GLiF Seminars

______________________________________________________________________ ______________________________________________________________________

Information Information

Time: Thursdays, 12h to 13.30h
(unless indicated otherwise)

Room: 52.737 (UPF-Poblenou)
(unless indicated otherwise)

Location: Universitat Pompeu
Fabra, Carrer Roc Boronat, 138

 

 

If you want to be kept informed
about GLiF activities, please sign
up for the GLiF google group, or
send an e-mail to:

______________________________________________________________________ ______________________________________________________________________

  • Thursday, May 26, 2022:  Paola Fritz-Huechante (Humboldt-Universität zu Berlin): Spanish change of state verbs: Focus on psychological verbs and their scalar structure

Abstract: The scalar analysis of transitive causative change of state (TCoS) verbs has classified them according to their scalar readings into: (i) upper-bounded verbs (e.g. Spanish secar ‘dry’), which denote a culminating change of state, and (ii) lower-bounded verbs (e.g. Spanish arrugar ‘wrinkle’), denoting a change from a zero to a non-zero value in relation to the property described by the semantic core of the verb (cf. Hay et al. 1999; Kennedy & McNally 2005; Kennedy & Levin 2008). This talk shows that Spanish transitive experiencer object (TEO) verbs, in their eventive reading, are also able to denote causative eventualities yielding scalar inferences. We proposed two types of TEO verbs (parallel to the TCoS classification): (i) lower-bounded TEO verbs, setting a minimum degree in their scale structure that the experiencer needs to fulfill (e.g. deleitar ‘delight’, alegrar ‘make happy’), and (ii) upper-bounded TEO verbs, setting a maximum degree which the experiencer needs to reach (e.g. serenar ‘calm down’, despreocupar ‘make free of worries’). In a forced-choice selection test, the availability of lower- and upper-bounded scale readings was tested for the two classes of verbs (TEO, TCoS). Results showed that the factor bound was significant to the extent that both lower-bounded TEO and TCoS predicates yielded a response compatible with a lower-bounded reading significantly more often than predicates that possess an upper bound, supporting the aforementioned distinction. Results also showed a significant effect of verb type, differentiating a change of state on the mental (TEO) vs. the physical (TCoS) levels.

At 12:00 in room 52.737 (Campus del Poblenou)

______________________________________________________________________ ______________________________________________________________________

  • Thursday, April 21, 2022:  Paula Rubio-Fernandez (University of Oslo): The psychophysics of referential communication: Discriminability vs informativity

Abstract: In this talk, I will present a pragmatic account of referential communication that stands in contrast with traditional accounts in the Gricean tradition by focusing on perceptual discriminability rather than informativity. According to this new account, referential over-specification can result in efficient speaker-hearer coordination, rather than being pragmatically infelicitous. The results of a series of language-production experiments support the view that speakers are efficient when producing redundant size and colour adjectives, doing so more often when modification could facilitate the hearer's search for the referent in the visual display. English speakers also used redundant colour adjectives more often than Spanish speakers, supporting the hypothesis that redundant adjectives are more efficient in prenominal than in postnominal position. Regarding comprehension, a series of cross-linguistic eye tracking studies confirm that redundant modification can facilitate target identification. Finally, I will present a probabilistic model of referential communication and contrastive inferences. Overall, the results of these studies confirm that redundancy helps efficient referential communication.

At 12:00 in room 52.737 (Campus del Poblenou)

______________________________________________________________________ ______________________________________________________________________

  • Thursday, April 7, 2022:  Salvatore Pistoia-Reda (Università degli Studi di Siena): Language, logicality, and pragmatics

Abstract: In this talk, I will focus on the logicality of language, i.e. the idea that the language system contains a deductive device to exclude trivialities. Puzzling evidence for logicality comes from acceptable contradictions and tautologies. The standard response in the literature involves assuming that the language system is interfaced with a natural logic which cannot access acceptable trivialities. In this talk, I will discuss evidence supporting an alternative response, which rejects the stipulation of a natural logic and assumes instead the pragmatic modulation of nonlogical terms.

At 12:00 in room 52.737 (Campus del Poblenou)

______________________________________________________________________ ______________________________________________________________________

  • Thursday, March 10, 2022:  Masha Esipova (University of Oslo): Performative expression of meaning and the architecture of grammar

Abstract: Performative expression of meaning hinges on the act of producing a certain form itself. For instance, when we want to let out our anger or frustration by swearing, the act of swearing itself is crucial for achieving our expressive goals. Performative expression of meaning is an important part of our linguistic and near-linguistic behavior and can involve meaning-form mappings of varied conventionalization and arbitrariness status as well as forms produced by a variety of articulators. For example, we can performatively express our immediate emotions by producing fully regularized and arbitrary forms like 'damn' or 'ouch', less regularized and less arbitrary vocalizations like grunts or huffs, different types of gesture (facial, manual, full body, etc.) or prosodic modulations, etc. We can also performatively produce various types of forms to build rapport with someone or, conversely, to insult them, to express aspects of our identity or our allegiance to a group or a cause, to perform a ritual of some sort, etc. In this talk, I will discuss the following issues pertaining to performative expression of meaning: (i) architectural distinctions between performatively expressed meaning and truth-conditional meaning (as glimpsed through recoverability in ellipsis/anaphora and compositionality); (ii) how these distinctions should inform our formal modeling of performatively expressed meaning; (iii) the mechanism of going from performative expression of meaning to demonstrations of such expression within truth-conditional content.

