S-Babble
Archived Schedule: 2021-2022
Fall 2021
9/28: Round Robin
All Babblers are invited to give short informal presentations of data, ideas, or questions related to syntax and/or semantics to kick things off for the year! We encourage you to use this opportunity for group feedback and discussion on any topics or puzzles that you've been musing over lately. Names of those who would like to present will be drawn randomly, and each presenter will have 5-10 minutes (depending on the number of presenters) for presentation and discussion. You are still welcome and encouraged to attend even if you do not have a topic you would like to present.
A hyperbole is an exaggerated statement that isn’t meant literally and is used for effect (typically, to convey affect or make an evaluation of either positive or negative valence). For example, I may utter ‘The water in the bath is boiling’, despite knowing that the water in the bath isn’t boiling, to communicate that the water in the bath is too hot—and that I’m frustrated as a result of this. Since Grice (1975), this phenomenon has been thought to be fully pragmatic in nature (cf. Sperber and Wilson 2008; Wilson 2011; Kao, Wu, Bergen and Goodman 2014). In this talk, I put forward a contrarian idea: hyperbole interpretation is partly a semantic phenomenon (come to the talk if you want to know more).
10/19: Nathaniel Imel (Washington) - The simplicity/informativeness trade-off in modal semantics
The meanings across the world's languages have been argued to support efficient communication. Evidence for this hypothesis has drawn on cross-linguistic analyses of vocabulary in semantic domains of both content and function words (Kemp and Regier, 2012; Steinert-Threlkeld, 2019). This project extends the simplicity/informativeness trade-off analysis to lexical modal semantics (e.g. can, ought, might). Two proposed universals in this domain from Nauze (2008) and Vander Klok (2013) are used for generating many artificial languages with varying degrees of quasi-naturalness as a proxy for real data. Preliminary results show Vander Klok's typology approximates the simplest languages, while the degree of satisfaction of Nauze's proposal is strongly correlated with informativeness.
11/9: Travis Major (USC) and Michael Diercks (Pomona) - Reanalyzing Bukusu "complementizer agreement" (joint work with Justine Sikuku (Moi University))
I take some initial steps toward a theory of real definition, drawing upon recent developments in higher-order logic. I demonstrate that the resulting account allows for extremely fine-grained distinctions (i.e., that it can distinguish between any relata that differ syntactically while avoiding the Russell-Myhill problem), has desirable logical features (in that it can consistently embrace three principles that initially appear incompatible - The Identification Hypothesis: If F is, by definition, G then F is the same as G, Irreflexivity: There are no reflexive definitions, and Leibniz's Law) and possesses the resources to resolve the paradox of analysis.
11/30: Yourdanis Sedarous (University of Michigan) - Investigating the extent of shared syntactic structures in the grammar of bilinguals
The present study investigates to what extent syntactic representations are shared between a bilingual’s languages, specifically when these structures overlap in varying degrees across those languages. In this talk, I focus on the bilingual knowledge and use of long-distance dependencies (e.g constituent questions) in Egyptian Arabic-English bilinguals. The results of a bilingual corpus analysis suggest that some structures of English trigger repetition of the same structure in the switch to Egyptian Arabic, while the results of an acceptability judgement task investigating code-switched LDDs suggest that speakers’ dominance of their two or more languages affects their sensitivity to illicit sentences in partially convergent structures. A third experiment is proposed to further test to what extent structures are shared.
12/7: JJ Lim (UCSD Linguistics) - Dependent accusative case in Khalkha Mongolian: Evidence from converbal adjuncts
In this talk, I argue for a dependent account of accusative case in Khalkha Mongolian. Mongolian embedded subjects may surface with accusative case marking, which is usually associated with direct/specific objects. There are two theories of how accusative case is assigned: (1) by entering into a relationship with some functional head e.g. transitive v; (2) by entering into a dependency with a higher nominal. Based on original fieldwork, I show that (2) covers more empirical ground than (1) in Mongolian. Crucial evidence comes from converbal adjuncts, whose syntactic properties and position make (1) unlikely to be true.
