Table of contents
1. The aim of this work
2. The research background
3. The theoretical foundation
3.1. The ‘mixed game model’
3.2. The quasi-universal semantic structure
3.3. The principle of meaning equivalence
3.4. The expression side
4. Investigating language
4.1. Preliminary remarks on the methodology
4.1.1. How to compare ‘see’ and ‘sehen’?
126.96.36.199. The implementation of the contrastive semantic analysis
188.8.131.52. The predicating fields of the analysis
4.1.2. Searching the corpus
4.2. The semantic analysis of ‘see’ and ‘sehen’
4.2.1. The contrastive semantic analysis
184.108.40.206. Correlating the expression side with the meaning side
220.127.116.11. The problematic cases
4.2.2. A quick glance at ‘the complex’
18.104.22.168. What does the corpus tell us?
22.214.171.124. The problematic cases
5. Interpreting the results
5.1. Units in contrast
5.2. Meaning in ‘the complex’
6. Conclusion and outlook
7. List of works cited
1. The aim of this work
In the light of de Saussure’s definition of the linguistic sign (2001: 76ff.) and Chomskyan generative grammar (e.g., Chomsky 1966, 1965, 1957), a great number of Linguists of the second half of the 20th century used to define language primarily in terms of an artificially constructed autonomous sign system. From this traditional point of view, the field of linguistics did not see beyond the construct it had created resulting in the deformation of the object of study, language-in-use.
The pragmatic turn of the 1960s brought along what one might call the ‘Copernican revolution’ in linguistics. In this connection, Martinet (1975: 10) pointed at one of the most fundamental prerequisites of well-conducted scientific research, in that he rejected firmly any scientific proceeding that sacrifices the integrity of the object of study to methodological exigencies. On the contrary, the object itself, language in use, dictates to a great extent the methodology of the researcher.
Taking into consideration the object’s integrity, pragmatic theories have finally dismissed the artificial sign system constructed by structuralist approaches and, consequently, center around the research on language-in-use. In this context, the speech act theory, for example, points out an essential aspect of language, which had been neglected by most traditional theories: Language-in-use is an integrated part of human interaction (cf., e.g., Austin 1976; Searle 1969). In “The language myth and linguistics humanized” (2002a), among others, Weigand defines language-in-use in terms of the so called ‘dialogic action game’ (64). She (cf., e.g., 2002a: 63; 1998: 30) assumes that human beings and their abilities are at the center of communicative interaction. Thereby, the linguist’s investigations are not to be restricted to the verbal text only; on the contrary, communication between human beings involves different abilities at the same time, which are used with a view to serve specific communicative purposes. Thus, in addition to verbal expressions, perceiving and thinking play also extensive roles within language-in-use (e.g., Weigand 2002a: 65; 2000: 6ff.).
Eventually, this pragmatic understanding of language is willing to account for the complexity of the object of study. In this new light, Weigand defines her model of dialogic action games as an open-ended model, which “contains not only the definite and quasi-definite of rules and conventions but also the indefinite and cases of probability […]” (1998: 27). The insight that “human beings are, in principle, different individuals, with different cognitive backgrounds” leads to the conclusion that “problems of understanding are therefore constitutive” (Weigand 2002a: 64). As meanings are not restricted to unambiguous definitions, problems of understanding inevitably arise. For that reason, the interlocutors are enforced to negotiate meaning and understanding in the process of their communicative interaction (Weigand 2002a: 64).
Seen from a semantic-pragmatic perspective, it is obvious that lexical descriptions in the tradition of structuralism are in need for a change, as they cannot possibly account for the communicative functions that the lexis of a language actually performs. Nevertheless, it becomes clear that even nowadays language is often described in terms of an autonomous abstract system. Especially with respect to lexicography, lexical descriptions still tend to center around the structuralist equation of the word and a unit of meaning (Sinclair 1998: 5; Weigand 1998: 42). Most of the time, the actual conditions of language-in-use are totally neglected, i.e. the lexical definitions of word meaning lack any systematic connection to the actual ways-of-use a word is embedded in when used by a speaker. Thereby, it is a common phenomenon that lexicographers list frequent senses of a word rather peripherally, whereas senses which are actually rarely applied in language-in-use are highlighted (cf., e.g., Biber et al. 2004: 40f.).
Especially the corpus-based linguistic research during the last decades has made it obvious that the question on the definition of the lexical unit needs to be discussed in a new light (cf., e.g., Sinclair 1998: 1, 3, 6). As a consequence, the syntactical dimension of language needs be looked at much closer than before in order to arrive at an adequate description of the vocabulary. The change from the paradigmatic to the syntagmatic perspective has already resulted in new lexicographic models like Sinclair’s collins cobuild English Dictionary (1987) which, in the context of its lexical descriptions, pays special attention to the surrounding language of a keyword.
With respect to the above-mentioned development of linguistics, one can say that the pragmatic-semantic approaches finally proclaim the end of the one-word lexical unit; instead, they center on multi-word phrases which they see as the basic starting point for adequate descriptions in the field of lexical semantics (cf, e.g., Sinclair 1998: 23; Weigand 2005: 2ff.).
In the course of this treatise, I discuss, in the context of the above-mentioned assumptions, the issue of the lexical unit. Thereby, I will concentrate on the two conflicting definitions given, on the one hand, in the form of the one-word lexical unit and, on the other, in the form the multi-word unit. During my argumentation, I attempt to provide sufficient evidence in favour of the multi-word lexical unit showing that the equation of the single word with a unit of meaning needs to be overcome ultimately.
This treatise comprises both a theoretical and an empirical part. The former lays the theoretical foundation for the empirical investigation put forward in the course of the methodological section. After the obligatory sketch of the research background, I introduce the reader to the basic assumptions of Weigand’s pragmatic model. Looking at lexical semantics, the universal level of contrastive studies is kept in the foreground. In this context, I primarily concentrate on the definition of predicating fields and meaning positions as well as the principle of meaning equivalence. In addition, the theoretical introduction will offer some preliminary remarks on the expression side.
The methodological part features either a contrastive semantic as well as a corpus analysis. Both methodological instruments are used with a view to verify the need for syntactically defined multi-word lexical units.
As a first step of the verification, dictionary entries of ‘see’ and ‘sehen’ are correlated with a quasi-universal semantic structure. Considering the result of the analysis, a closer look at the expression side should already show that semantically defined multi-word phrases are the minimal lexical units that function on the level of the utterance.
 Examples are conveniently available in, e.g., Lyons (1983 and 1980), Löbner (2003) or Matthews (2001).
 cf. also Weigand (2000)
 E.g., Sinclair (1998: 3) criticizes this point and thus rejects firmly the traditional conventions in lexicography.