At 12:00 online. Please contact sara[dot]amido[at]upf[dot]edu for the zoom link.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, March 3, 2022:  Sara Amido (Universitat Pompeu Fabra) and Sebastian Buchczyk (Universitat Pompeu Fabra): VERUM in inversion exclamatives: evidence from European Portuguese

Abstract: Zanuttini and Portner (2003), and Rett (2012) say that exclamatives are not suitable as answers to questions since they only denote a degree property and thus do not denote and assert propositional content (A: How tall is Tony’s child? B: *How very tall he is!). However, data from European Portuguese (EP) suggests that a particular type of exclamatives seem to be able to do so. We show that they fit Rett’s (2011) classification of inversion exclamatives, and we dub them se-exclamatives. When a se-exclamative is uttered after a polar question whether p (p or ~p?), the question is resolved with p. The phenomenon extends to British English inversion exclamatives (A: Can Adam cook steak? B: (Boy,) CAN he (cook steak)!). Even though Rett defends that exclamatives do not assert content, she speculates that they have a “verum interpretation” or an “adverbial interpretation” (2012: 420 f.). We show that in contexts in which whether p? is placed on the Table (Farkas & Bruce 2009), inversion exclamatives as follow-up moves to such questions, are indeed a verum strategy as per Gutzmann & Castroviejo Miró (2011). A verum operator within these exclamatives requires a proposition p, which we propose must either be presupposed by the exclamative or silently asserted before it, as evidenced by data from EP.

At 12:00 in room 52.737 (Campus del Poblenou)

______________________________________________________________________ ______________________________________________________________________

  • Thursday, February 24, 2022:  Dan Dediu (ICREA/Universitat de Barcelona): Languages change all the time, but not all things change equally fast or in the same ways -- (in)stability meets quantitative methods

Abstract: That languages change continuously is yesterday's news for linguists (and most people, really) and, with the exception of some purists, it is seen as a natural consequence of what language is. However, it is also old news that not all languages seem to change equally fast, and that even various components/aspects/levels of language seem to change at different speeds and in different manners. Likewise, the detailed causes of these changes are extremely varied, ranging from language-internal processes, to socio-political and (arguably) environmental and (possibly) biological. Here I will try to make two cases. First, that we need to approach these questions in an empirical, data-driven and quantitative manner, not being shy to borrow ideas and methods from other fields (in particular, evolutionary biology and psychology) and to be bold enough to adapt them to fit the specificities of language. Second, we must move away from a language-only view of language, and instead see it as a complex system that is continuously evolving in a wider environment composed not only of the climate, ecology, physics and chemistry in which the language users live and communicate, but also by their wider socio-cultural system and even their own biology.

At 12:00 in room 55.309 (Campus del Poblenou)

______________________________________________________________________ ______________________________________________________________________

  • Thursday, February 17, 2022:  Martina Wiltschko (ICREA/Universitat Pompeu Fabra) and Elizabeth Ritter (University of Calgary): Beyond pronouns

Abstract: Greenberg’s universal 42 states that all languages have pronominal categories involving at least three persons and two numbers. However, this characterization fails to capture the properties of so-called pronouns in languages that the World Atlas of Linguistic Structures characterizes as “avoiding pronouns for reasons of politeness”, such as Japanese. We propose that these languages have a distinct category of nominals, which we call paranouns. Paranouns are unlike pronouns because they are not bundles of phi-features, and they are unlike nouns because their content is interactional in nature; that is to say, it consists of socio-linguistic information about the interactants and the relations between them. We show that the lack of phi features and interactional content follow from the proposal that paranouns associate with the interactional structure, a set of categories dominating the functional structure (i.e, the DP layer). We then extend this proposal to develop a novel analysis of formality distinctions in languages with pronouns. More specifically, we argue that the content of formal pronouns is first associated with functional structure, but then recycled in the interactional structure.  This derivation ensures that the pronoun receives a marked interpretation with socio-linguistic consequences, while retaining its original phi feature specification.

At 12:00 in room 52.737 (Campus del Poblenou)

______________________________________________________________________ ______________________________________________________________________

  • Tuesday, February 1, 2022: Chenjie Yuan's PhD defense: Reasoning in Dialogue. Exploring the Inferential Articulation of Speech Acts. Directors: Laia Mayol Toll (UPF) and Enric Vallduví Botet (UPF). 

Committee: Stergios Chatzikyriakidis (Crete University), Louise McNally Seifert (UPF), Julie Hunter (Linagora Labs). 

At 11:00 online.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, January 13, 2022: Alexandra Navarrete González' PhD defense: Focus and contrast in Catalan Sign Language (LSC): Form and interpretation. Directors: Gemma Barberà Altimira (UPF) and Josep Quer Villanueva (UPF). 

Committee: Carlo Cecchetto (CNRS- Université Paris 8), Laia Mayol (UPF), Vadim Kimmelman (Bergen University). 

At 10:00 online.

______________________________________________________________________ ______________________________________________________________________

  • Friday, January 14, 2022: Raquel Veiga Busto's PhD defense: Person and Number in Catalan Sign Language Pronouns (LSC). Director: Josep Quer Villanueva (UPF). 

Committee: Markus Steinbach (Gottingen University), Gemma Barberà Altimira (UPF), Patricia Cabredo-Hofherr (CNRS- France 8). 

At 11:00 online.

______________________________________________________________________ ______________________________________________________________________

  • Monday, December 13, 2021: Gemma Barberà Altimira's tenure talk: Anaphoric reference in Catalan Sign Language (LSC) discourse. Analysing role shift features and signing space from a microdiachronic perspective

At 11:00 in room 55.309 (UPF-Poblenou).

This is a hybrid event. Please contact sebastian[dot]buchczyk[at]upf[dot]edu for the Zoom link.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, December 9, 2021:  Evelina Leivada (Universitat Rovira i Virgili): Cognitive adaptations to bilingualism: Where, why, and how

Abstract: It has been repeatedly argued that bi-/multilingualism confers certain effects: from better preservation of specific brain structures to enhanced performance in tasks that tap into a range of executive function measures. The effects observed on the behavioral front are the topic of an intense debate. Some scholars have found evidence in favor of a robust bilingual advantage, according to which bilinguals outperform monolinguals in certain tasks. A second line of research has produced equally strong results that reveal a disadvantage in bilingual populations. A third line of research has found evidence for a null, ‘neither advantage, nor disadvantage’ outcome, putting forth the conclusion that the advantage-hypothesis lacks the status of a robust phenomenon. To put this variety of findings in perspective, the situation is so complex that it has been argued that asking when and where the advantages occur amounts to treasure hunting by randomly digging holes in uninhabited islands. Against this background, I am going to present my recent and ongoing work that deals with three pivotal, interrelated questions that often stay in the margins: 1) What factors drive the observed effects? 2) Where are the effects expected to be found? 3) Why do the effects have the distribution they have? To address these questions, I will present results from a recent experiment, and then I will sketch out the early workings of a new theory of cognitive adaptations to bilingualism.

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, November 25, 2021:  Teodora Mihoc (Universitat Autònoma de Barcelona & Canonical, Juju Team): Aspectual operators: Temporality, evaluativity, and polarity sensitivity

Abstract: The aspectual operator still features interesting properties related to temporality, evaluativity, and polarity sensitivity. This is not unique to still but rather true of aspectual operators more generally—very similar properties can be found in already, anymore, and yet. And it is not unique to aspectual operators either—very similar facts have been reported for disjunction, indefinites, minimizers, or numerals. This paper starts from Beck (2020)’s proposal for temporality in still; generalizes it to already, anymore, and yet; and then generalizes it even further to capture evaluativity and, partially, polarity sensitivity also, drawing on Mihoc (2021)’s proposal for similar properties in superlative-modified numerals, which in turn draws on previous insights about disjunction, indefinites, minimizers, and other numerals. The result is a new solution to temporality, evaluativity and, partially, polarity sensitivity in still that unifies it not just with other aspectual operators but also with disjunction, indefinites, minimizers, and numerals.

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, November 11, 2021:  Agustín Vicente (Ikerbasque & Universidad del País Vasco/Euskal Herriko): "Thinking for speaking" revisited: a case against semantic representations.

Abstract: This talk is (mostly) about the problem of moving from concepts to language. It is a problem that has gone unnoticed, possibly because the problem is less obvious when we posit an intermediate representational system adapted to the expressive needs and constraints of language, which is what has been done (a) in order to account for relativity effects, and (b) in order to explain speech production. Slobin (1996) referred to such level as "thinking for speaking", and Levinson (2003) has called it "semantic representations". However, I explain that, far from solving the mapping problem, positing this intermediate system makes the problem even harder. Consequently, I try to account for attested relativity effects as well as to sketch a model of speech production without resorting to "thinking for speaking".

At 12:00 in room 52.737 (Campus del Poblenou).

______________________________________________________________________ ______________________________________________________________________

  • Wednesday, October 6, 2021: Josep Ausensi Jiménez' PhD defense: The Contribution of Roots: The Division of Labor between Grammar and the Lexicon in Meaning Composition. Director: Josep Maria Fontana Méndez (UPF). 

