North American Computational Linguistics Olympiad for high school students
Practice Round: January 12, 2017
Location: University of Utah Language & Communication building (LNCO) Room 2945
Address: 255 S. Central Campus Drive, Ste 2945
Salt Lake City, UT 84112
Open Round: January 26, 2017
Time: 10:00am - 1:00pm
Location: University of Utah Language & Communication building (LNCO) Room 2110
Address: 255 S. Central Campus Drive, Ste 2110
Salt Lake City, UT 84112
The University of Utah is a local host for NACLO! NACLO is a contest in which high school students learn about linguistics and language technologies by solving puzzles. The contest helps students identify potential careers related to languages and language technologies. They also exercise problem-solving skills and have fun!
Registration is open now! To register, please follow this link and list the University of Utah as your Site:
For questions, information, and practice problems, please visit http://nacloweb.org/.
If you would like to receive announcements about our events, please email firstname.lastname@example.org and ask to be added to the ling-announcements listserve.
Jan. 21 - Mykel Brinkerhoff Practice Talk
Mykel Brinkerhoff will be giving a practice talk for his upcoming UUSCIL presentation.
Jan. 28 - Miranda McCarvel & Cole Brendel
Miranda McCarvel and Cole Brendel will be giving practice talks for their upcoming conference presentations.
Mar. 3 - Allard Jongman (University of Kansas)
Phonological neutralization: Acoustics, processing, and acquisition
Phonological neutralization, the elimination of a phonemic contrast in certain contexts, is considered a cornerstone of phonological theory. In this talk, I will provide a survey of some of our research on the acoustics and perception of neutralization across a variety of languages. Neutralization of both segmental and suprasegmental contrasts will be considered. Our results show that while some instances of phonological neutralization are phonetically complete, others are incomplete. I will also provide data examining the extent to which second language learners may acquire neutralization. Rather than interpreting incomplete neutralization as evidence against formalist theories of phonology, I will argue that incomplete neutralization in fact supports the traditional phonological analysis of neutralization and the notion of underlying representations.
Mar. 24 - Dave Kush
Using structure to guide memory access during sentence processing
Natural language allows dependencies between items at a distance. Establishing dependencies between distant items in real-time sentence processing makes use of memory: previously-seen licensors must be stored and subsequently retrieved from memory when their dependents are encountered. Antecedent-pronoun dependencies provide an illustrative example. Pronouns usually find their antecedents in the preceding linguistic context. When a parser encounters a pronoun, it must access memory to find a suitable antecedent for that pronoun from among previously-seen NPs. In this talk I investigates the memory retrieval procedures that underlie this antecedent identification process. In particular, I investigate a tension that arises between linguistic constraints on antecedent-pronoun relations and well-motivated models of memory retrieval.
Antecedent-pronoun dependencies are governed by a number of structural constraints. These constraints are typically stated in terms of syntactic relations (e.g. c-command, Reinhart, 1983) between a pronoun and a potential antecedent. Recent research in psycholinguistics has motivated a model of processing that uses a cue-based retrieval mechanism (e.g. Lewis, Vasishth, & Van Dyke, 2006) in which item-to-item syntactic relations such as c-command are difficult to use as retrieval cues. As a result, these models predict that the earliest stages of antecedent retrieval should not display sensitivity to well-known syntactic constraints. That is, antecedent retrieval should erroneously access NPs in structurally inappropriate positions as potential antecedents. I present a series of experiments that show that this prediction is not borne out: antecedent retrieval shows immediate sensitivity to structural relations across a variety of configurations. I conclude by discussing ideas for how to reconcile these findings with the cue-based framework.
April 14, 2016
Bayesian Statistics for Linguists Workshop
Thursday, October 27 – Friday, October 28
Bayesian Statistics for Linguists Workshop with Prof. Noah Silbert, University of Cincinnati
Jan. 22 - Shannon Barrios-Peer Abstract Review Session
Feb. 5 - Adrian Palmer & Daniel Dixon - Language Assessment in the Real World
The state of the art in language assessment development is seen as the result of solutions to real-world problems facing the language assessment profession. In this presentation, we will trace some of the problems that language assessment developers have faced historically and how they have gone about creating solutions. We will then show how this history has led us to the current state of the art: justification-based language assessment development, an approach that begins with justifying the consequences of assessment use for stakeholders. We then show how the Test-ify® software we have developed solves some of the most recent real-world problems facing language testers. The presentation is designed to be engaging as well as informative. It will also be the first time we have demonstrated the software to non-assessment specialists.
