Gert-Jan Schoenmakers: The wonderful world of island experiments

An important discussion in syntactic research revolves around the question whether the phenomena we observe are due to specific constraints on language structure or due to non-syntactic components of human cognition. A central topic in these discussions are so-called ‘island configurations’ (Ross 1967), i.e. unacceptable long-distance dependencies such as in (1b).

(1)    a. Which Taylor Swift album do you think [that I like ___ best ] ?
b. * Which Taylor Swift album do you wonder [whether I like ___ best ] ?

Constructions like (1b) are traditionally analyzed as violating particular syntactic rules (Chomsky 1973, 1977), but alternative accounts argue instead that the source of unacceptability is rather conflicting discourse-packaging conditions (Erteschik-Shir 1973; Goldberg 2006, 2013; Abeillé et al. 2020) or limitations of the processing system (Kluender 1991; Hofmeister & Sag 2010). Furthermore, syntactic accounts that rely on feature-based intervention effects (Starke 2001, Rizzi 2018) that have been linked to sentence processing (Villata 2017) have also been employed to explain certain island effects.

Many of the predictions that follow from these theoretical frameworks have been tested experimentally (in different guises) for various languages. In this talk, I will discuss a series of experiments that I conducted with colleagues to contribute to these discussions. I will use data from Dutch, Romanian, and French (preliminary) to address different questions coming from these accounts and, time permitting, I will shed light on some methodological issues pertaining to the reliability of the datasets.

Teams
Meeting ID: 350 359 744 147
Passcode: gS3gM3qu
Link to join: https://teams.microsoft.com/l/meetup-join/19%3a_J7gGufoC4RZaTkFplCZLmMBAqUw_5gOE07TFeuMFSs1%40thread.tacv2/1742219080409?context=%7b%22Tid%22%3a%22d72758a0-a446-4e0f-a0aa-4bf95a4a10e7%22%2c%22Oid%22%3a%22aa1e7f0b-4732-46e2-accd-b7d55568cd85%22%7d