get ’em, stop it, gimme

Among the everyday examples of a phenomenon subjected to analysis in an awesome new paper by Joan Bresnan, “On Weak Object Pronouns in English”, which she will present at the Lexical Functional Grammar conference this July in Canberra (LFG2019, 8-10 July, sponsored by the ARC’s [Australian Research Council] Centre of Excellence for the Dynamics of Language, program on-line here).

Joan’s paper is a demonstration of what can be done with serious resources — really big databases, serious statistical tools, complex analytic tools — in investigating  very ordinary, but intricately structured, phenomena, and in how you might try to integrate the approaches of usage-based frameworks with those of formal grammar.

For me the paper has a special resonance, because the analysis develops some ideas of mine in a little note from 1986 that appeared only in a working papers volume and has mostly gone unnoticed since then: “The Unaccented Pronoun Constraint in English” (OSU Working Papers in Linguistics 32.100-113, 1986).

Examples off the street, all involving the verb give:

(#1) give ’em

(#2) give it

(#3) gimme

The abstract for Joan’s paper:

The ‘weak’ object pronouns in English are unstressed, unaccented pronouns which have restricted syntactic distributions adjacent to their head verbs or prepositions (Zwicky 1986, Wallenberg 2007): the pronouns in get ’em, stop it, gimme are examples. These pronouns are closely bound to their heads by phonology, allomorphy, and syntactic adjacency, yet they have semantic and syntactic scope typical of ordinary syntactic DPs in the VP, and they can occur with a small class of quantifier specifiers that indicate their phrasal status. Those verbs and pronouns with the highest probability of occurring together are the most likely to encliticize and fuse together. All of these properties of weak object pronouns yield to a coherent understanding within a hybrid theory (Bresnan 2018) that incorporates a probabilistic model of the mental lexicon from usage-based linguistics (Pierrehumbert 2001, 2002, 2006) as the lexicon of lfg with lexical sharing (Wescoat 2002, 2005). The formal representations of weak object pronoun encliticization in this theory can be thought of as depicting a synchronic state of partial grammaticalization of syntactically independent, prosodically reduced pronouns into morphosyntactic enclitics.

Two notes on linguists cited in the abstract:

Joel C. Wallenberg is a Stanford BA.(with honors) from 2003, with the thesis “The Story of the American –self: a case study in morphological variation” (I was his principal advisor); Joel went on to a Univ. of Pa. PhD in 2009

Michael T. Wescoat is a Stanford PhD from 2002, with the dissertation Lexical Sharing (Tom Wasow was his principal advisor, and I was on the committee; Geoff Pullum, then at UC Santa Cruz, was a significant influence on Michael’s work, though he wasn’t on the committee)

In any case, my little note posited prosodic attachment of weak pronouns to the left as requiring that the category of the host (i) be a sister to that of the pronoun, (ii) be a lexical category, and (iii) be a category that governs case marking. That’s then the basis for the lexical sharing schema Joan proposes.

I couldn’t find any place to publish the note properly, even back in 1986 — lots of my work even then was too quirky for publication in regular journals — so the piece has been cited almost entirely by me, my collaborators (like Geoff Pullum), my students (like Joel Wallenberg), and my close colleagues (like David Dowty at Ohio State and Ivan Sag at Stanford).  So it’s gratifying that Joan could find a way to embody its ideas and test them.


Leave a Reply

%d bloggers like this: