Multiple sluicing through an experimental lens: Amelioration effects, crosslinguistic variation, and processing

Abstract

Multiple Sluicing (MS) is a clausal ellipsis configuration involving two or more adjacent wh-interrogative phrases that are interpreted together as standard multiple wh-question (Takahashi 1994). In the literature, sentences that involve MS have been widely regarded as marginally acceptable (especially in English) or as a peripheral and marked construction; yet, various factors have been identified that improve the acceptability of these constructions.

This dissertation aims to present a comprehensive investigation into the syntactic derivation, acceptability factors, and crosslinguistic patterns of MS in English, German, and Spanish. The goal is to unravel the complexities surrounding this construction and provide a data-driven baseline of judgments for configurations that have often received diverging judgments from different scholars or for which no acceptability judgments have been provided in the existing literature. Moreover, the studies presented here will serve as the basis for evaluating the current syntactic approaches to MS and diagnosing their shortcomings. New insights into MS are obtained in three series of experiments in which, for the most part, side-by-side comparisons are drawn from parallel experiments on English, German, and Spanish MS data.

First, Experiment Series 1 studies the influence of prepositionhood and weight on the acceptability of MS in the three above-mentioned languages. The presence of a prepositional wh-phrase in the non-initial sluiced wh-remnant is found to enhance overall acceptability in English and Spanish, but not in German. Moreover, English shows a significant decrease in acceptability when the non-initial sluiced wh-remnant was made heavy, contrary to Lasnik’s (2014) claims. No additional weight effects are observed in the three analyzed languages. The differences observed in those factors between the three languages can be explained by implementing a cue-based retrieval approach to ellipsis; in particular, the presence of cues that enable the retrieval of the thematic relation between the wh-phrases enhances the overall acceptability.

Second, Experiment Series 2 addresses the clausemate condition (CC) and island insensitivity in MS. The experimental results across the languages tested indicate that MS originating from islands and non-islands is equally acceptable and does not exhibit processing differences. Violations of the CC lead to a significant penalty in the acceptability. These support the island evasion approach, i.e., they suggest that MS can be derived from a non-isomorphic short source elliptic clause.

Lastly, Experiment Series 3 further explores the viability of the “short source” approach to MS configurations, drawing on the generalizations proposed by Abels & Dayal (2017, 2023) that were already investigated in Experiment Series 2. To investigate this approach, instances of MS that display apparently acceptable obviations of the CC (i.e., with a biclausal antecedent and a bound pronoun in the embedded clause) were tested. The experiment, focusing on German and Spanish, analyzed the underlying structure of the elliptic clause by means of case-matching or preposition-matching. The results revealed that sluiced wh-remnants that are case/preposition matched with the antecedent elements in the embedded clause received higher ratings, indicating a preference for a short source interpretation. Additionally, these findings suggest that the CC cannot be obviated without a grammatical short source and support the idea that the CC has a syntactico-semantic explanation. Therefore, these studies challenge the discourse-based solution Barros & Frank (2023) proposed, suggesting that apparent obviations to the CC result from a syntactic locality phenomenon. The strict locality constraint of the CC has been observed in the three analyzed languages, and the putative obviations to the CC can be argued to be only superficial. When put under experimental scrutiny, conditions where an underlying short source was available received significantly higher ratings. Thus, CC obviations are only apparent, and when a short source is blocked, the acceptability MS configuration with apparent CC violations suffers a decrease in the acceptability.

The experimental methods used in the dissertation facilitate the collection of the first set of reliable crosslinguistic data on those factors identified in the previous literature as affecting the acceptability of MS. The results from parallel studies in different languages show that, ceteris paribus, the factors that affect the acceptability of MS constructions in non-multiple-wh-fronting languages can be captured under a single analysis. Such an analysis states that (i) ease of recoverability of the thematic domain of the ellipsis-site from the discourse directly positively correlates with acceptability, and (ii) recoverability is made easier by the presence of cues that help to guide the hearer in recovering the thematic structure of the elliptic clause. So, while the specific cues may differ from language to language, those cues that guide the retrieval of the thematic structure from antecedent have crosslinguistically improved the acceptability.

To summarize, this dissertation systematically investigates MS by employing data collected from acceptability judgment and self-paced reading studies. By conducting controlled experiments, the studies presented here provide compelling empirical evidence that sheds light on the amelioration effects observed in this construction across different languages. Moreover, these findings strongly suggest that the existing syntactic derivations of MS inadequately encapsulate the complexity of the data and underscore the necessity for appealing to supplementary processing mechanisms to capture the data. By being the first to conduct experimentally-grounded research on MS, this research paves the way for future investigations into the intricacies of multiple sluicing by advancing the understanding of this configuration’s underlying mechanisms.