Presents a method for objectively comparing identity resolution technologies across data sources. Describes evaluation metrics and testing infrastructure intended to assess system performance against different operational requirements.

Develops a unified model of predicate composition in which word and phrase properties follow from syntactic structure and interface conditions. Drawing especially on Persian and Armenian, the dissertation argues for shared grammatical primitives and parallel nominal and verbal structures.

Reviews differences between Persian heritage and second-language speakers in vocabulary, morphology, syntax, literacy, and functional competence. Argues that heritage instruction should build on learners' existing strengths while addressing their specific linguistic needs.

(not public released)

Compiles corpora and language analysis tools for languages of Afghanistan and Pakistan.

Examines how specificity and the internal structure of Persian wh-phrases relate to their position and interpretation. Proposes two focus positions that distinguish requests for new information from exhaustive or contrastive identification.

Evaluates automated translation quality metrics for English-Iraqi Arabic speech translation in the TRANSTAC program. Examines challenges posed by dialogue structure and Arabic, and explores metric variants intended to improve the reliability of evaluation scores.

Describes the components and conventions of Afghan personal names, including compound given names, honorifics, and tribal, geographic, and professional identifiers. Discusses variation in family names, dialects, and English transcription.

Provides Dari transcription guidelines for the TRANSTAC program, supporting the preparation of spoken-language data for evaluating tactical speech translation systems. (Not public released)

Presents metrics, test data, and an evaluation infrastructure for comparing identity resolution technologies across data sources and operational needs, including the challenges of sharing evaluation data while protecting privacy.

Examines telicity in Persian complex predicates through evidence from semelfactive constructions, challenging the claim that the nonverbal element alone determines whether an event has an inherent endpoint.

Surveys research on the Iranian blogosphere as of 2008, emphasizing gaps in linguistic and computational analysis. Considers how research on English blog language can inform the study of Persian blogs.