Oberseminar general & theoretical linguistics

The Oberseminar will take place in room 1.13 at Wilhelmstrasse 19 on selected Mondays from 4pm (ct) to 6pm.

dummy

Upcoming talks

Jan 28, 2018	Markus Werning (Bochum)
	“When Pragmatics Trumps the Lexicon: Reversing the N400 effect in the Predictive Completion Task”
	Due to learned statistical regularities, the lexical entries of nouns not only contain information about prototypical features (“tomatoes are red”), but also about typical purposes/functions (“books are for reading”) and ways of bringing about things (“cakes are made by baking”) (Pustejovsky, 1995). These are reflected by high semantic similarities between nouns and the respective verbs as measured by LSA or GloVe. When a verb follows a noun, greater semantic similarity should correlate with a higher probabilistic expectancy. However, according to Bayesian Pragmatics/Rational Speech Act Theory (Frank & Goodman, 2012), also pragmatic principles such Quantity and Relation should influence probabilistic expectations on how a sentence or discourse will be continued. Can pragmatics overturn lexical information - and if so - how and in which stage of the comprehension process?
	To investigate these questions, we introduce a Predictive Completion Task in which the hearer at every moment in a communicative situation generates a probabilistic expectation about how the speaker will continue a discourse (Cosentino, Baggio, Kontinen, & Werning, 2017; Werning & Cosentino, 2017). In two EEG experiments, we exploit the well-established observation that the conditional probability of a word given a preceding context is negatively correlated with the amplitude of its N400 component (Kuperberg & Jaeger, 2016). We observe that pragmatic factor can lead to a reversal of expectancy relative to predictions on the basis of lexical information.
Feb 5, 2018	Barbara Tomaszewicz (Cologne)
	Focus association of ‘most’ and ‘only’ in a Visual Verification Task
	click here for abstract

dummy dummy

Past talks

Jan 22, 2018	Harald Hammarström (Jena)
	“Deep and Shallow in Automated Cognate Recognition”
	Given a set of languages with wordlists, the problem of cognate identification is to decide which sets of forms are derived from a common source (List et al. 2017, Kondrak 2009 and references therein). Typically, cognate identification is performed by aligning words phonetically and checking if the similarity exceeds a human-tuned or pre-trained threshold value. However, if this threshold value is set too strict, only shallow cognates (drei vs three) will be discovered, while if set too loose, non-cognates along with deep cognates (aqua vs eau) will be retrieved. We will argue that the use of a threshold value inherently poses this dilemma. As an alternative we propose a threshold-free approach to cognate identification with a different conceptualization. Only shallow cognates are targeted in each phase of cognate identification. Once shallow cognates have been identified, a (preliminary) most shallow subgroup may be inferred. This subgroup implies a proto-language which can be preliminarily reconstructed using the sound correspondences from the shallow cognates as a starting point. Shallow cognate identification may then be performed again, this time with the proto-language of the subgroup, and a new subgroup posited, and so forth iteratively. This way deep cognates may be recognized if and only if the modern forms become more similar the deeper we go back in time.
Nov 27, 2017	Guido Seiler (Munich)
	“Dialects as a testing ground for theories of language change: closing the gap in a typology of mixed languages”
	The study of dialect variation has proven a powerful tool to deepen our understanding of the fundamental mechanisms underlying language change: Variation between very closely related languages/varieties often creates living laboratories where “What happens if…”-questions can be assessed. After an overview of recent developments within this approach I will focus on a newly emerging but rapidly growing field of historical linguistics, namely the study of change induced by contact between closely related languages/varieties, and in particular on the emergence of radically mixed languages. Whereas the existing literature suggests that the results of heavy language contact are qualitatively different depending on whether the involved languages are closely related genetically or not, I will argue that this is not (always) the case: Rather, radically mixed languages, including languages with their lexicon inherited from one parental language but grammatical structure from another, may emerge, too, even if the parental languages are close cousins, as long as the appropriate sociolinguistic conditions for mixed language emergence are met. The case study is based on my own fieldwork in an Amish community in Indiana (USA). Amish Shwitzer is a mixed language that emerged as a result of intensive contact between Pennsylvania Dutch and Bernese Swiss German. Whereas grammatical structure has largely converged with Pennsylvania Dutch, Bernese Swiss German features are strikingly concentrated in the lexicon. This kind of compartmentalization has not been reported so far for mixed languages with closely related parents. The case study sheds new light on theories of mixed language emergence, in particular on concepts such as dominating language or widespread bilingualism, which are not applicable to the Amish Shwitzer situation.