Winter 2022
1/4: Planning meeting
Please come ready to discuss plans for the upcoming quarter of S-Babble!
This paper investigates the adverbial use of confirm in Colloquial Singapore English (CSE, or Singlish) to mean ‘definite/surely/for sure’, as evinced by its interpretation and flexible position within the clause. I propose that the source of Singlish confirm is Mandarin kending ‘confirm’, given their semantic and syntactic parallelism. I further argue that Singlish confirm is best understood as a modal speaker-oriented adverb (SpOA) that has only a strongly subjective reading.
1/18: Patrick Muñoz (University of Chicago) - On the distinction between egophors and classical evidentials
Egophoricity encodes an agent’s ‘personal access’ to an eventuality, while evidentiality encodes an agent’s ‘source of information’ for the occurrence of an eventuality. As these two categories resemble one another truth-conditionally, and are formally encoded in similar ways, recent work has questioned to what extent they may be assimilable. I provide a first formal treatment of the distinction between egophors and classical evidentials, first by noting from a wide range of cross-linguistic data that they consistently diverge and converge with respect to a range of semantic features, and then proposing a simple formal distinction between them that accounts for their similarities and differences. Egophors and evidentials differ in four key respects: (i) evidentials can introduce novel time references, by ‘double-tensing,’ while egophors cannot; (ii) evidentials always encode information source distinctions, while egophors never do; (iii) egophors are incompatible with novel information, and cannot receive mirative readings, while evidentials are compatible with new information, and prefer mirativity in certain environments; and (iv) egohors often require an agent to be a participant in the reported eventuality, while evidentials never do. They in turn align in three key respects: (i) they exhibit ‘learning effects,’ such that their use is disallowed in circumstances that vitiate the learning of the agent (due to forgetting, delayed realization, etc.), (ii) they exhibit ‘origo shift,' where commitment attaches to agents other than the speaker, in the same environments, and (iii) they both introduce a P-not-at-issue commitment on the ‘epistemic authority’ of a speech act. I propose that egophors and evidentials are both learning operators, which signal not-at-issue how the epistemic authority has learned of the existence of the reported eventuality: they differ in that while evidentials are complex operators, introducing a secondary learning eventuality and relating it to the reported eventuality, egophors are simplex operators, introducing no new eventuality, but equating the learning eventuality with the reported one. I explain how this distinction derives the above observations, and compare this approach favorably to previous treatments of the two notions.
1/25: Grant Goodall (UCSD Linguistics) - The island/non-island distinction in long-distance extraction: Evidence from L2 acceptability (joint work with Boyoung Kim (KAIST))
Experimental studies regularly find that extraction out of an embedded clause results in a substantial degradation in acceptability but that the degradation is much greater when the embedded clause is an island structure. We explore these two facts by means of a series of acceptability experiments with L1 and L2 (L1 Korean) speakers of English, We find that the L2 speakers show greater degradation than L1 speakers for extraction out of non-islands, even though the two groups behave very similarly for extraction out of islands. Moreover, the L2 degradation with non-islands becomes smaller and more L1-like as exposure to the language increases. These initially surprising findings make sense if we assume that speakers must actively construct environments in which extraction out of embedded clauses is possible and that learning how to do this takes time. Evidence for this view comes from cross-linguistic variation in long-distance extraction, long-distance extraction in child English, and lexical restrictions on long-distance extraction. At a broader level, our results suggest that long-distance extraction does not come “for free” once speakers have acquired embedded clauses and extraction.
2/1: Scott AnderBois (Brown University) - Preverbs in Yucatec Maya (joint work with Grant Armstrong (University of Wisconsin) and Edber Dzidz Yam (University of New Mexico))
Cross-linguistically, adverbs are often thought of as an extremely flexible category both syntactically and semantically. In this talk, we argue that in addition to adverbs, Yucatec Maya has a distinct adverbial category we call "preverbs". Syntactically, preverbs can only occur in an immediately preverbal position, not available to other adverbs, discourse particles, etc. and distinct from superficially similar constructions in the language such as V-V compounds and incorporation. Semantically, we argue that they are restricted to a particular subset of adverbial meanings, "verb-related" ones in the sense of Maienborn & Schaefer 2011. Finally, we propose a division between two sub-categories of preverbs and propose a tentative syntactic analysis motivating the proposed semantic restrictions.
