Speaker Series
2007-2008
2006-2007
2005-2006
2008-2009
Dr. Katrin ErkDepartment of Linguistics, University of Texas- Austin
Tuesday, September 23, 2008 from 11:40am to 12:55pm
'Polysemy: Some corpus observations and computational models'
Determining the meaning of a polysemous word in a given context is a difficult task: It is difficult to do this automatically, and it is even difficult for human annotators to perform this task manually. It is possible that this is due to the underlying model of word meaning: Polysemy is usually framed as a list of dictionary senses, and the task as picking the one sense that is contextually appropriate. But there is evidence indicating that not all words may have clearly disjoint senses.
The first half of this talk describes two ongoing annotation tasks designed to determine the degree to which different words have disjoint senses. The second half of the talk discusses vector space models of word meaning, computational models of word meaning that does not take recourse to dictionary senses. Vector space models have successfully been used in language technology, especially, information retrieval, and cognitive science, and they can be induced automatically from corpora. However, they are mostly used to describe word meaning in isolation, rather than meaning in a specific context, and existing vector space models for meaning in context do not take syntactic structure sufficiently into account. We present a novel 'structured vector space model' that addresses these issues by incorporating the selectional preferences for words' argument positions. This makes it possible to integrate syntax into the computation of word meaning in context. In addition, the model performs at and above the state of the art for modeling the contextual adequacy of paraphrases.
Dr. John Williams
RCEAL, University of Cambridge
Tuesday, September 23, 2008 from 2:30pm to 3:30pm, ICC 462
'Sequence learning and incidental acquisition of word order regularities in natural language'
There is a long tradition of implicit learning research looking at learning of artificial grammars (finite state grammars that generate meaningless letter strings). Are the associative learning processes evident in these studies at work in learning word order in natural language? To what extent do meaning and prior linguistic knowledge need to be taken into account in the natural language case? In this talk I shall present a study in which incidental learning of natural language word order is compared directly to a meaningless analogue (in which the same regularities underlie meaningless syllable strings). The results of both experiments are compared to connectionist (simple recurrent network) simulations. The comparisons suggest that similar associative sequence learning mechanisms underlie learning of both the natural language and its meaningless analogue (with the result that there are certain limitations to what is learned). However, to achieve this alignment it is necessary that we take into account the linguistic categories and meaning structure that the participants are likely to impose on the natural language. I conclude that the initial incidental learning of word order can be explained in terms of associative (sequence) learning, and that linguistic knowledge is engaged to the extent that it defines the categories over which statistics are computed.
Dr. Veneeta Dayal
Department of Linguistics, Rutgers University
Friday, September 26, 2008 from 2:30pm to 3:45pm, ICC 450
'Free Choice and Indeterminacy'
I consider the range of FCIs in English: 'Any', 'Wh-ever' (Free Relatives) and 'Some N or other'. I claim that they each have a multi-dimensional meaning, sharing their truth conditional contribution with standard one-dimensional quantifiers like every, plain free relatives and some but encoding on top of that, a requirement of indeterminacy. I formalize the indeterminacy of 'any' as a grammatical constraint against there being a single set of individuals that have the relevant property in every relevant world. FC 'any' is predicted to be unacceptable in precisely those cases where the truth conditional meaning contradicts the implicature of indeterminacy. The formalization allows for a uniform and compositional explanation for two particularly recalcitrant problems in the semantics of FC 'any', the subtrigging effect and the partitive puzzle. It also accounts in a straightforward way for the propensity that FC any has been observed to have for taking wide scope over modals and negation. The analysis takes note of supplementary/numeral 'any' which display somewhat different behavior from regular FC any. In contrast to the indeterminacy requirement of 'any', the indeterminacy requirement of 'Wh-ever' and 'some N or other' is weaker, merely implying lack of knowledge on the part of the speaker about the set of individuals or implying lack of relevance of their identity. Such indeterminacy is compatible with there being no single set of individuals across all relevant worlds (as is the case with 'any') or with the existence of such a set.
Speaker: Dr. Barry Schein
Department of Linguistics, University of Southern California
Friday, October 17, 2008 from 2:30pm to 3:45pm, ICC 450
Talk Title: Simple Clauses: Conjoined
Dr. Jacqueline Messing
Kislak Fellow, Kluge Center for Scholars, Library of Congress
Assistant Professor, Department of Anthropology, University of South Florida
jmessing@cas.usf.edu
Tuesday, October 28, 2008 from 11:40am to 12:55pm, ICC 425
Being “Indian” in a Modern World:
Identity & Narrative in Colonial & Contemporary Tlaxcala, Mexico
Each year, indigenous languages around the world disappear with the death of their last living speaker. It is imperative for the social sciences and humanities to explain how and why people come to abandon their ancestral languages, as well as the role of colonialism, globalization and racism in this process. In my presentation I will share some ideas and data from a book-length study in progress that offers a new perspective on language shift, through an anthropological study of social change in two Central Mexico indigenous communities within the historically unique state of Tlaxcala. Using a linguistic-anthropological and ethno-historical approach, I analyze ideologies of language, identity and “progress” in three competing discourses emergent in everyday conversation andhistorical documents. My approach involves comparing contemporary discourses (taped during ethnographic research) to historical ones analyzed in Nahuatl court testimonies, wills and testaments. These first-hand narratives from the period of sixteenth century conquest in Tlaxcala demonstrate how “Indian” and “Colonial” identities were created and contested, to culminate in drastic social and linguistic change in the late twentieth century.
Dr. Cleo Condoravdi
Time: 2:30-3:45, Friday, Oct.31, ICC 462
Computing Textual Inferences
A measure of understanding a text is the ability to make inferences based on the information conveyed by it. Given a passage of text and a hypothesis, the task would be to automatically infer whether the hypothesis follows from the text, whether it is contradicted by it, or whether it is compatible with it. At PARC we have been working on a system for computing linguistically-based textual inferences such as the ones below.
Passage: Ed has been living in Athens for 3 years.
Mary visited Athens in the last 2 years.
Hypothesis: Mary visited Athens while Ed lived in Athens.
Answer: YES
Passage: The diplomat does not know that the president failed to
destroy the evidence.
Hypothesis: The president managed to destroy the evidence.
Answer: NO
Passage: No one stayed throughout the concert.
Hypothesis: No one stayed throughout the first part of the concert.
Answer: UNKNOWN
Texts are parsed to produce packed functional-structures and these are
rewritten and canonicalized, without unpacking, into abstract knowledge representations (AKR). An AKR representation is a flat set of facts that involves concepts, roles, temporal relations and contexts. In this talk I show how AKRs are derived from parsed text and discuss the system's algorithm for entailment and contradiction detection (ECD). ECD operates on the AKRs of the passage and of the hypothesis in order to detect a potential entailment or contradiction between them, without the need for disambiguation.
Dr. Michele Koven
University of Illinois at Urbana-Champaign
Tuesday, November 11, 2008 from 11:40am to 12:55pm, ICC 425
Performance and Experience of Identities in Two Languages:
How French-Portuguese bilinguals display and manage language-specific self-presentations in narrative and second-person address.
Bilinguals often report that they feel like a different person in
their two languages. In the words of one participant, 'when I speak
Portuguese, automatically, I'm in a different world…it's a different
color.' This talk provides an empirically grounded, theoretical account of how the same speakers perform, experience, and are perceived by others to take on different personas in their two languages. I do this by showing how the bilingual daughters of Portuguese migrants raised in France use their two languages in ways that establish them as different locally recognizable 'types' of people.