Nov 13, 2017	Matthias Urban (Tübingen)
	“Language contact, language shift and the (linguistic) prehistory of the ancient Central Andes: a research program”
	In this talk, I describe the research agenda of my newly established Emmy Noether Junior Research Group “The language dynamics of the ancient Central Andes” at Tübingen and illustrate its approach with a case study.
	To begin with I provide relevant background information on the linguistic landscape as well as the cultural trajectory of the Central Andes – coast and highland of Peru, highlands of western Bolivia, and the northern and northwesternmost parts of Argentina. I also sketch the state of the art of research on the linguistic prehistory and how it might relate to the cultural development of this center of pristine civilization in South America. Pointing out lacunae and neglected aspects in extant research, I move on to describe the general approach and “philosophy” of my group’s projected work, which focusses on language contact and language shift as a window to the prehistory of the Central Andes instead of expansion of the large Quechuan language family. I also provide short descriptions of planned subprojects.
	Finally, I zoom in on one particular area of the Central Andes, that of Chachapoyas on the eastern slopes of the Northern Peruvian Andes. In this area, a particular (now moribund) variety of Quechua is spoken which features structural characteristics that depart strongly from typical Quechan patterns. Given that Chachapoyas was the home of a particular ethnic group with a distinct material culture prior to Inca and then Spanish conquest in the late 15th and early 16th centuries AD, “substrate” influence from a local language conventionally known as Chacha is a real possibility. Unfortunately, however, Chachapoyas presents a “worst case scenario”: apart from placenames and personal names, however, the pre-Inca Chacha language is not documented at all. Based on an extrapolation of different lines of evidence –placenames and personal names, vocabulary items without Quechua cognates in Chachapoyas Quechua itself, and structural-typological considerations– I suggest that the language underlying Chachapoyas Quechua on the eastern slopes of Northern Peru could have been a close relative of the Cholón language, formerly spoken to the south of the Chachapoyas area.
Nov 06, 2017	Jack Tomlinson (Berlin, ZAS)
	“Be timely: how turn-taking bottleneck constrains the integration of QUDs into implicatures”
	Understanding pragmatic inferences can sometimes result in processing delays, but this depends on the availability of alternatives during processing (Degen & Tannenhaus, 2014, Bott & Chemla, 2016). One notable source of alternatives is the Question Under Discussion (QUDs) (Roberts, 2012) and this has been shown to predict whether a scalar implicature is derived or not (Cummins, 2015). However, even if alternatives are available to the listener, it is unknown whether integrating alternatives into implicatures is an effortless process (Zondervan, 2010). To answer this question, I present a novel method for testing the integration of QUDs into implicature processing; speicifically one that examines whether listeners can integrate alternatives at various turn-taking gaps (see Levinson & Torreiria, 2016). In this tasks, overhearers are given explicit QUDs prior to overhearing an addressee respond to several questions implicity answering this QUD (via implicatures). If QUDs are accommodated prior to the derivation of the implicature, turn-taking gaps should have marginal effects on whether listeners derive the implicatures and implicature rates should be higher if scalar terms are found in sentence-intial postion (predictive account). In contrast, if listeners need additional effort to integrate alternatives from QUD, then implicature rates should increase at longer turn-taking gaps (delayed intergration account) . To test this, I manipulated the sentential position of the scalar term, einige (some), (Einige haben bestanden vs. Bestanden haben einige) as well as the turn intervals (250, 750, 1250ms) between interlocutors question-answer sequences that participants overheard. Listeners had to respond to the intial QUD by clicking on “Yes” or “No” responses and their mouse path data were also collected and analyzed.
