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.
Do so has traditionally been seen as ungrammatical with stative antecedents, a requirement which has been traditionally derived from the presence of do.
*Josh believes in faeries, but Andy doesn't do so.
But...
Josh believes in faeries, and has done so for at least the last 30 years.
In this talk, I show that the traditional view is incorrect on several fronts. First, stative antecedents are grammatical; any unacceptability is pragmatic in nature. Second, do is not the source of unacceptability with stative antecedents. Rather, it can be derived from a clash between the discourse conditions of the do so construction and pragmatic knowledge about states.
Condition on reference: do so should refer to a highly salient and accessible referent
Condition on informativity: use of do so should be informative
These conditions often result in the co-occurence of do so with a non-contrastive adjunct phrase, or some other type of modification. This leads to anomaly with many states, which often resist many types of modification. Lastly, this clash with pragmatic knowledge is not specific to stative antecedents, but may arise with non-stative antecedents as well.
Unlike previous approaches, my analysis captures the full breadth of available data. First, it allows for stative antecedents to be acceptable to the extent that they can plausibly take modification. Second, it provides a means for explaining why non-statives are also sometimes unacceptable. Additionally, my account makes no reference to the stativity of the antecedent, casting doubt on both the need to equate do with non-stativity, as well as the need to have a grammatical category "state" in the first place.
What is the difference between a state (e.g., know) and an event (e.g., learn)? Intuitively, we might say it is about change, where events involve change but states do not. Linguistically, states (more properly stative predicates) are defined by their behavior. That is, there are a number of diagnostics that are used to categorically segregate stative from non-stative predicates (Vendler 1967; Dowty 1979; Bott 2010), and these diagnostics are definitional. They are seen as interacting with a grammatically-encoded, categorical distinction between stative and non-stative predicates.
I believe that this underlying belief, that there is a grammatically categorical distinction, is on the wrong track. "States" do not form a coherent group: one state will be more state-like than another, a third will be somewhere in between. The line between stativity and non-stativity is fuzzy, not discrete. Diagnostics for stativity are not sensitive to stativity, per se, but to more fine-grained characteristics of eventualities that cut across the traditional state/event divide, such as controllability. Notice the gradient behavior of persuade with both statives and non-statives:
Statives
??Feifei persuaded Josh to be happy
?Feifei persuaded Josh to be nice
Feifei persuaded Josh to be open-minded
Non-statives
??Feifei persuaded Josh to remember
?Feifei persuaded Josh to fall
Feifei persuaded Josh to sit
If persuade were sensitive to a grammatical distinction between stative and non-stative predicates, we would expect each predicate in (1) to be similarly bad, and each predicate in (2) to be similarly good, yet this is not the case. Instead, the gradient acceptability appears to vary depending on how plausibly under Josh's control the predicate is seen to be. Such variable and plausibility-based judgements are better handled by a pragmatic account than a grammatical one.
The space of phenomena involved or related to stativity and non-stativity is vast, too vast to be effectively tackled in one dissertation. My work therefore focuses one one corner of this landscape, namely English main verb do, a verb that has long been associated with issues of stativity/non-stativity. The goal of this talk is two-fold: 1) to motivate doubt about the nature of the traditional state/event distinction; 2) to motivate my choice of do as a point-of-entry into the larger issues at hand. To accomplish my first goal, I will discuss a number of pervasive problems with the diagnostics used to define stativity, thereby undermining the definition itself. From this discussion, it will emerge that English main verb do appears in a number of different environments that appear to be sensitive to stativity. I thus accomplish my second goal though noting that, in looking closely at do I must necessarily navigate a more diverse range of environments than would otherwise be the case. That is, more mileage can be gained through an investigation of do than through, e.g., an investigation of the imperative.
Colloquial Jakartan Indonesian behaves like a tenseless language, with no tense morphology on the verb stem; the context disambiguates between a present and a past tense interpretation. As in many tenseless languages (Bochnak, 2019), the future is obligatorily marked by an adverb or akan/bakal 'will'. We argue that mau 'want’ can be used to mark the future as well, as in the following:
Context: We are at a party, but it’s getting late. I need to leave.
