10.1 Multi word patterns

Over the past weeks, we have been focussing mostly on properties of individual lexemes. Of course, there was always an emphasis on the fact that lexical entries are not restricted to word level. We have encountered compounds and complex prepositions which, in some situation, behave like other single-word lexemes. However, we haven’t really focussed at the syntagmatic relationships between lexemes. In grammar, we would transition from morphology to syntax now. What we could see as equivalent in lexical semantics, is multi-word patterns, which range from simple adjective-nouns collocations, to clause-sized idioms.

10.1.1 From collocations to syntactic patterns

  1. tall building
  2. remarkable feat
  3. stark contrast
  4. thick accent

With collocations, we have encountered a common class of multi-word pattern already. Most of you should also be familiar with a range of syntactic patterns, such as Subject-Verb-Object, relative clauses, the double-object construction, tenses etc. Phrase structures (remember syntax trees) are inherently multi-word patterns. Many of these structures have aspects of multi-word patterns, and often include lexical elements. Tenses, such as the going-to future (57, 58) have a fixed element (e.g. the semi-modal going to/gonna) and a variable verb phrase slot. Not every lexical verb can be used with all tenses. As a result, you get the patterns that every learner of English is familiar with. For example, stative verbs don’t usually occur in the progressive (60). Any syntactic structure has a lexical component to it.

  1. They are going to open the gyms again.
  2. They’re gonna open the gyms again.
  3. I am eating at the moment.
  4. * I am knowing it at the moment, but I might forget.

going to is an interesting case because of the direction of the contraction to gonna. With phrase structure as the only model to explain the syntagmatic relationships, we would have to expect the particle to to be contracted to its closest constituent, which would be the following verb, with which it forms a unit (to-infinitive). What we actually see is going and to fusing together. This is easily explained just by looking at frequencies. Imagine that our brain works like a sort of computer that constantly tries to model and predict incoming stimuli and automate and reduce wherever possible. The probability that going is followed by to is very high, in certain uses you could consider it redundant information. The verb slot following to, however, is much more variable. Therefore, even the most frequent combinations of to + verb aren’t frequent and distinctive enough to trigger contraction.11

10.1.2 Lexical variability

If we think of lexical patterns and syntactic patterns as two sides of the same coin, we still have to account for their obvious differences. Syntactic patterns are much more general and have optional slots and highly variable slots. Collocations on the other hand are sometimes part of larger variable patterns, but there is always some lexically ivariant part. Below we have elements like ear off and into that don’t change. The other slots in those constructions, however, do. Nevertheless, the variable slots cannot be arbitrarily filled. The verb slot in (61) is largely restricted to communicative verbs, in (62) the noun slot only allows a very restricted set of swear words.

  1. [VERB] [SOMEONE]’s ear off
  2. [beat|kick|slap] the [crap|hell|shit] out of [SOMEONE]
  3. into-causative: [VERB] [SOMEONE] into [VERB]ing [something]

In conclusion, the most general and most variable structures would be found on the side of syntactic patterns, while the more fixed constructions are determined lexically. Lexical and syntactic patterns co-exist and readily mix and mingle.

  1. Of course, this is a strongly simplified explanation of how contractions happen. The exact mechanics behind grammaticalization processes responsible for things like the going-to-future are a topic in Historical Linguistics.↩︎