I discuss two sites where participants display and experience different types of personas in French and Portuguese: strategies of self-presentation in narrative, and use of non-equivalent forms of second-person address. In both cases, bilinguals wrestle with multiple frameworks for interpreting their available ways of talking. Their challenges to 'translate' such forms and their awareness of the challenges of translation raise questions of how transnational bilinguals can transport –'translate' - values and identities across the multiple interpretive frameworks of their distinct monolingual contexts.
2007-2008
Dr. Juan Manuel Hernández-Campoy and Dr. Juan Antonio Cutillas-Espino
Tuesday, September 18, 2007 from 11:40am to 12:55pm, ICC 450
Performative Approaches to Style-Shifting: Speaker External Factors
The study of style within the variationist tradition in sociolinguistics has traditionally received little attention in general terms. Some of the main introductory textbooks dealing with this discipline hardly mention style as a variable; and when they do, they usually understand style as a reflection of the speaker’s attention to his/her own speech. Increasingly, researchers have moved from viewing stylistic variation as a primarily reactive phenomenon, conditioned by matters external to speakers such as formality (Labov 1972) and audience (Bell 1984), to more proactive approaches, in which speakers use stylistic resources to project and create identity, as well as to accomplish conversational and longer-term goals (e.g. Coupland 2001a, 2001b, forthcoming; Eckert 2000; Schilling-Estes 2002). In other words, the Audience Design model conceives stylistic variation primarily as the result of adaptation to the features of a present or absent audience, whereas so-called Speaker Design approaches (e.g. Coupland 1996, Schilling-Estes 2002) look at stylistic variation as a process of identity building. In the present paper, we extend this constructivist approach to style shifting by demonstrating that even in seemingly highly constrained stylistic contexts, namely publicly broadcast radio or political speech, people make personal, strategic, and sometimes quite surprising, stylistic choices.
In the first of our two studies, we analyse the speech of a radio presenter in a local station in Murcia and compare it to the audience’s linguistic behaviour shown in phone calls received during the programme. We also analyse the data obtained in an interview with the radio presenter. Our results, which show a radical divergence between the presenter’s speech and that of his audience, are contrasted with both Audience Design and Speaker Design theoretical tenets, using the explicit knowledge of the presenter’s attitudes and opinions to contrast theory and fact. We conclude that neither model offers a completely satisfactory explanation of the patterns found. Finally, we reflect on the need to consider not only performance, but also the script (in the form of sociolinguistic norms and attitudes to language) that underlies the individual linguistic behaviour, thus suggesting the need to consider community-specific factors in the explanation of stylistic variation.
The second piece of research focuses on the unexpected (and controversial) use of many features of the local dialect by a female former President of the Local Government of Murcia, in southeastern Spain. The Murcian dialect is stigmatized within Spain but also carries covert prestige for Murcians as a marker of local identity and solidarity. The comparison of the President’s broadcast speech with that of other local speakers shows, surprisingly, that she has higher usage levels for dialect features than any of the other groups. Her hyper-use of Murcian dialect features indicates that she is not shifting her speech in reaction to formality, or even in accommodation to the many Murcians in her audience. Rather, she is purposely designing her speech to project an image that highlights her Murcian identity and her socialist ideals. The fact the even prominent politicians use stylistic resources in ways that are most fully explicated by appealing to speaker-internal as well as speaker-external, situational factors lends further support for viewing style as a matter of Speaker Design as well as Audience Design.
Dr. Juan Manuel Hernández-Campoy and Dr. Juan Antonio Cutillas-Espino
Thursday, September 20, 2007 from 11:40am to 12:55pm, ICC 450
Murcian Spanish: Diachronic and Synchronic Perspectives
In any process of linguistic standardisation, the promotion of one variety to the status of standard triggers the devaluation of the other linguistic varieties present within the boundaries of the nation-state and impinges upon their domains. Diachronically speaking, this process is a constant struggle between the standard and the non-standard varieties either to reach uniformity and invariance, or to avoid compliance and maintain local values and customs, always under the pressures of prestige of different kinds. The present study examines aspects of the dialect contact maintained between the standard Castilian Spanish and the non-standard variety of Murcia, in southeastern Spain. Such long-term contact situations most often result in dialect obsolescence together with standardisation and even levelling; conversely and less frequently, they may result in survival processes of dialect maintenance. Conclusions are reached from results obtained in two pieces of research: i) a longitudinal study of Murcian speech carried out in order to detect and measure the apparently increasing spread of standard Castilian features from northern Peninsular Spanish over Murcian Spanish, as well as its geolinguistic patterns of diffusion; ii) a cross-sectional study carried out in a local community considering socio-demographic factors such as gender and age.
Dr. Tonia Bleam
Thursday, October 4, 2007 from 10:15am to 11:30am, ICC 450
On the Status of Null Determiners Or: The Role of Information Structure in determining the distribution of bare nominals in Spanish
It has long been observed that the distribution of bare nominals in Romance languages such as Spanish and Italian differs from their distribution in English. Spanish and Italian bare nominals are much more restricted in their distribution, being barred from canonical pre-verbal subject position. Previous accounts of these facts have put the locus of variation in the presence or absence of a null determiner: Spanish and Italian bare nominals contain a null determiner which is subject to a syntactic constraint, whereas English bare nominals contain no such null determiner.
In this talk, I show that principles of information structure are crucial to accounting for the distribution and interpretation of bare nominals in Spanish, and that once we adopt these principles (which we need independently), the Null Determiner Hypothesis no longer does any explanatory work for us.
A second goal of this paper is to demonstrate the need for flexible composition in the semantics. In particular it will be argued that property-denoting nominals can appear in argument positions in the syntax, and that we thus cannot adhere to the neo-Montagovian position that function application is the only means of semantic composition. However, the Montagovian idea that there is a strict mapping between syntactic category and semantic type is maintained.
Dr. Draga Zec
Tuesday, October 23, 2007 from 11:40am to 12:55pm, ICC 450
Types of interactions in prominence based prosodic systems
According to Ivic 1958, the Stokavian dialect group, which includes Serbian and Croatian regional idioms, diverges into a range of pitch accent systems. The goal of this paper is to capture both the similarities and differences across these systems in terms of different rankings of a small set of OT constraints. The constraints will make reference to tone and stress, the two crucial components of the Stokavian pitch accent systems. There are several properties that all these systems have in common. First, the pitch accented syllable bears stress, and is also linked to a H(igh) tone (Lehiste and Ivic 1986, Inkelas and Zec 1988). Second, tone is associated with lexical forms, and serves as a basis for characterizing lexical classes. The lexical H may occur on any syllable in the word, and its locus influences the place of stress. The differences among the systems, while minimal, have the cumulative effect of giving rise to the prosodic partitioning of the Stokavian dialect group across several dimensions. The resulting Stokavian pitch accent systems crucially diverge with regard to the prosodic influences on the place of stress: those of the tonal variety, such as the locus of the lexical H, as well as those directly associated with stress systems, such as syllable weight effects, and edge orientation. In the proposed analysis, this dialectal divergence is expressed as a factorial typology of constraints on tone and stress. The resulting factorial typology exhibits a close fit with Ivic’s description of the dialectal variation.
Dr. Valentine Hacquard
Wednesday, October 31, 2007 from 4:15am to 6:45am, ICC 462
Various types of modality are expressed, cross-linguistically using the same modal words. In influential work, Kratzer (1981, 1991) proposed a unified semantics for modals, where the type of modality involved is contextually determined. A long standing problem for such a proposal is that, while the same lexical item can express 'epistemic' or 'root' modality, there is a systematic correlation across languages between the type of modality a modal can have and the position of that modal with respect to other functional elements: epistemics tend to scope high, while roots scope low (cf. Brennan 1993, Cinque 1999, a.o.). In this this talk, I explore a semantically motivated alternative explanation for this ordering, which does not stipulate different relative syntactic positions for different types of modals, but still maintains a unified semantics for modals in the spirit of Kratzer.
