Alan Prince Colloquium Next Friday (10/17) at 3:30 PM

Join us in the Greenberg Room next Friday as we welcome Alan Prince (Rutgers University), who will talk on “Testing the boundaries: aspects of typological structure in OT.” His abstract is given below, as well as a link to a PDF version of the abstract which includes additional figures.

Abstract: The formal typology is perhaps the central object of modern linguistics, where formal typology = the set of all grammars admitted by the premises of a theory. In OT, this object is both self-consciously placed in the foreground and amenable to study.

A formal typology classifies grammars in terms of the inner mechanisms of its theory. In OT, a typology classifies grammars in terms of shared and distinguishing ranking patterns (Alber & Prince) that collectively combine to give the entire set of grammars in the typology. What these are, and may be, depends on the structure of the typology (Merchant & Prince).

Typological structure in OT involves both geometry and order. A notion of adjacency between grammars leads to the ‘typohedron’ of a typology, where each grammar is represented by a single vertex. Adjacency comes from the linear orders (‘rankings’) that grammars consist of. Basic classification involves sets of grammars that are geometrically adjacent in this sense. Because of the way constraint ranking selects optima, order and equivalence relations between entire grammars, grounded in the geometry, also emerge. Basic classification respects these relations as well. (In Merchant & Prince, they are represented by the MOAT — ‘mother of all tableaux’ — which contains the essential OT properties of each constraint.) Both aspects of structure are representable graphically in ways that render them quite accessible (see PDF version).
The logic of OT ranking leads to two further developments that build from the basic structural elements. (1) Ranking properties may take limited scope, so that in grammars outside the scope of a property, certain distinctions are moot. This follows from the fact that constraints need not be crucially ranked with respect to each other in every grammar. (2) Constraints belong to classes as well, based on the symmetries of their ranking behavior. Constraints must often have an atomic character, but their behavior may echo that of symmetrical partners operating in distal regions of the typology.

Find the augmented abstract pdf here.

Alan Prince bonus talk, Thursday at 4PM

In addition to his colloquium talk on Friday, Alan Prince will speak at a Psycholinguistics Group meeting next Thursday at 4PM about OTWorkplace, a new computational framework for OT analysis.

Description:

OTWorkplace is an interactive environment for linguistic analysis using the tools of modern rigorous OT, written by A. Prince, B. Tesar, and N. Merchant. In the background lies the previous pioneering work of B. Hayes on OTSoft, and it belongs to a set of packages that includes work by A. Andrews, T. Biro, A. Hogan, W. Raymond, J. Riggle, R. Staubs et al., and M. Walther.

The basic concept is like that of Maple or Mathematica. The aim is to provide a relatively straightforward and easily manipulable user interface which produces output in a form where further calculations can be made, further ideas explored, manually or via software, which may be built-in or user-defined.The end result is an automated research notebook, which may be preserved, indexed, commented, edited, expanded, contracted, as the project flows and develops. 

OTWorkplace runs within Excel and produces Excel workbooks. The virtues of platform-independence have been set aside in favor of the virtues of usability, flexibility, and ease of further development. System requirement are Windows (32 or 64-bit) and Microsof Office (32-bit only). It has been run successfully on Macs using emulators like Parallels and VirtualBox. It is recommended that users install from the installer package on my webpage at http://ruccs.rutgers.edu/~prince/. Installation has been reported successful on operating systems ranging from XP to Windows 8.1. Updates are posted on the barebones OTWorkplace website at https://sites.google.com/site/otworkplace. Current version: OTWorkplace_X_70.xlsm. The installer package installs Graphviz, the programming language Ruby (with appropriate add-ons), the Ruby scripts (‘RUBOT’) that are called for various heavy-lifting tasks not covered in the VBA core of the program (which is contained in the OTWorkplace workbook itself), putting everything in the right places and setting the appropriate environment variables. You do none of this, and are responsible only for clicking on a few filenames and responding civilly to a few questions. There’s a readme file with the installer package that guides you through these clicks.

In this meeting, I will focus on using OTWorkplace to analyze OT typologies, which by something close to a fortiori covers OT analysis of single languages. No experience with the program will be assumed, though some prior acquaintance with OT and Excel ought to smooth the way. If you have installed the program on your laptop, you should be able to follow along. Documentation is imperfect at present. The material in the OTWorkplace workbook, where the program resides, is out of date. The OTWorkplace website contains various pieces of current documentation mostly relevant to automated violation counting. The hope is to have a general quick start guide up there shortly.

 

 

The Language of Food Featured in Boston Globe

Read the article here, which also includes an interview with the author himself!

SMircle Monday (10/13) at 3:15 PM: Baier

Join the SMircle workshop as they hear from Nico Baier (UC Berkeley), who will be presenting his work on adjectives in Noon (Senegal) on October 13, 3.15pm in the Greenberg Room.