Sebenarnya aku nggak mau, tapi aku mau pulang sekarang ya.
Actually I neg want but I fut go-home now ok
'I don't actually want to but I will go home now.'
We call this use of mau the future mau. Future mau can have a purely temporal use, is compatible with inanimate subjects, and with the negation of mau meaning 'want' (also spelled out as pengen). However, future mau cannot be directly negated. It is incompatible with clausemate negation, negative quantifiers, and with the implicit negation triggered by the alternatives of 'only’ (Rooth, 1985). Yet, it is compatible with negation and with a negative quantifier in a higher clause. It thus seems that we cannot negate future mau directly, but we can negate the proposition that contains it. We discuss the empirical picture, as well as argue that there is a dispositional requirement associated with future mau, namely a requirement that the subject is disposed to causing the eventuality of the verb. Finally, we propose that mau as a modal (Sneddon, 2010; Jeoung, 2020) is the dispositional will (Copley, 2002), having the meaning of bakal 'will' enriched with a presupposition about the dispositions of its subject. We also propose a solution to the negation puzzle, arguing that future mau behaves as a PPI, and following New (2023) in positing two adjunction sites for auxiliaries, one below and one above negP; being a PPI, future mau can only attach to the higher adjunction site. All in all, we show that mau in Indonesian can either quantify over buletic alternatives or simply over accessible worlds like 'will', with its 'desire' component being turned into a dispositional requirement, eliminating the need for an attitude holder. Indonesian shows that 'want' can synchronically mean 'will', a change which is diachronically attested in many languages (Heine, 2017).
In the first half of this meeting, I will do a practice poster presentation for the upcoming LSA Annual Meeting, titled "semantic evidence for the non-compositional nature of the echo question marker tako in Korean." The presentation examines the formal semantic contribution of seemingly compositional parts of tako, the Korean echo question marker, and then argues that it should be analyzed as a non-compositional unit to adequately account for any semantic analyses of echo questions. In the second half, I will briefly review two acceptability surveys I did over the summer, which were conducted to test whether pair-list and functional readings are available in Korean multiple-wh questions & echo questions. Previous studies on English and Spanish have argued that those readings are unavailable in echo questions (e.g., Beck & Reis 2018; Chernova 2017), but the results from my surveys seem to provide counterexamples to them. Any suggestions for interpreting the results are welcome!
Multiple center-embedding of clauses (as in “the rat the cat the dog worried chased ate the cheese”) is commonly assumed, on the basis of observational and experimental evidence, to present difficulties for processing (Chomsky & Miller 1963, Gibson & Thomas 1998) and therefore to be disfavored in actual language use. Furthermore, corpus research on multiple clausal center-embedding has found that it occurs very rarely in writing (Karlsson 2007) and is practically nonexistent in spontaneous speech (Karlsson 2020, Laury & Ono 2010). However, these corpus studies are generally limited to European languages with predominantly SVO order, rather than languages with stricter SOV order such as Japanese and Korean, in which center-embedding may be more prevalent. In addition, research on processing factors in center-embedding has focused on experiments featuring coined example sentences, rather than sentences occurring in real discourse. The current study addresses these assumptions about center-embedding using (a) cross-linguistic corpus data and (b) observations of naturalistic center-embedded sentences in English.
First I present results from a typologically broader survey of languages, and demonstrate that the prevalence of multiple clausal center-embedding is greater in verb-final languages. The results show a general correlation between basic word order and prevalence of center-embedding: multiple center embeddings, as defined in the search, are much more common in SOV languages (e.g. Japanese, Korean) than in SVO languages (e.g. English, French) or those with mixed order (e.g. German). While overall rates of center-embedding are still low even in the most embedding-rich language, Korean (~1 sentence in 1400), this is still much higher than the prevalence observed in languages with SVO order, where center-embedding was very rare (around 1 in 30,000 sentences) or nonexistent in the searched corpora. These results challenge the widespread assumption that multiple center-embedding is generally disfavored, and suggest that the prevalence of this construction is in fact contingent on language-specific syntactic factors rather than entirely on language-independent processing factors.