2006-2007
Dr. Jeffrey LidzTuesday, September 26, 2006 from 1:15pm to 2:30pm, ICC 462
Competence and Performance in the Acquisition of Quantification
Whereas adults can assign either scopal interpretation to the quantificational NP (QNP) in sentences containing QNPs and negation like (1), 4-year-olds show a massive preference for surface scope (2) over inverse scope (3). (Musolino, Crain and Thornton 2000, Lidz and Musolino 2002, Musolino & Lidz 2006, Lidz & Musolino 2006). 1) Every horse didn't jump over the fence 2) Every horse is such that it didn't jump over the fence (i.e., none did) 3) Not every horse jumped over the fence (i.e., some did and some didn't) In this talk we investigate the factors (prosodic, syntactic, pragmatic, parsing) responsible for children's overly narrow interpretations. We show that while children do have adult-like syntactic representations of such sentences, deficiencies in the domains of pragmatics and sentence parsing lead to their failure to access inverse scope interpretations.
Drs. Sjef Barbiers, John Beavers and Jan Pieter Kunst
Friday, November 10, 2006 from 10:15am to 5:00pm
First Workshop on Dialect Syntax
ICC 425, McCarthy Room
10:15 Raffaella Zanuttini, Georgetown University
Introductory remarks
10:30 Sjef Barbiers, Meertens Instituut (Amsterdam)
'Impossible and unrealized syntactic structures'
11:30 John Beavers, Georgetown University
'Expressive capability trumps syntax: A case study in colloquial English pronominals'
ICC 450
2:15 Jan Pieter Kunst, Meertens Instituut (Amsterdam)
'Technical aspects of dialect syntax research: The Syntactic Atlas of Dutch dialects'
3:15-3:30 Coffee Break
3:30-5:00 Discussion continues
Complete abstracts may be found here: http://linguistics.georgetown.edu/14467.html
This is the first workshop on dialect syntax hosted at Georgetown; it is sponsored by the Linguistics Department with help from the Meertens Instituut (Amsterdam). The day will be devoted to a series of discussions on the study of the syntax of non-standard varieties, or dialects. Each talk will last 45 minutes, followed by a question/answer period.
The first talk will be by Sjef Barbiers, from the Meertens Instituut, project manager of the Syntactic Atlas of the Dutch Dialects (http://www.meertens.knaw.nl/projecten/sand/sandeng.html), and now also of the European Dialect Syntax project sponsored by the European Science Foundation. He will briefly illustrate these projects and then present some of the empirical and theoretical findings that have emerged from the work on the syntactic atlas of Dutch dialects.
The second talk will be by John Beavers, visiting Assistant Professor in the Linguistics Department at Georgetown (http://www9.georgetown.edu/faculty/jtb44/index.html). He will present some research he's been doing (in collaboration with Andrew Koontz-Garboden) on a syntactic property that appears in many colloquial varieties of English, including his own Texas English.
The third talk will be by Jan Pieter Kunst, the software developer for the Syntactic Atlas of the Dutch Dialects project; he will discuss issues related to the design and implementation of the database that contains the empirical findings of the research on the syntax of Dutch dialects.
Dr. Silvina Montrul
Monday, November 20, 2006 from 4:15pm to 5:45pm
Incomplete Acquisition in Adult Bilingualism
In this talk, I address the morphosyntactic development of two types of adult bilinguals with low-intermediate proficiency in Spanish (i.e., incomplete learners)—L2 learners and heritage speakers—who differ with respect of age of onset of acquisition of Spanish. Heritage speakers received input at home in early childhood whereas L2 learners start acquisition after puberty. Comparison of the linguistic behavior of these two types of learners is crucial to inform classic—and ongoing—debates in SLA theory on the one hand, as well as the emerging field of heritage language acquisition, on the other. For example, some linguistic and cognitive theories of SLA maintain that the acquisition of an L2 by adults and the acquisition of an L1 by children differ in fundamental ways, invoking maturational constraints. A critical/sensitive period would explain why the outcome of L1 acquisition is complete and successful while that of L2 acquisition typically is not (DeKeyser 2003; Long 2000). Linguistic theories of L2 competence maintaining that adult L2 learners differ from monolingual and bilingual children with respect to access to Universal Grammar (Bley-Vroman 1989; Clahsen & Muysken 1989; Meisel 1997) predict that bilinguals who acquired two languages in childhood should show evidence of early parameter setting in their two languages, whereas postpuberty L2 learners should be unable to reset parameters in the L2. By contrast, for the Full Access position (White 2003), parameter resetting by adult L2 learners is possible regardless of age of acquisition. To address these hypotheses, I will discuss results of ongoing experiments on three aspects of object expression in Spanish: object clitic placement, clitic left dislocations, and differential object marking. Object clitics and their correct placement are acquired around the age of 2 by monolingual Spanish children, and so is differential object marking, while clitic left dislocations, being part of the left-periphery, emerge by the age of 3. It follows that heritage speakers who acquired and used Spanish early in childhood should have solid knowledge of clitics and word order, even if English is now their stronger language. By contrast, if L2 learners have no access to Universal Grammar, and cannot transfer clitic projections or differential object marking from English, they should be unable to acquire these structures and clitic left dislocations in Spanish. Results of 34 L2 learners, 34 proficiency-matched heritage speakers and a control group of 22 Spanish native speakers on two off-line grammaticality judgment tasks and a on-line processing task showed selective advantages for heritage speakers, suggesting that age of acquisition alone may not the best predictor for these findings. I discuss implications for theories of SLA and linguistic theory.
Dr. Colleen Cotter
Monday, November 27, 2006 from 3:30pm to 5:30pm
Professor Colleen Cotter (former journalist, GU Faculty member, and current Linguistics Professor at Queen Mary College, London) will be talking about her experiences in journalism, as a linguist, and how to combine the two.
Dr. Jenefer Philp
Tuesday, December 5, 2006 from 3:00pm to 4:00pm
Child's play? The role of maturity and context in second language acquisition'
This paper considers the distinctive features of child second language acquisition, as outlined through previous research. Some of these features are aspects of the nature of childhood: maturational factors associated with cognitive and social development. These change in nature through the different stages of childhood. Other features are associated with the particular social contexts of children, and their roles and experiences within these settings. This chapter reflects on the ways in which the nature of language learning differs for each of these contexts; for example, the interactions of home differ from those of school; and within each context, for example, language learning in foreign language differs from second language instructional settings. The chapter concludes with a discussion of differences between adult and child L2acquisition and problems associated with the application of findings of research on adult L2 acquisition to child L2 learning.
