Aarhus Universitets segl

Syntactic and semantic dependencies across adjunct boundaries

Guest lecture by Rob Truswell, University of Edinburgh, UK

Oplysninger om arrangementet


Onsdag 24. august 2022,  kl. 11:15 - 12:30


Building 1481-324


Ken Ramshøj Christensen

In this talk, I report on work in progress under the auspices of the LASER project (Locality and the Argument-Adjunct Distinction: Structure-building vs. Structure-enrichment, researchers Kenyon Branan, Thomas McFadden, Elise Newman, Sandhya Sundaresan, Rob Truswell, and Hedde Zeijlstra), concerning the "selective opacity" (Keine 2019) of adjuncts for syntactic dependencies, with particular focus on wh-movement.

Here's an illustration of the selective opacity of adjuncts. Some syntactic dependencies, like obligatory control, routinely cross adjunct boundaries (compare (1a), with control into a complement, with (1b), with control into an adjunct -- both are fine, see Landau 2021 for arguments that (1b) must be an instance of obligatory control). On the other hand, others, like long-distance agreement, apparently never do ((2a) illustrates LDA into a complement, but there is no equivalent into an adjunct like (2b); (2c) shows that the same sentence without LDA is grammatical). So adjuncts are selectively opaque: they block some kinds of syntactic dependency, but don't affect others.

1a)    The house of cards avoided [__ collapsing completely].
1b)    The house of cards wobbled a little [before __ collapsing completely].

2a)    Ram-ko [peṛ kaʈ-ne] dene the
          Ram-DAT tree cut-INF.OBL give.INF.M.PL be.INSTR.PST.M.PL
         "Ram had to let the trees be cut."
2b)    *Ram-ko [peṛ kaʈ-ne] Almoṛaa jaa-ne the
          Ram-DAT tree cut-INF.OBL Almora go-INF.MPL be.PST.MPL
         "Ram had to go to Almore to cut trees."
2c)     Ram-ko [peṛ kaʈ-ne] Almoṛaa jaa-naa thaa
          Ram-DAT tree cut-INF.OBL Almora go-INF be.PST
         "Ram had to go to Almore to cut trees." (Hindi, all examples from Rajesh Bhatt, p.c. to Sandhya Sundaresan)


The status of wh-movement with respect to the selective opacity of adjuncts is complex. On the one hand, it is now widely accepted that wh-movement out of adjuncts is sometimes possible (Belletti & Rizzi 1981, Chomsky 1982, Postal 1998, Truswell 2007, 2011, contra Cattell 1976, Huang 1982, Uriagereka 1999, etc.). On the other hand, wh-movement out of adjuncts is also constrained by a variety of factors. It is impossible altogether in many languages (Postal 1998, Truswell 2008 -- see (3)), and even in languages which permit it, it is subject to "weak island"-like constraints (only referential noun phrases can be extracted, (4)), and semantic restrictions on event structure, as in (5).

3a)    the director that she flew there [to confront __]
3b)    *le directeur qu' elle y est allée en avion [pour confronter __]
          the director that she there is gone.F in plane for confront.INF
          "the director that she flew there to confront" (French, Postal 1998)

4)      *the way in which she flew there [to speak to the director __]

5a)    What did she arrive [whistling __]? (Matrix VP = achievement)
5b)    *What does she work [whistling __]? (Matrix VP = activity, Truswell 2007)


A crucial step in making sense of this mess is to draw a clean dividing line between grammatical and extragrammatical effects. In Truswell (2011), I pushed for a mainly extragrammatical treatment of wh-movement out of adjuncts, placing as much explanatory burden as possible on semantics, pragmatics, and processing. But for all its appeal a priori, this extragrammatical approach has limited potential for situating wh-movement relative to the selective opacity demonstrated in (1-2). Worse, there are hints that wh-movement out of adjuncts shows "conditional opacity" (Grano & Lasnik 2018): it is only possible when some other dependency crosses the adjunct boundary (for instance, (6a), with obligatory adjunct control, is judged more acceptable than (6b), with arbitrary adjunct control).

6a)    ?Who is John_i well-dressed [PRO_i to greet __]?
6b)    *Who is the door open [PRO_arb to greet __]?


In this way, wh-movement from adjuncts raises acute (currently unanswered!) questions concerning the limits of syntax. This talk will evaluate the two most obvious approaches to these questions: developing partially extragrammatical accounts of selective and conditional opacity, or syntacticizing primarily semantic phenomena such as relations among events.

Participation is open to everybody

The event is financed by SCC’s research programme for language, linguistics, and cognition