1235 results
Search results
Publication Postverbal gwai in Cantonese: A Syntactic Approach to Rhetorical Questions(University of Pennsylvania, 2024-04-20) Choi, Tsun HeiRhetorical questions are traditionally considered a pragmatic and semantic issue. In this paper, I discuss the rhetorical expression gwai in Cantonese. By arguing that gwai, though base generated as a postverbal suffix, Agrees with the Speaker Phrase, Addressee Phrase, and Commitment Phrase located at the expressive component above CP, I suggest rhetorical questions may be analysed syntactically. The proposal may even be extended to other rhetorical expressions in Chinese. This study provides a new perspective not only to rhetorical questions and expressive components in the minimalist framework, but also to the study of SFPs and intonation in Cantonese.Publication Prosody Reveals Syntactic Structure: Secondary Predication in Finite Metrical Corpora(University of Pennsylvania, 2024-04-20) Caso, AnabelleThe mapping of syntax to prosody is regulated by correspondence requirements that hold between abstract syntactic structure and prosodic structure (Selkirk 2011, Elfner 2012, Ito and Master 2013, among others). Cross-linguistically, secondary predicates can be marked with special prosody indicative of their complex syntactic-semantic structure (Winkler 1997, Guzzo and Goad 2017). Serbian and Modern Irish provide two unique examples of this, demonstrating differential sentential stress and initial mutation application between secondary predicate and attributive adjectival constructions, respectively. In ancient languages with extant metrical corpora, prosodically marked structures are encoded via positions of isolation in a line (Hale and Kissock 2021). The goal of this paper is to demonstrate consistency in the distribution of secondary predicates in the RigVeda and the Homeric poems—secondary predicates are prosodically isolated structures, and are therefore prosodically marked. I show that the tendency to prosodically isolate secondary predicates in metrical texts is indicative of their treatment in non-metrical prose. I conclude that secondary predicates in the Vedic and Ancient Greek languages were prosodically marked structures with relative syntactic-semantic complexity, placing these languages into the cross-linguistic typology of secondary predication.Publication Against Evacuation Movement in NP-Ellipsis(University of Pennsylvania, 2024-04-20) Benz, Johanna; Mendes, Gesoel; Salzmann, MartinIn this paper, we provide evidence from NP-ellipsis and its remnants against the popular mechanism of evacuation movement, arguing that the mechanism wrongly predicts freezing effects for subextraction from PP-remnants in English, as well as being incompatible with genitive remnants of NP-ellipsis in German, which cannot undergo the movement required to escape ellipsis. We argue for an alternative analysis following a separate research tradition, according to which constituents can survive ellipsis without evacuation movement when they are focused. Modeling ellipsis as an instruction to forgo vocabulary and assuming that this procedure is blocked in focus environments reconciles the availability of certain types of ellipsis remnants with their inability to undergo evacuation movement.Publication Every Provides an Implicit Comparison Class When Each Does Not(University of Pennsylvania, 2024-04-20) Knowlton, Tyler; Schwarz, FlorianAlthough the English universal quantifiers "each" and "every" are highly similar, it's often noted that "each" is 'more individualistic' than "every" in some way. In a recent proposal, outlined in Knowlton (2021), this difference is explained in terms of distinct lexical meanings. The idea is essentially that the meaning of an expression like "every frog" has a semantic constituent that corresponds to the plurality "the frogs", whereas the meaning of "each frog" introduces no such plural element. The bulk of the evidence in support of this view comes from sentence verification tasks (e.g., participants recall group properties, like average frog color and the number of frogs, after evaluating sentences like "every frog is green" better than after evaluating corresponding sentences with "each"). To broaden the empirical landscape, we test a novel prediction of this semantic proposal: since the mental representation that serves as the meaning of "every" calls for grouping the things quantified over, that group representation should be available to serve as the comparison class for predicates like "is the same color" and as a plural antecedent for predicates like "was told that they should jointly select a winner". We show that this prediction is borne out in a simple forced choice judgment study: "every NP" is preferred over "each NP" in both cases, at least relative to baseline preferences for using either quantifier.Publication The Effects of Gender and Autism Spectrum Disorder on Prosody(University of Pennsylvania, 2024-04-20) Knutsen, Sten; Kuziemski, Alexandra; Kenny, Megan; Peppé, Susan; Stromswold, KarinTo date, surprisingly little research has investigated how people with autism spectrum disorder (ASD) use prosody, and no research has investigated whether there are gender differences in ASD prosody. To address these gaps in the literature, we studied the functional use of prosody in 116 native English-speaking college students, 25 of whom had ASD and 91 of whom did not. Participants were tested on their ability to comprehend and produce affective prosody, question/declarative