	Overall, participants had lower implicature rates (chance level) for some-initial and some-final at short latencies (250ms). In contrast, some-final responses yielded higher implicature rates than some-intial at both 750ms and 1250ms intervals, suggesting an effect of focusing the scalar term. Critically, implicature rates decreased significantly below chance at 1250ms intervals and this was more the case for some-initial responses. The later finding suggests that longer turn-taking intervals are understood as hesitations, which reduce implicature rates. An analysis of the mouse-path data showed similar implicature patterns as found in Tomlinson et al (2013) at all intervals, albeit only for some-intial positions. Together this supports the idea that QUD integration is not automatic and can depend on cognitive constraints such as turn-taking gaps. These data also show that turn-taking gaps affect both the integration of alternatives and derivation of implicatures by adding additional epistemic information to the speaker. I expand on this idea in proposing a “Be timely” maxim and discuss the reasons why theories of scalar implicatures should account for turn-taking data.
Oct 30, 2017	Bob van Tiel (Brussels & Berlin, ZAS)
	Processing “some” and “or”: evaluating theoretical models of pragmatic reasoning
	The standard view in natural language semantics holds that “or” expresses inclusive disjunction. In certain embedding contexts, however, “or” may take on a conjunctive meaning. Recent theorists have explained these conjunctive inferences as conversational implicatures. We evaluate this explanation by testing the processing profiles of four types of conjunctive inferences and comparing these with the processing profile of a bona fide conversational implicature that is hypothesised to be akin to conjunctive inferences, i.e., the scalar inference from “some” to “some but not all”. At face value, the results speak against the implicature-based explanation of conjunctive inferences. On a closer look, however, the results have more far-reaching consequences for the widespread assumption that processing data are indicative of the mechanism that underlies pragmatic inferencing.
July 31, 2017	Ciyang Qing (Stanford)
	“Unifying three Mandarin ‘dou’ constructions”
	In Mandarin Chinese, dou is a versatile focus-sensitive particle. I will discuss three types of dou constructions: (i) the scalar lian … dou construction, in which the focused constituent must appear to the left of dou, (ii) the universal wh … dou construction, in which the focused wh-constituent also must appear to the left of dou, and (iii) the dou … wh exhaustive question construction, in which the focused wh-constituent must appear to the right of dou. I will propose a semantics of dou that operates on a two-dimensional Roothian system to derive the exhaustive question reading, and show that the restrictions on the order between the focused constituent and the particle dou follow from this semantics of dou.
July 17, 2017	Robert D. Hawkins (Stanford)
	“Convention-formation in iterated reference games”
	Just as drivers depend on shared behavioral conventions to safely navigate traffic, successful communication depends on a set of shared linguistic conventions. All members of a language community share stable knowledge of global conventions, many of which evolved over long timescales. Yet communication partners are also remarkably flexible in their ability to adapt to local contexts and to dynamically coordinate on local, ad hoc conventions across a relatively small number of repeated interactions. In this talk, I???ll present results from several interactive communication games that investigate the cognitive mechanisms that support such flexibility. First, these results suggest that recent efforts to show failures in flexible social reasoning should in fact be reinterpreted as pragmatic successes. Second, they motivate a theory of local convention-formation where agents, though initially uncertain about word meanings in context, learn from and thus adapt to a partner who is assumed to be using language with such knowledge. We formalize both contributions in a computational model of language understanding as recursive social inference.
July 10, 2017	Anna Howell (Tübingen)
	“Alternatives and Disjunction through the lens of Disjunctive Questions in Yoruba” click here for abstract
July 3, 2017	Olivier Morin (Jena)
	“Cognition, history and chance in the evolution of visual culture”
	Worldwide visual culture provides ideal material to test and compare a range of hypotheses coming from cognitive science, cultural evolution, and cultural history. Images, designs and patterns are naturally well documented, and visual cognition is among the least controversial areas of cognitive science. Thus, questions that would receive tentative or debatable answers if asked about languages, or social norms, can be addressed with precision when visual culture is concerned. Two studies will be presented.
