280 likes | 632 Views
Jordan Zlatev. Semantics and Lexicology SVEM21 5. Neogenerativist and Neostructuralist Semantics. Projects (general). Summarize the main ideas/concepts in the text. Relate to the discussion of the tradition (and author) in the textbook (Geeraerts)
E N D
Jordan Zlatev Semantics and LexicologySVEM21 5. Neogenerativist and Neostructuralist Semantics
Projects (general) • Summarize the main ideas/concepts in the text. • Relate to the discussion of the tradition (and author) in the textbook (Geeraerts) • Analyze examples from another language (e.g. Swedish) using the concepts, categories, distinctions… discussed.
Geeraerts, end of Chapter 3 • Cognitive adequacy “a type of meaning description that paid less attention to formalization, but that explicitly opted for a maximalist, encyclopedic, psychologically realist form of semantics, and that thus broke radically with the legacy of structuralism” > Chapter 5, Cognitive Semantics • Formal adequacy “theories that continue the lines set out by structuralism, but that do so with specific attention to concerns issuing from generativist semantics: the demarcation of linguistic knowledge with regard to cognition in the broader sense, and the possibility of formalizing linguistic meaning” (: 121)
Neo-generativist and neo-structuralist semantics • “Neogenerativist” • 1.Conceptual Semantics (Jackendoff) • 2. Two-level semantics (Bierwisch) • 3. Generative Lexicon (Pustejovsky) • “Neostructuralist” • 4.WordNet (Miller, Fellbaum) • 5. Explanatory Combinatorial Dictionary (Mel’cuk) • 6. Distributional Corpus Analysis • Natural Semantic Metalanguage (Wierzbicka)
1.Conceptual Semantics • “There is no privileged level of “linguistic semantics” at which specifically linguistic effects of meaning can be separated out from more general cognitive effects such as categorization…” (Jackendoff 1996: 104), (:138) • Words as “interfaces” across modules.
Conceptual system Linguisticsystem Inter-subjective information Visual/3D format Body format Subjective information Based on Jackendoff (1992: 14)
Lexical entries (Figure 4.1, :139) run V _<PPj> [event GO ([THING]i, [PATH]j) put V _<NPj> <PPk> [event CAUSE ([THING], [event GO [THING]j, [PATH]k)
Universal conceptual primitives and structures EVENT GO THING PATH Path-function (TO, FROM, VIA) PLACE Place-function (IN, ON, ABOVE, BELOW) THING (Time, Property) STATE BE THING PLACE Place-function (IN, ON, ABOVE, BELOW) THING (Time, Property)
Implications • A strict separation between “conceptual” and “non-conceptual” information • Conceptual primitives and structures: “innate” and “universal” • More subtle differences of meaning, such as different “manner verbs”: run, jog +walk, crawl, fly…? – should be a matter of non-universal “perceptual representations”
Problems • Too universalist: not clear if motion verbs in all languages (Japanese, Mayan languages) have a semantic component (GO), as opposed to a pragmatic, “defeasable” implicature. • ROOM, TRAIN – “primitives”? • Information about jogging – purely non-conceptual? • “need criteria to determine what enters into a conceptual description and what can be relegated to the non-conceptual cognitive modules” (: 141) • (rather) static, with respect to context
2. Two-level semantics • “provides a model for the interaction of word knowledge and world knowledge in actual contexts of use” (:143) • “More explicitly than Jackendoff, the two-level approach deals with meaning variation… accounting for polysemy and semantic flexibility is a major focus in contemporary lexical semantics (: 143).
“Semantic form” and “conceptual structure” university • Level 1 (“semantic form”): λx [PURPOSE [x, w] & advanced study [w]] • Level 2 (“conceptual structure): The university offers scholarships. λx [INSTITUTION [x] & PURPOSE [x, w]] The university lies in the centre of the town. λx [BUILDING [x] & PURPOSE [x, w]]
Problems? • “McDonald’s University” – advanced? A counterexample, or just a “creative” use of the term university? • Contextualization requires encyclopedic knowledge (Taylor): why not? (The model does not deny this…) ? Der Palast hat die Frage bereits entschieden. The Palace has already come to a decision on the issue. • Language change – from pragmatic inference (Level 2) to semantic form (Level 1): rather an argument for keeping the levels distinct! This does requires however, more than one entry in the case when the old meaning is preserved. since(temporal) + since (causal), cf. 145
3. Generative Lexicon • “the most advanced approach among the formal componential theories…” (: 154) • Targetting “regular polysemy” (Apresjan), “logical polysemy” (Pustejovsky): • Building-Institution • Count noun – Mass noun • Product-Producer • Process-Result • Contents-Container • Telic-Atelic action • Emotional state – Expressing emotional state
Representational format • Argument structure • Event structure • Qualia structure (descriptive features) • Formal (“what something is”) • Constitutive (“what something consists of”) • Telic (“the purpose”) • Agentive (“how something came into being”) See Figure 4.2 (: 149) and example for novel, (: 155)
Predicate-argument matching • Type matching • Accommodation • Type coercion • Exploitation (using “dotted types”) • Introduction (making a “dotted type”) See examples, p. 151 Extensions • Lexical rules: operates upon rules • Metaphor Lexical Rule: “semantic type can be anything”, but preserves qualia structure
Problems • “profit from a broader empirical basis” (: 152) • Overgenerating • as with Two-level Semantics? (if not encyclopedic knowledge is included) • Sydney began a novel / a sweater. (TELIC = write) • Undergenerating • Waiting for a bus. (other reasons that taking it)But need the model account for such clearly pragmatic interpretations? • Primitives like “physical object”
4. WordNet • English, and other European languages • Nouns, verbs, adjectives, adverbs • Each entry: • Synset • Definition • Example • Synonym sets (synsets): president, chairman, chairwoman, chairperson
Synsets: related by sense relations • Hyperonyms • Hyponyms • Meronyms • Antonyms (for adjectives, adverbs, verbs) • Entailments • Hyponyms • “troponyms”: walk < stride • Presupposition: succeed < try • Causality: show > see
“Restrictions” • No differentiation between different kinds of antonyms • Definitions: “the network information does not completely replace such definitional information” (: 160) • Originally, psychological adequacy, but not anymore: “a machine readable dictionary… not a model of the mental lexicon” (: 160)
5. Explanatory Combinatorial Dictionary • Adding “lexical functions”, e.g. head of (dead-faculty, board-chair, ship-captain…): Cap • Syntagmatic, and not only paradigmatic, differ across languages: • question-ask (English) • Frage-stellen (German) • question-poser (French) See example of Revulsion, 162-163
“Restrictions” Practical • Applied mostly to Russian and French • Elaborate, but time-consuming (hence WordNet is preferred for practical lexicography) Theoretical • Again: entries contain an analytic definition • Does not include part-whole relations: world-knowledge, but Cap?
6. Distributional corpus analysis • “a collection of naturally occurring text, chosen to characterise a state or variety of a language” (Sinclair 1991: 171), (: 167) • Language on the level of parole, not langue • “a radical usage-based, rather than system-based approach” (: 168) • But note: “… of a language” (Sinclair)
Key notions • Collocation: “a lexical relation between two or more words which have a tendency to co-occur within a few words of each other in running text” (Stubbs 2002: 24) • “Node” + “collocate” (see Figure 4.3) • Colligation: syntactic pattern • Semantic preference: b/n the node and “a set of semantically related words” • Semantic/discourse prosody: positive vs. negative (emotive attitude)
Evaluation • Popular in cognitive science (quantitative, “objective”) • “the interaction between theoretical lexical semantics… and statistical lexical semantics is still rather restricted” • “the least structuralist of the ‘neostructuralist’ approaches” (: 176) • “Given the problems of demarcation and selection of primitives… distributional corpus analysis has the clear advantage of making contact with the probabilsitic paradigm in computational linguistics” (: 177)
“Crictical remarks” • “primarily a method, not a model” (: 177) • “has not yet reached the stage where it can present a stable set of methodological procedures coupled to specific descriptive questions” (: 178) • “whether all the relevant information that language users have about the reference of words, may be retrieved from a corpus” (: 178)
Conclusions • Note the biased terminology: models which aimed for distinguishing lexical meaning from general knowledge (“pragmatics 1”) and contextual usage (“pragmatics 2”) where first called “minimal”, then “parsimonious” – and then: “reductionist and exclusionary” (: 176) • Distributional corpus analysis is on other hand works in a “non-reductionist, usage-oriented way” (:177) • One could argue that the latter, especially if “radical”, abolished distinctions (semantics/pragmatics etc), reduces meaning to use, quality to quantity – and is in essence the truly reductionist approach!
Conclusions • In Chapter 4, Geeraerts shows the problems with conceptual or semantic “primitives” and making clear distinctions b/n lexical meaning and (a) encyclopedia and (b) usage – but does not show • That the search for universal semantic concepts is futile • That semantics/pragmatics distinctions are not necessary – even though “unclear”, and “dynamic”