Research

My work turns complex language data into evidence about comprehension and legal meaning. Each project below states the question, the data, the methods, my contribution, and its status.

Predicting Miranda-warning comprehension (dissertation)

What computational and linguistic features predict how difficult a Miranda warning is to understand?

Data / corpus: Miranda warning texts and elicited-imitation comprehension data.
Methods: NLP-driven feature extraction, corpus analysis, predictive modeling, mixed-methods design.
My contribution: Doctoral researcher — study design, feature engineering, modeling, and analysis.
Status: In progress (dissertation). Related paper in progress: Predictive Complexity Modeling of Miranda Warnings.
Collaborators: Supervisor: Scott Jarvis. Committee: Ainsworth, Finegan, Larsson, Egbert.

Model Utah Jury Instructions (MUJI)

How comprehensible are standard jury instructions, and how can plain-language redesign improve them?

Data / corpus: Large corpus of jury-instruction language.
Methods: Large-scale corpus analysis, NLP feature extraction and annotation, quantitative and mixed methods.
My contribution: Paid research assistant / computational language data analyst.
Status: Ongoing (2024–present). Multiple manuscripts under review (Applied Linguistics; IJSLL).
Collaborators: PIs: Scott Jarvis, Jesse Egbert. McKenzie Endowment for Democracy Grant, NAU.

Sense-tagging legal language

Can corpus-based disambiguation resolve ambiguous word senses in jury instructions?

Data / corpus: Jury-instruction corpus with sense annotations.
Methods: Corpus-based word-sense disambiguation, semantic gold-standard annotation.
My contribution: Lead author.
Status: Published 2026 - International Journal for the Semiotics of Law.
Collaborators: Solo (Akbary, M.).

Lexical diversity & American TV genres

Does lexical diversity predict genre across American television?

Data / corpus: 4.2M-word genre-balanced American television corpus.
Methods: Corpus construction, lexical-diversity metrics, genre classification.
My contribution: Lead author; built the corpus.
Status: Published 2023 - Digital Scholarship in the Humanities (OUP).
Collaborators: Scott Jarvis.

Elicited imitation & listening comprehension

Can elicited imitation validly measure listening comprehension in forensic contexts?

Data / corpus: L2 elicited-imitation and comprehension task data.
Methods: Experimental design, statistical analysis, mixed methods.
My contribution: Lead / co-author across a paper series.
Status: Published 2023–2024 (Research Methods in Applied Linguistics; IJSLL).
Collaborators: Benzaia, Jarvis, Park.