Feb. 12 - Aaron Kaplan- Off the Edge of the Phonological Map
Languages often restrict certain elements--high vowels, say--to strong positions like stressed syllables. For example, in the Romance variety spoken in Central Veneto, a post-tonic high vowel triggers raising of the stressed syllable and any vowels between it and the stressed syllable: /órdeni/ --> [úrdini] 'order (2sg.)'. This is the result of a requirement that high vowel features at least partially overlap with the stressed syllable. Positional Licensing constraints provide an account of these patterns in Optimality Theory, but I show that they make erroneous predictions in Harmonic Grammar, a relative of OT that uses constraint weights instead of OT's rankings. In Harmonic Grammar, Positional Licensing predicts a language in which the spreading seen in Central Veneto occurs over short distances but is blocked at longer distances; in fact, it predicts that every language is like this! Such a pattern is unattested crosslinguistically. I argue that correcting this flaw requires a significant reimagining of the Positional Licensing formalism and even Harmonic Grammar itself. Currently, Positional Licensing penalizes elements that do not overlap with the target strong position, but in HG it must instead reward elements that do overlap with that position.
But "positive" constraints of this sort are widely regarded as extremely problematic in frameworks like HG, and repairing their defects requires grafting elements of another theory, Harmonic Serialism, onto HG. This leaves us in a very strange position:
positive constraints in a theory that combines the defining characteristics of Harmonic Grammar and Harmonic Serialism.
Feb. 19 - Aniko Csirmaz- My lexicon is dumber than yours
In generative syntactic research, a large number of characteristics were encoded as properties of lexical items. These properties distinguished between count and mass nouns and between different types of verbs. The latter determine the difference between stative / nonstative verbs (sleep vs run) and atelic / telic (run vs fall) (see Vendler 1967 and Dowty 1979, a.o.). One problem with the telic/atelic property is that it is not verbs per se that are telic or atelic, but larger pieces of structure. This is shown by the fact that 'Jill ran' is atelic but 'Jill ran to the fire escape' is telic. It seems more attractive to assume that telicity is not a lexical property, but it is perhaps determined by a functional head; this head can be relevant for case licensing as well. One problem with a naive implementation of a functional head is that telicity is compositional; in general, overt elements unambiguously determine the telicity of the structure in question.
Feb. 26 - Brian Dillon (Umass) - Which noun phrases is this verb supposed to agree with... and when?
The study of agreement constraints has yielded much insight into the organization of grammatical knowledge, within and across languages. In a parallel fashion, the study of agreement production and comprehension have provided key data in the development of theories of language production and comprehension. In this talk I present work at the intersection of these two research traditions. I present the results of experimental research (joint work with Adrian Staub, Charles Clifton Jr, and Josh Levy) that suggests that the grammar of many American English speakers is variable: in certain syntactic configurations, more than one NP is permitted to control agreement (Kimball & Aissen, 1971). However, our work suggests that this variability is not random, and in particular, optional agreement processes are constrained by the nature of the parser. We propose that variable agreement choices arise in part as a function of how the parser stores syntactic material in working memory during the incremental production of syntactic structures.
March 5 - Sean Redmond (Communication Science & Disorders)
What Attention-Deficit/Hyperactivity Disorder (ADHD) can teach us about Specific Language Impairment (SLI)
Everybody’s heard of ADHD but outside of the research community, very few people are aware of the existence of SLI. This is interesting because these disorders occur in the population at roughly the same rates and both compromise the academic, social, and vocational outcomes of affected individuals. This presentation highlights some key findings associated with an ongoing line of inquiry comparing the linguistic and socioemotional profiles associated with ADHD and SLI. What happens to children’s language development when both disorders are present will be discussed as well. In addition to informing clinical practice, these data speak to ongoing debates about the “specific” nature of the language impairments associated with SLI and whether information processing deficits of the sort associated with ADHD contribute to children’s language symptoms.
