Constructions in interaction
The final section of the volume is dedicated to research conducted at the crossroads of construction grammar and conversation analysis. As a matter of principle, and in line with the theoretical presuppositions of Cognitive Linguistics, (most) constructionist approaches to grammar do not distinguish between semantic and discourse-functional properties of constructions, since both of these may become conventionally associated with specific forms. Moreover, in Cognitive Linguistics at large a development can be observed from an almost exclusive focus on the individual, subjective construal of reality to incorporating the social, intersubjective dimension of speaker-hearer-interaction (Verhagen 2005; Croft 2009). Still, in the words of Linell (2009: 97), “it is undeniable that many variants of CxG suffer from an interactional deficit”. The absence of interactional approaches from, for instance, Hoffmann and Trousdale’s (2013) handbook, should, however, not lead one to conclude that there have not been attempts to fill this hiatus, see, e.g., the German-language volumes edited by Deppermann, Fiehler, and Spranz Fogasy (2006), Gunthner and Imo (2006), and Gunthner and Brucker (2009). The chapters in this section likewise deal with discourse functions of selected grammatical patterns - viz. prepositional phrases (Brucker), appositions (Imo), and complement clauses (Wide) - that can be attested only in detailed qualitative analyses of situated interaction. All three authors explore the possibilities and limitations of combining a methodological commitment to conversation analysis with the degree of generalization that is inherent to the notions of “construction” and “constructional network”.
In his contribution on ‘Tying constructions with the preposition mit in German talk-in-interaction’, Jorg Bucker demonstrates that not only particles, adverbs and conjunctions, but also prepositional phrases may be used as discourse-structuring devices. Specifically, for German mit (‘with’) + Noun Phrase it is shown that in spoken talk-in-interaction it often functions as a “tying construction”: in such cases, the speaker uses it to refer to a topical antecedent in the preceding discourse that is considered to be accessible, to different degrees. Bucker distinguishes between two patterns - attributive and non-attributive uses of “tying” mit + Noun Phrase - that differ slightly in both form and in function. These differences, however, can be explained fully compositionally as resulting from their different positions in the clause and different degrees of subordination. Bucker argues, therefore, that only one encompassing mittying + Noun Phrase-construction needs to be assumed, modeled along the lines of Goldberg (1995), that can be realized as two different “constructs” in conversation.
Wolfgang Imo, in the chapter entitled ‘Appositions in monologue, increments in dialogue? On appositions and apposition-like patterns in spoken German and their status as constructions’, deals with patterns in conversation that, more or less, resemble what have traditionally been called appositions. He shows that in his non-monological data, not a single instance can be found of the classic, “wide scope” NP + NP apposition (as in I met John, my old friend, in London). Instead, interactional data feature a kind of pattern that resembles an apposition but that exhibits quite different properties, both in syntax and in function. Whereas in typical appositions the two NPs are juxtaposed, the second NP in the interactional pattern occurs at quite a distance from the first one. It may in fact be used after the so-called right verb brace, which is a strong signal for syntactic closure in German. In this respect, the second NP behaves more like a syntactic unit expansion or increment. As for its function, the “interactional” type of apposition is added “on line” to respond to a real or potential problem of understanding and is thus used to re-focus, paraphrase or repair a previous utterance. The NPs in such “very wide scope” or “peripheral” appositions do not necessarily refer to persons; the latter is a typical feature of wide scope appositions in written and monologic discourse. In addition to a formal and a functional entry for characterizing constructions, at least a sequential entry is needed to capture this type of information. In order to determine the exact relation between these apposition-like patterns and other syntactic constructions in a constructional network, more empirical data are required about the actual use of related syntactic patterns in conversation.
In the final chapter, ‘Constructions as resources in interaction: syntactically unintegrated att ‘that’-clauses in spoken Swedish’, Camilla Wide analyses instances of insubordination with the complementizer att ‘that’ in spoken Swedish (both Swedish Swedish and Finnish Swedish): these “subordinate” clauses are syntactically unintegrated since they are used without a main clause. Wide focuses on the phenomenon of discourse insubordination, that has also been observed for independent complement clauses in German and Dutch. Rather than being licensed by the syntactic context, the use of such clauses is licensed by the interactional and pragmatic context. Wide distinguishes between two types: one with a rephrasing function and one with a reasoning function. In the first type, the speaker adds more specific information or makes the intended speech act, such as a question or a suggestion, more explicit. The second type equally relates to something said in the preceding discussion, but here the speaker expands the line of reasoning by expressing a consequence of what another speaker has said. The two types clearly fulfill different functions in interaction and they may have developed from a different source. However, whether or not they instantiate two different constructions in the sense of construction grammar depends on the weight that is given to semantic/functional differences, as compared to formal differentiation. If the relationship to the prior context is itself regarded as a formal or form-related feature, it would be possible to treat the two types as different nodes in a constructional network since their “external syntax” (Linell 2009) is different.