Zero-Adjective contrast in much-less ellipsis: the advantage for parallel syntax

ABSTRACT

This paper explores the processing of sentences with a much less coordinator (I don’t own a pink hat, much less a red one). This understudied ellipsis sentence, one of several focus-sensitive coordination structures, imposes syntactic and semantic conditions on the relationship between the correlate (a pink hat) and the remnant (a red one). We present the case of zero-adjective contrast, in which an NP remnant introduces an adjective without an overt counterpart in the correlate (I don’t own a hat, much less a red one). Although zero-adjective contrast could in principle ease comprehension by limiting the possible relationships between the remnant and correlate to entailment, we find that zero-adjective contrast is avoided in production and taxing in online processing. Results from several studies support a processing model in which syntactic parallelism is the primary guide for determining contrast in ellipsis structures, even when violating parallelism would assist in computing semantic relationships.

KEYWORDS:

• Ellipsis
• parallelism
• contrast
• coordination

Acknowledgments

Thanks to Joseph Tyler, Dallas Cox, and Benjamin Lee for assistance with Experiments 2A-B and 4; Sarah Nelson and Blake Clark for assistance with Experiment 1B; Katherine Griffitts, Joe Castle, and Torianne Crouch for assistance with Experiment 2B; Megan Ison and Will Hunt for assistance with the corpus analysis; Masaya Yoshida and David Potter for general discussion of FSC structures; and several anonymous reviewers for helpful comments.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Katy Carlson http://orcid.org/0000-0001-6372-4389

Notes

1. Where possible, we give examples of much less ellipsis found in the Corpus of Contemporary American English (COCA; Davies, 2008). This source is indicated parenthetically.

2. For reasons of space, we have omitted syntactic tests that indicate clausal ellipsis; see Hulsey (2008) for arguments that FSCs always involve some kind of ellipsis, and Harris (2016) for arguments that the type of ellipsis involved is of the clausal variety.

3. As noted by a reviewer, this comparison is weakened by the fact that FSCs often appear in the scope of explicit negation, even though a fair number of examples observed in corpora are licensed by implicit negation, negative adjective and adverbs, questions, and pragmatic adversity. Even when the FSC structure appears beneath clausal negation, fronting is still prohibited for FSC structures (a-b), and not subordination (c-d).

• (i) a. It's not the case that John would go to the party, much less the after party.

•   b. * It's not the case that, much less the after party, John would go to the party.

•   c. It's not the case that John will go to the party, even if Mary came.

•   d. It's not the case that, even if Mary came, John will go to the party.

4. We are following previous literature in assuming that these structures coordinate constituents (e.g. Hulsey, 2008; Toosarvandani 2009, 2010), though a reviewer questions whether the proper relation might be that of subordination. We continue to use the term coordination because parallelism effects are also observed for subordinate clauses (e.g., Sturt et al., 2010), and would not therefore confound our findings, even if the mechanisms for parallelism are not necessarily the same between coordination and subordination.

5. Important exceptions include dialects that allow so-called “positive” let alone, in which the scalar relation between remnant and correlate are reversed (Fillmore et al., 1988; see also Mark Liberman's commentary on Language Log, November 21, 2007, accessible as http://itre.cis.upenn.edu/~myl/languagelog/archives/005142.html), and the afterthought use, which appears to be devoid of a scalar component and can be paraphrased along the lines of not to mention (Cappelle, Dugas, & Tobin, 2015).

6. This strength of this generalization depends on the syntactic category of the remnant. NP and VP remnants, the most common categories, displayed total contrasts (with NP or VP correlates) in approximately 80% of cases, whereas Adverb remnants showed a near even split between Adverb (55%) and PP (45%) correlates. Nevertheless, the majority of remnant classes showed total or near total matching between remnant category and contrast.

7. Also, the lack of a strong processing cost for VP over NP remnants in both types of FSC structure further supports the idea that the processor projects an ellipsis structure at the coordinator, obviating structural economy preferences. It would otherwise be a puzzle why a larger and more structurally complex VP remnant would fail to elicit processing costs compared to an NP remnant.

Additional information

Funding

We also extend thanks for partial financial support to Pomona College, the Eunice Kennedy Shriver National Institute of Child Health and Human Development of the National Institutes of Health under [grant number R15HD072713], and an Institutional Development Award (IDeA) from the National Institute of General Medical Sciences of the National Institutes of Health under [grant number 5P20GM103436-13]. The research described here is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or any other institution.