2/8: Round Robin
All Babblers are invited to give short informal presentations of data, ideas, or questions related to syntax and/or semantics. We encourage you to use this opportunity for group feedback and discussion on any topics or puzzles that you've been musing over lately. Names of those who would like to present will be drawn randomly, and each presenter will have 5-10 minutes (depending on the number of presenters) for presentation and discussion. You are still welcome and encouraged to attend even if you do not have a topic you would like to present.
2/15: Milad Mayel (UCSD Linguistics) - The Function of Accented Pronouns
The felicitous use of prosodic accent has been shown to have syntactic, semantic, and pragmatic implications. However, the motivations for accenting, especially in the case of pronouns, are not always clear and, consequently, current hypotheses have been unable to uniformly predict these accented referring expressions. Furthermore, empirical results, to date, have not been mutually corroborative. Through the use of written data, where such accenting is notated orthographically, this project seeks to provide a basis for an empirically adequate model of pronoun accentuation.
Spring 2022
In this talk, I investigate the syntax of pronominal clitics in Modern Greek and propose that pronominal cliticization involves the clitic left dislocation (CLLD) of a null pronoun. Following Iatridou (1991, 1995), CLLD is argued to involve predication, with the dislocated expression serving as the subject of the predication and the clitic functioning as a predicate variable, creating an open position in the predicate, making predication possible. A novel idea in this talk is that the subject of the predication can be null. Iatridou also observed that a locality condition governs predication such that the predicate variable must be quite local to the subject of predication--in fact, the clitic must be contained in the minimal maximal projection serving as predicate. I further argue, following Schneider-Zioga 1994, 1998, that clitics have pronominal properties and thus the clitic is also subject to an anti-locality requirement. Given the locality condition that governs the relation between clitics and the subjects of predication (recall the CLLD configuration), the anti-locality conditions on clitics, and the fact that the clitic is also co-indexed with the (CLLDed) subject of the predication, it is predicted that cliticization will only be possible if the minimal maximal projection (the predicate) containing the clitic constitutes a binding domain. Pronominal cliticization in Greek is found not just with the complement of verbs, but also of prepositions and adjectives. Objects of verbs can freely cliticize. However, the complements of prepositions and adjectives cannot--only certain prepositions and adjectives allow for cliticization of their arguments. This distribution can be captured once we establish whether or not a TP, prepositional or adjectival phrase constitutes a binding domain as demonstrated by the distribution of tonic pronouns and reflexives within that domain. Further, I note that a special relation exists between clitics and genitive case in Greek and demonstrate through the lens of the predication analysis that clitics participate in dependent case assignment in Greek.
4/12: Ebru Evcen (UCSD Linguistics) - Making the Question under Discussion explicit shifts counterfactual interpretation
The comprehension of counterfactual statements (‘If there had been zebras, then there would have been lions’) has been subject to much research, but two key questions remain: Can comprehenders interpret counterfactuals without relying on causal inferences? And can comprehenders reach the actual state interpretation relying only on grammatical cues, or is this interpretation triggered by communicative goals? We answer these questions by relying on non-causal counterfactuals, and by manipulating the Question under Discussion between experiments: In Exp. 1, we replicate Orenes et al. (2019), using a web-based eye-tracking paradigm. In Exp. 2, we make the QuD explicit by asking about the actual state of affairs. The results reveal that making a contextually relevant alternative explicit via the QuD shifts counterfactual interpretation, but in general, the suppositional state interpretation is preferred in non-causal counterfactuals. These results imply that the driving forces behind counterfactual processing are pragmatic, not syntactic.
