Overview • Introduction & Background aka ’Theory’ Complex phenomena deserve • Goals, Corpora, & ’Quantitative’ Methods complex explanations – choosing • Results how to think in Finnish • Discussion • Conclusions Antti Arppe University of Helsinki QITL 2 , Osnabrück, 2.6.2006 Introduction – Modelling of lexical Theory and concepts choice in computational theory • contextuality of usage and meaning: ”You shall know a word by the • In the case of semantically similar words, company that it keeps!” (J. R. Firth) – words are in a structural and semantic relationship with others in their context especially near-synonyms, at least three – the choice (i.e. usage) and meaning of words is interconnected with their context – in a language with a free word order such Finnish, (functional) dependency levels have been suggested (Edmonds grammar (Tesnière) is a practical way to explore such strucural relationships • non-modularity of language – constructionality and Hirst 2002) – regularities in co-occurrence and structure can be observed at a continuum of levels from individual words and synonym groups to general semantic groupings or parts-of-speech -> Construction Grammar 1) conceptual-semantic level • synonymy – some word pairs or groups have relatively similar meanings 2) subconceptual/stylistic-semantic level, and – in some contexts such words can be interchanged with each other without an essential change in the meaning of the entire utterance 3) syntactic-semantic level Factors influencing lexical choice Critical assessment of these results on the syntactic-semantic level – monocausality • (mainly) lexicographically motivated corpus-based studies show • The mentioned studies are typically differences in the use of semantically similar words, i.e. synonyms, in e.g. monofactorial/monocausal, focusing on one their: 1) lexical context linguistic category or one feature within a • e.g. English powerful vs. strong in Biber et al. 1998 2) syntactic structures of which they form part of category (at a time) • e.g. English begin vs. start in Biber et al. 1998 3) semantic classification of some particular argument – HOWEVER Jantunen (2002) does go through a wide • e.g. English shake verbs in Atkins & Levin 1996 range of categories, but does not quantitatively 4) style-associated text type , in which they are used evaluate their interactions • e.g Biber 1998 • while the above studies have focused on English, with minimal – With justification, Gries (2003) has argued morphology, it has also been shown in languages with extensive morphology such as Finnish that similar differentiation is evident convincingly for a holistic approach using 5) wrt the inflectional forms and the associated morphosyntactic features in which multifactorial (i.e. multivariate) statistical methods synonyms are used • Finnish miettiä and pohtia ‘think, ponder, reflect, consider’ in Arppe and Arppe & • HOWEVERm these multivariate methods build upon Järvikivi 2002 univariate and bivariate analysis • tärkeä vs. keskeinen ‘important, central’ in Jantunen 2002 1
Critical assessment – dichotomous Subsequent goals, methods and setups corpora in this study • The mentioned studies concern typically synonym pairs instead of • Explore and develop corpus-based and statistical groups with more than two members (quantitative) methodology with an aim to: – powerful vs. strong , start vs. begin , miettiä vs. pohtia , tärkeä vs. – Extend from dichotomous to polytomous (more than two) setups keskeinen • Inclusion of other members of the THINK synonym groups, with – BUT ALSO Gries’ own study of particle placement concerns a roughly similar magnitudes of frequency (common translations in dichotomous choice between two alternative constructions boldface ): – this has been noted earlier by also Divjak and Gries (forthcoming), – ajatella : 1 intend 2 plan 3 imagine, fancy, conceive (conceive of sth) 4 motivating their exceptional study of nine Russian verbs meaning ’try’ ponder 5 reflect 6 think , think of, give a thought to, figure 7 consider • However, lexicographical reality, clearly evident in both dictionaries 8 take from some perspective 9 regard, make of (sth) and in language use, often indicates that there are more than just – miettiä : 1 think 2 meditate, ponder (meditate on sth) 3 reflect 4 contemplate, conceive (conceive [of] sth), consider , mull [over], two members to a synonym group wonder (wonder about sth), give a thought to muse, cast about for 5 – THOUGH full interchangability for more than two synonyms may be think twice, thoroughly prima facie rarer, there are probably at least some contexts where any – pohtia : 1 deliberate, consider , ponder , think over 2 contemplate, one in a group of more than two synonyms could be substituted with discuss (discuss sth), debate, talk over, puzzle, think in terms of 3 each other without a major reservation wonder (wonder about sth) 4 turn over, chew over 5 kick around / about 6 (think out loud) talk about – Consequently, the differences observed between some pair might – harkita : 1 contemplate 2 ponder , deliberate, think over 3 weigh, weigh change, diminish or even disappear when studied within the entire up 4 consider , think of, think in terms of 5 think, entertain 6 think out 7 group be considering [doing sth] Goals (cont’d) … Goals (cont’d) – Extend from (simple) monofactorial to (complex) multifactorial • Extend from traditional written corpora such as newspapers or models of explanation of lexical choice published literature (formal, standardized and monologic in nature) to more informal material with a dialogic character • Inclusion of all practically available linguistic and extra-linguistic contextual information – In addition to two months of Helsingin Sanomat , Finland’s largest daily newspaper from January-February 1995 – Morphological features and inflectional structure – Syntactic arguments (according to dependency grammar as • 3,3M words with 1750 instances of the studied verbs implemented in the Functional Dependency Parser for Finnish by – Inclusion of six months of Finnish Internet discussion group material Connexor, influenced by Tesnière 195X) from 2002-2003 – Semantic classifications of syntactic arguments (according to WordNet • sfnet.keskustelu.ihmissuhteet (human relationships) and in the case of nominal lexemes and loosely adapting semantic sfnet.keskustelu.politiikka (politics) semantic primitives of Wierzbicka in the case of non-nominal adverbs) • 400K words with 1654 instances of the studied verbs • Building upon Gries’ framework (2003) of combining various – Newspaper/Newsgroup section, author and quotation/body information statistical methods available from both sources as extralinguistic context – X 2 test, Cramér's V , lambda (Goodman-Kruskal), correlation and – In addition, various aspects of repetion were also included as uncertainty coefficient (UC, Thiel) for discovering significant individual extralinguistic context (first use within article/posting, repetition of the features preceding verb, individual preceding verbs of the same group) – Regression analysis for studying the simultanous influence and interaction of significant features Current descriptions of Finnish Current descriptions ... THINK verbs • Pajunen (2001: 313-319) • Pajunen: Verbien argumenttirakenne ‘Argument – [käsittää], ajatella : • x-arg: subject A :ab: Structure of [Finnish] Verbs’ (2001: 62-63) • y-arg: object, clause argument=subordinate – “Primary-B verbs [i.e. mental verbs ], with the clause, participle, infinitive exception of speech verbs and some descriptive • A :gentivity: volitional participation in state or event, perception verbs, in general have a flat sensing and/or perceiving [classificatory] structure . … In classes with very flat – harkita structure these relationships [hyponym-hypernym] are • x-arg: subject:A:(a)b: y-arg: object, clause rare and classificatory structure consists of minor sets argument which are in loose co-hyponymic relationhips to • Agentivity: (volitional participation in state or each other (i.e. contrast groups) event), sensing and/or perceiving 2
Recommend
More recommend