Adjective Agreement in Noon: Evidence for a Split Theory of Noun-Modifier Concord

In this paper, I argue that two distinct mechanisms are needed to account for the pattern of noun-adjective agreement in Noon (Cangin, Senegal). Adjectives in Noon exhibit two types of agreement. First, they take a prefix that encodes the class and number of the noun they modify. Second, attributive adjectives must agree in definiteness with the noun they modify. Nouns and adjectives take an identical definiteness suffix that marks by a suffix on the the head noun’s class/number along with three degrees of deixis. The class/number prefix is obligatory for both attributive and predicative adjectives, while the definiteness suffix is available only for attributive adjectives. Previous work presupposes that the same mechanism derives noun-modifier agreement (concord) and argument-predicate agreement: usually a modified version of Chomsky’s (2001) Agree. However, others have argued that noun-modifier agreement is morphological in nature and should be derived post-syntactically (Norris 2014). Here, I argue for a split approach: prefixal agreement on Noon adjectives is derived by valuation of Phi-probe on a, while another is derived via a post-syntactic process of Feature Copying.

Look Who’s Talking!

Eve V. Clark is giving the plenary talk “Variation and Experience in First Language Acquisition” at the conference Diferencias Individuales en la adquisicion del lenguaje, Universidad Nacional Autonoma de México, Mexico City on Thursday, October 9.

Vera Gribanova presented “On certain manifestations of polarity focus in Russian” at the LRC Workshop on Syntax and Information Structure.

Andrea Beltrama will present at the Berkeley Syntax Circle next Friday, October 17th.

Dan Jurafsky will present on Extracting Social Meaning from Language: The Computational Linguistics of Food and the Spread of Innovation at MIT’s CSAIL Thursday, October 9th.

VPUE Undergraduate Intern Presentations Today (10/3) at 3:30!

Come hear from our undergraduate VPUE interns as they present their summer research projects today in the Greenberg Room. The presentations start at 3:30, with detailed schedule below.

3:30pm Yesid Castro and Som Mai Nguyen: “Effects of talker variability and reported talker traits”

3:50pm Kayland Harrison and Breanna Williams: “Post vocalic r-retention in reading by African Americans in the Moving to Opportunity Corpus”

4:10pm Jonathan Engel “Crowd Sourcing Meaning: Statements With Evaluative Adjectives”

P-Interest Workshop Meeting today at noon

Join the P-Interest group today (10/3) at noon in the Greenberg Room, where Sunwoo Jeong will present the results of her first QP, “Iconicity in Suprasegmental Variables: The Case of Archetypal Hollywood Characters of the 1940s-50s”. Abstract:

Films are potent vehicles that not only reflect common linguistic practices, but also create new social meanings for linguistic variables and actively shape dominant language ideologies of the era. This was especially the case for films made during the Golden Age of Hollywood in which several distinctive film genres, featuring highly stylized female characters, emerged as important cultural phenomena: femme fatales in film noir, independent brunettes in screwball comedies, and dumb blondes in musical comedies. This paper argues that systematic variation in suprasegmental linguistic cues like pitch, prosody, and voice quality was employed by the actresses to index the three prominent archetypes mentioned above, and more importantly, that the realizations of these variables were not arbitrary in that they created an iconic tie with the archetype that they indexed. Combined with other cinematic devices that fortified this iconic relation, the underlying ideologies behind these linguistic variables were more easily naturalized, resulting in wider dissemination.

The supporting evidence for this argument comes from an in-depth analysis of pitch and voice quality variables of the three archetypes in 15 films. Actresses and films that were highly representative of each archetype/film genre (e.g. Barbara Stanwyck as an iconic femme fatale), as well as actresses that portray multiple archetypes across different films (e.g. Marilyn Monroe as a dumb blonde in Gentlemen Prefer Blondesvs. a femme fatale in Niagara) were maximally chosen. Utterances with no background noise were exhaustively extracted from each film, and relevant acoustic features (maximum F0, minimum F0, F0 standard deviation of each utterance and H1-H2 of each vowel) were measured.

A series of mixed effects models shows that the dumb blonde is characterized by significantly higher pitch, higher F0 standard deviation, and breathy voice; the femme fatale by lower pitch, lower F0 standard deviation and breathy voice; and the independent brunette by lower pitch, higher F0 standard deviation and modal to creaky voice. Such systematic stylistic differentiation is reliably observed both within a single actress portraying different archetypes and also across multiple actresses, demonstrating that the variation is transparently realized both at the level of intra-speaker and inter-speaker variation.

Crucially, the quantitative analyses mentioned above combined with qualitative analyses of pitch contours and spectrograms, show that the variables iconically represent the archetypes themselves. For example, the femme fatale resorts to intensity rather than pitch variation to convey emphasis (although pitch is typically the most prominent linguistic cue for stress), and this usage of non-normative acoustic cues reflects her transgressive character. Also, her strikingly monotonous and horizontal intonational contour without a natural declination pattern iconically reflects her composed, unperturbed nature and strengthens the angular compositions and somber, monochromatic imagery of the film noir. Linguistic variation is situated within the broader semiotic system of the film, and the implicit ideological message it conveys is fortified by the film’s imagery, facilitating its propagation as easily accessible stylistic resources.