs, for example the telic/atelic (bounded/unbounded) distinction and the punctual/durative distinction. Telic verbs describe processes which are seen as having a natural completion, which atelic verbs do not. A telic example is ‘Matthew was growing up,’ and an atelic example is ‘Matthew was drinking.’ If these procsses are interrupted at any point, we can automatically say ‘Matthew drank,’ but not ‘Matthew grew up.’ However, atelic verbs can form telic phrases and sentences by combining with other grammatical elements, so that ‘Matthew was drinking a pint of beer’ is telic. Durative verbs, as the term suggests, describe processes that last for a period of time, while punctual describes those that seem so instantaneous that they have no detectable internal structure, as in the comparison between ‘The man slept’ and ‘The light flashed.’ As has often been observed, if an English punctual verb is used with a durative adverbial, the result is an iterative meaning, as in ‘The light flashed all night,’ where we understand the event to be repeated over the time mentioned.

Situation type typically interacts with aspect. Aspect is the grammatical system that allows the speaker choices in how to portray the internal temporal nature of a situation. An event, for example, may be viewed as closed and completed, as in ‘Joan wrote a book,’ or as an ongoing process, perhaps unfinished, as in ‘Joan was writing a book.’

The latter verb form is described as being in the progressive aspect in English, but similar distinctions are very common in the languages of the world. In many languages we find described a distinction between perfective and imperfective aspects, used to describe complete versus incomplete events; see Bybee et al. (1994) for a survey. As mentioned above, aspect is intimately associated both with situation type and tense. In Classical Arabic the perfective is strongly associated with past tense (Comrie 1976, Binnick 1991). In English, for example, stative verbs are typically not used with progressive aspect, so that one may say ‘I know some French’ but not ‘I am knowing some French.’ Staying with the progressive, when it is used in the present tense in English (and in many other languages) it carries a meaning of proximate future or confident prediction as in ‘We’re driving to Los Angeles’ or ‘I’m leaving you.’ The combination of the three semantic categories of tense, situation type, and aspect produces a complex system that allows speakers to make subtle distinctions in relating an event or describing a situation.

3.2 Modality

Modality is a semantic system that allows speakers to express varying attitudes to a proposition. Semanticists have traditionally identified two types of modality. One is termed epistemic modality, which encodes a speaker’s commitment to, or belief in, a proposition, from the certainty of ‘The ozone layer is shrinking’ to the weaker commitments of ‘The ozone layer may/might/could be shrinking.’ The second is deontic modality, where the speaker signals a judgment toward social factors of obligation, responsibility, and permission, as in the various interpretations of ‘You must/can/may/ought to borrow this book.’ These examples show that similar markers, here auxiliary verbs, can be used for both types. When modality distinctions are marked by particular verbal forms, these are traditionally called moods. Thus many languages, including Classical Greek and Somali, have a verb form labeled the optative mood for expressing wishes and desires. Other markers of modality in English include verbs of propositional attitude, as in ‘I know/believe/think/doubt/that the ozone layer is shrinking,’ and modal adjectives, as in ‘It is certain/probable/likely/possible that the ozone layer is shrinking.’

A related semantic system is evidentiality, where a speaker communicates the basis or source for presenting a proposition. In English and many other languages this may be done by adding expressions like ‘allegedly,’ ‘so I’ve heard,’ ‘they say,’ etc., but certain languages mark such differences morphologically, as in Makah, a Nootkan language spoken in Washington State (Jacobsen 1986, p. 10):

wiki caxaw: ‘It’s bad weather’ (seen or experienced directly);

wiki caxakpi d: ‘It looks like bad weather’ (inference from physical evidence);

wiki caxakqad/ I: ‘It sounds like bad weather’ (on the evidence of hearing); and

wiki caxakwa d: ‘I’m told there’s bad weather’ (quoting someone else).

3.3 Semantic Roles

This term describes the speaker’s semantic repertoire for relating participants in a described event. One influential proposal in the semantics literature is that each language contains a set of semantic roles, the choice of which is partly determined by the lexical semantics of the verb selected by the speaker. A characteristic list of such roles is:

agent: the initiator of some action, capable of acting with volition;

patient: the entity undergoing the effect of some action, often undergoing some change in state;

theme: the entity which is moved by an action, or whose location is described;

experiencer: the entity which is aware of the action or state described by the predicate but which is not in control of the action or state;

beneficiary: the entity for whose benefit the action was performed;

instrument: the means by which an action is performed or something comes about;

location: the place in which something is situated or takes place;

goal: the entity towards which something moves;

recipient: the entity which receives something; and

source: the entity from which something moves.

In an example like ‘Harry immobilized the tank with a broomstick,’ the entity Harry is described as the agent, the tank as the patient, and the broomstick as the instrument. These roles have also variously been called deep semantic cases, thematic relations, participant roles, and thematic roles.

One concern is to explain the matching between semantic roles and grammatical relations. In many languages, as in the last example, there is a tendency for the subject of the sentence to correspond to the agent and for the direct object to correspond to a patient or theme; an instrument often occurs as a prepositional phrase. Certain verbs allow variations from this basic mapping, for example the we find with English verbs such as ‘break’: ‘The boy broke the window with a stone’ (subject = agent); ‘The stone broke the window’ (subject instrument); ‘The window broke’ (subject = patient). Clearly verbs can be arranged into classes depending on the variations of mappings they allow, and not all English verbs pattern like ‘break.’ We can say ‘The admiral watched the battle with a telescope,’ but ‘The telescope watched the battle’ and ‘The battle watched’ sound decidedly odd.

From this literature emerges the claim that certain mappings are more natural or universal. One proposal is that, for example, there is an implicational hierarchy governing the mapping to subject, typically such as: agent > recipient/benefactive > theme/patient > instrument > location. In such a hierarchy each left element is more preferred than its right neighbor, so that moving rightward along the string gives us fewer expected subjects. The hierarchy also makes certain typological claims: if a language allows a certain semantic role to be subject, it will allow all those to its left. Thus if we find that a language allows the role instrument to be subject, we predict that it allows the roles to the left, but we do not know if it allows location subjects.

One further application of semantic roles is in lexical semantics, where the notion allows verbs to be classified by their semantic argument structure. Verbs are assigned semantic role templates or grids by which they may be sorted into natural classes. Thus, English has a class of transfer, or giving verbs, which in one type includes the verbs ‘give,’ ‘lend,’ ‘supply,’ ‘pay,’ ‘donate,’ ‘contribute.’ These verbs encode a view of the transfer from the perspective of the agent and may be assigned the pattern < agent, theme, recipient > , as in ‘The committee donated aid to the famine victims.’ A second subclass of these transfer verbs encodes the process from the perspective of the recipient. These verbs include ‘receive,’ ‘accept,’ ‘borrow,’ ‘buy,’ ‘purchase,’ ‘rent,’ ‘hire,’ and have the pattern < recipient, theme, source >, as in ‘The victims received aid from the committee.’

3.4 Entailment, Presupposition, And Implication

These terms relate to types of information a hearer gains from an utterance but which are not stated directly by the speaker. These phenomena have received a lot of attention because they seem to straddle the putative divide between semantics and pragmatics described above, and because they reveal the dynamic and interactive nature of understanding the meaning of utterances. Entailment describes a relationship between sentences such that on the basis of one sentence, a hearer will accept a second, unstated sentence purely on the basis of the meaning of the first. Thus sentence A entails sentence B, if it is not possible to accept A but reject B. In this view a sentence such as ‘I bought a dog today’ entails ‘I bought an animal today’; or ‘President Kennedy was assassinated yesterday’ entails ‘President Kennedy is now dead.’ Clearly these sentential relations depend on lexical relations: a speaker who understands the meaning of the English word ‘dog’ knows that a dog is an animal; similarly the verb ‘assassinate’ necessarily involves the death of the unfortunate object argument. Entailment then is seen as a purely automatic process, involving no reasoning or deduction, but following from the hearer’s linguistic knowledge. Entailment is amenable to characterization by truth conditions. A sentence is said to entail another if the truth of the first guarantees the truth of the second, and the falsity of the second guarantees the falsity of the first.