Committee: Rafael Marín Gálvez (Université de Lille), Jaume Mateu Fontanals (Universitat Autònoma de Barcelona), Alexandra Spalek (Universitetet i Oslo). 

At 12:00 in room 55.309 (UPF-Poblenou).

______________________________________________________________________ ______________________________________________________________________

  • Thursday, June 17, 2021:  Peter Sutton (UPF): The Problem of the Many and the semantics of countability

Abstract: The problem of the many (Geach 1962/1980; Lewis 1993/1999; and Unger 1980) relates to a paradox-generating line of reasoning. The problem is usually formulated in terms of nouns such as cloud and mountain:  

There are entities that can clearly be truly described as one mountain or one cloud, but it is less clear what is in the extension of singular count expressions such as mountain and cloud even in situations in which it is clear that there is one mountain or one cloud.  Take mountain, suppose we select some region of land that is a good candidate to demarcate the boundaries of the mountain. A problem arises, because there are many other marginally different, but equally valid ways of drawing the boundary. Since these competing extensions are genuinely as good as one another, then either all of these extensions count as one mountain or none of them do. Either way, we do not have one mountain, contrary to what we know to be true.


The problem of the many is most frequently discussed in the context of philosophical metaphysics. However, in this talk, I argue that the problem is directly relevant to semantic, mereological analyses of countability and numerals. First, I argue that vagueness is orthogonal to the problem of the many, since the problem arises for nouns that are not vague in the way that cloud and mountain are. Second, I claim that problem-of-the-many cases are genuinely problematic for the leading mereological theories of count nouns (e.g., Chierchia 2015; Landman 2016, Rothstein 2017), all of which, I argue, falsely predict either that count nouns such as mountain are not count, or that there are many mountains in situations such as the one above.

To address this challenge to theories of countability, I propose that we should opt for a logically weaker mereological criterion than proposed in the above theories. The criterion I propose is weak quantization. A predicate P is weakly quantized iff at most one of any two non-overlapping parts of a P is a P. Count nouns denote weakly quantized sets relative to a world and a context.

The wider relevance of this proposal is what it implies for the formal underpinnings of counting and individuation in natural languages. In particular, it suggests that counting based on extensional non-identity, as formalised in the cardinality function of classical, extensional set theory (in which | X | is the number of unique members of X), only approximates the cardinality function encoded by natural language grammars.

When we carry these considerations through into the semantics of numerals, we arrive at a semantic solution to the problem of the many. In natural languages, provided that the set of entities being counted is weakly quantized, multiple entities that are not identical to one another, such as those aligned with the different ways of drawing the boundary of a mountain, can still count as one.

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, June 10, 2021:  Martina Wiltschko (UPF/ICREA): Why do we talk? A syntactician’s point of view.

Abstract: The question regarding the origins and nature of human language has long occupied scholars in various fields. A common view within the generative enterprise is that human language is comprised of “recursive mechanisms for discrete infinity along with mappings to the interfaces with the conceptual-intentional and sensory-motor systems” (Hauser et al. 2014). While modern in its cognitive outlook, this view still follows traditional views of grammar in its conceptualization of language: the unit of analysis of a generative syntactician is typically the sentence. Sentences appear to have a particular form (however language-specific) and a particular function (e.g., to express a thought). The task of a syntactician is to explore how sentences are generated and how their form relates to their meaning, and how language variation comes about and is restricted. This roughly defines the sub-disciplines of linguistics (phonology, morphology, syntax, and semantics), which have for a long time defined the field.

In this talk I wish to make the case for exploring the question regarding the nature of language from a different angle, namely from the point of view of talking. That is, when people talk to each other language takes on a specific form and function: on the one hand speakers do not always communicate with full sentences, and on the other hand, utterances are enriched with elements that do not contribute to expressing thought, but instead are used to regulate the interaction itself. Moreover, the forms used to communicate do not only consist of sound-meaning bundles of the Saussurian type (morphemes or words) but instead, in face-to face interaction, language is enriched with embodied forms such as eye-gaze, facial expressions, gestures, and posture. And language is not only used to communicate what we think and know about the world, but also to convey our attitudes and emotions, for example. These aspects of language have traditionally been ignored within the generative enterprise; they have been relegated to performance phenomena and hence not to be considered in our exploration of language competence. This is despite the fact that within scholarly traditions that explore language in interaction, it has long been shown that there is systematicity in the language of interaction. For example, eye-gaze is systematically correlated with particular aspects of the human language faculty such as turn-taking and deixis. Similarly, gestures have been recently argued to be integrated into utterances and that syntax and semantics are modality blind. Thus, forms beyond words and functions beyond expressing thoughts are no less part of our human-specific competence for communication as is language in the narrow sense.

Given these facts about talking the traditional model according to which the computational system interfaces with the cognitive-intentional system (roughly thought) and the articulatory-perceptual system (roughly sound) should be revised. Rather, I conclude that the computational system which derives sentences also has to interface with systems that regulate embodied forms of language as well as non-epistemic functions of language. This leaves us with a practical question: how do we explore these non-traditional form-function pairings in order to draw conclusions about the computational system and its place in human cognition?