Second, I examine 190 observed English examples of multiple center-embedding, with regard to a facilitating semantic-syntactic factor observed in experiments: type of subject in each clause. NPs/DPs can be ranked according to how discourse-accessible they are. Definite descriptions introducing new referents are the most specific and least accessible (see Ariel 2014, Gundel et al. 1993); names more accessible; bare plurals, quantifiers/indefinite pronouns, and third-person pronouns more accessible than names; and indexical pronouns and zero subjects most accessible of all. Experimental evidence (Warren and Gibson 2002) suggests that center-embedded sentences are easier to process if the embedded clause subjects are higher accessibility, because subjects that introduce new specific referents are harder to integrate into the sentence while keeping multiple open dependencies in memory (Gibson 1998, 2000). The observed English data reveals the same factors in attested examples: there is a general trend for low-accessibility subjects (e.g. definite descriptions) in main clauses, and more accessible subjects such as pronouns or quantifiers in embedded clauses, producing an overall correlation between clause embedding depth and accessibility.
Overall, these results demonstrate two important findings. First, center-embedding is not uniformly disfavored across all languages, but varies with syntactic type. Second, even in a language with a low incidence of multiple center-embedding, namely English, genuine examples are found – and these examples generally conform to the factors shown by experimentation to facilitate the processing and comprehension of center-embedded sentences. Even sentences with three levels of center-embedding can occasionally be observed, and nicely exemplify these facilitating factors. Together, these findings reveal that the difficulty of center-embedding is shaped by both universal factors of working memory and language-particular features, and that center-embedding in itself is not inherently problematic.
This is a practice talk for the upcoming LSA Annual Meeting. In our talk, we analyze restrictions on the position of possessive clitics in two varieties of Mongolian, Khalkha and Alshaa. We document intricate patterns of possessive clitic placement that is sensitive to case, quantifiers, and postpositions. We derive these patterns from syntactic movement constrained by antilocality. We propose that variation stems from i) where quantifiers are Merged, ii) the size of nominal complements, and iii) specific properties of K, e.g. whether it probes for person features or whether it undergoes Lowering to D.
In many Mayan languages, combinations of subjects and objects are restricted by relative animacy hierarchy effects: objects must be at least as high as subjects in terms of animacy. Building empirically on a novel description of Chuj, as well as reported data for nine additional Mayan languages from across the family, we offer a new approach to these effects. Our analysis builds theoretically on recent work tracing person/animacy restrictions to the nature of featural representations and the operation Agree, bringing this literature together with current understandings of Mayan syntax and the high-/low-absolutive parameter. We argue that the cross-Mayan data—relative hierarchy effects holding in the same way across both high-absolutive and low-absolutive languages—are best handled by, and brings new support for, an interaction/satisfaction approach to Agree and hierarchy effects (Deal 2023). Our analysis also casts new light on key topics in Mayan syntax, including the proper analysis of ergativity and the nature of obviation effects (Aissen 1997).
This is a practice talk for the upcoming LSA Annual Meeting. Our study analyzes wh-reduplication patterns in Korean and Khalkha Mongolian inspired by Davis 2015, which proposes a reduplication morpheme RED to account for Yaeyaman wh-reduplication. We will go over how the wh-reduplication patterns in Korean, Khalkha Mongolian, and Yaeyaman vary and propose that the variation across those languages depends on (i) the attachment site of RED and (ii) whether RED agrees with the highest or all wh-items by modifying and extending Davis' original account for Yaeyaman.
All Babblers are invited to give short informal presentations of data, ideas, or questions related to syntax and/or semantics so prospective linguistics grad students can taste a bit about who we are & what we do!