	One describes widespread cognitive constraints on the shape of letters in 116 scripts (Tha??, Greek, Af??ka, etc.). Orientation anisotropy, vertical symmetry dominance, and orientation-based segregation of letters, all left a clear mark on most of the world’s writing systems. Contrary to the received view in cultural evolution, scripts did not gradually evolve to fit these cognitive constraints, nor did cultural selection play any role in the process.
	Our second case study focuses on the diffusion of heraldic designs on coats of arms (between the late middle ages and early modern times). These designs could spread by imitation, but given the limited number of motifs and tinctures, the chances for some of them of being independently reinvented were substantial. A model which assumes that heraldic designs diffuse by independent reinvention only was tested on two heraldic corpus. The model predicts the appearance of actual designs in the historical record with surprising accuracy. These two studies suggest ways that a quantified history of visual culture can shed light on mechanisms of cultural change on a vast scale.
June 28, 2017	Judith Degen (Stanford)
	“Mentioning atypical properties of objects is communicatively efficient”
	What governs how much information speakers include in referring expressions? Atypical properties of objects are more likely to be included in referring expressions than typical ones (Sedivy 2003; Westerbeek et al 2015). For example, speakers are more likely to call a blue banana a ???blue banana??? and a yellow banana a “banana”. A unified account of this phenomenon is lacking. We ask: when should a rational speaker mention an object???s color? Reference production is modeled within the Rational Speech Act framework (Frank & Goodman 2012). Utterances (e.g., ???banana???, ???blue???, and ???blue banana???) are taken to have a graded semantics: rather than assuming all bananas are equally good instances of ???banana???, we empirically elicited object-utterance typicality values for all possible utterances. Pragmatic speakers select utterances proportionally to the probability that a literal listener using a graded semantics will select the intended referent. We evaluate the proposed model on a dataset of freely produced referring expressions collected in an interactive reference game experiment via the web. We conclude that the systematicity with which speakers redundantly mention color implicates a system geared towards communicative efficiency rather than towards wasteful overinformativeness, and discuss potential extensions of this approach to other production phenomena, such as optional instrument mention.
May 29, 2017	Suzanne Stevenson (Toronto)
	“Knowledge mismatch in communicative interaction: Probabilistic weighing of perspectives”
	(Joint work with Daphna Heller, Mindaugas Mozuraitis, and Chris Parisien)
	Knowledge mismatch is inherent to communicative interaction, since conversational partners naturally differ in their knowledge and beliefs. Effective conversational moves, such as asking a question or referring to an entity, crucially depend then on each participant acting appropriately with respect to which knowledge is privileged (unique to that participant) and which is shared among the participants. For example, much work in linguistics suggests that a definite referring expression such as the knife must be interpreted in the context of mutual knowledge (???common ground???) in which there is a uniquely identifiable knife relevant to both conversational partners.
	What has been less clear is the moment-by-moment cognitive processing by which people reconcile knowledge mismatch in language production and interpretation. We propose a theory and computational model in which differing views of relevant knowledge, both privileged to the individual and shared with the conversational partner, are simultaneously and probabilistically considered at the earliest moments of language processing. This approach arises from a view of communicative interaction as a process requiring continual and rapid accessing and updating of both privileged and shared knowledge.
May 24, 2017	Ralf Vogel (Bielefeld)
	“Experimental explorations on grammatical taboos in Standard German”
	Since about 20 years, the use of experimental methods in syntactic research has been growing steadily. Recently, Sprouse et al. (2013) nevertheless showed that linguists’ expert judgements have a very high degree of accuracy as compared to the outcomes of elicitation experiments.
	Experimental methods are therefore best applied when complementing, not replacing the traditional method of expert judgements. Such a situation occurs when a phenomenon exceeds the capacities of the expert’s judgements. I see (at least) two sources for such excessive demands:
	A psychological one: differences between expressions might be too subtle to assess reliably from the linguists’ armchair (e.g. gradient acceptability, degrees of markedness).
	A sociological one: certain expressions might have a controversial status within the speech community. The expert judgement (perhaps then even controversial among linguists) in such a situation will only be one opinion among others.
	Grammatical taboos have both features: they are socioculturally induced instances of grammatical markedness.
	Grammatical taboos have a paradoxical nature. They are structures or constructions that are part of the language insofar as they are used by the speakers and therefore obviously a natural consequence of the language’s grammatical apparatus. Nevertheless speakers at the same time believe (to varying degrees) that they are not part of their language. This is usually due to prescription that can occur in various ways and comes along with a particular ideology of standard language that is dominant within the speech community and therefore also an important influence of participants’ behaviour in grammatical experiments.
	I will present several explorative experimental studies on grammatical taboos. My example cases from standard German are auxiliary “tun”, V2-Sentences with “weil”, d-Pronouns and the double perfect. I am trying to find answers to the following questions, a.o.:
	- Can we adjust our experimental methods in order to minimize the influence of prescription?
	- Can grammatical taboos be distinguished empirically from other cases of syntactic markedness?
	- Can we decide for very strong taboos, using experimental methods, whether those are instances of markedness or ungrammaticality?
	- Can we identify sociolinguistic factors that correlate with subjects’ evaluation of grammatical taboos?
	Two main results of my studies are:
	i) There is reason to be optimistic with respect to the issue of successfully establishing and evaluating the category of grammatical taboos, using the observational means provided by experimental syntax;
	ii) There is good reason and perhaps even an urgent need for theoretical and empirical grammarians to include the sociocultural dimension of grammar into their models. (“la langue est un fait social”, de Saussure)
May 8, 2017	Heather Burnett (Paris)
	“Signaling Games, Sociolinguistic Variation and the Construction of Style” In this presentation, I introduce social meaning games (SMGs), which are developed for the analysis of the strategic aspect of sociolinguistic variation (in the sense of Labov 1963, Labov 1966, et seq.). While remarks have been made (eg. Goffman 1970, Dror et al. 2013, Clark 2014) about the potential usefulness of game theory in the analysis of the meaning of variable linguistic phenomena (for example, variable use of the English (ING) suffix (1)), a general framework uniting variationist sociolinguistics with game theoretic pragmatics has yet to be developed.   (1) I???m workin??? on it vs I???m working on it..   I propose that such a unification is possible through the integration of the Third Wave approach to the meaning of sociolinguistic variation (see Eckert 2000, 2008, 2012) with signalling games (Lewis 1969) and a Bayesian approach to speaker/listener reasoning (see Oaksford & Chater 2007 for a review). The combination of ???signaling games + Bayesian reasoning??? has previously been argued to be particularly useful in the analysis of a large class of pragmatic phenomena, including scalar implicatures, manner implicatures and context-sensitive reference (see Franke & J??ger 2016 for an overview)..