In this talk, I show how the interactional spine hypothesis in the sense of Wiltschko 2021 can serve as a heuristic for discovery and comparison of the forms and functions of talking. I present two case studies: one relating to the form and function of eye-gaze and the other relating to the expression of emotions in language in interaction.

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, June 3, 2021:  Julia Kolkmann (University of York) & Ingrid Lossius Falkum (University of Oslo): The pragmatics of possession

Abstract: English pre-nominal possessives (N’s N) express a relation between two nominals, e.g. kinship (John’s sister), time (yesterday’s newspaper), control (John’s car) and many others given that this relation is contingent upon context. However, the degree to which the linguistic context overtly explicates this relation varies from utterance to utterance, ranging from low-information contexts (e.g. John walked into the pub. His coat was white from the snow.) to contexts which are far more telling (e.g. John is a first-time author. His book has been a great success). Given that possessive interpretation is a non-issue from a communicative point of view, the question arises where the possessive relation originates from: is it provided by the (linguistic or extra-linguistic) context, does it come from the semantics of the head noun, or do we need to appeal to both? The locus of the possessive relation has been a matter of contention among formal semanticists on the one hand (e.g. Barker 1995, Vikner & Jensen 2002) and cognitively-oriented pragmaticists (e.g. Sperber & Wilson 1986/1995, Blakemore 2002, Aitken 2009) on the other: where semantic accounts often argue in favour of a default semantics for pre-nominal possessives and thus predict a rather minor contribution of the context, pragmatic accounts converge on the idea of an underspecified semantics enriched by means of a pragmatic process that operates over decoded linguistic meanings and contextually available assumptions. In this paper, we present the results of a corpus analysis of 3,000 pre-nominal possessive NPs which were analysed in terms of what kind of relation they expressed and how much contextual explication each relation received. The results show that the quantitative reality is far more mixed than what is predicted by default semantic accounts, and that co-textual support of possessive interpretations is the norm, even in cases where the semantics of the head noun indicates a possible possessive relation. We take our results to be compatible with an account where contextually appropriate possessive relations are derived mainly pragmatically, on the basis of accessible encyclopaedic or situational knowledge as well as the linguistic context, from highly underspecified relations encoded by possessive NPs. This approach has the advantage of accounting uniformly not only for possessive interpretations in high-information contexts but also for readily available interpretations in low-information contexts.

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, May 27, 2021: Steffen Heidinger & Edgar Onea (University of Graz): What you interpret as focus and why: Focus affinity and the accessoriness hierarchy in Spanish

Abstract: The notion “focus affinity” refers to the likelihood with which a grammatical role (subject, object, etc.) is the narrow focus of a sentence. In our experimental study, we investigate the focus affinity of three adjunct types (instruments, locatives, depictive secondary predicates) in Spanish and show that these grammatical roles indeed differ with respect to focus affinity. Depictives show the highest degree of focus affinity, followed by instruments and finally locatives. We also present an account of the differences between grammatical roles in terms of focus affinity − and thus also answer the question why some grammatical roles have a stronger affinity to focus than others. It is argued that the more accessory the information expressed by a grammatical role, the greater the focus affinity of that grammatical role. Why should this be this way? The more accessory a grammatical role is, the more it violates the economy principle (Avoid unimportant information!) when it is non-focal (i.e., part of the background). As a consequence, if a grammatical role A is more accessory than another grammatical role B, A is more likely to be interpreted as narrow focus than B, because under this interpretation the economy principle is violated to a lesser degree.

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, May 13th, 2021: Frank Sode (Goethe University Frankfurt): On the conditional nature of embedded V2-clauses in counterfactual wish reports of German

Abstract: This talk addresses embedded V2-clauses under “preference predicates” that typically are marked by subjunctive mood (“Konjunktiv II”), cf. Frank (1998); Meinunger (2007).

a. Ich {wollte / wünschte}, ich wäre schon zu Hause.
  I {want.SUBJ / wisht.SUBJ} I be.SUBJ already at home
  ‘I wish I was already at home.’
b. Ich wäre froh, Ich wäre schon zu Hause.
  I be.SUBJ happy I be.SUBJ already at home
  ‘I would be glad if I was already at home.’
c. Es wäre gut, Ich wäre schon zu Hause.
  It be.SUBJ good I be.SUBJ already at home
  ‘It would be good if I was already at home.’

These embedded V2-clauses dont’t really fit into the classical picture of the licensing conditions of embedded V2-clauses since predicates like “wollen” (‘want’), “gut” (‘good’) and “froh” (‘happy’) are not assertive and don’t license embedded root phenomena. I argue that, first, the embedding enviroments form a natural class from a syntactic and semantic point of view – not only in German but cross-linguistically – and that, second, this class of embedded V2-clauses are a grammaticalized form of “complement fulfilling conditionals” and therefore fall out of the scope of the classical picture of embedded V2-clauses in German.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, May 6, 2021: Daniel Gutzmann (University of Cologne): Context? Contexts? Multicontexts!

