This paper investigates the syntactic structure of multiple sluicing (MS) in English and German. The acceptability status of this construction in English remains controversial, and the judgments reported in the literature vary significantly (cf., Takahashi 1994; Nishigauchi 1998; Merchant 2001; Lasnik 2014; Kotek and Barros 2018). Therefore, obtaining experimental results on the acceptability status of MS is crucial for successful future investigations of this configuration. Here, I focus on two factors that have been reported to influence the acceptability of English multiple sluicing: prepositionhood, which is the presence or absence of a prepositional wh-phrase as the non-initial remnant, and weight, which refers to the number of heads following the wh-word in a non-initial remnant. The results of my experiments show that, despite multiple sluicing generally being a marked construction in English, speakers exhibit clear preferences for some configurations over others. First, the presence of a prepositional wh-remnant improves the overall acceptability of multiple sluicing sentences, which confirms claims from the literature (e.g., Richards 2010; Lasnik 2014). I argue that this preference is observed because prepositions provide additional cues which facilitate the retrieval of the information presented in the antecedent (cf. Martin and McElree 2008, 2011). Second, the results presented here reveal a significant negative effect on the overall acceptability of English multiple sluicing in the presence of a heavy wh-phrase, contra Lasnik (2014). I will interpret these results as motivated by information-structural factors, namely the reiteration of given material causing a penalty in the perceived acceptability of multiple sluicing sentences. Moreover, my experiment on German confirms that multiple sluicing in this language is not modulated by the factors of prepositionhood and weight. I claim that the differences observed for prepositionhood and weight in the two examined languages can be captured by an implementation of a cue-based retrieval approach to ellipsis (Martin and McElree 2008, 2011; Nykiel, Kim, and Sim 2023), once coupled with stipulations about the differing grammaticality status of MS in English and German.
This paper investigates the information structure and discourse properties of fragmentary subordinate clauses (FSCs) in German, focusing on FSCs introduced by the causal complementizer weil (because), such as Ein Drittel der Bevolkerung ist vulnerabel, weil vorerkrankt ('A third of the population is vulnerable because [they are] pre-existing condition sufferers'). We analyze weil-FSCs as a common ground (CG) management tool, showing that their contribution to discourse goes beyond answering explicit or implicit questions under discussion (QUDs). Instead, weil-FSCs function as a speaker strategy for updating the common ground without inviting negotiation or challenge. Depending on context, FSCs may provide background information, reactivate shared knowledge, or introduce new content as if it were already mutually accepted. This rhetorical strategy allows speakers to increase shared information with a syntactically reduced structure while preserving the core proposition of the main clause that contains the FSC. For the hearer, the weil-FSC functions as a request to parse the information and add it to the common ground. We argue that in those instances where the proposition of the FSC is presented as if it were already in the common ground, the causal content must be accommodated on the go without objection.
Barros and Kotek (2019) group the semantic identity conditions on clausal ellipsis that have been developed in the transformational generative literature into two groups, those that restrict antecedents of clausal ellipsis to questions or questionlike semantic entities, and those that do not. They refer to these as Q-equivalence and focus-based approaches, respectively. Barros and Kotek present strong arguments to suggest that an issue-based Q-equivalence theory such as AnderBois' (2011, 2014) cannot be correct. They also demonstrate that one QUD-based Q-equivalence approach - namely, Barros (2014) - has shortcomings in its current form. From this, Barros and Kotek conclude that, even if a Q-equivalence approach can be developed that does not exhibit the same shortcomings that AnderBois' and Barros' theories do, this approach will require supplementary assumptions and technical machinery that are/is not required under a focus-based approach. Thus, a focus-based approach is favoured on conceptual grounds, as it is more parsimonious. In this response paper, I contend that Barros and Kotek's conclusion is unwarranted. I argue that it arises from an unbalanced comparison, in which (i) the least feasible QUD-based approach is treated as the exemplar of QUD-based approaches more generally and (ii) data that supports QUD-based approaches and/or undermines Barros and Kotek's focus-based approach are neglected. I will show that, once a more balanced comparison is made, any QUD-based approach that is more fleshedout than Barros' (2014) is superior to the focus-based approach.
The majority of previous research on ellipsis has broadly agreed that sluicing requires some type of identity (syntactic, semantic, or some combination of both) between the elided material and its antecedent in order to be felicitous. By way of three experiments, we investigate the possibility of inferentially-resolved sluicing, and demonstrate (i) that it can be highly acceptable even when the elided material deviates substantially from its antecedent and (ii) that the acceptability of such sluices varies considerably across items. While we argue that both findings are inherently more consistent with referential theories of ellipsis than identity theories, our investigation also highlights new challenges for referential theories in explaining the variability associated with inferential ellipsis resolution.
This paper investigates the ambiguity of wh-questions at the Syntax-Discourse Interface by comparing German and Swedish. It focuses on three questions: i. What kinds of ambiguities occur and how can the strategies of the speaker and the interpretive strategies of the listener be characterized in both languages? ii. What kinds of linguistic devices are used in marking information-eliciting questions vs. rhetorical questions in each language? iii. How can we describe situations where wh-questions are not disambiguated and serve further functions? And what role does the common ground of the interlocutors play in these situations? In answering these questions, we propose two hypotheses: the Complex Ambiguity Hypothesis and the Cross-Linguistic Hypothesis. The Complex Ambiguity Hypothesis claims that there is an ambiguity between the information-eliciting interpretation and the rhetorical interpretation that is also interrelated with an information-structural ambiguity. The observation that there are language-specific linguistic means that a speaker can select to disambiguate these two layers of ambiguity is specified in the Cross-Linguistic Hypothesis. Empirical evidence supports the observation that Swedish and German show considerable differences in the syntactic realization of wh-questions and in their mapping to discourse strategies. Swedish speakers prefer the cleft construction to unambiguously mark the focus of argument wh-questions and the need for a referential specification of the wh-element. The disambiguation of information structure by clefts in Swedish also has consequences for other discourse-related aspects of wh-questions and influences their potential use as rhetorical questions. In contrast to Swedish, German speakers tend to use different types of modal particles both for disambiguation of the information structure and for the explicit marking of the rhetorical character of wh-questions. Finally, we discuss the complexity of the ambiguity issue with reference to special instances of Swedish and German wh-questions in which ambiguity is not eliminated despite the use of clefts and modal particles. In these cases, which often have a rhetorical flavor and are closely related to emotions, the listener must rely on further contextual clues or propositions in the common ground that guide the interpretation.information structure, ambiguity, wh-questions, rhetorical questions, common ground that guide the interpretation.
This paper discusses a set of verbal adjuncts in English that is characterized by syntactic, semantic, and information-structural complexity. I will focus on extraction asymmetries reported for this adjunct class and argue that these asymmetries do not necessarily require additional grammatical licensing principles. Instead, I will present experimental evidence showing that the reported extraction asymmetries are actually unrelated to extraction because the same effects can be observed in non-extraction constructions. Factors like the aspectual class of the embedding verb and its grammatical verb type lead to identical relative acceptability differences in the presence and absence of extraction, so that the reported extraction asymmetries investigated in this paper can be explained independently without the requirement for grammatical licensing principles. This points in the direction of independent effects on acceptability which can be linked to different degrees of processing complexity in the adjunct type under investigation.
The central claim of the paper is that focus-background structure always reflects the occurrence of one or more of a variety of focusing operators in semantic structure, i.e., of truth-conditional or use-conditional operators which split their scope into two parts. It will be shown that this view accounts for the semantic and pragmatic non-uniformity of focus-background structure, especially in regard to the role of alternatives and to the givenness of background information. Further emphasis will be laid on the interplay of focusing operators, which is highly flexible but subject to general scope constraints. In combination with these constraints, the multiple operator view also explains restrictions on the focus-background structure of clauses with expressive elements and on the inclusion of dependent clauses into the focus-background structure of the main clause.