Dr. Adele Goldberg
Wednesday, February 28, 2007 at 5:45pm
Constructions at Work: The Nature of Generalization in Language
We clearly retain a great deal of specific information about how individual lexical items can be used. At the same time, it would be a mistake to conclude that speakers do not form generalizations, that such generalizations are merely epiphenomenal. Far from being an arbitrary collection of stipulated descriptions, our knowledge of linguistic constructions, like our knowledge generally, forms an integrated and motivated network. The constructionist approach to grammar allows both broad generalizations and more limited patterns to be analyzed and accounted for fully. In particular, constructionist approaches are generally usage-based: facts about the actual use of linguistic expressions such as frequencies and individual patterns that are fully compositional are recorded alongside more traditional linguistic generalizations (Langacker 1988). Instances are represented at some level of abstraction due to selective encoding, and generalizations over instances are made as well. Insights gained from research in general categorization can shed light on how learners go from the specific to the more general. My presentation will focus on several such factors that promote and constrain generalizations: a) skewed input, b) degree of coverage and c) statistical preemption. Profile A prominent cognitive linguist, Dr. Adele Goldberg is one of the principal founders of (Cognitive) Construction Grammar approaches to language. Her most recent book, Constructions at Work: The Nature of Generalization in Language, presents compelling evidence in favor of a 'usage-based' model of language acquisition that incorporates both item-based knowledge and broad generalizations manifested in learning grammatical constructions--conventionalized parings of form and semantic/discourse function. After receiving her Ph.D. at UC Berkeley in 1992, Dr. Goldberg taught at UC San Diego and the University of Illinois, Urbana-Champaign, before moving to Princeton in 2004. She was honored to be a fellow at the Center for Advanced Study in the Behavioral Sciences (Stanford, CA) and won the Gustave O. Arlt Book Award in the Humanities for her first book, Constructions: A Construction Grammar Approach to Argument Structure. Her research focus is on the psychology of language, including theoretical and experimental aspects of grammar and its representation, acquisition of form-function correspondences, and syntactic priming. http://www.princeton.edu/~adele/
Dr. Christine Mallinson
Thursday, March 1, 2007 from 11:40am to 12:55pm
'Katrina That Bitch!': Hegemonic Representations of Women's Sexuality on Hurricane Katrina Souvenir T-Shirts
This study examines hegemonic representations of women's sexuality on Hurricane Katrina souvenir t-shirts displayed in New Orleans, Louisiana, in March 2006. Previous cultural research has conceptualized the t-shirt as an 'open text,' (Crane 2000) upon which messages are a type of 'personal graffiti' (Heeren 1980) that reveal attitudes and norms of the individual and associated groups. Adopting this approach, this paper analyzes 25 slogans that depict a sexually aggressive and destructive 'Katrina' to the public. In the analysis, four linguistic strategies are prominent. The use of sexual slang (Schultz 1973, Sutton 1995), combined with the use of the active voice to attribute sex acts to 'female' (but not 'male') hurricanes, constructs a gendered Hurricane Katrina as aggressive and promiscuous. At the same time, the use of expletives helps cast the presumed wearer or 'voice' of the t-shirts as being that of a man (de Klerk 1992, 1997; Coates 1993; Sutton 1995; Sapolsky and Kaye 2005), thereby demonstrating the invisibility of woman as both subject and victim. Finally, the jokes made as part of the Hurricane Katrina slogans are shown to parallel other joke cycles in the genre of disaster humor (Smyth 1986, Oring 1987), and results from brief interviews with 26 passersby on Decatur Street in New Orleans support this claim. Results from this study thus speak to the construction of hegemonic discourses about gender and sexuality via media messages that normalize women's sexual degradation, establish the heterosexual man as the universal subject and victim, and engage linguistic 'therapies' like jokes to cope with uncertainty, including disasters such as hurricanes.
Dr. Andrew Salway
Thursday, March 15, 2007 from 11:40am to 12:55pm
Contributions of a Computational Linguistic Approach to Narrative Analysis
Recent developments in narratology have shifted interest onto the kinds of information that are required to understand a story. In this context, this talk will consider how a computational linguistic approach can contribute to narrative analysis, both in the development of narrative theory, and in the development of software tools that extract story-related information from narrative texts. The talk will concentrate on the example of audio description which is emerging as an intriguing new text type with social import. Audio description is a spoken account of on-screen action that accompanies an increasing number of films – in cinemas and on DVD releases – for the benefit of blind and visually impaired audiences; in the USA it is also referred to as 'described video'. In effect, the part of the story told by the moving image is re-told in words which are integrated with the film's dialogue. Audio description refers to a restricted domain, i.e. whatever is depicted in films, and it is scripted and recorded by trained professionals following style guidelines. The work presented in this talk considers audio description as a sublanguage and proceeds to investigate the kinds of information about a film's story that it provides. The approach taken is an automated analysis to identify unusually frequent and characteristic word sequences and paradigms in a corpus of audio description scripts, with reference to a general language corpus. On the basis of this formal analysis, common kinds of information required for story understanding are suggested, and templates and heuristics for an information extraction system are defined. The findings from this work, and from similar analyses of other types of narrative texts, will be discussed with respect to the development of narrative theory using corpus analysis methods. We will also outline some computer applications for narrative-based film retrieval and browsing, and for assisting in the production of audio description.
Dr. Maria Polinsky
Thursday, March 22, 2007 from 11:40am to 12:55pm
A-Complementous Contemplation
This talk addresses the following general question: What is the range of options available for expressing an embedded proposition and how are these options constrained by the grammar? We argue that there are languages like Adyghe (NW Caucasian) that lack complement clauses and make use of a relativization strategy to convey propositional content. What would be a complement clause in English (bracketed string in (1)) is replaced in Adyghe by what descriptively appears to be a complex DP with overt case and tense marking (bracketed string in (2)). (1) I told you [that the boy had gone to school]. (2) [B’aler SkolEm zEre-KWe-Re]-r we-S-{Wa-R boy to_school EMBEDDED-go-PAST-ABS 2SG.OBL-1SG.ERG-say-PAST ‘I told you that the boy had gone to school.’ We propose an analysis of such complex DPs as containing relative clauses. In particular, we argue that (a) the overt case marker takes a complex NP with a null head combining with a relative clause; (b) the relative clause contains a silent operator and trace (the head raising analysis is shown to be untenable). We present empirical arguments in support of both of these analytical components and against alternative analyses. Our analysis also accounts for the ban on sluicing in Adyghe and, since Adyghe allows for only relative complementation, lends further empirical support to the notion that relative and non-relative complementation are different (Rizzi 1990; Lasnik and Saito 1992). In addition to expressing embedded propositions, Adyghe also uses complex DPs with a relative clause attached to express embedded questions (these will also be discussed in the talk). On the semantic side, we tentatively suggest that Adyghe complex DPs with a null head and a relative clause have the same denotations as complex DPs with a complement clause in English and other familiar languages. Thus, a more literal translation of (2) above would be I told you [the fact that the boy had gone to school]. However, in Adyghe the relationship between the null head and the embedded clause is one of modification, not complementation. If this analysis is on the right track, the range of DPs available for expressing propositions is broader than is usually thought: it includes regular nominalizations, complex DPs with complement clauses and complex DPs with relative clauses attached. We conclude with a tentative proposal on the ways to constrain parametric variation between English-type languages and Adyghe-type languages.
Dr. Zoltan Kovesces
Monday, March 26, from 1:15 to 2:30
Universality and Cultural Variation in Metaphor
Cognitive linguists have so far paid a great deal of attention to the remarkable universality of many conceptual metaphors. However, their theories fail to account for the equally impressive diversity of metaphorical conceptualization both across and within cultures. This talk presents an attempt to lay down the foundations of a theory of metaphor that is capable of simultaneously accounting for both universality and variation in metaphor.
Handout in PDF
Dr. Ron Butters
Wednesday, April 11, 2007 from 4:00pm to 5:15pm
The discourse of operatives working to catch sexual predators in IM messages
This presentation builds upon established discourse analysis models and methodology (in the work of, e.g, Roger Shuy and Malcolm Coulthard) for analyzing putatively incriminating linguistic evidence, in this case Instant Message exchanges between (a) unsuspecting adult men of various ages (b) adult males who were pretending to be minors. The cases, many of which led to felony sexual enticement and solicitation charges and convictions, are among hundreds of 'mark-and-decoy' IM conversations instigated by agents of Perverted Justice, a vigilante organization dedicated to the exposure, arrest, and conviction of 'sexual predators'. 'Dateline', the NBC news magazine, has paid Perverted Justice hundreds of thousands of dollars for the use of the material collected by the online decoys; 'Dateline' reporters have then conducted surprise 'interviews' upon some of the marks as they arrived at what they were led to believe were the homes of the 'youths'. The web site Perverted-Justice.com publishes transcripts of over 150 such IM exchanges.
