polysemy from a semantic and non-semantic...
TRANSCRIPT
Polysemyfrom a semantic and non-semantic viewpoint
Alain PolguèreNancy-Université & ATILF CNRS
........
March 24, 2010 — LORIA, Vandœuvre-lès-Nancy
jeudi 1 avril 2010
Topics
Summary We will examine the notion of polysemy from the viewpoint of combinatorial properties of lexical units: free combinatorics (that can directly be inferred from meaning) and restricted combinatorics (that cannot directly be inferred from meaning).
1. Let’s first name those “things”2. Rough characterization of the notion of polysemy3. Current perspectives4. Free combinatorics and semantic derivations5. Restricted combinatorics6. Polysemy from a theoretical and descriptive viewpoint
2jeudi 1 avril 2010
1. Let’s first names those “things”
• Language and especially “words” are our object of study• To reason properly, we need to first agree on some basic
terminology• It doesn’t matter so much, at first, if we do not understand
perfectly well the concepts we are referring to
3jeudi 1 avril 2010
No more thing, no more word
4
• To analyze lexical content of sentences, let’s not discuss “words”
• From a linguistic point of view, sentences are not made up of wordsWould a chemist say that water is made up of drops?
• And now, the time has come to talk of many things…
jeudi 1 avril 2010
5
What does it take to take the lead that you now take?
Four levels of analysis that entail four different types of entities1. One single signifier (Fr. signifiant), occurring three times
๏ take2. Three different linguistic signs, that are wordforms
๏ take1 (V; ‘to require’) infinitive
๏ take2 (V, collocate; ‘to cause to be in a given situation’) infinitive
๏ take2 (V, collocate; ‘to cause to be in a given situation’) 2nd pers., sing., present
3. Two different lexical units, more precisely lexemes๏ TAKE1 (V ‘to require’)
๏ TAKE2 (V, collocate; ‘to cause to be in a given situation’)
4. One single vocable, that happens to be polysemic๏ TAKE
Revised version of the infamous transparency no. 5!
jeudi 1 avril 2010
Form now on
• The term word is prohibited• Depending on what we are talking about, we use either
๏ signifier, — linguistic form, regardless of the meaning it conveys
๏ wordform, — association between a form and a given meaning it conveys ⇒ it is a linguistic sign
๏ lexeme, — lexical unit that manifests itself as wordforms (= its inflected forms)or
๏ vocable — grouping of lexical units, lexemes or idioms, (i) that have distinct but related meanings and (ii) that are expressed by the same forms
6jeudi 1 avril 2010
Longman Dictionary of Contemporary English (LDOCE)
7jeudi 1 avril 2010
2. Rough characterization of the notion of polysemy
• There are hot debates on polysemy• They will be mentioned, but I am here to provide you
with solutions, not to expose problems only• My presentation may be partisan, at time, and non-
consensual, often• Here is a definition, as if there were a consensus on the
notion of polysemy
8jeudi 1 avril 2010
Defining the term polysemy
• polysemy๏ polysemy of what?
• polysemy of a signifier/a worform/a lexeme/a vocable?๏ polysemy of a vocable!
• Polysemy of X = formal property of the vocable X, which is such that it contains more than one lexical unit
9jeudi 1 avril 2010
Semiotic perspective
• Polysemy is also a semiotic resource of natural languages๏ Lexical signs are basically non-iconic at a primary level
(except for onomatopoeic signs, etc.)๏ However, many linguistic signs are iconic at a secondary level
๏ THINK (V) → THINKER (N, ‘person who thinks’)
๏ KNIFE (N) → KNIFE (V, ‘to stab with a knife’)
๏ MOUSE (N, ‘animal’) → MOUSE (N, computer device’)
• Polysemy is a practical means for creating new lexical units๏ savings on signifiers๏ gains on iconicity
10jeudi 1 avril 2010
Beware ! Ambiguity is prowling
• By contrast, ambiguity is neither a property of vocables nor a means for creating lexical units
• Ambiguity is a characteristic of linguistic forms that, potentially, impairs the interpretation process
• It’s a problem for automatic analysis and so-called word sense disambiguation (WSD) — cf. Kilgarriff (1997)
• It is by no means a defining notion for the notion of polysemy
• Polysemy, first of all, has to do with the structure of linguistic (lexical) knowledge
11jeudi 1 avril 2010
3. Current perspectives
• For some (illogical?) reasons, polysemy is a controversial notion
• Let’s summarize a few well-known perspectives
12jeudi 1 avril 2010
Dictionaries:the no-nonsense perspective
• For lexicographers, accounting for the polysemic structure of vocables is a difficult but feasible task
• Description of each copolyseme: meaning, grammatical properties, etc.
• Opponents call this sense enumeration
13jeudi 1 avril 2010
Self-evidence in lexicology:word senses are not just word senses
• In English, the various lexemes of a vocable are referred to as senses
๏ Terminological problem: in French acception, but in English word sense
• But they are more than just senses, they are lexical units, i.e. senses
๏ that entertain strong semantic links with specific lexical units of the lexicon: TO SING → SONG, BREAD → BAKER, etc.
๏ that are expressed by specific forms and have specific combinatorial properties
A lexical unit is a complex cluster of semantic, formal and combinatorial properties; it is neither just a form nor just a meaning!