Conditional statements often convey implied meanings beyond their literal content: The conditional ‘If you mow the lawn, you’ll receive $5’ is literally true even when the lawn is not mowed and the person receives $5 anyway (e.g., for a different chore). However, listeners often judge this sentence as false in those situations, treating it exhaustively with an ‘if and only if’ meaning, known as Conditional Perfection (CP). In the present study, we investigate how people arrive at pragmatic interpretation of conditional statements. In particular, we ask whether, when people are reasoning about conditional statements, they begin by considering a literal meaning that they subsequently ‘perfect’ through pragmatic reasoning, or if instead they begin with a pragmatically enriched meaning, and only consider a literal meaning when required by context. We first present reaction time data that strongly support the latter sequence, as evidenced by the longer time required to arrive at literal interpretations compared to pragmatic ones. Further, our findings indicate that transitioning from pragmatic to literal meanings demands additional cognitive effort, suggesting a default tendency toward pragmatic interpretations.
--> S-Babble lunch after the meeting at 12:30
The generalization that A'-movement cannot form binding relations has been argued for some time now (see Mahajan 1990, Müller & Sternefeld 1994, Safir 2019). However, challenges to this generalization have simultaneously been brought up by apparent exceptions which appear to demonstrate that A'-movement can in fact bind (see Barss 1986, Sportiche 2006, Chomsky 2023). The goal of this investigation is to assess the legitimacy of constructions which seem to demonstrate A'-movement forming binding relations. I will argue that these constructions do not obey binding conditions through traditional avenues, but instead, uniformly behave logophorically due to the presence of a null logophoric binder (following Charnavel 2020). As a result, the generalization may remain intact, and I will conclude this investigation with a comparison of two different approaches, Safir (2019)'s insulation account, and Chomsky (2023)'s box theory which attempt to deal with the generalization. The analysis will seek to demonstrate that the box theory not only can provide more straightforward solutions, but also account for a wider set of data than the insulation account.
REFERENCES
Barss, A. (1986). Chains and anaphoric dependence: On reconstruction and its implications. Ph.D. diss, MIT.
Chomsky, N. (2023). The Miracle Creed and SMT. http://www.icl.keio.ac.jp/news/2023/Miracle%20Creed-SMT%20FINAL%20%2831%29%201-23.pdf
Mahajan, A. K. (1990). The A/A-bar distinction and movement theory. Ph.D. diss, MIT.
Müller, G. & Sternefeld, W. (1994). Scrambling as Α-bar movement. In N. Corver & H. Riemsdijk (Ed.), Studies on Scrambling: Movement and Non-Movement Approaches to Free Word-Order Phenomena (pp. 331-386). Berlin, Boston: De Gruyter Mouton. https://doi.org/10.1515/9783110857214.331
Safir, K. (2019). The A/Ā distinction as an epiphenomenon. Linguistic Inquiry, 50(2), 285-336.
Sentences with syntactic movement out of sentential complements of manner-of-speaking (MoS) verbs (e.g., whisper, shout) are degraded in acceptability, an effect called the “manner-of-speaking (MoS) island effect.” Accounts variably attribute the MoS island effect to the violation of the subjacency condition, to the low frequency of MoS verbs taking sentential complements, or to a general information structural constraint that discourse-backgrounded constituents cannot be extracted. In three acceptability judgment experiments, we find that the MoS island effect can be modulated by foregrounding or backgrounding the extracted constituent, suggesting a causal relationship between discourse backgroundedness and the MoS island effect. Our findings challenge syntactic and frequency accounts of the MoS island effect. Joint work with Jiayi Lu and Judith Degen.
In the early 60s, Richard Montague (1930-1971) still believed that: “[The] systematic exploration of the English language, indeed of what might be called the ‘logic of ordinary English’, […] would be either extremely laborious or impossible. In any case, the authors of the present book would not find it rewarding.” Just a few years later, he radically changed his mind: “There is philosophic interest in attempting to analyze ordinary English” and “I reject the contention that an important theoretical difference exists between formal and natural languages.” At the very same time¾not by chance¾ Montague changed his beliefs in what should be the adequate framework for philosophy: from “set theory with individuals and the possible addition of empirical predicates” to higher-order “intensional logic.” These changes coincide with the beginning of Montague’s seminal work on natural language semantics that culminated with the last three papers he published before his sudden and violent death. In this talk, I examine these changes with the goal of understanding what motivated them and what light these motivations may shed on Montague’s subsequent work on natural language. This investigation is presented as part of a broader ongoing project of an intellectual and personal biography of Montague.