Anyone who has seen the 'Dateline' broadcasts may wonder what sort of defense could be possible in the face of such overwhelming-seeming evidence of sexual crimes against underage youths. There may be no disputing the legitimacy of the convictions in the cases under analysis here (in which the marks were typically found guilty and sentenced to as much as six years in federal prison). Even so, linguistic interpretation of the of the IMs and the 'Dateline' broadcast--none of which was presented to any of the courts--might well have altered the outcome of at least some of the cases.
Dr. Jason Kandybowicz
Thursday, April 12, 2007 from 11:40am to 12:55pm
Historically, the notion of EDGE as a grammatically sensitive domain has received strong support from processes and interactions occurring in both the morphological and phonological wings of grammar. Recently, edges have come to play a prominent role in syntactic analysis as well. As with any theoretical innovation or paradigm shift, two crucial issues are raised. First, how principled is the innovation? Second, what advantages does the new conceptualization offer over existing accounts? This talk addresses the second issue by way of two case studies of edge sensitivity in Nupe, a Benue-Congo language spoken in central Nigeria. We show that reference to edges in both the narrow syntax and at the syntax-phonology interface provides principled explanations of two long-standing unsolved puzzles in the Nupe literature. The argument is thus that a syntactic theory that embraces edges is indeed desirable from an analytical standpoint in so far as it provides for satisfactory formulations of anomalous phenomena that were previously unavailable.
The first case study showcases narrow syntactic edge sensitivity. A puzzle perennially observed in the Nupe literature is that extraction from tensed clauses is possible, but extraction from perfect clauses is not (Smith 1967, Kandybowicz & Baker 2003).
(1) a. Ke Musa è/à pa __ o?
[Present/Future TP]
what Musa PRS/FUT pound o
‘What is Musa pounding?’/‘What will Musa pound?’
b. *Ke Musa á pa __ o?
[Perfect TP]
what Musa PRF pound o
‘What has Musa pounded?’
We argue that the existence of EDGE FEATURES (Chomsky 2005) allows for an elegant solution to this empirical problem. At the same time, we show that the problem of Nupe perfect extraction sheds light on the very nature of the edge features that are borne by strong phase heads, i.e. those features responsible for driving cyclic movement to phase edge positions in compliance with the PHASE IMPENETRABILITY CONDITION (Chomsky 2001). Contra Chomsky (2005), we argue that edge features are not inherent properties of strong phase heads (as least for the v phase), but are rather derivative properties inherited by way of edge features already present on the lexical verb which are transmitted to the phase head via head movement.
The second case study showcases sensitivity to the edge at the syntax-phonology interface. A-bar movement in Nupe shows clear Comp-trace effects.
(2) *Zèé Musa gàn [gànán __ nì enyà] o?
who Musa say COMP beat drum o
‘Who did Musa say beat the drum?’
The puzzle in this case is that a seemingly unrelated range of options exist in the language for averting Comp-trace effects. These options include the following: phonological reduction of C0; existence of TP-adjoined adverbials; resumption of the displaced occurrence; and spelling out of tense markers. We show that Nupe Comp-trace effects reduce to a violation of the INTONATIONAL PHRASE EDGE GENERALIZATION (An 2006), which requires that the mapping from syntax to phonology result in an output in which the edge of every intonational phrase is phonetically marked. The seemingly disparate cases of Comp-trace resolution previously mentioned follow naturally as a consequence.
Dr. Trude Heift
Tuesday, April 17, from 11:40 to 12:55
Learner Responses to Corrective Feedback for Spelling Errors in CALL
Communicating linguistic errors to foreign language learners has been subject to fairly long-standing controversies over whether the provision of corrective feedback is necessary or helpful for L2 development. Ferris (2004), in reviewing the grammar correction debate in L2 writing suggests that longitudinal studies are needed to provide evidence for the efficiency of error correction over time. Russell Valezy and Spada (2006) also conclude that too little empirical research has been done on the effectiveness of corrective feedback but that the evidence so far supports the assumption that corrective feedback does work. However, many variables are involved and assessing their impact only adds to the puzzle.
This presentation focuses on error correction and corrective feedback in a Computer-Assisted Language Learning (CALL) environment. More specifically, it describes a study that investigates learner responses to three distinct types of corrective feedback for misspellings produced by English learners of German. 28 beginner and intermediate students used The E-Tutor, an online parser-based CALL system for German that recorded student interaction with the software over 15 weeks. The study considered a corpus of 1268 misspellings for which one of the feedback types provided correction suggestions for the misspelling in addition to highlighting the error in the learner’s input. Study results indicate that, while the number of correct responses was significantly higher when the system provided a correction list, there was also significantly less learner uptake for the feedback type that did not provide any correction suggestions. Moreover, learners were far more successful in submitting the target word if it appeared among the correction suggestions. Finally, the order in which the words appear in the suggestion list seems to be a determining factor for students favouring one word over another.
This study supports previous findings (Heift. 2004, 2006) that suggest that there are significant differences in the way students respond to different types of corrective feedback in a CALL environment although this study did not find any significant differences with respect to gender and proficiency level. With the ultimate goal of understanding how corrective feedback intersects with learners’ working styles, satisfaction and success in CALL, this presentation will suggest areas for future development of error correction and corrective feedback for CALL.
References
Ferris, D. R. (2004). The "Grammar Correction" Debate in L2 Writing: Where Are We, and Where Do We Go from Here? (and What Do We Do in the Meantime ...?). Journal of Second Language Writing, 13, 49-62.
Heift, T. (2004). Corrective Feedback and Learner Uptake in CALL. ReCALL, 15(2), 416-431.
Heift, T. (2006). Context-sensitive Help in CALL. Computer Assisted Language Learning, 19(2+3), 243-259.
Russell Valezy, J., & Spada, N. (2006). The Effectiveness of Corrective Feedback for Second Language Acquisition: A Meta-Analysis of the Research. In J. M. Norris & L. Ortega (Eds.), Synthesizing Research on Language Learning and Teaching (pp. 133-164). Amsterdam/ Philadelphia: Benjamins.
Dr. Michael Israel
Wednesday, April 18, from 5:45 to 8:10
Putting It All Together: Analogical Processes in the Acquisition of Argument Structure
Recent work on syntactic development has made the case that children's knowledge of abstract grammatical patterns emerges gradually from their experience with a rich assortment of item-based constructions. (Tomasello 2000, 2003; Goldberg 2006).
Most of the evidence for this view has come from observations of what young children do not do with language\either from experiments showing that very young children lack some of the productivity typical of older speakers' language use (Abott-Smith et al. 2001; Akhtar & Tomasello 1997), or from corpus studies suggesting that young children's complex utterances may be produced by relatively simple processes of merging or combining previously experienced utterances (Lieven et al. 2003; Dabrowska & Lieven 2005).
In this talk I seek to bolster the case for a usage-based theory of language acquisition by examining some of the ways that children actually do use language creatively. In particular, I will argue that the types of errors which children make with argument structure constructions reflect roughly three stages of development consistent with the predictions of a usage-based grammar. In the earliest stage, children's errors include strange word orders and patterns of combination, which suggests that that children have not yet acquired any entrenched patterns on which to model their utterances. Later, once children begin to use constructions more productively, their utterances tend to be organized around one or a few item specific constructions. Finally, as children attain some critical mass of item-specific constructions in their fourth year, novel utterances emerge which reflect low level analogies across highly entrenched formulae. One consequence of this is that there are striking family resemblances in the idiosyncratic creative uses of different children.
