Computer Science

The emerging field of synthetic biology moves beyond conventional genetic manipulation to construct novel life forms which do not originate in nature. We explore the problem of designing the provably shortest genomic sequence to encode a given set of genes by exploiting alternate reading frames. We present an algorithm for designing the shortest DNA sequence simultaneously encoding two given amino acid sequences. We show that the coding sequence of naturally occurring pairs of overlapping genes approach maximum compression. We also investigate the impact of alternate coding matrices on overlapping sequence design. Finally, we discuss an interesting application for overlapping gene design, namely the interleaving of an antibiotic resistance gene into a target gene inserted into a virus or plasmid for amplification.

Identifying and assaying the relative abundance of members of complex microbial communities is an important problem in ecology. Sandberg et al. 11 investigated the usage of genomic signatures to provide high identification percentages from short sequence samples. In this paper we present an improved naive Bayesian classification method using conditional probabilities, which can be used to classify unsequenced bacterial species, as well as identify and predict the frequency of the dominant species in mixed microbial populations.

What progress prevention research has made comes through strategic partnerships with communities and institutions that host this research, as well as professional and practice networks that facilitate the diffusion of knowledge about prevention. We discuss partnership issues related to the design, analysis, and implementation of prevention research and especially how rigorous designs, including random assignment, get resolved through a partnership between community stakeholders, institutions, and researchers. These partnerships shape not only study design, but they determine the data that can be collected and how results and new methods are disseminated. We also examine a second type of partnership to improve the implementation of effective prevention programs into practice. We draw on social networks to studying partnership formation and function. The experience of the Prevention Science and Methodology Group, which itself is a networked partnership between scientists and methodologists, is highlighted.

We would like to thank Judith Aissen, Barbara Citko, Hana Filip, Adam Milton Morgan,andtheaudiencesoftheUCBerkeleySyntaxCircle,FASL19,FDSL8.5and the 2011 LSA Annual Meeting for their constructive comments and suggestions. Also, thank you to the anonymous FASL reviewers for their constructive insights. This research was supported in part by funding from the National Heritage Language Resource Center at UCLA and from the Faculty of Arts and Sciences at Harvard University and the Davis Center at Harvard. All errors are our sole responsibility. 12 We assume that syntax does not provide linearization statements. 13 Note that additional faithfulness constraints are needed to prevent operations such as deletion from being a licit PCC repair strategy.

This paper presents the first-ever processing experiment on relativization in Avar, an ergative language with prenominal relatives. The results show no processing difference between the ergative subject gap and the absolutive object gap. The absolutive subject gap, however, is processed much faster. We propose a principled explanation for this result. On the one hand, Avar has a subject preference (cf. the Accessibility Hierarchy, Keenan and Comrie, 1977), which would make the processing of the ergative and the absolutive subject gap easier than the processing of the absolutive object gap. On the other hand, the ergative DP in a relative clause serves as a strong cue that allows the parser to project the remainder of the clause, including the absolutive object DP (cf. Marantz, 1991, 2000); such morphological cueing favors the absolutive object gap. Thus, two processing preferences, the one for subject relatives and the other for morphologically cued clauses, cancel each other out in terms of processing difficulty. As a result, reading time results for the ergative subject and absolutive object relative clauses are very similar. The overall processing results are significantly different from what is found in accusative languages, where subject preference and morphological cueing reinforce each other, leading to a strong transitive subject advantage.

We describe a new task-based corpus in the Spanish language. The corpus consists of videos, transcripts, and annotations of the interaction between a naive speaker and a confederate listener. The speaker instructs the listener to MOVE, ROTATE, or PAINT objects on a computer screen. This resource can be used to study how participants produce instructions in a collaborative goal-oriented scenario, in Spanish. The data set is ideally suited for investigating incremental processes of the production and interpretation of language. We demonstrate here how to use this corpus to explore language-specific differences in utterance planning, for English and Spanish speakers.

""Language production research has focused on planning at the lexical and syntactic levels. Consequently, little is known about speakers’ strategies for distributing information across multiple utterances. We present evidence that (a) the overall amount of information in an intended message affects how speakers distribute that information across clauses, and (b) speakers have relatively early access to (at least an estimate of) message complexity, specifically before word retrieval. The observed effect is unexpected for competing theories of sentence production based on availability or macro-propositional
accounts. We discuss these findings in an information theoretic context coupled with a sketch of a model of limited resources at the level of macro-propositional planning.""

We describe a new multimodal corpus currently under development. The corpus consists of videos of task-oriented dialogues that are annotated for speaker’s verbal requests and domain action executions. This resource provides data for new research on language production and comprehension. The corpus can be used to study speakers’ decisions as to how to structure their utterances given the complexity of the message they are trying to convey.

