Lewis: language is grammaticalised lexis, not lexicalised grammar. Thornbury (1998) said it was an idea without a methodology; by 2010, usage-based theory and corpora had moved him towards Lewis.

Claim

Lewis (1993, p. 34): “language consists of grammaticalized lexis, not lexicalized grammar”. Thornbury’s 1998 critique: no syllabus guidance, eclectic chunk-awareness activities, risk of premature lexicalisation — “all chunks, no pineapple” (lots of retrievable lexis, no generative grammar). In 2010 he is “more inclined to Michael’s position” because corpus linguistics and usage-based theory supply the missing rationale.

Mechanism / theory

  • Corpus linguistics / COBUILD (Sinclair); Hoey’s lexical priming (comments); Simpson-Vlach & Ellis (2010) on academic formulas.
  • Richards & Rodgers (2001, p. 138): “still an idea in search of an approach and a methodology.”
  • Skehan’s lexical → syntactic → relexicalisation cycle (comment) explains the fossilisation worry.
  • Lewis’s “frequency fallacy”: frequency ≠ utility.

Classroom moves

  • Highlight multi-word chunks and collocations as units to notice and reuse.
  • Combine a pre-selected lexical syllabus (frequent, pragmatically useful formulae) with emergent language.
  • Teach chunks from texts/transcripts, not decontextualised lists.
  • Reformulate learner output at the point of need.

Limits & critique

  • Selection problem: “where do you begin?” (Pickles, comment).
  • Dave Willis’s lexical syllabus (1990) predates Lewis (Peppard, comment).
  • Lewis himself later disowned “lexis is the answer”.

Apply in my context (Claude)

  • FCE Use of English Parts 2 and 4 (open cloze, key-word transformations) and IELTS Lexical Resource (collocation, “less common lexical items” at Band 7) both test chunk knowledge directly. A lexical syllabus pays off quickly here.
  • The “no pineapple” risk is real with IELTS learners who memorise templates. Watch for chunks they can retrieve but can’t vary or extend.

Connections