Abstract: Most semantic theories of context-sensitive interpretation are based on the (mostly implicit) assumption that there is the utterance context; i.e. that there is one utterance context per utterance. In this talk, I will challenge this “one utterance, one context” assumption by collecting many problematic cases. This will lead to abandoning this basic assumption. However, instead of throwing semantic theories of context dependency like Kaplan’s framework over board because of these challenges, I suggest an expansion of the Kaplan-style approaches by what may be called “multicontexts”, by which I understand a supercontext that included a multitude of “classical” context and thus can solve the problems while keeping the over interpretational strategies the same.

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, April 22, 2021: Xuping Li (Zhejiang University): Counting and countability in classifier languages: evidence from a Tai-Kadai language

Abstract: This research addresses ways of counting in classifier languages by investigating how countability is encoded in nominal phrase structures in Donglan Zhuang, a Tai-Kadai language spoken southwest China. This language features three types of classifier phrases, including the canonical numeral classifier phrase “Num-Cl-N”, the classifier doubling construction “Num-Cl-Cl-N”, and the bare classifier phrase “Cl-N”. To account for the syntactic variability, we propose a two-layer analysis that includes two distinct heads of classifiers in the extended projection of nouns, in the form of [DP D [CLP CLnum [CLP CLnom NP]]]. The upper classifier is the numeral classifier CLnum, whose function is to count (Scontras 2014). The lower classifier is the nominal classifiers CLnom, whose semantic function is to encode countability and derive taxonomic kinds. No matter what status is NP, [CLP CLnom NP] will be invariably used as count nominals. Donglan Zhuang thus offers us more straightforward empirical data than Chinese that a specific type of classifiers is fully dedicated to marking the mass/count distinction (Cheng and Sybesma 1998). Assuming that bare nouns are property-denoting, we argue that nominal classifiers CLnom in Donglan Zhuang, on the one hand, encode countability by imposing a ‘dividing’ structure onto the nominal predicates denoted by NPs in the sense of Borer (2005), and on the other hand, create kind expressions from the count nominal predicates derived. It follows that bare nouns in classifier languages are not uniform with respect to the [±argument] parametric setting (Chierchia 1998, Dayal 2004).

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, April 15, 2021: Anna Paradís (University of Oxford): Catalan clitics are climbers!

Abstract: The aim of this talk is to provide a descriptive and formal approach to linguistic variation regarding the different Clitic Climbing (CC) patterns attested in Catalan varieties and how CC correlates with respect to restructuring and control. The data collected support two major claims: i) the absence of CC does not entail the absence of restructuring not only in those languages in which the phenomenon displays an optional nature but also in those languages in which CC is ruled out; ii) restructuring is universal; what differentiates languages is the way in which it becomes visible (i.e. the set of transparency phenomena). The analysis presented here locates the variation regarding the different CC patterns on the nature of embedded v*. Ultimately, our proposal leads to a redefinition of restructuring.

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, March 25, 2021: Lauren Fonteyn (Leiden University): Schematically similar, but distributionally distinct? Using contextualized embeddings to study prepositions

Abstract: The term ‘meaning’, as it is presently employed in Linguistics, is a polysemous concept, covering a broad range of operational definitions. Focussing on two of these definitions, meaning as ‘concept’ and meaning as ‘context’ (also known as ‘distributional semantics’), this paper explores to what extent these operational definitions lead to converging conclusions regarding the number and nature of distinct senses a polysemous form covers. More specifically, it investigates whether the sense network that emerges from the principled polysemy model of over as proposed by Tyler & Evans (2003; 2001) can be reconstructed by the neural (or predictive, see Baroni et al. 2014) language model BERT. The study assesses whether the contextual information encoded in BERT embeddings can be employed to successfully (i) recognize the abstract sense categories and (ii) replicate the relative distances between the senses of over proposed in the principled polysemy model.

What emerges from these explorations is that BERT clearly captures fine-grained, local semantic similarities between tokens. Even with an entirely unsupervised application of BERT, discrete, coherent token groupings can be discerned that correspond relatively well with the sense categories proposed by linguists. Furthermore, embeddings of over also clearly encode information about conceptual domains, as concrete, spatial uses of prepositions, as in (1) are neatly distinguished from more abstract, metaphorical extensions (into the conceptual domain of time, or other non-spatial domains), as in (2):

1. I noticed a painting hanging over the piano (COHA, 2006)

2. a. The war on witchcraft intensified over the next 200 years, sending millions of cats, not to mention humans, to their deaths. (COHA, 2001)

b. But Mike had seemed okay with it, as if he was completely over Lindsey (COHA, 2009).