	I define the games and then show the predictions of this framework for both linguistic production and interpretation, as exemplified by the modeling of six empirical studies: .   PRODUCTION   1. Labov (1966)???s study of the social stratification of (ING, i.e -in??? vs -ing) in New York City. 2. Labov (2012)???s study of President Obama and Sara Palin???s use of (ING) in formal vs informal settings. 3. Gratton (2015)???s study of the use of (ING) by non-binary individuals (i.e., individuals whose gender identity does not respect the male/female binary) in their home vs a public coffee shop..   INTERPRETATION   4. Campbell-Kibler (2007)’s experimental study of the interpretation of (ING) in the United States.. 5. Podesva et al. (2015)’s experimental study of the interpretation of /t/ release in the speech of 6 American politicians. 6. Levon (2014)’s experimental study of the relationship between gender stereotypes and the interpretation of high/low pitch by men in the UK..   Based on these examples, I argue that SMGs have potential to provide a new, precise understanding of how we use our linguistic resources to communicate information and carve out our place in the social world..   SELECTED REFERENCES.   Campbell-Kibler, K. (2007). Accent,(ing), and the social logic of listener perceptions. American speech, 82(1):32???64. Clark, R. (2011). Meaningful Games. Cambridge: MIT Press. Eckert, P. (2012). Three waves of variation study. Annual Revue of Anthropology. 41:87-100. Eckert, P. (2008). Variation and the indexical field. Journal of sociolinguistics, 12(4):453???476. Eckert, P. (2000). Language variation as social practice: The linguistic construction of identity in Belten High. Wiley-Blackwell. Franke, M. & J??ger, G. (2016). Probabilistic pragmatics, or why Bayes??? rule is probably important for pragmatics. Zeitschrift f??r Sprachwissenschaft, 35:3???44. Goffman, E. (1970). Strategic interaction, volume 1. University of Pennsylvania Press. Gratton, C. (2015). Indexin??? Gender: Variable (ING) and the Creation of Non-Binary Trans Identities. Best poster at New Ways of Analyzing Linguistic Variation 44. U. Toronto Labov, W. (1966). The Social Stratification of English in New York City. Washington DC:Center for Applied Linguistics. Labov, W. (2012). Dialect Diversity in America: The Politics of Language Change. Charlottesville: University of Virginia Press. Levon, E. (2014). Categories, stereotypes, and the linguistic perception of sexuality. Language in Society, 43(05):539???566. Lewis, D. (1969). Convention: A Philosophical Study. Harvard University Press. Oaksford M. & N. Chater (2007). Bayesian Rationality: the probabilistic approach to human reasoning. OUP. Podesva, R. J., Reynolds, J., Callier, P., and Baptiste, J. (2015). Constraints on the social meaning of released/t: A production and perception study of us politicians. Language Variation and Change, 27(01):59???87.
April 24, 2017	Bill Thompson (Nijmegen)
	“Mapping Semantic Networks Across Languages: First Impressions” Words carve up human experience into richly structured categories of events, objects, relationships, emotions, and ideas. Categories like these subserve human communication and interaction, but it has long been unclear how similar these categories are for speakers of different languages. Recent progress in applied machine learning and artificial intelligence has resulted in large scale machine-readable representations of semantic networks across many languages. I???ll present some preliminary results of a collaborative project whose goal is to leverage these resources, in combination with more traditional linguistic resources, to map patterns of regularity and diversity across semantic domains in many languages. I???ll also discus ways in which these inferred networks can be fused with agent-based simulation and experimental techniques to model the evolution of semantic systems.
Jan 30, 2017	Timo Roettger & Mathias Stoeber (Cologne)
	“Manual Response Tracking during Intonation-Based Intention Recognition” During the perception of an unfolding speech signal listeners use prosodic information to guide their interpretation of what a speaker intends to communicate. This process can take place long before disambiguating lexical information becomes available, allowing the comprehender to make rapid inferences about what a speaker intends to say, even if these inferences are based on partial information. The inference process can only be fully understood using experimental techniques that measure the real-time integration of prosodic information to resolve temporally ambiguous interpretations. While eye tracking experiments have advanced our knowledge about the real-time integration of prosodic information tremendously, we will propose that tracking hand movements complement existing methodologies and offer low-cost alternatives to eye tracking: It has been demonstrated that continuous nonlinear trajectories recorded from the streaming of x,y coordinates of computer mouse movements can serve as an informative indicator of cognitive processes. This talk will introduce mouse tracking as a tool for speech scientists, capable of unravelling the temporal dynamics of speech processing. Additionally, we will present preliminary results of a mouse tracking paradigm attempting to illuminate prosody-based intention recognition. In light of the presented data, we will discuss both methodological advantages and challenges of the proposed paradigm. We hope to convince the audience that mouse tracking is a methodology which holds great promise for low-cost but detailed exploration of fine-grained temporal aspects of speech perception in general and prosody-based intention recognition in particular.
