Why multi-word information is key to language learning and use

This is a FREE lecture given by Associate Professor Anna Siyanova-Chanturia, School of Linguistics and Applied Language Studies, Te Herenga Waka – Victoria University of Wellington

Human language is original and highly creative. A seemingly unlimited number of phrases and sentences can be constructed using a finite number of words and rules of grammar. Interestingly, however, while we undoubtedly can exercise the creative potential of language, we rarely do so. Much of the language we use on a daily basis is, in fact, ‘formulaic’ or ‘prefabricated’, rather than newly assembled word-by-word.

In English, we have fast food yet a quick meal, rather than the other way around. We recognise that bride and groom sounds better than the nearly identical groom and bride. And we are much more likely to offer a friend a cup of tea than a mug of tea (even though we may well end up giving them a mug!). So, despite the potentially infinite creativity of language, many words tend to co-occur with some words more often than with other, seemingly identical ones. Such co-occurrences are known as multi-word expressions (MWEs).

MWEs encompass a large set of sequences above the word level, such as collocations (strong tea), binomials (fish and chips), multi-word verbs (put up with), idioms (tie the knot), and so on. These sequences differ in several ways. However, what they have in common is that they are conventional and, hence, highly familiar to proficient language speakers. Importantly, MWEs are predictable, or even uniquely predictable. For example, excruciating only ever evokes pain, while toss a is very likely to be completed by coin. In the present talk, I will review key empirical evidence attesting to easier semantic integration of MWEs compared to novel language, and to the activation of a template-matching mechanism for uniquely predictable phrasal configurations that MWEs are. Once marginalised and largely disregarded, MWEs have been shown to have far-reaching implications for theories of first and second language acquisition, processing, and use.