I will argue that these results provide strong support for the usage-based view that abstract grammatical knowledge may be acquired in much the same way that other conceptual categories are learned. At the same time, however, I will argue that the developmental patterns observed here suggest that children rely on certain sorts of abstract semantic categories and simple syntactic abilities from a very early age.
Profile
Dr. Michael Israel has published extensively on issues of motivated form-function relationships and first language acquisition in such journals as Cognitive Linguistics, Journal of Pragmatics, Berkeley Linguistics Society, and Linguistics and Philosophy , as well as in numerous edited books. His particular areas of research center on the phenomenon of polarity sensitivity and the study of child language acquisition from a functional perspective. His recent and forthcoming papers include: "Mental spaces and mental verbs in early child English" (to appear in Tyler, Kim & Takada (eds.), Language in the Context of Use: Usage-based Approaches to Language and Language Learning ), "Saying less and meaning less" (in Birner & Ward (eds.), Drawing the Boundaries of Meaning , 2006), and "Common sense and 'literal meaning'" (in Coulson & Lewandowska-Tomaszczyk (eds.), The Literal and Non-literal in Language and Thought , 2005).
Dr. Israel received his Ph.D. from the University of California, San Diego in 1998, with his dissertation, The Rhetoric of Grammar: Scalar Reasoning and Polarity Sensitivity . Before joining the Department of English at the University of Maryland, he served as director of the ReVerb project at the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany, from 1999 to 2001. His first book, The Least Bits of Grammar: Pragmatics, Polarity, and the Logic of Scales, is forthcoming later this year from Cambridge University Press.
2005-2006
Dr. Frederick Ericson
Friday, September 30, 2005, 2:00 pm, ICC 462
Musicality in Classroom Talk and Listening: Possible Consequences for Cognition and Learning
Speech prosody-- tempo, pitch, and volume emphasis in the real-time performance of talk-- togther with changes in voice quality, provide auditors with cues to new or repeated information, and to shifts in emotional key, thus signaling crucial "next" moments in discourse at which special attention needs to be paid by auditors. The cues are implicit-- an aspect of the "contextualizing cueing" identified by Cumperz. The workings of this cueing system will be discussed and illustated through video examples of classroom talk, with particular empasis on analyzing the musicality of speaking in relation to issues of cognition and learning.
Dr. Edward Smith
Thursday, October 6, 5:00 pm, New Research Building Auditorium
Can Medial-Temporal-Lobe Patients Learn A Category Implicitly?
Can a person with a damaged medial-temporal lobe learn a category implicitly?To address this question, we compared the performance of participants with mildAlzheimer’s disease (AD) to that of age-matched controls in a standard implicit learning task. In this task, participants were first presented a series of objects,then told the objects formed a category, and then had to categorize a longsequence of test items (Knowlton & Squire, 1993). We tested the hypotheses that: (1) both Control and AD participants would show evidence for implicitlearning after the unwanted contribution of learning during test is removed; (2)the degree of implicit learning is the same for AD and Control participants; and (3) training with exemplars that are highly similar to an unseen prototype will lead to better implicit category learning than training with exemplars that are less similar to a prototype. With respect to the first hypothesis, we found that both AD and Control participants performed better on tests of implicit learning than couldbe attributed to just learning on test trials. We found no clear means forevaluating our second hypothesis, and argue that comparisons of the degree of implicit learning between patient and control groups in this paradigm areconfounded by the contribution of other memory systems. In line with the thirdhypothesis, only training with similar exemplars resulted in significant implicit category learning for AD participants.
About the Speaker
Edward E. Smith is the William B. Ransford Professor of Psychology at ColumbiaUniversity. Prior to his move to
Dr. Miok Pak
Friday, Octover 7, 11:00 -12:30 pm, ICC 450
Agreement in Korean Syntax: The case of Sentence Final Particles
In this talk, I discuss sentence final particles in Korean, exemplified in (1) below:
(1) Na-nun cemsim-ul mek-ess-e/eyo/so/supnita.
I-TOP lunch-ACC eat-PAST-Intimate/Polite/Semiformal/Formal
‘I ate lunch.’
These sentence final particles mark speech style (as well as clause type, the fact which will not be discussed in this talk). Hence, there are at least 6 different speech style particles that can occur in declarative alone. These speech style particles reflect a relationship between the speaker and addressee.
Recently, there have been some attempts to treat the topics related to the notion of ‘speaker’ -- such as evidentiality and logophoricity -- from a syntactic point of view. (Cinque 1999, Tenny 2000, Speas 2004, and Speas and Tenny 2004 among others) Speas (2004), for example, argues for Speech Act Phrase bearing speaker’s points of view of a reported event, basing her argument on the morphological reflex of different kinds of evidentiality available across languages. Most recently the notion of ‘addressee’ has also been proposed to play a role in syntax. For example, in their study of imperatives, Mauck et al 2005, Brandstetter 2005 and Zanuttini 2005 have argued that the notion of addressee is represented in the structure of the clause, discussing the fact that there is a tight relationship between the subject of imperatives and the addressee.
Building on these proposals, in this talk I discuss speech style particles in Korean, which have never received a syntactic analysis. I argue that they involve the notions of speaker and addressee, giving them a representation in the syntax. The main contentions of this talk are the following:
(i) The meaning of speech style particles can be expressed using the notions of speaker and addressee, hierarchy, and intimacy/formality.
(ii) The grammatical patterning of speech style particles can be explained by postulating that these notions are represented in syntax.
(iii) The speech style particles are involved in a formal agreement relation with the vocative phrase, which I assume to be present in all clauses.
If point (iii) is valid, then we can conclude that agreement relations can be found in a non-agreement language like Korean, that is, a language that exhibits no phi-feature agreement.
Dr. Yael Sharvit
Friday, October 14, 2005, 11:00- 12:30 pm, ICC 450
Some Pragmatic Thoughts on NPI Licensing
In this talk I argue (following Bhatt and Sharvit, to appear) that superlative expressions provide support for Kadmon and Landman's (1993) theory of NPI licensing. These constructions are an interesting test case for any theory of NPI licensing, because, as observed by Bhatt (2002), the superlative morpheme cannot license a (weak) NPI across an intensional verb such as 'say' (while negation can). This shows that we need a theory of NPI licensing that allows us to place constraints on any of the constituents c-commanded by the NPI-licensor. Kadmon and Landman's theory provides a way to do precisely that. In defending this theory, I also examine some of its weaknesses, and respond to some criticisms raised by various authors (Chierchia 2001, among others).
The Pennsylvania State university
Tuesday, November 1, 1:00- 2:00 pm, ICC 450
Investigating the Role of Working Memory in L2 Processing: Methodological Issues
Why do some people learn a second language (L2) more efficiently, easily and rapidly than others after the same amount of exposure to the language? One of the factors that has been put forth to explain adult inter-learner variability is working memory (WM), defined as the cognitive resources required to process and store incoming information (Baddeley, 1986, 2003). These resources differ from person to person (Just & Carpenter, 1992) and, because they are limited, can affect L2 reading and listening (e.g., Geva & Ryan, 1993; Harrrington & Sawyer, 1992), L2 syntactic processing and development (e.g., Ellis & Sinclair, 1996; Vos, Gunter, Schriefers,& Friederici, 2001; cf. Juffs, 2004); L2 lexical processing and development (e.g., French, 2003; Papagno & Vallar, 1995); noticing of feedback on L2 errors (e.g., Mackey, Philp, Egi, Fuji, & Tatsumi, 2002; Sagarra, 2004); and L2 proficiency (e.g., Kroll, Michael, Tokowicz, & Dufour, 2002; Payne & Whitney, 2002; Service & Kohonen, 1995). The first goal of the present study is to investigate whether the short-term effects of WM capacity found in these cross-sectional studies also reflect a long-term influence. The second objective is to examine whether processing information “strategically” can compensate for lack of capacity (capacity theory vs. processing efficiency hypothesis).