Jan 16, 2017	Maribel Romero (Konstanz) “On the readings of many” click here for abstract
Nov 28, 2016	Thomas M??ller (Jena)
	“‘I see what you mean!’ - The Influence of Common Perceptual Context and Perspective-Taking on the Evolution of Graphic Codes” click here for abstract
Nov 14, 2016	Daniel Gutzmann (Cologne) “Expressive, much?” (Joint work with Robert Henderson, The University of Arizona) click here for abstract
Oct 31, 2016	Andreas Bischoff (Freiburg)
	“Lautwandelketten und Maximisierung akustischer Distinktivit??t”
	Ein besonderes Ph??nomen des Lautwandels sind Lautwandelketten (engl. chain shifts). Als solche bezeichnet man den Effekt, dass ein Lautwandel einen weiteren Lautwandel ausl??st. So kann ein Lautwandel /e/ > /??/ einerseits die Ursache f??r einen folgenden Lautwandel /??/ > /i/ (sog. Phonologischer Schub, engl. pushing) darstellen, andererseits auch einen Lautwandel /a/ > /e/ nach sich ziehen (sog. Phonologischer Sog, engl. pulling). Mit Hilfe von solchen Verschiebungsketten kann die akustische Distinktivit??t der Phoneme aufrecht erhalten werden und sind daher u. a. f??r die Theorie der self organization relevant. Der Vortrag wird das Ph??nomen aus akustischer, theoretischer und empirischer Sicht thematisieren und die Fragen stellen, wie oft diese erscheinen und wie wichtig sie als Ursache f??r Lautwandel sind.
July 5, 2016	Ryan Cotterell (Johns Hopkins)
	“Modeling Word Forms Using Latent Underlying Morphs and Phonology”
	The observed pronunciations or spellings of words are often explained as arising from the ???underlying forms??? of their morphemes. These forms are latent strings that linguists try to reconstruct by hand. We propose to reconstruct them automatically at scale, enabling generalization to new words. Given some surface word types of a concatenative language along with the abstract morpheme sequences that they express, we show how to recover consistent underlying forms for these morphemes, together with the (stochastic) phonology that maps each concatenation of underlying forms to a surface form. Our technique involves loopy belief propagation in a natural directed graphical model whose variables are unknown strings and whose conditional distributions are encoded as finite-state machines with trainable weights. We define training and evaluation paradigms for the task of surface word prediction, and report results on subsets of 7 languages.
June 24, 2016	Bodo Winter (Merced, Birmingham) & Andrew Wedel (Arizona)
	“Robustness in phonetic systems: Evidence from phoneme frequencies and iterated learning”
	When speakers of a language communicate with one another, they frequently do so in noisy and sometimes rapidly changing acoustic environments. Language is structured so as to resist these perturbations. I will present two projects on the robustness of spoken language. First, a typological study showing that languages prefer phonemes that are perceptually distinct. Second, an iterated learning experiment showing how robust movement-to-acoustic mappings change the course of language evolution. Together with computational models, these studies show that languages evolve under communicative pressures to maintain function in spite of noise.
	Andrew Wedel (Arizona) & Bodo Winter (Merced, Birmingham)
	“The interaction of functional pressure, redundancy and category variation in phonetic systems”
	At the level of individual speakers, evidence from natural speech corpora and laboratory investigations shows that the existence of similar lexical items promotes hyperarticulation of phonetic cues that distinguish them. In parallel, at diachronic time-scales evidence suggests that phoneme inventories evolve under the influence of a bias toward maintenance of existing lexico-semantic distinctions. How might these be causally related? Here we present simple exemplar-based simulations that help us explore hypotheses about a general linking mechanism between pressure to maintain lexical contrasts in individual usage events, and maintenance of phonemic contrasts over diachronic time scales. We show that redundancy in sound categories interacts with communicative pressure to modulate ???cryptic variation??? in sound categories and show that within the model, this sub-phonemic variation shapes future sound change.