However, there are no indications that BERT embeddings also encode information about the abstract image schema resemblances between tokens across those domains. These findings highlight the fact that such imagistic similarities may not be straightforwardly captured in contextualized embeddings. Such findings can provide an interesting basis for further experimental research (testing to what extent different operational models of meaning representation are complementary when assessed against elicited behavioural data), as well as a discussion on how we can bring about a "greater cross-fertilization of theoretical and computational approaches" to the study of meaning (Boleda 2020: 213).

Baroni, Marco, Georgiana Dinu & Germán Kruszewski. 2014. Don't count, predict! A systematic comparison of context-counting vs. context predicting semantic vectors. In Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 238–247.

Boleda, Gemma. 2020. Distributional Semantics and Linguistic Theory. Annual Review of Linguistics 6(1). 213–234.

Devlin, Jacob, Ming-Wei Chang, Kenton Lee & Kristina Toutanova. 2019. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In Proceedings of NAACL-HLT 2019. 4171–4186.

Tyler, Andrea & Vyvyan Evans. 2001. Reconsidering Prepositional Polysemy Networks: The Case of Over. Language 77(4). 724–765.

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, March 18, 2021: Valentin Rose (Osnabrück University): Biased subject pronouns? – A comparative analysis of Italian and Spanish 3rd subject pronouns

Abstract: In this talk I want to discuss the variation of pronominal resolution between Italian and Spanish. Both languages are classically seen as null-subject languages in which null subjects (NS) and overt subject pronouns (OS) coexist, but display different pragmatic functions (Cardinaletti & Starke 1999; Carminati 2002). NS are used to refer to highly prominent antecedents (mostly subjects); OS are used to switch the attention to less prominent antecedents (mostly non-subjects). As these two languages are relatively close related to each other, we would not expect to find variation in pronominal resolution. However, research on this topic shows that only the pronouns in Italian show a rather clear division of labor with NS preferably referring to subjects and OS preferably referring to non-subjects. The same cannot be said about Spanish in which OS can refer relatively free to both subject and non-subject antecedents. More recently, Leonetti Escandell et al. (2019) argue, based on data of an offline experiment, that this variation between Italian and Spanish is due to word-order differences (SVO vs. VSO) and thus to the different level of prominence subject and object exhibit. Although quite appealing, their test items do not contain word-order variation. The goal of this presentation is two-fold: first, by presenting data from an offline experiment with Italian monolinguals, I want to show that Leonetti Escandell et al. (2019) are on the right track, at least for Italian. However, word-order variation only plays a role in resolving NS, but not OS. Also, the data indicate that NS seem to be less biased than their overt counterparts. Secondly, I will present an approach which tries to capture the variation between pronominal resolution and word-order variation in Italian and Spanish. I follow Sheehan (2015) that Italian and Spanish check the EPP in different ways. EPP in the former needs to be checked by a XP; the latter allows EPP to be checked by either XP or X˚. As I will show, this mirrors the bias for Italian OS and simultaneously guarantees the relative freedom of Spanish OS.

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, March 11, 2021: Reading session (students)

For the first reading group session among students in this academic year, we will discuss a paper proposed by Sebastian: Benincà, P., & Poletto, C. (2004). Topic, Focus and V2: defining the CP sublayers.

Everyone is welcomed and please feel free to propose other papers that you want discuss with the group!

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, March 4, 2021: Tobias Gretenkort (RWTH Aachen University): Language and technology – graph theory in cognitive and social network analysis

Abstract: This presentation is going to explore the benefits of the application of formal graph theory to the analysis of language. The past decades in both society as a whole and linguistics as an academic field are characterised by a steadily increasing relevance of technology. While society faces challenges in dealing with the resulting enhanced interpersonal connectivity, the work of linguists nowadays faces a myriad of new research methods based on these technologies. Such technologies also have a deep impact on theory. This is, however, not a recent phenomenon. The work of Noam Chomsky is deeply influenced by the mathematical advances in computation and automata theory at the time, most notably by Alan Turing. The technological revolution that is the Internet has produced a new tool and method of conceptualisation of language – networks. Networks are widely studied in informatics and communication under the mathematical umbrella graph-theory, but have hitherto only started to impact the study of linguistics proper. The presentation will focus on highlighting the different applications of graph theory to human language, specifically on the applicability of networks on different levels of linguistic inquiry, such as typology, grammar, cognition, and interaction networks. It will be shown that all these applications of networks to language operate under fundamentally different axiomatic assumptions, the advantages of pitfalls of which are going to be discussed in each context. Finally, the formal analysis of interaction networks will be characterised drawing on empirical data from a network corpus from online social media, hinting to some of the properties of such networks.