March 12 - Tanya Flores (Languages & Literature)- Accommodating to Stereotypes
This presentation will focus on the Sociolinguistic-Speech Accommodation analyses of three cases of phonetic variation in Chilean Spanish. Previous literature (Lipski 2009, Hualde 2005, Silva-Corvalan 1987, Oroz 1966) reports that the innovative variants of the /tr/cluster, the affricated /tf'/, and the velar fricative /xx are associated to specific social traits of speakers who produce them. Most of the studies on these sound changes are descriptive and/or sociolinguistic studies focusing on the sex, age, and social class of the speakers.
March 23 - Akira Omaki (Johns Hopkins)- Linking parser development and acquisition of syntax
Traditional work on syntactic development has paid little attention to the developing parsing mechanism in language learners. In this talk, I will argue that a proper understanding of syntactic development requires an understanding of how language learners encode and represent syntactic structures of the input. In the face of cross-linguistic variations in syntactic structures, language learners must rely to some extent on information in the input in order to acquire the target grammar. By definition, input is merely an external signal, and it must be assigned a mental representation before it could be used for the purpose of language acquisition. This view predicts that the learning outcome or developmental trajectory should vary as a function of (possibly incorrect) syntactic representations that learners assign based on their immature knowledge and parsing mechanisms. I will present novel evidence for this prediction from word learning behaviors in Japanese infants, as well as 5-year-old children's incremental parsing mechanisms (and errors) in comprehension of English, Japanese and French wh-questions. While these findings demonstrate the importance of parsing constraints on language acquisition, it also raises a new research question: how do parsing mechanisms themselves develop in language learners? In the second part of this talk, I will discuss on-going investigations of predictive sentence processing mechanisms (or lack thereof) in 5-year-old children. I will propose that predictive parsing procedures may need to develop through years of accumulation of skewed distributional information in the input. As such, research on parser development not only informs theories of language acquisition, but also provides theoretical implications for models of sentence processing.
March 26 - Ben Slade- Iyaric Iformations: an Optimality Theoretic Analysis of Rastafari morphology
The speech of members of the Rastafari community (originating in Jamaica) exhibits various linguistic innovations, including garden-variety extensions of productive morphological patterns to produce neologisms like _upful_ "positive" or _livity_ "lifestyle", as well as examples of punning/word-play like _politricks_ "politics". But we also find examples of more unusual linguistic innovations, such as _outiquity_ and _Ital_, both part of larger systems of morphological transformations. In this talk I will focus on the latter type.
Words like _Ital_ (<_vital_) and _Iver_ (<_ever_) are part of a larger group of commonly used "I-words", which often involve the replacement of the initial syllable with _I_ [ɑɪ]. However, the pattern of _I_-replacement is considerably more complex than that, as can be seen from _Iration_ (<_creation_) and _Irous_ (<_desirous_)---for which we would expect _Iyation_ and _Isirous_, respectively. The morphological process involved here then is not simple replacement of the first sound or syllable, but is rather akin to the process of blending (e.g._spork_, _brunch_).
One factor involved is maximizing overlap, e.g. the blend _chillax_ (<_chill_ +_relax_) takes advantage of an overlap of _l_, which does not occur in the hypothetical alternative *_relill_. Likewise, _Irous_ maximizes overlap between the _I_ component and _desirous_ by aligning _I_ with the position of the homophonous [ɑɪ] sound in _desirous_. Further evidence of the complexity of I-word formation can be seen in the case of monosyllable base-words, e.g. _food_ become _yood_ in Rasta Talk, and yet another type of formation is found in _Iyadda_ (<_father_).
I present an Optimality Theoretic analysis of I-words, which treats them as akin to blending.