A pool of 110 beginning L2 learners of Spanish completed L2 performance tests during their second- and fourth-semester study of Spanish. WM was measured with Daneman & Carpenter’s (1980) Reading Span Test in the subjects’ native language (L1). Regressional analyses revealed no relation between WM and score gains in the L2 performance tests (grammar, reading), except for listening comprehension. Because other researchers have also found a lack of relation between the Reading Span Test and some aspects of L2 processing (e.g., Juffs, 2004), it has been suggested that the Reading Span Test may have biased the results as it may not measure processing for meaning (central executive). However, if this is the case, how can we explain the correlation between WM and L2 reading found in studies where the Reading Span Test was employed? A close look at these studies reveals that the test was always altered to force semantic processing (e.g., by adding a grammaticality judgment or a verification task).
Experiment 2.
To determine whether lack of relation between WM and L2 reading and grammatical development is caused by the Reading Span Test, 761 beginning L2 learners of Spanish completed a series of L2 performance tests (grammar, reading) at the beginning and at the end of a semester, Daneman & Carpenter’s (1980) Reading Span Test (processing = phonological loop), and Waters & Caplan’s (1996) WM test (processing = central executive). Regressional analyses showed that WM, when measured with a test that forces semantic processing, predicts L2 grammatical development and L2 reading. The use of particular processing strategies did not predict L2 performance. These results support the Capacity Theory and suggest the need for caution in using the Reading Span Test to measure WM’s central executive.
Dr. Shawn Loewen
The University of Auckland, New Zeland
Tuesday, November 8, 12:00- 2:00 pm, ICC 450
Incidental focul on form: its occurrence, characteristics, and effectiveness in the L2 classroom
Incidental focus on form draws learners' attention to linguistic items as they arise spontaneously, without prior planning, in meaning-focused L2 classroom activities. Such a combination of form and meaning is argued to be beneficial for L2 learners. In the present study, thirty-two hours of naturally-occurring, meaning-focused L2 lessons were observed in 12 different classes of young adults in a private language school in
Dr. Noam Chomsky
MIT
Tuesday, November 8, 2005, 2:00 pm, Gaston hall
Language Design and Origins
The “biolinguistic” approach that has taken shape since the 1950s regards a person’s language as a “cognitive organ,” one of the subsystems that interact in human life. A language generates an infinite variety of structured expressions, each of which can be taken to be a set of instructions for the systems within which the language is embedded: the semantic systems that make generated expressions available for thought and for actions, such as referring to the world in certain ways; and the sensorimotor systems that produce and interpret external events. For any such system, we can identify three factors that enter into its growth in the individual: genetic endowment, experience, and independent principles that hold more generally. Insofar as the third factor is involved, the language will be efficiently designed to satisfy conditions imposed by these interface systems. We can regard an account of linguistic phenomena as principled if it derives them by third-factor principles satisfying interface conditions. Recent work indicates that principled explanation can go well beyond what had been assumed. We thereby learn more about what is distinctive to language and human capacities more generally, and may be able to approach the study of evolution of language in more productive ways.
Dr. Susi Wurmbrrand
University of Connecticut
Wednesday, November 9, 2005, 10:15- 11:30 am, ICC 462
Infinitives are tenseless
A common view since Stowell (1982) holds that infinitival complements can be tensed or tensless, and that the presence vs. absence of infinitival tense correlates iwth different syntactic structures: infinitives with an irrealis future interpretation involve control, whereas infinitives that lack this interpretation involve ECM/raising. In this talk, I argue that this view is empirically and theoretically untenable. First, I show that the presence vs. absence of an irrealis future interpretation does not correlate with the control vs. ECM/raising distinction, since there are i) future ECM infinitives and ii) control infinitives disallowing a future interpretation. Second, assuming that future is not a simple tense but composed of PRESENT tense plus a future modal woll, I arguethat all infinitives (including irrealis future infinitives) are tenseless; i.e., infinitives lack any semantic tense value (in particular, the PRESENT tense part of the future) and only involve the modal woll if they are interpreted as irrealis future infinitives. Evidence for the lack of tense will come from the semantic properties of infinitival 'tense': lack of double access readings and the locality of sequence-of-tense.
Dr. Julie Anne Legate
University of Delaware
Friday, November 18, 2006, 11:00- 12:30 pm, ICC 450
On Case and Case
Consider the following analysis of the English case system: in the syntax, T assigns structural nominative Case and v assigns structural accusative Case; these abstract Cases are realized morphologically through a (zero) default, because (for DPs) English lacks nominative case morphology and lacks accusative case morphology. This seems a thoroughly unremarkable analysis. However, this type of analysis has not been applied to ergative-absolutive languages. I argue that this has significantly impeded our understanding of ergativity, or rather "absolutivity". I show that a large class of ergative-absolutive languages differ from English only in having an inherent Case for the thematic subject, i.e. ergative. What is referred to as "absolutive" is in fact the default morphological realization of abstract nominative Case on intransitive subjects and abstract accusative Case on transitive objects. I demonstrate that recognizing this class of languages provides an explanation for a range of behavioural distinctions among ergative-absolutive languages. Finally, I show that it also provides an explanation for a striking pattern found in many Pama-Nyungan languages, in which the subparts of a single DP appear to bear different cases.
Dr. Deborah Chen Pichler
Galludet University
Tuesday, February 21, 2006, 10:15- 11:15 am, ICC 450
Acquiring ALS as an L1 (and a Few Thoughts on Acquiring ASL as an L2)
Interest in the acquisition of American Sign Language (ASL) has grown steadily over the last fifty years, leading to investigation of a wide variety of language development phenomena. However, ASL acquisition studies are still overwhelmingly based on very limited data, and the lack of a standardized coding/transcription system for signed languages often leads to contradictory conclusions and difficulties in comparison across studies. Furthermore, acquisition research has focused almost exclusively on L1 acquisition, with very little attention paid to the acquisition of ASL as an L2. Given the recent explosion in interest in ASL at high schools and universities across the US, this state of affairs is clearly problematic.
This talk will discuss the handful of studies on development of word order in L1 ASL, an area that nicely illustrates the above-mentioned in ASL L1 research. Different data collection methods and incomplete disclosure of coding criteria led to a situation in which the two existing studies on the topic appeared to fully contradict one another. Chen Pichler (2001) attempts to reconcile some of the apparent contradiction, while adding new data to the very limited corpus on which all ASL word order research to date has been based.
The talk will conclude with a brief report of new research on L2 acquisition of ASL by hearing learners. Results are from a recent pilot experiment run at Gallaudet as part of an on-going study of beginning ASL L2 phonology. While the findings of this first experiment are quite interesting, and look promising for continued investigation, their implications are still rather unclear, at least to the speaker. It is hoped that members of the audience, many of whom doubtless possess far greater expertise in the area of L2 acquisition than the speaker, will have helpful comments to contribute to the discussion.Dr. Mark Baker
Rutgers Univeristy, Department of Linguistics
Tuesday, March 14, 2006, 2:40- 4:00 pm, ICC 662
Two Syntactic Parameters of Agreement
Despite various core similarities, agreement in Niger-Congo (NC) languages turns out to be systematically different from agreement in Indo-European (IE) languages in two respects. First, the agreed with NP must be higher in the structure than the agreeing head in NC languages, whereas this requirement does not hold in IE languages. Second, agreement in IE languages is subject to a case-valuation condition that does not hold in NC languages: a head can agree with an NP only if the head values the case of that NP or vice versa (IE only). I show how these two parameters account for systematic differences in the behavior of subject agreement, object agreement, agreement on prepositions, agreement on complementizers, and agreement on determiners in the two language families. I then discuss preliminary results of studying the typological distribution of these two parameters across a sample of 100 languages. In addition to other languages that behave like NC or like IE, I ! present Turkish as a possible language in which both conditions on agreement hold, and Georgian as a possible language in which neither condition on agreement holds. Since these conditions apply to the full inventory of functional heads in the languages in question, this investigation supports the idea that there are true syntactic parameters (“macroparameters”) that do not reduce to stipulations about the features borne by particular items in the lexicon.