intonation, phrase boundaries, lexical stress, phrase stress, and contrastive stress. Bayesian ANOVAs revealed that women generally outperformed men, particularly in prosody production. Participants who did and did not have ASD generally performed similarly on most subtests, a finding that likely reflects that the participants with ASD were college students with no history of written or spoken language impairments. The one subtest where participants with ASD performed substantially worse than those without ASD was the contrastive stress production subtest, which is the only production subtest that specifically taps discourse pragmatics. On this test, women who did and did not have ASD performed similarly, whereas men with ASD performed substantially worse than men who did not have ASD. Previous studies have shown that people with ASD sometimes use camouflaging techniques to hide their ASD symptoms in public, and that women with ASD camouflage their behavior more than men with ASD. We believe that, taken as a whole, our findings provide the first evidence of gender differences in prosodic camouflaging in ASD.Publication New and Changing Social Evaluations of All-lowercase and Exclamation Points(University of Pennsylvania, 2024-04-20) Abirou, Lena; Kerrigan, Aly; Michell, Jay; Wade, LaceyThis study examines the social evaluations of two variables---all lowercase (compared to standard capitalization) and exclamation point usage (!, !! or none)---in Computer-Mediated Communication (CMC). We present the results of a matched guise experiment investigating social perceptions of various combinations of these two variables when used in SMS text messages. Results reveal a range of social meanings associated with both features. All lowercase is associated with a younger, trendier, more online texter, as predicted based on previous literature, and is also evaluated as more feminine and queer, novel social meanings that are further supported by participants' open-ended judgments of all lower-case. The open-ended responses reveal more nuanced associations of lowercase, specifically indexing a stance of trying hard to appear like you're not trying hard or carefully-calibrated casualness. Exclamation points reveal an incremental evaluative pattern, such that 1 exclamation point is judged as more extroverted, image-oriented, laidback, queer, and feminine than no exclamation points, and evaluations are strengthened for 2 exclamation points. Additionally, 2 exclamation points are seen as particularly trendy and tech-savvy, unlike a single exclamation point. We find limited evidence for significant interactions between all-lowercase, exclamation point usage, and perceived texter gender, but we do observe a general trend for ratings to be stronger, especially for lowercase, among younger participants. We interpret these results as indicating emerging and changing social meanings of all lowercase and exclamation point usage.Publication Phorhépecha Clitics: Person Splits and Omnivorous Number(University of Pennsylvania, 2024-04-20) Aitha, Akshay; Kurtz, NaomiWe investigate the complexities of clitic agreement patterns in Phorhépecha (language isolate, Michoacán). We show that clitics in the language display a preference for agreeing with participant objects in monotransitive sentences: whenever the object is a participant, regardless of the person features of the subject, clitics invariably agree with the object and not the subject. Further, we demonstrate that clitics exhibit omnivorous number agreement (Nevins 2011). A single morphological marker of plurality on clitics surfaces when two conditions hold: (1) whenever a subject, object, or both are plural, and (2) if the object is a participant. If the object is third person, clitics do not display omnivorous number agreement, and instead agree solely with the subject. We argue that these complex and interrelated patterns of agreement are due to relatively simple syntactic mechanisms. Firstly, we submit that clitics themselves are functional heads (following Sportiche 1996) that probe for person and number. Secondly, we propose that participant objects undergo object shift to a position above subjects and closer to the person probe on clitics, which, we maintain, captures the preference of clitics to agree with participant objects. Thirdly, we attribute omnivorous number agreement to a relativized number probe on clitics, which searches for any plural argument (subjects and objects) within the probe's c-command domain. Finally, we adduce independent evidence for our movement-based analysis from verbal agreement. In Phorhépecha, verbs agree with plural objects only if those objects are third person. We claim that verbal object agreement arises with third-person objects, and not participant objects, because only the former do not undergo object shift and instead remain in their base positions, from which they are accessible to a number probe within the extended projection of the verb. In contrast, participant objects shift to a higher position and are consequently inaccessible to said number probe.Publication Adults Behaving Childishly: Errors in Adult Responses to Wh-Questions(University of Pennsylvania, 2024-04-20) Lutken, Jane; Stromswold, KarinEnglish speaking kindergarteners sometimes incorrectly answer questions with medial wh-relativizers by apparently responding to the relativizer. For example, they might answer How did Lewis tell Sally what he picked? by saying what was picked (e.g., apples). In other words, they interpret it