See the program at http://jeschull.com/uuslaw/.

Organized by Prof. Kyle Johnson. See workshop website

The ICESL Launch Workshop will take place on April 1, 2011 on the campus of the University of Massachusetts, Amherst. Confirmed speakers are Brian Dillon and Alexandra Jesse. There will also be a poster session that we hope will represent the full range of computational and experimental research on language being done at our university. If you would like to present a poster, please send the names of the authors and title to us by e-mail by March 12, 2011. See http://people.umass.edu/icesl/ for more information.

For a detailed program, see http://dl.dropbox.com/u/1329068/Treebank-Hyderabad/sass-program.html.

Sponsored by John Kingston (Linguistics) and Rebecca Spencer (Psychology). Registration begins 8:30 a.m. Please visit the website for schedule details, student-friendly fees and other information: http://cognaclab.com/wp/nest/.

WSCLA (Workshop on Structure and Constituency in Languages of the Americas) will be held at UMass Feb.11-13 2011. Invited speakers include Emmon Bach, SOAS, U. of London, Benjamin Bruening, University of Delaware, Amy Rose Deal, Harvard University, and Lisa Matthewson, University of British Columbia. Organized by Profs. Seth Cable and Peggy Speas. For more information see http://people.umass.edu/wscla.

The Economy of Encoding and Anaphoric Dependency with Relational Nouns:Evidence from Child Grammar

Terue Miyashita, CLS Practice Talk

and:

Preterite had production in typically-developing AAE speakers and AAE speakers with specific language impairment

Frances Burns

"What are the stages in the acquisition of each and every: From generic to specific or vise versa?"

Infants are faced with the challenge of discovering abstract phonological categories from the real-valued speech signal. This is a potentially prohibitively complex high-dimensional problem, not only for the infant language learner, but for the linguist attempting to model the infant’s task.

In this talk, I argue that there are inductive biases that constrain how the infant construes the speech signal and phonological categories. To understand what these biases are, I use cross-linguistic fieldwork and psychological experimentation: they provide evidence about what human learners do. I complement these methods with computational modeling to tease apart how such learning could be possible: I compare models with different parameterizations of the speech signal and see how well they separate the data into appropriate categories.

In my model system of lexical tone based on data I collected from a range of tonal languages, this research has yielded the following results: (1) coarse temporal resolution is sufficient for tonal classification; (2) the classification of the tone of a syllable requires contextual information from neighboring regions; (3) there are no simple invariant pitch-based cues such as pitch velocity for tones (Gauthier et al. 2007); (4) in fact, tone is not defined over only pitch, but over voice quality cues as well. In summary, while the discovery of tones takes place in a higher dimensional space than a pitch-based 2-dimensional space (Barry and Blamey 2004), the space is not prohibitively high-dimensional. These results will serve as the foundation for a cognitively motivated computational model of phonological category acquisition.

This talk presents a series of experiments that explore the phonetic bases of intervocalic lenition – specifically, voicing and spirantization of intervocalic stops. The first experiment tests the traditional understanding of phonological patterns like these as articulatory effort reduction, with speakers substituting an easy sound for a hard one; surprisingly, this study provides no evidence that lenited productions are truly easier than their unlenited counterparts. Three additional experiments investigate whether an alternative account of lenition based on perception is viable; the results suggest that attested alternations such as spirantization of voiced stops are preferred on perceptual grounds to unattested alternations such as intervocalic devoicing. Thus, the hypothesis of the P-map (Steriade 2001) can explain the broad strokes of lenition, although there are differences by place of articulation that do not match well with the typology. I conclude with an analysis of intervocalic spirantization couched within Optimality Theory, and particularly Dispersion Theory. Unlike previous accounts of lenition, this analysis invokes no constraints that directly favor lenited forms over unlenited ones, since no such constraints are motivated by the experimental evidence. The constraints that are made available by the experimental results are nevertheless able to account for a sizeable portion of the typology of lenition. I conclude that articulatory factors say less about lenition than traditionally thought, and that perceptual factors say more – and that theories of phonology that are committed to taking phonetics seriously must take notice.

Research on syllable weight in the generative tradition has focused almost exclusively on systems in which weight is treated as an ordinal (strict domination) scale of clearly delineated categories (e.g. heavy and light). As I discuss, stress and poetic meter can also treat weight as a gradient interval scale in which (1) differences between types are matters of relative degree and (2) there is no clear segregation of types into categories. I propose modeling such systems in a probabilistic Harmonic Grammar framework (as in e.g. Goldwater & Johnson 2003, Hayes & Wilson 2008, Ryan 2010) with gradiently violable constraints (cf. Flemming 2001, Zhang 2007, Pater to appear). Investigating gradient weight systems allows the phonetics/phonology of weight to be put under a microscope, so to speak, yielding new evidence supporting and extending phonological universals: First, the scales that emerge language-internally in the gradient realm recapitulate the universal hierarchy inferred from the crosslinguistic categorical typology (Gordon 2006). Second, factors in weight that are relatively marginal in the categorical typology (e.g. properties of onsets) are shown to exert statistical effects even in languages in which such factors play no role in categorization.

In this talk, I argue that long-distance phonological dependencies are grounded in perceptual asymmetries. The analytical claims are 1. That laryngeal cooccurrence restrictions are restrictions on the perceptual strength of contrasts between roots, as opposed to restrictions on laryngeal configurations in isolated roots, and 2. That laryngeal cooccurrence restrictions are restrictions on auditory, as opposed to articulatory, features.

Both long-distance laryngeal dissimilation, where roots may have one but not two laryngeally marked stops (MacEachern 1999), and assimilation, where stops in a root must agree in laryngeal features (Hansson 2001; Rose and Walker 2004) are given a unified account based on a grammatical pressure to neutralize indistinct contrasts. The contrast based analysis is supported by the empirical finding that certain non-adjacent sounds interact with one another in perception. Specifically, the perception of a contrast in ejection or aspiration is degraded in roots with another ejective or aspirate, as compared to roots with another plain stop (e.g. the pair k'ap'i-kap'i is more confusable than the pair k'api-kapi). Roots that are minimally distinguished by having one vs. two laryngeally marked stops are confusable with one another (e.g. k'ap'i is confusable with kap'i), and thus languages may avoid having both types of forms.

The analysis integrates long-distance phonological neutralizations with analyses of local neutralizations based on phonetic cues and contrast strength (Flemming 1995, 2004, 2006; Steriade 1997), showing that both local and non-local phenomena are driven by grammatical constraints against perceptually indistinct contrasts.

Perception of Exuberant Exponence in Batsbi: Funtional or Incidental?

Alice C. Harris (work done with Arthur Samuel)

The term “exuberant exponence” refers to the occurrence of more than two markers (exponents) of a feature or bundle of features within a single word, as in this example from Batsbi (Nakh-Daghestanian, severely endangered): d-ex-d-o-d-anŏ ‘evidently they are tearing down’, where each instance of d- realizes the gender and number of the object. While multiple exponence is seen in this example, some verbal lexemes in the language govern no gender-number marking at all, some one exponent, and some two. In a series of three experiments, we compare verbs that have no agreement marker with ones that have a single marker, and we compare verbs with one agreement marker with ones that have two. We find that word recognition is slower with agreement than without it; words with two agreement markers are recognized more slowly and with more errors relative to verbs with a single marker. For grammaticality judgments, subjects were generally slower to respond when the verb carried more markers. For verbs with no marker versus verbs with one marker, this extra cognitive effort yielded improved accuracy; however, this advantage did not extend to multiple exponence, as the extra processing time did not produce much improvement in accuracy. In cued recall, the presence of one marker conferred a clear advantage in accuracy, but the presence of two agreement markers actually resulted in decreased accuracy. Overall, multiple exponence was found not to confer a functional advantage.

Special semantics seminar organized by Prof. Angelika Kratzer

Hans Kamp and Chris Kennedy will be guest speakers in Angelika Kratzer's seminar. Chris Kennedy: Possible scales: Gradable modals vs. graded modality. Hans Kamp: "I" and "you" as vehicles for "de se" and "de re"

Dr. Franck Salameh is Assistant Professor of Near Eastern Studies and Slavic Studies at Boston College and author of Language, Memory, and Identity in the Middle East: The Case for Lebanon (Lexington Books, 2010). Interview with the author

Elie Kedourie--a pillar of modern Middle East Studies--once described his academic field as a “bore,” a narrative that claimed there to be no non-Arab “others” in the Middle East, and no cultures, languages, or histories beyond those of Arabs. This talk seeks to unpack these prevalent assumptions, parse the language and ideologies behind them, and demonstrate that despite its many religious, cultural, and linguistic similarities, the modern Middle East--like its ancient Near East precursor--lacks the requisite historical uniformity and continuity to warrant the reductive--and ultimately misleading--appellation “Arab world.”