In his influential book, Potts (2005) analyzed conventional implicatures, e.g., the content expressed by parentheticals, appositives and expressives, as nonat-issue content. He drew a distinction between the asserted proposition (at-issue content), which is typically taken to be an answer to a question under discussion (e.g., Roberts 2012; Tonhauser 2012), and a range of other entailments and presuppositions conveyed in an utterance (non-at-issue content). Only recently, it has been shown that this division is relevant for syntactic processing as well (see e.g., Dillon, Clifton and Frazier 2014). A similar divide within an utterance seems to be evoked by certain adverbials. It has been observed that modifiers of manner trigger a 'pre-suppositionlike' inference.
This paper tests the predictions of the QUD approach (Roberts 1996) with respect to the processing of discourse-initial fragments. To this effect, we present three empirical studies, two production studies and an acceptability rating study. In a pretest, we first estimate the salience of potential QUDs in a given utterance context c (= context-based likelihood). In the first experiment, we then estimate the salience of QUDs given both the utterance context c and the utterance a itself (= answer-based likelihood). In the second experiment subjects rate discourse-initial fragments and sentences a that relate to a QUD with an either high (= predictable) or low (= unpredictable) context-based likelihood. Our results show that the answerbased likelihoods do not predict the ratings, which is surprising given the assumptions of the QUD approach. At the same time the data suggests (i) that subjects generally prefer utterances whose QUD is already salient in the utterance context, and (ii) that retrieving the QUD of fragments with a rather low context-based likelihood requires higher processing effort, which is in turn reflected in degraded ratings.
A substantial literature has shown that Condition C (Lasnik 1976; Chomsky 1981) is essentially correct in ruling out cases where a non-pronominal NP is c-commanded by a coreferential pronoun. At the same time, many have noted a number of counterexamples to Condition C (CCCs). These suggest that there may be factors other than syntactic configuration that govern whether two expressions in the same sentence may be used to corefer. In this chapter, we explore the hypothesis that both Condition C effects and CCCs are explained by a conspiracy of three non-syntactic factors: (i) interpretive economy, which favors binding over coreference (Reinhart 1983; Reuland 2001), (ii) antilogophoricity, which we take to be a semantic property of r-expressions (Patel-Grosz 2012; Yashima 2015) and (iii) processing complexity, which embodies factors related to the accessibility of discourse referents (Ariel 1990, 2001; Arnold 2010, i.a.).
Alrenga et al.(2012) (AKM) propose that in English, both the predicate-marking morpheme -er/more and the standard-marking morpheme than introduce the semantics of comparison. AKM's primary goal is to account for correlations between the scope of the standard phrase and the scope of comparison, but they also gesture at some potential implications of their analysis for the typology of comparatives. This paper examines these typological implications in more detail, by exploring the hypothesis that a point of cross-linguistic variation has to do whether the semantic content of comparison is introduced by an expression that composes with the predicate, by one that composes with the standard, or in both places. Taking AKM as a starting point, I will outline a set of properties that we expect a language to have depending on how it fits into this typology. I will then provide some initial evidence in support of the typology by showing that if English shows the expected properties of a "both" language, Mandarin Chinese shows the properties of a language in which comparison is exclusively in the predicate, and Japanese shows the properties of a language in which comparison is exclusively in the standard.
