Multiword Expression Reading Group

23 Oct, 2001
Review MWE definition, take closer look at MWE examples and the classification thereof
  • Fabre's comments on the applicability of the existing MWE definition ("any expression realisable with at least one space that is not entirely predictable on the basis of standard grammar rules and lexical entries") to French discussed, and the decision made to limit the definition and general discussion to English for the time being, and worry about other languages at a later point.
  • Main object of discussion my classification of a range of MWEs according to the "qualities" (features) of MWEs listed above
  • Distinction between lexicalised and institutionalised MWEs brought into question. Suggestion made that institutionalised MWEs are simply compositional (syntactically and semantically), and all other MWEs are lexicalised (due to idiosyncratic syntax or semantics, or the presence of words that cannot occur in isolation). Hence, if we have (non-)compositionality as a feature, the lexicalised/institutionalised distinction does not give us any extra insight into the nature of a given MWE.
  • In relation to in the meantime, doubts were raised as to the pertinence of the "compositionality" feature. I.e. is it meaningful to talk of compositionality with respect to words such as meantime which cannot occur in isolation? Case for YES: it is in cases like this where the semantics are obvious and uncontroversial despite a simplex usage not existing (through analogy with the derivationally-equivalent meanwhile/meanwhilst); case for NO: compositionality is based on simplex sense, and loses meaning when no such sense exists. The isomorphism between non-compositional and lexicalised MWEs is preserved if we take the latter stance.
  • "Institutionalised" suggested to be an unfortunate term, given its "Big Brother" connotations. Perhaps "conventionalised" is more appropriate.
  • The notion of conventionalisation is brought out more clearly in the context of incorporation, in languages such as Dutch and Danish, where it is possible to say house-buying but not pencil-buying, for example. We would not like to say, however, that to buy a house is a conventionalised MWE, and treat it instead as a (fully productive) collocation of high frequency due to social factors.
  • Do conventionalised MWEs such as mail man belong in the lexicon? In terms of implementation, definitely, but in terms of our mental lexicon, are they stored as lexicalised units or elsewhere in world knowledge? While it is possible to construct a template along the lines of X man, as a means of expressing the agents of particular services, it doesn't explain why we can't readily say bread man, for example, and how you explain the fact that we (you) say mail man not post man in US English. With the latter case, it is further possible to set up mail as the default lexeme when referring to postal activities, and get mail man in this way, but this wouldn't appear to work as cleanly with other synonym pairs (e.g. road/street).

Last modified: Thu Aug 21 12:44:14 PDT 2003