Lexical Grammar

Corpus-based examinations of the extent of modal marking (modal load) in if-conditionals in the written BNC have revealed that they have a significantly higher modal load (ML) than average, 1 as well as a higher ML than indirect interrogatives with if and whether, concessive conditionals with even if and whether, and non-conditional bi-partite constructions with when and whenever. Crucially, if-conditionals show a higher ML than conditionals with other subordinators (e.g. assuming). Also, the ML of ifconditional protases (i.e. their subordinate parts) is comparable to that of the baseline, despite their being already modally marked by if. 2 Explanations for the emerging ML patterns are sought in the tenets of two recent linguistic theories: Construction Grammar (CxG) (e.g. Fillmore 1998) and Lexical Grammar (LG) (e.g. Sinclair 1996). The juxtaposition was motivated by the significant overlap in their tenets. Both theories take into account meaning (semantic and pragmatic) as well as lexical and grammatical factors. Constructions are symbolic units with particular features pertaining to their form and meaning. The former specify morphological, phonological and syntactic properties; the latter specify semantic and pragmatic attributes (e.g. Croft and Cruse 2004: 257-258). In LG, the unit is the lexical item, comprising the core (a word/phrase) and its collocates, semantic preference, semantic prosody and colligations . The core difference between the two theories is that LG gives clear prominence to lexis over grammar (the component of colligation is optional), whereas CxG posits no distinction between them.

The aim of this study is to create a bilingual electronic glossary, which is based on corpora of cooking recipes either in Portuguese and French. Gaston Gross's model of object classes is the one in which this research is based on. Using the analysis of three dictionaries, a gap was found in terms of equivalence criteria, which could make them inefficient for the dictionary user. This problem of equivalence was what motivated this study. As a result, this glossary is expected to efficiently cover a larger range of the proposed equivalence as an electronic language tool.

Corpus based examinations of the extent of modal marking (modal load) in if-conditionals in the written BNC (Gabrielatos 2007, 2010) have revealed that they have a significantly higher modal load than average, as well as a higher modal load than indirect interrogatives with if and whether, concessive conditionals with even if and whether, as well as non-conditional bi-partite constructions with when and whenever. More importantly, if-conditionals show a higher modal load than conditionals with other subordinators (assuming, in case, on condition, provided, supposing, unless). It is also noteworthy that the protases of if-conditionals have a modal load at least equal to that of the baseline – despite protases being already modally marked by if (not calculated in the modal load). The picture becomes more complicated, and more interesting, when examining the modal load of the subordinate and matrix parts of the constructions in focus – particularly those containing if as their subordinator. The explanations for the emerging modal load patterns are sought in the tenets of two recent linguistic theories: Construction Grammar and Lexical Grammar. More precisely, what is sought is whether the modal load of if-conditionals (and other constructions with if) can be explained by the collocations (and semantic preference) of the word if (Lexical Grammar), or by the nature (and composition) of individual constructions (Construction Grammar).

The weak claim motivating this study is that if-conditionals are strong modality attractors, due to the conditional (i.e. modal) meaning of if, with modality appearing in the if-clause, the main clause, or both. The strong claim is that if-conditionals can be regarded as modal colligations. The weak claim can be supported if it is shown that if-conditionals contain modality in a significantly higher than average frequency. Before examining the conditions under which the strong claim can be supported we need to turn our attention to the notions of modality, collocation, colligation and semantic preference, which inform the notion of modal colligation introduced in this paper.