4/26: Seoyeon Jang (UCSD Linguistics) - A semantic proposal for the clause-type marker in Korean echo questions
Echo questions are a type of interrogative clause that requires a previously uttered sentence as an antecedent and (partially) repeat ("echo") it to convey a question asking for a confirmation or repetition of what has been said. While most-studied Indo-European languages like English and German do not have an overt marker that distinguishes echo questions from other types of clauses like ordinary interrogative and declarative clauses, Korean characterizes echo questions using the marker tako and a high rising final intonation, since the language requires a clause-type marker (which indicates the type and the speech-level (register) of a clause) in every clause. In this talk, I discuss the morpho-syntactic properties of Korean echo questions, argue that existing semantic approaches to echo questions do not capture those properties, and propose an alternative semantic analysis of Korean echo questions and their clause-type marker.
Despite their resemblance to subject relative clauses, pseudo-relatives are commonly analyzed as a nominal subject and CP predicate of a small clause, with the ‘gap' inside the CP being a null PRO controlled by the subject. Previous work on pseudo-relatives has almost entirely been restricted to European languages, and has primarily concentrated on their usage as complements of perception verbs. Based on original fieldwork on (Eastern Canadian) Inuktitut, this talk argues for a pseudo-relative analysis of an understudied existential construction in Inuit, in which a clause (rather than a noun) is morphologically incorporated into an existential verb 'have'. I propose that ‘have' syntactically embeds a small clause, whose predicate is the incorporated control CP and whose subject is the nominal pivot of the existential. This analysis is motivated by strong structural parallels with pseudo-relatives in other languages, and is shaped by well-established morphosyntactic properties specific to Inuit. If correct, the finding that pseudo-relatives exist in Inuit is notable, as it shows that they are not solely a European phenomenon. Moreover, given Inuit’s polysynthetic nature, its sentence-level syntax is not straightforward to ascertain; however, the analysis developed here shows that Inuit clause structure is composed of cross-linguistically familiar pieces.
5/10: Himidan Hassen (Independent Scholar), Peter Jenks (UC Berkeley) and Sharon Rose (UCSD Linguistics) - A'-satisfaction with φ-interaction in Tira
Recent literature has described several cases of hybrid A/A′-movement (van Urk 2015, Erlewine 2018, Colley & Privoznov 2021), including the observation that there may be several subcases of such movement (Scott 2021). This paper presents novel evidence from Tira (Kordofanian:Sudan) for two different types of hybrid A/A′-movement characteristic of topic vs. focus clauses in the language. Topic clauses target an XP for movement with both φ- and [TOPIC]-features. Focus clauses, built on relative clauses, only target an [A′]-feature, although they nevertheless show φ-agreement, hence constituting a novel case of overagreement. This distinction is derived with the interaction-satisfaction model of Agree (Deal to appear): φ-features are part of the interaction condition of the focus head but not its satisfaction condition.
5/24: Vincent Rouillard (MIT) - Maximal informativity and the licensing of temporal in-adverbials
The focus of this presentation is on temporal in-adverbials (TIAs) of the form in d M, where the measure phrase d M is composed of a degree d and a measure function M, e.g. in three days. I call attention to an ambiguity between so-called VP-level TIAs and perfect-level TIAs. VP-level TIAs modify VP predicates, and restrict their extensions to eventualities of a duration specified by the TIA. A well-known restriction on the use of VP-level TIAs is that they cannot combine with atelic predicates like be sick, whereas they are happy to combine with telic predicates like write up a paper.
(1) The student wrote up a paper in three days.
(2) *The student was sick in three days.
Sentences that are in the perfect admit perfect-level interpretations of TIAs. While these can appear in sentences with atelic predicates, they behave like negative polarity items (NPIs). In (4), the TIA specifies the amount of time that has gone by since the last sickness eventuality ended.
(3) *The student has been sick in three days.
(4) The student hasn't been sick in three days.
I argue that a unified semantics for VP- and perfect-level TIAs reveals that the unavailability of the former with atelic predicates and the polarity sensitivity of the latter follow from a more general principle. I show that TIAs are always ungrammatical when it is impossible to define a maximally informative degree for their measure phrase.