This paper introduces an extension of collocational analysis that takes into account grammatical structure and is specifically geared to investigating the interaction of lexemes and the grammatical constructions associated with them. The method is framed in a construction-based approach to language, i.e. it assumes that grammar consists of signs (form-meaning pairs) and is thus not fundamentally different from the lexicon. The method is applied to linguistic expressions at various levels of abstraction (words, semi-fixed phrases, argument structures, tense, aspect and mood). The method has two main applications: first, to increase the adequacy of grammatical description by providing an objective way of identifying the meaning of a grammatical construction and determining the degree to which particular slots in it prefer or are restricted to a particular set of lexemes; second, to provide data for linguistic theory-building.

The most frequent statistics in corpus linguistics are frequencies of occurrence and frequencies of co-occurrence of two or more linguistic variables. However, such frequencies in isolation may sometimes be misleading since they do not take into consideration the degree of dispersion of the relevant linguistic variable. Many dispersion measures and adjusted frequency measures have been suggested but are neither widely known nor applied. Another unfortunate aspect of such measures is that many also come with a variety of problems. I pursue three objectives with this article. First, I want to raise awareness of this issue and make the available measures more widely known, so I present an overview of many measures of dispersion and adjusted frequencies. Second, I propose a conceptually simple alternative measure, DP, explain and exemplify it, and compare it to previously discussed measures. Third and most importantly, I urge corpus linguists to explore the notion of dispersion in more detail and outline a few proposals which steps to take next.

Much recent work in Cognitive Linguistics and neighbouring disciplines has adopted a so-called usage-based perspective in which generalizations are based on the analysis of authentic usage data provided by computerized corpora. However, the analysis of such data does not always utilize methodological findings from other disciplines to avoid analytical pitfalls and, at the same time, generate robust results. A case in point is the strategy of using corpus frequencies. In this paper, we take up a recently much debated issue from construction grammar concerning the association between verbs and argument-structure constructions, and investigate a construction, the English as-predicative, in order to test the predictive power of different kinds of frequency data against that of a recent, more refined corpus-based approach, the so-called collexeme analysis. To that end, the results of the application of these corpus-based approaches to an analysis of the as predicative are compared with the results of a sentence-completion experiment. Concerning the topic under consideration, collexeme analysis is not only shown to be superior on a variety of theoretical and methodological grounds, it also significantly outperforms frequency as a predictor of subjects’ production preferences. We conclude by pointing out some implications for usage-based approaches.

Adopting the perspective of construction grammar and related frameworks, this paper introduces a corpus-based method for investigating correlations between lexical items occurring in two different slots of a grammatical construction. On the basis of three case studies dealing with the into-causa- tive, English possessive constructions, and the way-construction, we show that such correlations are determined by semantic coherence. We identify three kinds of coherence: one based on frame-semantic knowledge, one based on semantic prototypes, and one based on image schemas. We con- clude by proposing a method that can potentially enhance the precision of our results and that allows us to identify ever-finer contrasts by adopting a multidimensional perspective towards co-occurrence patterns.

In Construction Grammar, the ultimate grammatical unit is the construction, a conventionalized form-meaning pairing. We present interrelated evidence from three different methods, all of which speak in favor of attributing an ontological status to constructions for non-native speakers of English. Firstly, in a sentence-fragment completion study with German learners of English, we obtained a significant priming effect between constructions. Secondly, these priming effects correlate strongly with the verb-construction preferences in native speaker corpora: verbs which are strongly associated with one construction resist priming to another semantically compatible construction; more importantly, the priming effects do not correlate with verb-construction preferences from German translation equivalents, ruling out a translational explanation. Thirdly, in order to rule out an alternative account in terms of syntactic rather than constructional priming, we present semantic evidence obtained by a sorting study, showing that subjects exhibited a strong tendency towards a construction-based sorting, which even reflects recent explanations of how constructions are related.