Yesterday, Laura Michaelis (Univ. of Colorado-Boulder co-author of, among other things, the 2020 Cambridge book Syntactic Constructions in English) alerted me to an article by Philip Miller (Université Paris Cité) & Peter W. Culicover (Ohio State Univ. & Univ. of Washington), “Lexical be“, Journal of Linguistics (2025), 1-24 — truly, hot off the press — which argues, in elegant detail, for a constructional approach to syntactic description, involving the positive licensing of constructions (rather than (negative) constraints on syntactic structures), and also honors Geoffrey K. Pullum (Professor Emeritus of General Linguistics at the University of Edinburgh).
So, yes, a fair amount of technical stuff, showing a bit of how (some) linguists approach the description of the syntax of one language and how they dispute with one another over the form of such descriptions. (I’m mostly just an observer here, but you should know that everyone I just mentioned — Michaelis, Miller, Culicover, Pullum — has been a departmental colleague or co-author of mine, so I have to be seen as a participant-observer, as we say in sociolinguistics.)
The abstract (in scholarly-abstract style, but bear with it):
We explore the surprising lexical be construction in English (e.g. Why don’t you be quiet? ). After an overview of previous discussions, an investigation of the use of lexical be in the COCA and SOAP corpora is provided. It is shown that its distribution is highly skewed and that it is completely felicitous only under a very limited set of conditions. An account of lexical be is then provided showing that the conditions that license it are inherited from more general constructions, most importantly the negative imperative [‘Don’t (you) V’] construction and the ‘Why don’t you’ construction. In this light, it is suggested that the lexical be construction, with its special properties, provides strong evidence for a constructional approach to linguistic competence along the lines of Goldberg (1995), Culicover and Jackendoff (2005), Sag (2012).
And then the introduction to the paper, which serves as an encomium to GKP, making this piece a suitable item for a Festschrift for him:
Two hallmarks of Geoff Pullum’s remarkable career are (i) careful attention to the fine-grained properties of English constructions (as seen, for example, in Huddleston et al. (2002) [the monumental Cambridge Grammar of the English Language, in one nearly 1900-page volume] and (ii) compelling arguments for a model-theoretic approach to the licensing of grammatical form, as contrasted with familiar derivational approaches (e.g. Pullum and Scholz 2001). In this paper, we discuss an understudied set of constructions that nicely exemplify both of these threads, what Huddleston et al. (2002, 114) (H&P) call ‘lexical be’
A central element of the model-theoretic approach — if you’re not a specialist in the syntax-semantics interface, you really don’t need to know about model-theoretic approaches to get through this posting — is that the conditions of syntax are positive licensing conditions, saying that certain bits of syntactic structure (with an associated semantics and pragmatics — meanings and contexts of use) are well-formed elements of the language. Allowing, for example, in English, a VP construct with a BSE-form head V to serve as a positive imperative sentence:
Touch my monkey, instructing an addressee to touch the monkey belonging to the speaker
Touched my monkey is a well-formed English sentence ((What did he do that was so bad?) Touched my monkey), but it’s not an imperative
Touch fabulous is not a well-formed English sentence, because touch fabulous is not a well-formed English VP, so it can’t be an imperative; and, alas, no other construction licenses it either
Here you see a bit of the interaction of constructions alluded to in Miller & Culicover’s abstract: the positive imperative construction calls for a VP, so it inherits the conditions on VPs in English. (Their interactional logic is more complicated than this, but here you can get the flavor.)
As for positive licensing, the idea goes back a long way in GKP’s thinking, as in the handout for G.K. Pullum & A.M. Zwicky’s 1997 LSA talk “Licensing of prosodic features by syntactic rules: The key to Auxiliary Reduction”:
2. As in Zwicky (1994), there is support for an account that involves only licensing conditions, saying what is in the language, never exclusion conditions or well-formedness constraints. We do not take the default to be that auxiliaries will encliticize everywhere so that constraints have to step in and prevent this in the bad environments; we assume no clitic can appear anywhere unless a rule of grammar provides a license for it.
(The reference is to my 1994 Berkeley Linguistics Society paper on construction grammar, available on-line here.)
February 2, 2025 at 8:20 pm |
I think the construction approach makes sense and comports well with first-language acquisition data.