Talk and masterclass by Aaron Ecay

Posted on October 9, 2014 by

Aaron Ecay (University of Pennsylvania) is in Manchester on 14th-15th October, to give a talk and a masterclass.

The talk will be on Tuesday 14th October at 4.15pm in Samuel Alexander A7, with the title The evolution of do-support: lexical and lexical class effects. In it, Aaron will revisit a well-studied change from the history of English, drawing on a substantial new corpus study, and investigate the effects of individual words and classes of words on the progress of the change as a whole. Here’s the abstract:

In this talk, I will be discussing the history of do-support in English. This phenomenon emerged in the Early Modern English (EME) period (post-1500) and over the course of 200+ years gradually became obligatory in certain morphosyntactic contexts. I’ll argue that do-support in English passes through an intermediate stage, wherein “do” is neither a lexical verb (as in Middle English) nor last-resort morphology to support a stranded affix in T (as in present-day English). Rather, at this intermediate stage, “do” serves to mark the presence of an external argument. The evidence for this conclusion is:
1. examples of EME “do” co-occurring with modals, “have”, and the “ing” of gerunds
2. patterns of co-occurrence between EME “do” and certain adverbs
3. divergent patterns of evolution of “do”-support in unaccusatives vs. other predicate types (unergative, transitive, experiencer-subject)

Having established the conclusion, I’ll examine its implications for the Constant Rate Hypothesis (Kroch 1989, LVC). Considering “do”-support as a series of interrelated changes rather than a single monolithic change obscures its relationship to a putative verb-raising parameter. I’ll present data on the evolution of do-support and loss of verb-raising with individual lexical items and lexical classes from a large (10⁹ words) corpus of EME data which demonstrate that the trajectories of the change are nonetheless highly regular. The usual methodology for quantitative historical syntax in the CRH tradition (null hypothesis significance testing, with the expectation/hope that the null hypothesis will survive) was developed to suit datasets of a particular size; the use of a much larger dataset in this work presents a methodological challenge but does not, by itself, cause the core insight underlying the CRH to disappear.

The masterclass, on Wednesday 15th October, 11-1 in the Ellen Wilkinson Graduate School Training Room, is on Constructing quantitative grammatical arguments. Here’s the abstract:

The competence/performance distinction is one of the fundamental concepts of modern generative grammar. One of the most interesting aspects of historical syntax is that it provides a means to bridge this gap and make conclusions about speakers’ grammar based solely on their production. Indeed such inference is vital to the endeavour of historical syntax, for otherwise it would founder in the absence of native speakers of past language varieties. Our primary point of departure for this seminar will be the Constant Rate Hypothesis (CRH; Kroch 1989), one theoretical framework which makes these bridging inferences possible.

Changes in construction frequency do not happen immediately, rather they progress to completion gradually in corpus data. Since a single abstract parameter can have effects in a variety of surface contexts, it is possible to capture several different traces of a single linguistic change in a diachronic corpus. The CRH tells us that, if these independent traces move in the same way (in a sense to be made precise in our discussion), we should believe that they are indeed attributable to a single underlying grammatical parameter, and not several (quasi-)independent individual changes.

When working with corpus data, it is important to ensure that the construction frequencies we are measuring faithfully represent underlying grammatical realities. Kroch (1989) contains some examples of corrections which are applied to corpus data in order to factor out influences unrelated to the change being studied. Frisch (1997) contains another example of the application of the CRH methodology and the necessity of correcting for distortions in corpus data. Frisch’s paper should be read in combination with Wallage (2008), which critiques some of Frisch’s methodology and advocates for somewhat different conclusions.

The seminar will develop along the following lines, broadly speaking:

  1. background discussion and theoretical motivation of the CRH
  2. discussion of case studies from Kroch 1989, as a window to the theoretical and methodological environment for applying CRH analysis
  3. discussion of the dialog between Frisch and Wallage, as a window to understanding potential pitfalls of a CRH analysis
  4. (time permitting) discussion of other paradigms in quantitative historical argumentation


  • Frisch, Stefan. “The Change in Negation in Middle English: a NEGP Licensing Account.” Lingua 101 (1997): 21–64.
  • Kroch, Anthony. “Reflexes of Grammar in Patterns of Language Change.” Language Variation and Change 1.3 (1989): 199–244.
  • Wallage, Phillip. “Jespersen’s Cycle in Middle English: Parametric Variation and Grammatical Competition.” Lingua 118 (2008): 643–674.

All welcome. Contact George to get hold of the readings!