For decades, lexical semantic research was bound up with structuralist notions of language, leaving it rather divorced from trends in syntax, semantics and pragmatics in the 1970s and 1980s. Lexical semantic (or lexicological) studies to this point often focussed on paradigmatic relations such as synonymy and antonymy. As the lexicon became more central in approaches to syntax and related areas and as corpus methodologies and computational requirements developed, the nature of lexical semantics has changed, as have notions of what a mental lexicon is. Consequently lexical-semantic paradigms have been pushed out of the field's centre stage. For example, whereas the bulk of Cruse's (1986) textbook Lexical Semantics
(CUP) concerned synonymy, opposition and hyponymy, twenty years later CUP's forthcoming textbook (Lexical Meaning
by ML Murphy) devotes only one of 14 chapters to the subject. This raises the question that is central to this workshop: are lexical-semantic paradigms relevant to modern linguistics?
- Many (if not most) approaches to language (e.g., Cognitive Linguistics-Langacker inter alia, Conceptual Semantics-Jackendoff) equate lexical meaning with conceptual structures. How does this rejection of linguistic-semantic modularity affect notions of lexical relation? Are there such things as lexical relations, or just conceptual-semantic relations?
- Classical notions of componential meaning allow for clear definitions of lexical paradigms, but these notions have largely been abandoned. How are lexical paradigms accounted for in theories that represent meaning through prototypes, schemata or more complex componential metalanguages?
- The traditional grammar-lexicon has been compromised by the evidence that constructions, rather than "rules and words", are the basic building blocks of language. In constructionist approaches, the mental lexicon has thus morphed into a "constructicon" that represents syntagmatic relations directly in phrasal constructions. What is the role of lexical-semantic paradigms in such theories? (How) are they represented in the constructicon?
- Corpus methodologies have made the study of syntagmatic relations--through collocations and co-occurrences--both easy and insightful. However, paradigmatically related items generally do not collocate and are not (with the exception of antonymy/contrast) known for co-occurring. What have corpus methodologies added to the study of lexical-semantic paradigms, and how might they be extended to give further insight into the roles of such relations in language?
- English and (to varying extents) other European languages are rich in lexical synonyms, antonyms and hyponyms. However, there is no logical reason for languages to do so. Synonyms, it has been argued, are abhorred by language, just as nature abhors a vacuum (Cruse 1986), and languages have morphological and phrasal means to express antonymous and hyp(er)onymous notions. (In fact, most artificial languages do away with lexical antonyms in favour of morphologically derived opposites.) According to structuralist notions, lexical gaps in semantic fields tend to be filled, and so the lexicalisation of one concept encourages the lexicalisation of opposite, subordinate and superordinate concepts. Yet there is some evidence that not every language acts this way (e.g. Lancy and Strathern (1981) on Melpa's non-hierarchically structured lexicon; M. J. Hardman [cited in Murphy 2003] on a language without antonyms--Jaqi). Are relations like antonymy and hyponymy universal, and do they have similar significance cross-linguistically? How does such evidence contribute to the post-Whorfian debate on language and culture?
- Paradigmatic relations receive a fair amount of attention in Natural Language Processing, where they have implications for information retrieval, text generation and analysis and machine translation. What insight into the human lexicon can such projects give?
- Many claims are made about the importance of lexical-semantic paradigms to child language development and second language acquisition. Are these claims backed up by empirical evidence? Does such evidence argue for particular views on how the mental lexicon is structured?