Theories of underspecification assume that grammatical meanings consist of incomplete and highly abstract “core meanings” which are “enriched” in actual communication by contextual and encyclopaedic information to yield observable surface effects. These surface effects will henceforward be referred to as usages. According to most underspecification approaches, all usages of a given element can be traced back to a single invariant underlying core meaning, which, unlike the item’s usages, is not directly observable. Usages are always computed and derived ad hoc – they are neither entrenched nor con-ventionalized. Theories of underspecification normally have little to say about the nature of the derivational/computational processes involved; they aim at descriptive economy rather than cognitive plausibility.
Polysemy models, in turn, assume that linguistic meanings are conventionally given, stable entities which are organized in networks. These networks reflect the associative relations (metaphor, metonymy, taxonomic relations, Blank 2003, Nerlich & Clarke 2003) which the various meanings of a given item entertain with each other. As conventionally given objects, meanings are not (synchronically) derived from other underspecified (core) meanings, but are stored as such in the mental lexicon. Importantly, polysemy is the result of linguistic change; Bréal has characterized it as the “synchrony of semantic change” (see also Blank 2003).
In recent work on discourse markers and modal particles, an interesting debate on models of meaning representation has arisen. Traditionally, the underspecification model has been the more influential approach (examples are Weydt 1999 and Ollier 1995, Fischer 2005), but serious arguments have recently been put forward in favour of the polysemy model (see Hansen 2002, Waltereit 2006). The view that the polysemy of discourse items is the result of diachronic change is currently “in the air” – programmatic sketches have been formulated in Waltereit & Detges (2007) and in Hansen & Strudsholm (2008). Against the backdrop of such work, the project proposed here would not only provide new insights on the diachrony and synchrony of certain discourse markers. What is more important is its aim to make a contribution to a general theory of the polysemy of procedural items.
A central problem of polysemy approaches, which is of special interest for the project I propose here, is the controversy between semantic minimalism and semantic maximalism (Nerlich & Clarke 2003, 14). Essentially, this debate revolves around the same issue as the controversy between underspecification theories and polysemy approaches, namely the questions of how to distinguish conventionalized meanings from mere usages.
For the solution of this problem, language change provides a robust heuristic tool: functions which can empirically be shown to have arisen as a result of language change (e.g. by analyses of corpus data) must by definition be entrenched and conventional ob-jects, i.e. stable elements of a polysemous structure. Functions which cannot be shown to be the result of change must conversely be assumed to be mere usages. The aim of the project is to provide a plausible model for the cognitive organization of grammatical mean-ing, which – similar to already existing theories of polysemy in the lexicon – brings together synchrony and diachrony.
Depending on the nature of the linguistic item under investigation, the theoretical models discussed here, i.e. the underspecification approach and the polysemy model, exhibit varying degrees of plausibility. Thus, it seems obvious that e.g. the functions of certain converb constructions, such as the ablativus absolutus in Latin, can most likely be adequately captured by some kind of underspecification model, while lexical items will be most efficiently described in terms of polysemy (see Blank 2003). In the project under discussion here, I want to show that discourse markers, i.e. items with near-grammatical meanings, normally possess various functions which have been entrenched and conven-tionalized as a consequence of frequent usage, but can exhibit a certain degree of under-specification at the same time. Discourse markers lend themselves to analyses of this kind, because on the one hand, they have procedural meanings, i.e. they are near-grammatical items, while on the other hand, they have word-status, i.e. they can easily be identified in electronic corpora. What is more, discourse markers sometimes instantiate diverging developments (heterosemy and split), as it has been the case for ben and bien, where different (but metonymically related) functions correspond to differences in form.
Blank, A. (2003), "Polysemy in the lexicon and in discourse", in: B. Nerlich et al., eds., Polysemy. Flexible Patterns of Menaing in Mind and Language, Berlin & New York: Mouton de Gruyter, 267-293.
Hansen, M.-B. Mosegaard (2002), "La Polysémie de l'adverbe encore", Travaux de Linguis-tique 44, 143-166, 172.
Nerlich, B. & D. D. Clarke (2003), "Polysemie and flexibility", in: B. Nerlich et al., eds., Polyse-my. Flexible Patterns of Menaing in Mind and Language, Berlin & New York: Mouton de Gruyter, 3-29.
Ollier, M.-L. (1995), "Or, opérateur de rupture? " LINX 32, 13-31.
Waltereit, R. (2006) "Comparer la polysemie des marqueurs discursifs", in : M. Drescher & B. Frank-Job, eds., Les marqueurs discursifs dans les langues romanes. Approches théoriques et méthodologiques, Frankfurt: Lang, 141–151.
Waltereit, R. & U. Detges (2007), "Different functions, different histories. Modal particles and discourse markers from a diachronic point of view", Catalan Journal of Linguistics 6, 61-81.
Weydt, H. (1999), "Emploi infini de moyens finis: formation de mots, particules, expressions idiomatiques", Nouveaux cahiers d’allemand 17, 307-315.