Several UM linguists will be presenting their work at the 158th meeting of the Acoustical Society of America in San Antonio, Oct. 26-30.

Presenters and abstracts listed below

The perceptual time course of coarticulatory nasalization.
Patrice S. Beddor,  Julie E. Boland, Andries Coetzee, Kevin McGowan
Abstract:
Listeners’ moment‐by‐moment processing of anticipatory vowel nasalization and a following nasal consonant was investigated. English‐speaking participants’ eye movements were monitored as they heard instructions to look at one of two pictured objects on a computer screen. Trials included pictured pairs for naturally produced words of the form CVNC‐CVC (e.g., bend‐bed), CVNC‐CVNC (bend‐bent), and CVC‐CVC (bed‐bet). Vowels in CVNC words were coarticulatorily nasalized. Results to date show that, when participants heard a CVNC word (bend), they visually fixated the correct picture earlier when the competing picture was CVC (bed)—that is, when the vowel in the competitor would be expected to be non‐nasal—than when the competitor was another CVNC word (bent). Results also suggest that participants often fixated the target CVNC picture in CVNC‐CVC trials after onset of vowel nasalization but before N onset. However, although vowel nasalization facilitated early selection of CVNC over CVC, a non‐nasalized vowel was not similarly helpful for selecting CVC over CVNC. When participants heard CVC (bed), they did not fixate the correct picture earlier when the competing picture was CVNC (bend) than when the competitor was CVC (bet). Findings are interpreted in light of production data for English and perceptual theories.

Nasal coarticulation in clear speech
Anthony Brasher
Abstract:
This study tests whether speakers, when trying to speak clearly, employ variable enhancement strategies as a function of phonetic environment. Using aerodynamic and acoustical methods, this study examines the effects of phonemic context and speaking modality and on the spatial and temporal extent of anticipatory nasal coarticulation in English. Target words are English (C)VNCvoiced (e.g., bend) and (C)VNCvoiceless (e.g., bent) words spoken in either clear or citation speech modes. In order to enhance the percept of ／n／ in clear speech, speakers increase the duration of the nasal consonant in CVNCvoiced words but marginally increase, or even decrease, ／n／ duration in CVNCvoiceless words. While highly variable, airflow results suggest little difference on anticipatory nasalization as a function of speech mode. These results argue against models predicting a global reduction in coarticulation in clear speech.

Effects of prosodic structure on the relative timing of articulators in English lateral production.
Susan S. Lin
Abstract:
Previous research has established that American English speakers tend to produce syllable‐final ／l／ with movement of the tongue dorsum preceding movement of the tongue tip. However, the results of these studies differ with respect to the articulator timing in syllable‐initial ／l／, with some claiming synchrony (Browman and Goldstein, 1995) and others claiming asynchrony in the direction opposite that of syllable‐final ／l／ (Gick, 2003). This study uses ultrasound imaging to investigate the relative timing of the tongue tip and dorsum during production of syllable‐initial and syllable‐final ／l／ in multiple prosodic contexts. Prosody has a significant effect on both duration and extent of articulator movement in speech production—onsets of larger prosodic units involve larger and longer movement than onsets of smaller prosodic (Keating, 2006). The explanation that these effects result from speakers’ attempts to render perceptually more clear the segments that initiate phrases and utterances suggests that examining these segments at varying prosodic positions may provide insight into speakers’ knowledge of speech perception. Current preliminary results show that American English speakers may utilize at least two distinct timing relations in initial laterals, supporting a position that speaker knowledge may be variable between speakers.

Aerodynamic modeling for concatenative speech synthesis.
Kevin B. McGowan
Abstract:
Listeners can perceive and use a wide array of fine‐grained phonetic details, including the detailed coarticulatory influences of adjacent sounds, when perceiving speech. Details like anticipatory nasalization can, for example, potentially provide the listener with a rich network of informative cues and are a key to understanding listeners’ ability to disambiguate speech sounds from seemingly ambiguous input. Unfortunately, these coarticulatory cues are generally missing or contradictory in the output of speech synthesis systems. These systems work by concatenating variable‐length sound units chosen from a large database of recorded speech. Units are chosen to minimize two functions: the cost of aligning a particular unit with the desired speech output (target cost) and the cost of adjoining the next sound to the most recently selected unit (join cost). Generally, these costs are calculated using features which can be automatically extracted from the acoustic speech signal. A unit selection database is created, automatically segmented and automatically labeled with nasal and oral airflow feature vectors. These aerodynamic features are used as a proxy for articulatory information in the calculation of join and cost functions. Listeners’ mean opinion scores are obtained on output from this system and a baseline acoustic system for comparison.

Marlyse Baptista and Eric Brown spent time this summer in Cape Verde recording Cape Verdean Creole speakers on the islands of Santiago and São Vicente.  This trip also launched a long term project to digitize and and post recordings of Cape Verdean Creole on the web.

Working with linguists Emanuel de Pina from the University of Cape Verde and Saidu Bangura from the University of Santiago, the Michigan linguists collected comparative morpho-syntactic data from informants all over the two islands.  Eric remained in Cape Verde for an additional 3 weeks conducting sociolinguistic interviews and collecting recorded word lists.  Both researchers are grateful to their collaborators and colleagues who made the long trip worthwhile and look forward to continued work in the area.  Keep an eye out for future presentations and publications!

Marlyse and Eric also want to acknowledge and express gratitude for the support of the Department of Linguistics, the African Studies Center and the International Institute.  Marlyse received an African Heritage Initiative Seed Grant from the African Studies Centerto help support this work

Tim Chou presented his paper, “Feature Inheritance and Object Raising in Epistemic Modal Constructions in Mandarin Chinese” at the 2009 Mid-America Conference on Linguistics held at the University of Missouri, Oct. 9-11.

Excerpt from the abstract:

Epistemic modals like yinggai ‘should’ in Chinese have been assumed to be main predicates taking a TP complement as in (1a) since the subject can undergo A-movement to matrix Spec-T as in (1b) (cf. [8]). Besides, Lin (to appear) argued that the embedded object can raise to the matrix clause as illustrated by (2). What makes (2) interesting is that it appears to violate the Minimal Link Condition (MLC). Lin (to appear) suggests that the object raising in (2) is an instance of A’-movement, and the MLC violation is only apparent. However, this paper presents arguments challenging an A’-movement analysis of (2). First, given that Weak Crossover Effect (WCO) is a typical diagnosis of A’-movement (*Whoi does [hisi mother] like ti?), if object raising in (2) is an instance of A’-movement, we should expect it to be ruled out by WCO. However, this prediction is not borne out as shown in (3). Second, if object raising in (2) is actually A-movement, it should be able to feed binding condition A (Johni seems to himselfi [ti to like Mary]). This prediction is borne out as evidenced by (4). (3) and (4) jointly indicate that the object in (2) is an A-movement targeting the matrix Spec-T, rather than an instance of A’-movement. However, we need to explain why object raising in (2), as A-movement, is not excluded by MLC in Chinese.
Following Chomsky’s idea of feature inheritance ([3] and [4]), the unvalued φ-features on T are inherited from the phase head C. I assume that the embedded T in (2) does not contain any unvalued φ-features due to the lack of the CP-layer. It only has the inherent EPP feature, which needs to have only its edge filled. Crucially, in the absence of unvalued φ-features, the EPP does not impose any restriction on minimality. It only needs to attract some active goal with unvalued features to Spec-T, regardless of minimality. What determines the minimality effect in the derivation is the probe-goal relation between unvalued features, not the EPP. For this reason, the embedded object wancan ‘dinner’ may raise to embedded Spec-T for EPP requirements, producing (5a). Next, the matrix T with the unvalued φ-features inherited from C probes for the closest goal wancan ‘dinner’ at the embedded Spec-T and attracts it to its specifier for EPP as in (5b) (= (2)). In sum, the EPP alone does not force the applicability of the MLC, and the apparent MLC violation of object raising in (2) results from the lack of unvalued φ-features on the embedded T. Note that the effect of the MLC in Chinese is still evidenced elsewhere, as in (5c). The probe-goal relation between the matrix T and Zhangsan at Spec-v* cannot be established since the head of the A-chain (wancani, ti) at embedded Spec-T intervenes. As a result, Zhangsan at Spec-v* cannot move to matrix Spec-T.