	Adam Ussishkin (Arizona) “Maltese root priming is morphological, not phonological” click here for abstract
June 13, 2016	Marie-Christine Meyer (Berlin)
	“Structural conditions on implicature computation: The latest news”
	This talk has two goals. First, to introduce the audience to an ongoing and open-ended debate in the field of formal pragmatics. The debate builds on seminal work by R. Katzir (2007, 2008, 2014) and concerns the role of structure in the definition of the set of alternatives that can be used in the computation of (scalar) implicatures. Structural constraints on alternatives have been shown to provide a solution to the so-called Symmetry Problem (Kroch 1972, von Fintel & Heim 1997), and more general well-formedness conditions on alternatives have been argued to underlie otherwise unrelated phenomena (e.g. Meyer 2015). But the idea that alternatives are structural objects which may as such be subject to structural constraints has recently come under attack in view of data that Katzir???s original proposal cannot account for (Swanson 2010, Trinh & Haida 2015, Romoli 2013). These new data will bring us to the second goal of the talk: To present even more new, i.e., newest data. I will argue that most of the seemingly problematic issues have been insufficiently analyzed and that factors which are completely independent of structural constraints on alternatives, viz.,the presupposition of contrastive connectives (e.g., Toosarvandani 2013) and principles of information encoding (e.g., Parikh 2001) have to be taken into account. Once this is accomplished, a new classification of the data ensues which makes it possible to disentangle genuine issues of the structural approach on the one hand, and confounding factors on the other.
June 6, 2016	Sven Lauer (Göttingen & Konstanz)
	“(Un)conditional imperatives, (un)conditional modals, and (un)conditional endorsement” (Joint work with Cleo Condoravdi, Stanford University)
	A modal declarative like (1) can simply report that something is necessary, given a body of rules/laws/regulations, etc. without the speaker taking any position on the matter.       (1) (In view of what the law provides,) you have to file a tax return. Things are different with imperatives. A sentence like (2) apparently indicates that the speaker wants the addressee to file a tax return.       (2) File a tax return! This ‘endorsement’ implication is unsurprising for ‘directive’ uses of imperatives as orders, requests, pleas, and so forth, but it can also be detected for other uses of imperatives, e.g. as advice, invitations, permissions and concessions (Schwager, 2006; Kaufmann, 2012; Condoravdi & Lauer, 2012). This indicates that ‘endorsement’ is a conventional feature of the imperative form. We probe into the nature of the endorsement inference by examining conditional sentences with modal and imperative consequents, in particular those of the ‘anankastic’ type:       (3) a. If you want to go to Harlem, you have to / should take the A train.              b. If you want to go to Harlem, take the A train. We show that (i) conditional imperatives give rise to a form of ‘conditional endorsement’, while conditionalized modal sentences do not; and (ii) that a robust endorsement implication surprisingly ‘creeps into’ the interpretation of unconditional modal sentences in certain circumstances. We argue that a Condoravdi & Lauer (2012)-style analysis of imperatives is well-equipped to account for (i), and explore several avenues for accounting for the puzzling (ii).
April 25, 2016	Thomas Brochhagen (Amsterdam) “Ambiguity and alignment without a common prior”
	While ambiguity is a pervasive property of natural language communication, it is often characterized as suboptimal. Notwithstanding, recent investigations have proposed ambiguity’s vindication to lie in the exploitation of contextual information. Building on these insights, I propose a signaling model that takes context, as well as interlocutors with heterogeneous contextual expectations, i.e. different priors, into consideration. The talk will deal with the question how prior heterogeneity can be overcome, and on the conditions that ensure successful ambiguous communication. Put differently, we will focus on the question whether and when the claim that context can be exploited is justified if interlocutors do not share the same prior. Our main goal is to put forward a model to evaluate predictions about ambiguity’s putative advantages and the involved risk of misunderstanding. In doing so, we will touch upon issues related to lexical ambiguity, linguistic convergence in dialogue, and, if time allows, Horn’s division of pragmatic labor.

dummy dummy

contact: Michael Franke