About

Sandiway Fong is a professor in the Department of Linguistics at the University of Arizona, where he is also affiliated with the Department of Computer Science, the Cognitive Science Program, and SLAT. He is a computational linguist with interests spanning all aspects of language and computation. Fong directs the Human Language Technology (HLT) Master's program and serves as an associate professor, contributing to research at the intersection of linguistics and computation. His work involves exploring how computational methods can be applied to understand and process human language, emphasizing the integration of linguistic theory and computational techniques.

Research topics

• Computer Science
• Artificial Intelligence
• Natural Language Processing
• Psychology
• Information Retrieval
• Linguistics
• Cognitive science
• Neuroscience
• Philosophy
• Programming language

Selected publications

• A New Era with Artificial Intelligence Powered Learner Corpora for Teaching and Learning of English as a Second Language in Primary School Education

  Applied Language Sciences · 2025-12-06

  articleOpen accessSenior author

  This innovation reports on the design, implementation and evaluation of an AI-powered learner corpora to support English writing in a Malaysian national primary school. The intervention targeted Year 4 to Year 6 pupils learning English as a second language within the CEFR-aligned KSSR Semakan curriculum, in which a significant portion struggled with vocabulary, cohesion and sentence boundaries writing around A1 to A2 level. This work is in response to two competing tensions: i) pupils' increasing use of generative AI, which produces fast but less human-involved and locally-grounded language, and ii) the underutilisation of corpus linguistics in primary classrooms due to technical and training demands. A 152,187 word corpus of 560 anonymised pupil essays was compiled and uploaded into a retrieval-augmented custom GPT. This AI-powered learner corpora provided examples and feedback drawn only from local pupil texts and was embedded into a three-stage writing cycle: guided drafting, AI-mediated exploration and targeted revision. The design was informed by the Technology Acceptance Model and the Expectation-Confirmation Model, focusing on perceived usefulness, ease of use, satisfaction and continuance intention. Classroom observations and teacher reflections showed that pupils became more willing to revise and exhibited clearer control of basic sentence structure and cohesion. Teachers reported reduced example-generation workload and more focused conferencing. However, the impact on idea development was modest, some weaker writers engaged in formulaic borrowing and digital literacy differences shaped access to benefits. This paper concludes with practical recommendations for pedagogy, equity-focused scaling and future mixed-methods research on AI-powered learner corpora in school-based ESL writing.

  PublisherOA PDFDOI

• Large language models are not about natural language

  ArXiv.org · 2025-12-15

  preprintOpen accessSenior author

  Large Language Models are useless for linguistics, as they are probabilistic models that require a vast amount of data to analyse externalized strings of words. In contrast, human language is underpinned by a mind-internal computational system that recursively generates hierarchical thought structures. The language system grows with minimal external input and can readily distinguish between real language and impossible languages.

  PublisherOA PDFDOI

• On the nature of FormSet

  Linguistic Variation · 2025-05-13

  articleOpen access1st authorCorresponding

  Abstract FormSet, proposed by Chomsky (2021) , is one of two primitive set formation operations on Workspace items in the theory of I-Language, the other primitive is Merge. In this paper, we investigate the particular properties of FormSet, distinct from Merge, across phenomena in the noun phrase and verb phrase domains. In particular, Workspace items input to FormSet must be a coherent collection of items that obey natural conditions on parallelism. We define what parallelism means, both in terms of pre-conditions for FormSet, and as conditions on subsequent operations such as Merge applying to a set built by FormSet. In doing so, we obtain new, yet simple analyses for classic data in accordance with the Strong Minimalist Thesis (SMT).

  PublisherDOI

• Towards a Biologically-Plausible Computational Model of Human Language Cognition

  Proceedings of the 14th International Conference on Agents and Artificial Intelligence · 2024 · 3 citations

  Senior authorCorresponding
  • Computer Science
  • Computer Science
  • Natural Language Processing
  PublisherDOI

• Three reasons why AI doesn’t model human language

  Nature · 2024-03-19 · 10 citations

  letter
  PublisherDOI

• Merge and the Strong Minimalist Thesis

  2023 · 70 citations

  • Computer Science
  • Artificial Intelligence
  • Computer Science

  The goal of this contribution to the Elements series is to closely examine Merge, its form, its function, and its central role in current linguistic theory. It explores what it does (and does not do), why it has the form it has, and its development over time. The basic idea behind Merge is quite simple. However, Merge interacts, in intricate ways, with other components including the language's interfaces, laws of nature, and certain language-specific conditions. Because of this, and because of its fundamental place in the human faculty of language, this Element's focus on Merge provides insights into the goals and development of generative grammar more generally, and its prospects for the future.

  PublisherDOI

• Modeling Syntactic Knowledge With Neuro-Symbolic Computation

  2023-01-01 · 1 citations

  articleOpen access
  PublisherDOI

• On the computational modeling of English relative clauses

  Open Linguistics · 2023 · 4 citations

  1st authorCorresponding
  • Computer Science
  • Computer Science
  • Natural Language Processing

  Abstract Even in this era of parameter-heavy statistical modeling requiring large training datasets, we believe explicit symbolic models of grammar have much to offer, especially when it comes to modeling complex syntactic phenomena using a minimal number of parameters. It is the goal of explanatory symbolic models to make explicit a minimal set of features that license phrase structure, and thus, they should be of interest to engineers seeking parameter-efficient language models. Relative clauses have been much studied and have a long history in linguistics. We contribute a feature-driven account of the formation of a variety of basic English relative clauses in the Minimalist Program framework that is precisely defined, descriptively adequate, and computationally feasible in the sense that we have not observed an exponential scaling with the number of heads in the Lexical Array. Following previous work, we assume an analysis involving a uT feature and uRel feature, possibly simultaneously valued. In this article, we show a detailed mechanical implementation of this analysis and describe the structures computed for that , which , and who/whom relatives for standard English.

  PublisherOA PDFDOI

• Even deeper problems with neural network models of language

  Behavioral and Brain Sciences · 2023-01-01 · 1 citations

  article

  We recognize today's deep neural network (DNN) models of language behaviors as engineering achievements. However, what we know intuitively and scientifically about language shows that what DNNs are and how they are trained on bare texts, makes them poor models of mind and brain for language organization, as it interacts with infant biology, maturation, experience, unique principles, and natural law.

  PublisherDOI

• Simple Models : Computational and Linguistic Perspectives

  Institutional Repositories DataBase (IRDB) · 2022-03-14

  articleOpen access1st authorCorresponding
  PublisherOA PDF

Frequent coauthors

• Robert C. Berwick

  10 shared

• Jason Ginsburg

  Kyoto University

  9 shared

• C. Lee Giles

  4 shared

• Anna Maria Di Sciullo

  Université du Québec à Montréal

  4 shared

• Beracah Yankama

  Vassar College

  2 shared

• Hilton Alers-Valentín

  University of Puerto Rico-Mayaguez

  2 shared

• Thomas G. Bever

  University of Arizona

  2 shared

• Sandra Lawrence

  Griffith University

  2 shared