Franck Salameh is Assistant Professor of Near Eastern Studies, Arabic, and Hebrew in the Department of Slavic and Eastern Languages and Literatures at Boston College. This talk is partly based on his new book, Language, Memory and Identity in the Middle East: The Case of Lebanon (Lexington Books, 2010).

This event, sponsored by the Department of Judaic and Near Eastern Studies and co-sponsored by the Departments of Linguistics and Political Science at the University of Massachusetts Amherst, is free, open to the public and wheelchair accessible.

In this talk, I discuss the two ways of expressing indefiniteness in Hausa, namely by means of bare NPs or the structurally more compelx wani-DPs. I argue that the two kinds of indefinite NPs involve different modes of semantic composition, i.e. RESTRICT (Chung & Ladusaw 2004) and choice functions plus existential closure at various levels. The Hausa data would thus seem provide evidence in favour of Reinhart's (1997) flexible choice function approach, as opposed to the more restricted choice function accounts in Kratzer (1998) and Matthewson (1999). The final part of the talk looks in more detail at how the Hausa data fare with respect to these alternative analyses and other criticism levelled against the flexible choice function approach by Schwarz (2001).

This presentation will focus on Spanish poder, a modal verb standardly understood as an existential modal. Spanish poder can target three types of ‘modal flavors’: epistemic, counterfactual, and ability. My objective in this presentation will be to explore a unified account of the three types, focusing on the case of ability readings. Bhatt (1999, 2006) notes that aspectual morphology affects the range of interpretations associated with modals. This has been taken up more recently by Hacquard (2006, 2010), who discusses in detail the case of French pouvoir. The key observation is that perfective aspect can give rise to ‘actuality entailments’ in ability readings. In my discussion of Spanish, I will note differences between Spanish and French and then proceed to provide an account of the Spanish case in terms of a modal framework that appeals to Kratzer-style situations (Kratzer 1989, 2009, etc.). Tense and aspect will play a crucial role in the proposal, anchoring the modal claim to facts in the actual world.

A special talk about Quechua language revitalization efforts, by Martin Castillo Collao, a Quechua speaker and local language activist and Susan Kalt, a professor at Roxbury Community College. They have been working together on a project sponsored by the Endangered Languages Fund, and will be talking about community participation and curriculum development. All are welcome.

Part of a lecture series organized by Prof. Angelika Kratzer

• • English non-deontic 'should' is always future-oriented.
• • 'Should' only allows a circumstantial modal base. There is no epistemic 'should'.
• • The circumstantial restriction on 'should' is stated in the lexicon, and the circumstantial restriction derives the future-orientation.

I present the first proposal for the internal subgrouping and higher order structure of the Pama-Nyungan family of Australian languages. Previous work has identified more than 25 primary subgroups in the family, with little indication of how these groups might fit together. Some work has assumed that reconstruction of higher nodes in the tree would be impossible, either because extensive internal borrowing has obscured more remote relations, or because relevant languages are not sufficiently well attested. Here I show that the Pama-Nyungan tree has considerable internal structure, and that language contact and missing data do not impede reconstruction unduly. This work shows the power of combining historical reconstruction with computational approaches to phylogenetic inference and provides an illustration of the way in which language can give us new insights into unsolved problems in prehistory.

Ever since Rizzi's (1997) paper on the fine structure of the left periphery it has often been taken for granted that an explanation of the order of elements in the left periphery should take the form of a fine-grained, stipulated hierarchy of dedicated functional heads. In this talk I suggest that the order of elements in the left periphery instead follows to a large extent directly from considerations of locality without the need to invoke a cartographic template for the left periphery. I claim that both empirical and explanatory gains can be made by this shift in perspective.

In this talk, I discuss the realisation of focus in Ngamo, which like many (West) African languages does not require the explicit marking of focus on non-subjects in terms of absolute prominence (pitch, movement, markers). I discuss various analyses of the realisation of focus in Ngamo and show that only two are compatible with the observable facts: (i) Analysis I assumes that focus in Ngamo is consistently marked on all constituents, but not in terms of absolute prominence, but in terms of alignment with major prosodic phrases(Féry , submitted); (ii) Analysis II assumes an asymmetry in the focus marking system of Ngamo in that only subject foci must be marked. In the final part of the talk, I show that the association behaviour of focus-particles (exclusive 'only' vs additive 'also') provides us with evidence in favour of analysis I.