This paper concerns different conceptualizations for degree constructions of sufficiency and excess in English, with enough and too. According to many prior works, a sufficiency conveys that a measured degree meets or exceeds a minimum degree for some purpose (meeting-the-minimum), whereas an excessive conveys that a measured degree strictly exceeds a maximum degree for some purpose (exceedingthe-maximum). I will instead advocate for a causation-based conceptualization for sufficiency and excess, building on Schwarzschild(2008) and Grano(2022). Together with the monotonicity property of gradable predicates, I show that the causation-based descriptions can derive much of their respective truth conditions without stipulating the connection between sufficiency and meeting-the-minimum and between excess and exceeding-the-maximum as in previous accounts. I then present the facts from certain edge cases where the proposals diverge in their predictions; these prove to be problematic for the classic descriptions for these constructions, but are unproblematic for the causation-based formulation.
This chapter offers an answer to four main theoretical debates regarding the syntax and semantics of particle comparatives that have been the subject of ample discussion since the early 70s, focusing on Spanish and English data. Debate 1 addresses the issue of the linkage type between the comparees: is there a coordinate or a dependent (subordinate) relation between them? Debate 2 examines the process or processes responsible for Comparative Ellipsis, an ellipsis operation claimed to silence one or several constituents in clausal standards of comparison. Debate 3 focuses on the issue on the nature of the standard of comparison: can standard markers select for both clausal and phrasal complements? Debate 4 concerns the nature and mechanisms responsible for Comparative Deletion and Comparative Subdeletion: that is, the obligatory omission of a measure and a gradable predicate or a quantified nominal from the standard (as illustrated by the unacceptability of "More women than [I was expecting (*thirty men/*many women)] came."). What is the nature and mechanisms responsible for Comparative Deletion and Comparative Subdeletion, and why are these processes obligatory? In sum, this chapter offers a formal diagnosis protocol that allows us to distinguish the underlying structure of the major comparative subclasses: coordinate or dependent comparatives, with either clausal or directly phrasal standards. The present analysis allows us to dispense with the ad hoc rule of Comparative Ellipsis: its effects are derived from the common deletion operations that apply to either coordinate or subordinate structures. Moreover, Comparative Subdeletion is shown not to always be the result of wh-movement (Chomsky, 1977), nor an ad hoc deletion rule (Bresnan, 1975); but the result of multiple and symmetric quantifier raising.
Two oppositions in the grammar of property concept lexemes are: the positive-comparative opposition (e. g., wise-wiser) and the adjective-noun opposition (e. g., wise-wisdom). We ask: How does each of these oppositions' morphosyntactic relationship relate to its semantic relationship, and how (if at all) do these morphosyntax-semantic relations vary cross-linguistically? Reviewing and building on previous work on this question, we argue for three main points. First, comparative semantics is provided by a morpheme in syntax whereas positive semantics is achieved via covert contextual domain restriction of a quantifier. This quantifier is supplied by a possessive morpheme in the case of nominal property concept words and by the property concept word itself in the case of adjectives. Second, an investigation of the word-internal compositional semantics of derivationally complex property concept words like joy-ful and wisdom suggests that adjectives like wise and their possessed nominal counterparts like have wisdom are truth-conditionally and in some cases compositionally equivalent. Third, we show that many Mandarin phrasal possessed property predicates like you jiazhi 'have value' and you pinwei 'have taste' closely parallel English derivationally complex predicates like valu-able and taste-ful, respectively, and we argue that this reflects a broader (analytic vs. synthetic) typological distinction between Mandarin and English. In this way, we aim to provide a first step toward connecting cross-linguistic variation in the grammar of property concepts to a broader theory of parametric variation.
This paper focuses on the Degree Semantics Parameter (DSP), which predicts the clustering of a range of properties tied to the relevant parameter setting. Building on several studies on individual languages, it is shown that the clustering is not supported and the macro-parametric DSP should be dispensed with. Instead, crosslinguistic variation in degree constructions arguably derives from variation in languages' inventories of functional degree morphology, which introduces degree functions that measure the states introduced by property concept lexemes. This model dispenses with a semantic parameter over open-class lexical items, such as property concept lexemes, which can thus be assigned a universal semantics, and places crosslinguistic variation solely in the functional lexicon.