The analysis  shown in this paper contributes to the long-standing debate on the motivation for A-movement in linguistic theory. Specifically, it presents evidence against the universality of a checking-based theory of A-movement (cf. [1], [5], and [6]) because even though the embedded T does not contain any unvalued features except for its inherent EPP structural requirement, argument raising still occurs. Moreover, the fact that the embedded object can raise to matrix Spec-T via embedded Spec-T in (2) indicates that Mandarin Chinese employs the delayed version of the Phase Impenetrability Condition in [2] as in (6) so that the object within the domain of v* is accessible to EPP on the embedded T. We argue that the employment of delayed PIC is related to the lack of Case-valuation by v* in Chinese (in contrast to English) as implied by the analysis of Icelandic in [7].

Anna Babel presented a paper entitled “Valley Spanish: Origin and orientation” at the First International Workshop on Andean Spanish at the Freiburg Institute for Advanced Studies.

From Erica:

I attended the first three week session of the LSA Institute in Berkeley California with the intention of taking a number of courses on language acquisition and psycholinguistics. (I had initially registered for Prosody and Language Comprehension, Auditory Word Recognition, Infant Language Acquisition and Cross Linguistic Language Acquisition.) However, there were so many other interesting topics and lecturers, that I found myself sitting in on a lot of other courses just for my own edification. This wide exposure to a lot of varied subject matter was very stimulating and inspiring.

Living in the dorms allowed me to meet quite a few other students of Linguistics from all over the world, and we had quite a bit of fun during off-hours hiking in the Berkeley Hills, and seeing the local sites.

I wouldn’t hesitate the recommend participation in future LSA Institutes to anyone who is interested!

From the series:  Linguists in the Woods

From Joseph:

“This summer at the Linguistics Summer Institute, I was able to take courses in Prosody, Pragmatics, Intonational Typology and more. Furthermore, I was able to work closely with scholars from around the world on my discourse prosody research project, bringing perspective and insight into my current work and ideas for valuable follow-up studies. Particularly valuable were conversations with Delphine Dahan from Penn and Carlos Gussenhoven from the Netherlands. After talking with each of them, I was just flush with excitement!

The institute was a ton of fun. The people were super friendly and linguistics permeated everything. Where else would the average person know about glottal stops and conversational implicatures?”

The Department welcomes five new graduate students this year.

Tridha Chatterjee is interested in language contact in India

Yan Dong is interested in syllable structure and other aspects of phonological theory

Harim Kwon is interested in the phonetic-phonology interface

Candice Scott is interested in language contact, sociolinguistics and historical linguistics

Yiwen Zhou is interested in sociolinguistics and minority languages in China

Lauren Squires presented her paper, Voicing ‘Sexy Text’: TV News Representations of the Detroit Text Messaging Scandal at the Language in the (New) Media conference in Seattle, Sept. 2-5.