Why do some quantifiers presuppose their domain to be non-empty, and others not necessarily so? We propose an answer based on the interplay of dynamicity and a ban on idleness. Joint work with Yasu Sudo.
Wh-reduplication refers to the doubling of wh-items to indicate plurality in questions. Following Davis 2015, we assume that wh-reduplication is due to the attachment of a RED morpheme, which we treat as a modifying plural (Wiltschko 2008), to a wh-item. We discuss two profiles of wh-reduplication in Korean and Turkish that differ in: i) whether case markers and plural markers are included in the reduplicant, ii) the availability of associative plural readings, and iii) the productivity of wh-reduplication. To account for these patterns, we argue that RED attaches to the wh-root in Korean, while RED attaches to the DP in Turkish. Our findings suggest that both modifying plurals and head plurals may be found in a single language, and provide further support for Wiltschko's (2008) prediction that modifying plurals may attach to different sites within the nominal phrase cross-linguistically.
In global case splits, the case marking on one nominal depends on the relative features of multiple nominals in the clause. For example, in Shawi (Kawapanan; Peru) ergative case is marked on the subject only when the subject is at least as high as the object on the person hierarchy. Such global split patterns might suggest that a global evaluation of all of the nominals in the clause simultaneously is necessary to determine case marking. This might in turn be taken as support for configurational models of case assignment that take into account multiple nominals in the structure to determine dependent case marking. However, in this talk, I argue that not only is an Agree-based model of case assignment better suited to model the typology of hierarchy patterns found in global case splits but also an Agree-based model allows us to preserve a highly local computation of case assignment in global split patterns.
[practice for dissertation defense presentation]
This dissertation is the first-ever systematic investigation of the semantics and pragmatics of Echo Questions (EQs) in Korean. EQs are an understudied type of interrogative clause that conveys a question seeking confirmation or repetition of what has been previously said. EQs in Korean provide new insights into the typology and theories of interrogative clauses, as previous analyses on EQs in the most-studied (i.e., Indo-European) languages fail to accurately capture the characterizing properties of EQs in Korean. I propose a novel analysis that treats EQs as semantically identical to and pragmatically distinct from ordinary questions. I show that, while both EQs and ordinary questions denote a set of possible answers, EQs bear the presupposition that there exists at least one possible answer that has already been introduced in the discourse by uttering a sentence conveying it. Ultimately, this dissertation highlights the importance of cross-linguistic examination by enriching the formal theory of questions through previously unnoticed data.
The aim of the talk is to 1) introduce aspects of Diné Bizaad word and sentence structure to the group, 2) present data on Diné Bizaad concerning the optionality of wh-fronting in the language, and 3) present two ideas, from Denham (2000) and Cable (2007), on perspectives concerning wh-word activity in two other languages within the Na-Dene language family. I welcome feedback and ideas that will help steer the research in the right direction for exploring optional wh-movement in a language with overt question elements.
In Indonesian/Malay, there is an optional verbal prefix meN- which is widely taken to be the subject voice marker since it correlates with subject voice SVO word order, and when it is present, only the subject may be A’-extracted. In object voice OSV word order, meN- is obligatorily absent, and only the object may be A’-extracted. The well-known "meN-deletion" generalisation is, therefore, that movement of the object over the verb in Indonesian/Malay results in deletion of meN (Saddy 1991, Fortin 2006, Aldridge 2008, Cole et al. 2008, Sato 2012, Georgi 2014 among others). From the perspective of meN-deletion, the optional absence of meN- in subject voice is derived from a separate mechanism from the obligatory absence of meN in object voice. Most previous analyses remain silent on the optionality of meN in subject voice. In this talk, I argue against a view where “meN-deletion” is triggered by movement of a DP across the verb. Drawing from data in Jakarta Indonesian, I propose that the choice between flavours of functional Voice/v head is one-to-one with the overt presence/absence of meN- prefix on the verb. In doing so, I argue that word order in Indonesian/Malay is but an epiphenomenal correlate of voice in the language.