to mean: What did Lewis tell Sally (that) he picked? These errors are reminiscent of the wh-scope marking (WSM) construction found in languages like German, where the true wh-phrase appears medially while the scope of the wh-phrase is marked by an initial, contentless wh-phrase (Lutz, Muller, & von Stechow, 2000). deVilliers and Roeper (1995) argue these errors are due to children having an immature WSM-like grammar, whereas Lutken, Legendre, and Omaki (2020) argue they are due to immature processing. If children’s errors are due to an immature grammar, adults should not make WSM errors because their grammars are fully developed. If children’s WSM errors stem from their language processors being overtaxed, adults might also make WSM errors if they are similarly overtaxed. We present the results of two experiments in which we taxed adults’ processing mechanisms and examined their responses to questions with medial wh¬-relativizers. Adults read short, yet complex stories and answered questions following them. We also included a working memory task which adults did simultaneously in Experiment 1. We found that adults did make WSM errors in both experiments and they did so at a rate similar to children. Furthermore, adults were more likely to make errors – and WSM errors, specifically – when they simultaneously did the WM task. We suggest that our findings are consistent with an immature processing account of these errors.Publication Does Telugu have Wh-movement? Surprising Findings from Native Speakers(University of Pennsylvania, 2024-04-20) Lutken, Jane; Dharmpuri, Anhiti; Stromswold, KarinTelugu is a Dravidian language with subject-object-verb (SOV) word order. It is classically described as a wh-in-situ language (Bruening 2007, Subba Rao 2012), and consistent with this, object questions always have SOV word order. However, interviews with native Telugu-speaking adults suggest that Telugu may allow subject questions to have either SOV or OSV word order. Because object topicalization is allowed in Telugu declaratives (Davis, 2005), OSV subject questions could be the result of object-fronting in topicalized contexts. Alternatively, if Telugu has partial wh-movement (Jayaseelan 2001, 2008, Balusu, 2016), this would license OSV subject questions. We conducted a yoked judgment/production experiment in which 17 Telugu-speaking adults read brief stories that either did or did not support object topicalization. After each story, participants chose whether they thought the SOV or OSV version of a sentence fit the story context best, and then they said their choice aloud. Bayesian analyses of the judgment results revealed that, although Telugu-speakers chose both SOV and OSV options for subject questions, they preferred OSV subject questions after both topicalizing and non-topicalizing stories. In contrast, they preferred SOV declaratives, particularly after non-topicalizing stories. Acoustic analyses suggest that, whereas OSV declaratives involved object topicalization, OSV subject questions did not. Taken as a whole, these results indicate that Telugu OSV subject questions are not the result of object topicalization, but may be the result of partial wh-movement. While further research is needed to determine why SOV and OSV subject questions coexist in Telugu, this study demonstrates that OSV subject questions are an acceptable option – and perhaps even the preferred option – in Telugu.Publication Two Classes of ‘anyone’: A Contrastive Analysis of amwu-to and nwukwu-to in Korean(University of Pennsylvania, 2024-04-20) Sim, Rok; Dubinsky, StanleyThis paper investigates two Korean expressions, amwu-to and nwukwu-to, which are seen to be roughly equivalent in meaning to the English word anyone (although nwukwu-to has several other senses). The first of these two was characterized in Sells 2001 as a Negative Sensitive Item (NSI), where it is distinct from a Negative Polarity Item (NPI) in that an NSI must occur within the scope of negation – that is, it cannot appear in questions or as the antecedent of a conditional, as NPIs can. NSIs themselves have a wider distribution than Negative Concord Items (NCIs) which require clause-mate negation. Some (such as Park et al. 2021) have analyzed both amwu-to and nwukwu-to as NSIs and others (such as Tieu & Kang 2014) have claimed that they both belong to the even more restricted class of NCIs. In this paper, we will show, contrary to prior claims, that only amwu-to has the narrow distribution required of an NSI or NCI. We will further show that nwukwu-to has a wider distribution – not being an NCI, an NSI, or even an NPI. In closely examining their distribution and interpretations, we will demonstrate that these two items are far more distinct from each other than has previously been assumed. Their contrastive properties, in addition to the aforementioned negation restrictions, include differences in (i) syntactic category, (ii) position within the clause, and (iii) dependency with regard to the postpositional particle -to. They will also be shown to differ in their interpretations in certain contexts. We claim that these differences are attributable to amwu (as an NSI/NCI) needing to be interpreted in the scope of NEG to check a Neg Feature, and thereby being characterized as an “absolute non-existential quantifier” (in contrast with Giannakidou’s 2011 “dependent existential quantifier”). For its part, nwukwu can be interpreted either inside or outside of the scope of NEG and is characterized as an “existentially quantified head noun”.