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, February 18, 2021: Isidora Stojanovic (Institut Jean-Nicod), Bianca Cepollaro (Univ. San Raffaele, Milan) and Filippo Domaneschi (Univ. of Genoa): Literally a jerk: an experimental study of the meaning of expressive and evaluative terms

Abstract: The semantic literature on expressive terms, such as 'bastard' and 'jerk', converges on two claims. The first one is the idea that the use of such terms is felicitous as long as the speaker has a negative attitude toward the person for whom they are using the pejorative, and does not place any special constraints on the conversational context. We have argued against this idea in our "When is it OK to call someone a jerk? An experimental investigation of expressives" (Synthese 2020, doi 10.1007/s11229-020-02633-z). The second claim is that the content associated with expressives is subjective, i.e. it amounts to something like 'the speaker feels negatively about the target'. We have argued against this idea in a new study in which we have focused on the predicative - rather than referential - uses and investigated what is the content associated with expressives. Our new study replicates the results of the first one: it shows that expressives in their predicative use are sensitive to contextual information, and that they are judged less acceptable than other negatively valenced, evaluative terms, such as 'unbearable'. We observe that negatively valenced terms in general (expressives and non-expressives alike) are, ceteris paribus, judged to be less acceptable than positive evaluative terms, such as ‘smart’ and ‘kind’. As for how to spell out the expressive content, speakers tend to take the meaning of a sentence such as "X is a jerk" to be objective ("X must have done something bad") rather than subjective ("The speaker dislikes X"). This tendency shows up both in a selection study, in which participants have been asked to choose the most appropriate paraphrase, as well as in a rating study, in which the subjective vs. objective conditions are used in the description of the context of utterance. While negative terms (expressives and non-expressives alike) tend to be interpreted in a more objective manner, there is a strong preference to interpret positive terms in a more subjective, speaker-oriented manner. Note that within negative terms, expressives are in general judged less acceptable than non-expressive evaluatives, yet the latter get to be associated with objective contents to a higher degree. This suggests that the relationship between negative valence, tendency toward objective interpretation, and acceptability, is more complex than one may have thought.

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, February 11, 2021: Fabienne Martin (Humboldt University Berlin): On non-literal readings of noun phrases: the case of pancake subjects

Abstract: This paper explores some properties of so-called pancake sentences in Brazilian Portuguese and French. Pancake sentences are copular sentences built with a (postcopular) adjective, but differ from run-of-the-mill copular sentences in two respects: (i) the nominal expression in the subject position, which is individual-denoting in its literal meaning, is reinterpreted non-literally as an event type involving the original referent; (ii) there is agreement mismatch between this nominal expression in subject position and the adjective. Following Greenberg (2008), we propose that in Brazilian Portuguese and in French, it is the agreement feature mismatch which triggers the reinterpretation mechanism of the nominal expression, which stands for a non-overt semantic structure. However, the exact output of the reinterpretation mechanism (and the meaning of the covert semantic structure the nominal expression stands for) depends on the building blocks of a non-agreeing copular sentence, which are different in the two languages under discussion. These differences explain why French non-agreeing copular sentences may have more than one meaning and are thus not necessarily pancake sentences. We also observe that in general, only predicates of personal taste are licensed in pancake sentences in the languages under discussion, which we also account for. This is joint work with Janayna Carvalho (Federal University of Minas Gerais) and Artemis Alexiadou (HU & ZAS).

At 12:00 on Collaborate.

______________________________________________________________________ ______________________________________________________________________

  • Friday, January 15, 2021: Sara Cañas Peña's PhD defense: Polar Interrogatives in Catalan Sign Language (LSC). A Comprehensive Grammatical Analysis. Director: Josep Quer Villanueva (UPF). 

Committee: Annika Herrmann (Universität Hamburg), Yasutada Sudo  (University College London), Elena Castroviejo Miró (Universidad del País Vasco). 

At 10:00 on Zoom.

______________________________________________________________________ ______________________________________________________________________

  • Thursday, January 14, 2021: Zi Huang (UPF): POSS-ing and ACC-ing as complement of with(out)

Abstract: English has two types of gerunds that take direct complements: POSS-ing ("Clay's/his winning the match") and ACC-ing ("Clay/him winning the match"). Theories of English gerunds differ in whether POSS-ing and ACC-ing are interpreted in the same way. In this talk, I will present data from the BNC which involve POSS-ing and ACC-ing as complement of "with(out)". While ACC-ing is compatible with both "with" and "without", there is an asymmetry in the distribution of POSS-ing as complement of "without" and "with":
          (1) She had been suddenly taken to hospital
              a. without Darren knowing why/Darren's knowing why
              b. with Darren being fully informed/#Darren's being fully informed
I first provide an interpretation for "without + POSS-ing" using generic incausality (Zieleke 2020). Then I argue that the asymmetry in (1) can be attributed to different referential properties of the two gerunds: POSS-ing is anaphoric and cannot be anchored to the time of the matrix clause; ACC-ing is non-anaphoric and can be anchored to matrix tense. This distinction supports those analyses that assign different interpretations to POSS-ing and ACC-ing, like Portner (1992) and Grimm & McNally (2015).

At 12:00 on Collaborate.