Presupposition, on the other hand, is a more complicated notion. In basic terms, the idea is simple enough: that a speaker communicates certain assumptions aside from the main message. A range of linguistic elements communicates these assumptions. Some, such as names, and definiteness markers such as the articles ‘the’ and ‘my,’ presuppose the existence of entities. Thus ‘James Brown is in town’ presupposes the existence of a person so called. Other elements have more specific presuppositions. A verb such as ‘stop’ presupposes a preexisting situation. So a sentence ‘Christopher has stopped smoking’ presupposes ‘Christopher smoked.’ If treated as a truth-conditional relation, presupposition is distinguished from entailment by the fact that it survives under negation: ‘Christopher has not stopped smoking’ still presupposes ‘Christopher smoked,’ but the sentence ‘I didn’t buy a dog today’ does not entail ‘I bought an animal today.’

There are a number of other differences between entailment and presupposition that cast doubts on the ability of a purely semantic, truth-conditional account of the latter. Presuppositions are notoriously context sensitive, for example. They may be cancelled without causing an anomaly: a hearer can reply ‘Christopher hasn’t stopped smoking, because he never smoked’ to cancel the presupposition by what is sometimes called metalinguistic negation. This dependency on context has led some writers to propose that presupposition is a pragmatic notion, definable in terms of the set of background assumptions that the speaker assumes is shared in the conversation. See Beaver (1997) for discussion.

A third type of inference is Grice’s conversational implicature (1975, 1978). This is an extremely contextsensitive type of inference which allow participants in a conversation to maintain coherence. So, given the invented exchange below,

A: Did you give Mary the book?

B: I haven’t seen her yet.

It is reasonable for A to infer the answer ‘no’ to her question. Grice proposed that such inferences are routinely relied on by both speakers and hearers, and that this reliance is based on certain assumptions that hearers make about a speaker’s conduct. Grice classified these into several different types, giving rise to different types of inference, or, from the speaker’s point of view, what he termed implicatures. The four main maxims are called Quality, Quantity, Relevance, and Manner (Grice 1975, 1978). They amount to a claim that a listener will assume, unless there is evidence to the contrary, that a speaker will have calculated their utterance along a number of parameters: they will tell the truth, try to estimate what their audience knows, and package their material accordingly, have some idea of the current topic, and give some thought to their audience being able to understand them. In our example above, it is A’s assumption that B’s reply is intended to be relevant that allows the inference ‘no.’

Implicature has three characteristics: first, that it is implied rather than said; second, that its existence is a result of the context i.e., the specific interaction. There is no guarantee that in other contexts ‘I haven’t seen her’ will be used to communicate ‘no.’ Third, implicature is cancelable without causing a contradiction. Thus the implicature ‘no’ in our example can be cancelled if B adds the clause ‘but I mailed it to her last week.’

These three notions—entailment, presupposition, and implicature—can all be seen as types of inference. They are all produced in conversation, and are taken by participants to be part of the meaning of what a speaker has said. They differ in a number of features and crucially in context sensitivity. The attempt to provide a unified analysis of them all is a challenge to semantic and pragmatic theories. See Sperber and Wilson (1995) for an attempt at such a unified approach.

4. Future Developments

Although semantics remains theoretically a very diverse field it is possible to detect some shared trends which seem likely to develop further. One is a move away from a static view of sentences in isolation, detached from the speaker writer’s act of communication, toward dynamic, discourse-based approaches. This has always been characteristic of functional approaches to meaning but has also been noticeable in formal approaches as they move away from their more philosophical origins. Among examples of this we might mention discourse representation theory (Kamp and Reyle 1993) and dynamic semantics (Groenendijk et al. 1996).

Another development which seems likely to continue is a closer integration with other disciplines in cognitive science. In particular, computational techniques seem certain to make further impact on a range of semantic inquiry, from lexicography to the modeling of questions and other forms of dialogue. A subfield of computational semantics has emerged and will continue to develop; see Rosner and Johnson (1992) for example.

Bibliography:

1. Beaver D 1997 Presuppositions. In: Van Bentham J, Ter Meulen A (eds.). Handbook of Logic and Language. Elsevier, Amsterdam, pp. 939–1008
2. Binnick R I 1991 Time and the Verb: A Guide to Tense and Aspect. Oxford University Press, Oxford, UK
3. Bybee J, Perkins R, Pagliuca W 1994 The Evolution of Grammar: Tense, Aspect, and Modality in the Languages of the World. University of Chicago Press, Chicago
4. Chomsky N 1995 The Minimalist Program. MIT Press, Cambridge, MA
5. Comrie B 1976 Aspect: An Introduction to the Study of Verbal Aspect and Related Problems. Cambridge University Press, Cambridge, UK
6. Comrie B 1985 Tense. Cambridge University Press, Cambridge, UK
7. Fauconnier G 1994 Mental Spaces: Aspects of Meaning Construction in Natural Language. Cambridge University Press, Cambridge, UK
8. Fisiak J (ed.) 1985 Historical Semantics—Historical Word Formation. Mouton de Gruyter, Berlin
9. Frege G 1980 Translations from the Philosophical Writings of Gottlob Frege [ed. Geach P, Black M]. Blackwell, Oxford, UK
10. Gardenfors P 1999 Some tenets of cognitive semantics. In: Allwood J, Gardenfors P (eds.). Cognitive Semantics: Meaning and Cognition. John Benjamins, Amsterdam, pp. 12–36
11. Givon T 1995 Functionalism and Grammar. John Benjamins, Amsterdam
12. Grice H P 1975 Logic and conversation. In: Cole P, Morgan J (eds.). Syntax and Semantics, Speech Acts. Academic Press, New York, Vol. 3 pp. 43–58
13. Grice H P 1978 Further notes on logic and conversation. In: Cole P (ed.). Syntax and Semantics 9: Pragmatics. Academic Press, New York, pp. 113–28
14. Groenendijk J, Stokhof M, Veltman F 1996 Coreference and modality. In: Lappin S (ed.). The Handbook of Contemporary Semantic Theory. Blackwell, Oxford, UK, pp. 179–214
15. Halliday M A K 1994 An Introduction to Functional Grammar, 2nd edn. Edward Arnold, London
16. Harder P 1996 Functional Semantics: A Theory of Meaning, Structure and Tense in English. Mouton de Gruyter, Berlin
17. Hornstein N 1990 As Time Goes By: Tense and Universal Grammer. MIT Press, Cambridge, MA
18. Jackendoff R 1990 Semantic Structures. MIT Press, Cambridge, MA
19. Jackendoff R 1996 Semantics and cognition. In: Lappin S (ed.). The Handbook of Contemporary Semantic Theory. Blackwell, Oxford, UK, pp. 539–60
20. Jacobsen W H Jr 1986 The heterogeneity of evidentials in Makah. In: Chafe W, Nichols J (eds.). E identiality: The Linguistic Coding of Epistemology. Ablex, Norwood, NJ, pp. 3–28
21. Johnson M 1987 The Body in the Mind: The Bodily Basis of Meaning, Imagination, and Reason. University of Chicago Press, Chicago
22. Kamp H, Reyle U 1993 From Discourse to Logic. Kluwer, Dordrecht, The Netherlands
23. Lakoff G 1987 Women, Fire, and Dangerous Things: What Categories Reveal About the Mind. University of Chicago Press, Chicago
24. Langacker R W 1987 Foundations of Cognitive Grammar. Stanford University Press, Stanford, CA
25. Lappin S (ed.) 1996 The Handbook of Contemporary Semantic Theory. Blackwell, Oxford, UK
26. Lehmann W P 1992 Historical Linguistics, 3rd edn. Routledge, London
27. Montague R 1974 Formal Philosophy: Selected Papers of Richard Montague [ed. Thomason R H]. Yale University Press, New Haven, CT
28. Newmeyer F J 1998 Language Form and Language Function. MIT Press, Cambridge, MA
29. Partee B H 1996 The development of formal semantics in linguistic theory. In: Lappin S (ed.). The Handbook of Contemporary Semantic Theory. Blackwell, Oxford, UK, pp. 11–38
30. Pinker S 1989 Learnability and Cognition: The Acquisition of Argument Structure. MIT Press, Cambridge, MA
31. Rosner M, Johnson R (eds.) 1992 Computational Linguistics and Formal Semantics. Cambridge University Press, Cambridge, UK
32. Sperber D, Wilson D 1995 Relevance: Communication and Cognition, 2nd edn. Blackwell, Oxford, UK
33. Talmy L 2000 Toward a Cognitive Semantics. MIT Press, Cambridge, MA
34. Verkuyl H J 1993 A Theory of Aspectuality: The Interaction Between Temporal and Atemporal Structure. Cambridge University Press, Cambridge, UK
35. Wierzbicka A 1992 Semantics, Culture, and Cognition. Universal Concepts in Culture-specific Configurations. Oxford University Press, Oxford, UK