Drs. Gennaro Chierchia, Robert M. Harnish, and Wayne Davis
Friday, March 17, 2006, 10:00- 3:00 pm
Workshop on Implicature
Dr. Bonnie McElhinny
Tuesday, March 28, 2006, 11:30- 1:00 pm, ICC 450
Language, Gender ane Economies in Global Transitions: Provocative and Provoking Questions about How Gender is Articulated
The study of gender as socially constructed has been central to feminist discussions which distinguish between what is biologically shaped and socially shaped, between sex and gender. Increasingly, however, the utility of a focus on "social construction" has come under critique, with scholars calling for a fuller attention to sociohistorical and economic processes. This paper is devoted to systematically exploring ways to articulate studies of language and gender with studies of political economy, especially global economy, as it interrogates what notions of language, gender and global economy facilitate such articulations. It offers a critical overview of recent work on language, gender and globalization, and identifies five or six key contributions that sociolinguists and linguistic anthropologists can offer to the growing social scientific literature on gender and globalization.
Dr. Mark Light
University of Iowa, Department of Linguistics; School of Library and Information Science; Computer Science Department
Friday, March 31, 2006, 11:40 am, ICC 450
Relation Extraction for Bioscience Literature Using Just Enough Parsing
This talk presents a method for extracting protein interactions that is accurate and efficient. It relies on a novel set of grammatical phrases that can be extracted using efficient partial parsing methods. This set of grammatical phrases has utility for other extraction tasks such as general relation extraction and scope of affect predicates (e.g., speculation).
Dr. Nkoko Kamwangamalu
Howard University
Tuesday, April 5, 2006, 11:40- 12:55 pm, ICC 450
Language Policy and Planning in Post Colonial Africa
This paper reports on the current state of language policy and planning in post-colonial Africa. It points out that although colonialism ended years ago, its legacy together with elite closure continues to impact on the language policies of most African states. The paper considers the prospects for indigenous African languages in the light of current theories of language economics, a field of study whose focus, as Grin (2001) explains, is on the theoretical and empirical analysis of the ways in which linguistic and economic variables influence one another. It concludes with a call for language policies that not only consider the development and promotion of African languages as an integral part of Africa's economic development program, but also ensure that former colonial languages function in addition to rather than at the expense of African languages and the majority of their speakers.
Dr. ZhaoHang Han
Columbia Univeristy, College of Arts and Humanities
Tuesday, April 18, 2006, ICC 450
Recasts and Grammatical Morphemes
SLA research has, over the past four decades, evolved into a pluralistic science with circumscribed boundaries. Within each of its paradigms (e.g., the generative linguistic, the connectionist, the interactionist), and driven by a particular theoretical orientation, researchers have pursued distinct paths towards identifying, describing, and explaining what often turns out to be a common set of phenomena. One such phenomenon that has recently drawn much attention cross-paradigmatically is L2 learners' lack of acquisition of grammatical morphemes - often irrespective of the typological distance of their L1s from the target language. In the research domain of interactional feedback, a similar phenomenon has also been noted and reported, that learners tend not to attend to corrective feedback on morphosyntactic errors, but they do attend to such feedback on lexical and phonological errors. As yet, however, interactional feedback vis-à-vis grammatical morphemes has not been directly examined. In this talk, I would therefore like to lead a discussion on this issue, with specific reference to articles and plurals. Following a brief overview and interpretation of the major explanations advanced from other theoretical camps, I will turn to the research on the efficacy of interactional feedback, in particular, of recasts, establishing and illustrating my arguments with episodes extracted from a recent corpus of EFL classroom interactions (Kim & Han, in preparation), and from other, published studies, where pertinent.
Dr. Bonnie Webber
University of Edinburgh, School of Informatics
Wednesday, April 19, 2006, 11:40- 12:55, ICC 450
Discourse Grammar from a Lexical Perspective
To date, Language Technology has derived its greatest success from words and word-level techniques. Since discourse is so much more than words, will it prove to be beyond the promises of this technology? This talk suggests that the answer is "no", arguing that the lexicon provides a robust basis for low-level discourse grammar. I start by reviewing some previous proposals regarding discourse structure and discourse grammar, and then describe a lexicalised discourse grammar modelled on Lexicalised Tree-Adjoining Grammar. What is attractive about this approach from a linguistic perspective, is the range of examples it is able to explain. On the other hand, interesting examples are not necessarily common examples. So to provide empirical grounding for such work on discourse, I am working with colleagues at the University of Pennsylvania on what we call the "Penn Discourse TreeBank" (http://www.ircs.upenn.edu/~pdtb/). I will conclude the talk by describing features of this resource as of its first release.
Dr. Detmar Meurers
Ohio State University
Tuesday, April 25, 11:40- 12:55 pm, ICC 450
Using Natural Language Processing for Foreign Language Teaching: Pitfalls and Opportunities
Research since the 90s has shown that awareness of language forms and rules is important for an adult learner to successfully acquire a foreign language. But given the limited amount of time a teacher/tutor can spend with a student and the communicative focus of instruction, there are only few opportunities for fostering linguistic awareness through focus on form and providing individual feedback on errors. The situation appears ideally suited for Computer-Aided Language Learning (CALL) systems to step in, but traditional CALL systems lack the ability to analyze language and provide feedback on that basis. The more recent research on using natural language processing to obtain "intelligent" CALL (ICALL), however, has largely focused on specific technical issues without a clear link to the pedagogical needs and objectives in foreign language teaching. As a result, there are virtually no ICALL systems used in foreign language teaching today.
In this talk, I want to discuss some opportunities for using NLP for foreign language teaching that try to avoid this pitfall. I will primarily focus on an intelligent web-based workbook we are developing in support of the Portuguese Individualized Instruction program at the OSU Foreign Language Center.
Dr. Markus Dickinson
Georgetown University, Visiting Professor
Thursday, April 27, 2006, 11:40- 1:00 pm, ICC 450
Error Detection in Treebanks
Annotation errors in large corpora are harmful for both the training and evaluation of natural language processing technologies, but how do we systematicallly locate such errors? Using the idea that variation in annotation is likely erroneous, I will show how two different methods can successfully identify errors in syntactic annotation, and will demonstrate the impact of such errors on the training of parsing technologies.
The first method is the variation n-gram method, applicable to a variety of annotation types. By searching for identical strings which vary in their annotation, we are able to find erroneous mark-up. I will focus on how to adapt the method for syntactic annotation and the results from the Wall Street Journal (WSJ) corpus, as part of the Penn Treebank 3 release.
The second method for detecting errors in syntactic annotation searches for variation in local trees, as potential violators of endocentricity (the idea that a head projects to a phrase). The immediate dominance (ID) variation method indexes every rule in a treebank by its daughters and identifies the daughters which have multiple mothers. Such ID variation is flagged as a potential error. After further identifying specific rule occurrences which are erroneous, we demonstrate the impact such errors have on the effectiveness of a grammar induction algorithm and subsequent parsing.
Upcoming Events
- Nov 25, All day: Thanksgiving Recess Begins after Last Class
- Nov 27, All day: Fall Dissertation Proposals due to Department
- Nov 30, All day: Classes Resume