April 2 - Tania Ionin (UIUC)- Focus on Russian score: an experimental perspective
This paper examines the scope readings of Russian double-quantifier sentences like (1) in both native and non-native Russian, focusing on the relative contributions of word order, prosody, and information structure. Corresponding English sentences are ambiguous between surface-scope and inverse-scope readings, which are commonly derived by covert QR of either the subject QP or the object QP to a higher position at LF (e.g., May 1977, Heim & Kratzer 1998). For Russian, Ionin (2003) argues that scope of (1) is frozen in prosodically neutral sentences: the preverbal QP is in Topic position, with covert QR above Topic position being impossible. Ionin ties this claim to the ability of Russian objects to move overtly, resulting in OVS order (cf. Bailyn 1995), and argues that scrambled OVS sentences, just like SVO sentences, have only surface-scope readings: in the scrambled sentences, the scrambled object is in Topic position, and topics cannot reconstruct. In contrast, Antonyuk (2006) argues that covert QR does apply in Russian, and that both SVO and OVS orders are scopally ambiguous. Finally, contrastive focus on the preverbal QP in Russian has been tied to availability of inverse scope (Ionin 2003, Neeleman & Weerman 2009; cf. Bobaljik & Wurmbrand 2012 on the relationship between scope and focus in other languages).
April 9 - Karlos Arregi (U Chicago)- How to sell a melon: Postsyntactic mesoclisis in Spanish imperatives
Harris and Halle (2005) present a framework (hereafter, GeneralizedReduplication) that unites the treatment of phonological reduplication and metathesis with similar phenomena in morphology, thereby accounting for the apparently spurious placement of imperative plural inflection -n in non-standard Spanish. For instance, alongside standard "vénda-n-me-lo"
("Sell it to me!"), where -n precedes enclitics, one also finds forms such as "vénda-me-lo-n" and "vénda-n-me-lo-n", in which the plural suffix follows enclitics, with an optional copy of the suffix before them. More recently, Kayne (2010) has challenged their analysis, arguing that such cases should be uniformly treated in the syntax. In this talk, I reassess some of Kayne’s arguments, agreeing with his conclusion that the most important desiderata of any general analysis of these sorts of phenomena is restrictiveness, but contending that greater restrictiveness can be achieved through metaconstraints on postsyntactic Generalized Reduplication rather than through byzantine syntactic derivations. I conclude with general remarks about the division of labor in word-formation.
April 16 - Shannon Barrios -Predicting discrimination behavior across L2 development
Models of nonnative speech perception make predictions about listeners' discrimination performance for a pair of phones on the basis of how the sounds are assimilated to native language (L1) phonetic and phonological categories (Best, 1995; Best & Tyler, 2007). It is less clear whether such models also make accurate predictions about second language (L2) learner's discrimination performance when L2 categories, which may also influence the listener’s identification and discrimination behavior, are taken into account. To this end, we investigated the relationship between L1 English L2 Russian learners’ assimilation patterns and discrimination accuracy across two proficiency levels for two continua (i.e. a native-native continuum, /i/-/ɪ/ and a native-nonnative continuum, /ɪ/-/ɨ/). As expected, we found no difference in either assimilation patterns or discrimination performance between the two proficiency levels along the native continuum (/i/-/ɪ/). Additionally, while we find different assimilation patterns for high and low proficiency groups along the native-nonnative continuum (/ɨ/-/ɪ/), the predicted difference in discrimination performance was not observed. Interestingly, we find that discrimination accuracy for each assimilation type varies with proficiency levels along the native-nonnative continuum only. Possible explanations of this unexpected finding will be discussed.
April 23 - Alexis Wellwood (Northwestern)- Finding meaning in formal semantics
Linguistic meanings are not exhausted by their contributions to truth-conditions. Judgments of truth cannot therefore be the only data for semantic theory, especially if natural language semantics is a part of cognitive science. In this talk, I discuss how other behavioral data can allow us to infer the semantic properties of expressions, relying on assumptions about how they interact with other cognitive systems. As a test case, I discuss speaker understanding of "more" and "most", combining an analysis of their grammatical properties with evidence from adult performance and child language acquisition.
Check out the Event Calendar for other Linguistics Events!