The goal of my research is to understand speech input in a continuous manner by treating the input stream as fragmental utterances. This allows us to use various approaches to predict what comes downstream. Possible interpretations are trimmed by such predictions which in turn also allow us to complete information not readily available in the fragmental utterance. Semantic frames can encode all possible arguments for domain actions. As utterances are processed continuously, appropriate frames can be activated so that fragment interpretations can fill, correct or extend frames under consideration. In turn, feedback can be provided to the parser as the frames are manipulated possibly based on the completeness of the semantic frame construction.

We describe an annotation scheme aimed at capturing continuous understanding behavior in a multimodal dialogue corpus involving referential description tasks. By using multilayer annotation at the word level as opposed to sentence level, we can better understand the role of continuous understanding in dialogue. To this end, we annotate referring expressions, spatial relations, and speech acts at the earliest word that clarifies the speaker’s intentions. Word-level annotation allows us to trace how referential expressions and actions are understood incrementally. Our corpus has intertwined language and actions which help identify the relationships between language usage, intention recognition, and contextual changes which in turn can be used to develop conversational agents that understand language in a continuous manner.

Current dialogue systems generally operate in a pipelined,modular fashion on one complete utterance at a time. Evidence from human language understanding shows that human understanding operates incrementally and makes use of multiple sources of information during the parsing process, including traditionally “later” components such as pragmatics. In this paper we describe a spoken dialogue system that understands language incrementally, provides visual feedback on possible referents during the course of the user’s utterance, and allows for overlapping speech and actions. We further present findings from an empirical study showing that the resulting dialogue system is faster overall than its nonincremental counterpart. Furthermore, the incremental system is preferred to its nonincremental counterpart – beyond what is accounted for by factors such as speed and accuracy. These results indicate that successful incremental understanding systems will improve both performance and usability.

"Current dialogue systems generally operate  in a pipelined, modular fashion on one  complete utterance  at a time. Converging evidence shows that human understanding operates incrementally and makes use of multiple sources of information during the  parsing process, including traditionally “later” aspects such as pragmatics. We describe a spoken dialogue  system  that understands language incrementally, gives visual feedback on possible  referents during the course  of the user’s utterance, and allows for overlapping speech and actions. We  present findings from an empirical study showing
that the  resulting dialogue  system  is faster overall than its nonincremental counterpart. Furthermore, the incremental system  is preferred to its counterpart – beyond what is accounted for  by factors such as speed and accuracy. These resultsare the first to indicate, from a controlled user  study, that successful incremental
understanding systems will improve both performance and usability."

The prevalent state of the art in spoken language understanding by spoken dialog systems is both modular and whole-utterance. It is modular in that incoming utterances are processed by independent components that handle different aspects, such as acoustics, syntax, semantics, and intention / goal recognition. It is whole-utterance in that each component completes its work for an entire utterance prior to handing off the utterance to the next component. However, a growing body of evidence suggests that humans do not process language that way. Rather, people process speech by rapidly integrating constraints from multiple sources of knowledge and multiple linguistic levels incrementally, as the utterance unfolds. In this paper we describe ongoing work aimed at developing an architecture that will allow machines to understand spoken language in a similar way. This revolutionary approach is promising for two reasons: 1) it more accurately reflects contemporary models of human language understanding, and 2) it results in empirical improvements including increased parsing performance. Index Terms: dialogue systems, speech understanding, psycholinguistics, parsing, incremental understanding.

The language production system is assumed to be universal, yet surprisingly few languages have been investigated to date. We tested the effects of accessibility on constituent order variation in Yukatek Maya, an indigenous language of Mexico, which is typologically very different from Indo-European and other languages. Through this study, we contribute to the emerging discipline of cross-linguistic psycholinguistics

Abstract Current dialogue systems generally operate in a pipelined, modular fashion on one complete utterance at a time. Converging evidence shows that human understanding operates incrementally and makes use of multiple sources of information during the parsing process, including traditionally “later” aspects such as pragmatics.

Abstract We describe a new task-based corpus in the Spanish language. The corpus consists of videos, transcripts, and annotations of the interaction between a naive speaker and a confederate listener. The speaker instructs the listener to MOVE, ROTATE, or PAINT objects on a computer screen. This resource can be used to study how participants produce instructions in a collaborative goal-oriented scenario, in Spanish.

Islands have long been known to differ in strength (weak, eg, WH, CNPC, vs. strong islands, eg, Subject, Adjunct). Strong islands have been shown to be penetrable by experimental researchers beginning with Hiramatsu 1999 and Snyder 2000 (see also Hiramatsu 2000, Braze 2002, Goodall 2004b, Kravtchenko et al. 2009, Francom 2009, Jurka 2009, among others).