14jeudi 1 avril 2010
“Contextualists” in denial
• DENIAL: ‘failure to acknowledge an unacceptable truth or emotion or to admit it into consciousness, used as a defense mechanism’ [Oxford American Dictionaries]
• Analytic perspective (WSD, etc.) : it is too difficult to achieve satisfactory form → lexical unit matching in case polysemy is acknowledged
• Much easier to perform form → vocable matching and state that we are faced, in most cases, with a continuous rather that discrete phenomenon๏ The credo: “Lexical units do not have specific meanings; they
possess a meaning core and get their actual semantic content only in context”
15jeudi 1 avril 2010
Polysemy from an explanatory combinatorial viewpoint
• Identifying the polysemic structure of a vocable boils down to identifying all the semantically-related lexical units it contains
• Each lexical unit will have to be characterized along three axes๏ Its meaning๏ The set of semantically derived lexical units it is related to๏ Its restricted combinatorics
• While characterizing the copolysemes, it is a methodological mistake to focus on meaning only
16jeudi 1 avril 2010
4. Free combinatorics and semantic derivations
• Combinatorics of a lexical unit is the set of properties of this unit that determines its ability to combine with other lexical units in an utterance
• Free combinatorics of a lexical unit is that part of its combinatorics that can be directly inferred from its meaning (= definition)
17jeudi 1 avril 2010
Free combinatoricsas a means for distinguishing
between copolysemes
• Well-known tests for the identification of the polysemic structure of vocables exploit free combinatorics (Cruse 1986; Kilgarriff 1997:6-11; Mel’čuk et al. 1995)
• Example: Green-Apresjan criterion — or criterion of compatible cooccurrence — in Mel’čuk et al. (1995: 64–65)๏ Felipe scribbled in the latest book published by Patrick.
Compatible cooccurrence of one occurrence of book with to scribble and to publish⇒ one single lexeme BOOK in to scribble in a book and to publish a book
๏ *Felipe took a dry towel from the shelves and then a shower.Zeugma⇒ two different lexemes TAKE in to take a towel and to take a shower
18jeudi 1 avril 2010
Semantic derivations
• Strong paradigmatic links that connect lexical unit:synonymy, antonymy, conversivity, nominalization, verbalization, standard actant names, etc.
• A rather small (~25) set of universal semantic derivations has already been proposed: standard paradigmatic lexical functions(Mel’čuk 1996)
• It provides a descriptive frame for the identification and description of semantic derivations
19jeudi 1 avril 2010
5. Restricted combinatorics
• Combinatorics of a lexical unit that cannot be inferred from its meaning (nor its form)
• Grammatical combinatorics๏ (in-)compatibility with grammatical meanings๏ passive valency๏ active valency
• Lexical combinatorics = controlled collocations๏ Intensifiers, support verbs, verbs of “realization”, etc.๏ Accounted for by means of standard syntagmatic lexical
functions
20jeudi 1 avril 2010
6. Polysemy from a theoretical and descriptive viewpoint
• Tests such as the Green-Apresjan criterion are working with symptoms of polysemic relations, not with the actual lexical nature of each copolyseme
• Once we consider that a full lexicographic description of a lexical unit accounts for all its semantic, relational and combinatorial properties, we can follow guidelines that are lexicographic in nature
• Are L1 and L2 two distinct lexical units or just one (vaguer) unit L?๏ Lexicographic descriptions of L1 and L2 are almost redundant
except of a minor or very regular difference in definition⇒ only one unit L
๏ Lexicographic description of L is full of restrictions referring either to the meaning ‘L1’ or to the meaning ‘L2’⇒ two separate units L1 and L2
21jeudi 1 avril 2010
Illustration from the Lexique actif du français LAF (Mel’čuk & Polguère 2007)
22jeudi 1 avril 2010
Regular polysemy
• In each language, one can find patterns of polysemy๏ Regular and systematic polysemy (Apresjan 1974)๏ Cf. polysemy patterns for French (Barque 2008)
• Definitions of copolysemes should account (directly or indirectly) for polysemic patterns that are being activated — cf. semantic bridges
23jeudi 1 avril 2010
24
Metaphorical linkbased on physical resemblance
jeudi 1 avril 2010
Causativity link
25jeudi 1 avril 2010
References
‣Apresjan J. (1974) Regular Polysemy. Linguistics, 142, 5-32.‣Barque L. (2008) Description et formalisation de la polysémie régulière du français.
Ph.D. dissertation, Université Paris 7.‣Cruse A. (1986) Lexical Semantics. Cambridge U.-K., Cambridge University Press.‣Kilgarriff A. (1992) Polysemy. Ph.D. dissertation, University of Sussex.‣Kilgarriff A. (1997) I don't believe in word senses, Computers and the Humanities,
31, 91-113.‣Mel’čuk I. (1996) Lexical Functions: A Tool for the Description of Lexical
Relations in the Lexicon. Dans L. Wanner (dir.) : Lexical Functions in Lexicography and Natural Language Processing, Amsterdam/Philadelphia, Benjamins, 37-102.‣Mel’čuk I., Clas A., Polguère A. (1995) Introduction à la lexicologie explicative et
combinatoire, Paris/Louvain-la-Neuve, Duculot.‣Mel’čuk I., Polguère A. (2007) Lexique actif du français. L’apprentissage du
vocabulaire fondé sur 20 000 dérivations sémantiques et collocations du français, Bruxelles, De Boeck & Larcier.
26jeudi 1 avril 2010