Abstract:

The print media have tended to represent computer‐mediated communication, including text messaging, in a negative light and as a youth‐based practice (cf. Thurlow 2003, 2006; Crystal 2008). Yet as CMC’s use continues to expand, so does its media representation. This paper addresses the representation of text messaging as practiced by adults, through a case study of TV news coverage of text messages. In 2008, Detroit Mayor Kwame Kilpatrick resigned after thousands of text messages implicated him in a range of offenses, including an affair with his Chief of Staff. Messages sent by the Mayor and other city employees became the centerpiece of local public and political discourse about the events, variously named “text‐gate,” “text messaging scandal,” and “sexy text scandal.” The scandal thus compelled local media to talk about particular adults’ text messaging practices and to represent the language used therein. This case affords an exploration of how text messaging is represented in non‐youth contexts, and moreover of how TV broadcasts use multiple modalities as representational resources when “translating” texts for a TV audience. The data comprise over 100 instances of text messages that are read aloud by TV news anchors in scandal coverage, from three Detroit stations. In analyzing how the fundamentally visual language of the text messages is represented through both visual and oral modalities, several representational variables at different levels of language and discourse structure will be discussed, including:

1) replication (the messages’ reproduction on‐screen and aloud);

2) intonation (the messages’ intonational marking);

3) organization (the messages’ sequential presentation);

4) framing (the messages’ introductions).

Preliminary analysis shows that in general, the act of texting is represented as ordinarily conversational, involving rapid back‐and‐forth exchange between two participants. However, the language used within texting is represented through a “read speech” style, and broadcasters gloss stereotypical “CMC” lexical items (e.g., lol) into standard language or omit them altogether. Hence, while the broadcasters’ presentation treats the fact of these messages’ transmission through text as generally unremarkable, novel features or uses of the medium tend to be either metalinguistically highlighted or erased.

Li Yang successfully defended his dissertation,  Re-evaluating and Exploring the Contributions of Constituent Grammar to Semantic Role Labeling, on Sept. 4.

Committee:  Steve Abney (Chair), George Michailidis, Drago Radev, Rich Thomason

Li will continue his work for Janya in Buffalo, NY, a company that develops information extraction software.

Congratulations, Dr. Yang!

Abstract:

Since the seminal work of Gildea and Jurafsky (2000), semantic role labeling (SRL) researchers have been trying to determine the appropriate syntactic/semantic knowledge and machine learning algorithms to tackle the challenges in SRL. In search of the appropriate knowledge, SRL researchers
shifted from constituency grammar to dependency grammar around 2007 due to the suspension in improvement in the systems relying on features based on constituency grammar. However, the results from the CoNLL-2008 SRL systems, all of which utilized dependency grammar-based features, did not support the hypothesis that dependency grammar was more suitable for SRL. Therefore, determining the right syntactic/semantic knowledge for SRL still remains an open question. This entails that finding the right syntactic/semantic knowledge to create features that generalize across the syntactic variations that a verb appears in and involve argument movement or displacement remains a challenge as well.

The current dissertation continues the effort to discover the appropriate syntactic/semantic knowledge for SRL. Specifically, while seeking the proper features to solve the SRL problem in general, the present work focuses on tackling the syntactic variation challenge by integrating three types of less thoroughly explored knowledge in constituency grammar-based SRL systems, including context dependence among the semantic roles of core arguments, syntactic structures involving argument movement or displacement, and dependency grammar relations. Integrating such knowledge leads to the following novel approach.

The system identifies the core and non-core semantic arguments of a verb. To classify a non-core argument, the system uses a set of generic features. For a core-argument, the system relies on the preceding types of knowledge to extract the base argument configuration (BAC) feature in which the core arguments’ positions overlap with those of an argument structure of the verb. As a result, BAC features generalize across the syntactic variations a verb appears in. Together with the two levels of backoff features dealing with unrealized core arguments and unknown verbs respectively, BAC features effectively solve the argument classification task and successfully handles the preceding challenge. However, the experimental results indicate that  the overall performance is affected by the argument identification module. The immediate future work would be to improve the identification module.

Michael Marlo (Ph.D.  2007) has been offered a position as research scientist in African languages and Pashto at the Center for the Advanced Study of Language at the University of Maryland.

Shortly before he received the offer, Michael and his wife welcomed Jayden Marlo to the world.

Many congrats Mike and Jacinta