Please join us in celebrating the wonderful news that Professor Brian Dillon has been selected as one of this year’s three recipients of the prestigious Samuel F. Conti Faculty Fellowship Awards.

In addition to being specially recognized for the excellence of his research, Professor Dillon will receive a one-year release from all teaching and service duties. This will allow him to devote his creative energies to an exciting research project with Tal Linzen (NYU), investigating and deepening our understanding of how Large Language Models differ from humans in how they resolve syntactic and semantic ambiguity.

In his selection for the Conti Fellowship, Professor Dillon joins the ranks of previous illustrious Conti Fellows in our department, such as Lisa Green (2017), Peggy Speas (2006), Angelika Kratzer (1999), John McCarthy (1997), Lyn Frazier (1993), Lisa Selkirk (1991), Tom Roeper (1989), Emmon Bach (1982), and Barbara Partee, who was among the very first Conti Fellows in 1981.

Congratulations, Brian!

Congratulations to Alessa Farinella who has won a coveted Summer Dissertation Fellowship from the UMass Graduate School for the 2025 summer!

Below is information about some recent work by Ali Nirheche (with short descriptions by Ali).

Momayiz, Imane., Outchakoucht, Aissam., Choukrani, Omar., & Nirheche, Ali. (2024). TerjamaBench: A culturally specific dataset for evaluating translation models for Moroccan Darija. AtlasIA. Published on Hugging Face (machine learning and data science platform)

• “What we did is we compared prominent AI models (GPT4o, Claude 3.5 sonnet, Gemini) as well as few models that my colleagues have worked on for Moroccan Arabic (e.g., Terjman-Large-v1.2). We compared the performance of these models with respect to Moroccan Arabic (how well their output resembles how Moroccan Arabic is actually used).”

Nirheche, Ali. (2025). Moroccan Arabic Plurals Corpus [Data set]. Zenodo. https://doi.org/10.5281/zenodo.14642330

• “This is a corpus of Moroccan Arabic plural that I published back in January. The corpus contains 1,166 singular-plural noun pairs in Moroccan Arabic.”

Nirheche, Ali. 2024. Shiny App for MaxEnt with Hidden Structure. Shiny Application. Amherst, MA: University of Massachusetts Amherst. https://alingwist.shinyapps.io/HGR_app/

• “This is a user-friendly Shiny application, MaxEnt with Hidden Structure in R, that I developed. It’s designed to assist linguists in generating phonological grammars (weights) using a Maximum Entropy model.” (more info about the project here, funded by NSF grant with PI Joe Pater).

Sam Tilsen (Cornell University) will give a colloquium talk on Friday April 4 at 3.30 in ILC S221.

Systems, states, trajectories, and forces: on a dynamical vocabulary for understanding phonological  patterns
Many theories of phonological representation make use of hierarchical structure schemas comprised of segments, syllables, and syllable-internal constituents such as onsets, nuclei, codas, and rimes, or moras. An examination of the conceptual metaphors that underlie such schemas calls into question their utility for understanding the temporal organization of action in speech. This talk contrasts the metaphors of this more traditional, structural conception with those of a dynamical vocabulary, in which the states of hypothesized systems evolve in time according to change rules that incorporate system interactions. From this perspective, it is argued that the temporal organization of action can be understood parsimoniously with just two types of systems, gestural systems and vocal tract control systems, in combination with the incorporation of internal and external feedback mechanisms in change rules. This theory is applied to explain cross-linguistic variation in phonological and phonetic patterns associated with the moraic status of codas and word-initial consonant clusters. A notable consequence of the theory is that hierarchical structure schemas do not describe temporal organization in any particular utterance; rather, they summarize a recurring developmental progression in which the learning of internal predictive sensorimotor models allows actions that are competitively controlled early in development to become coordinatively controlled.

At this year’s Human Sentence Processing conference, Maayan Keshev was awarded the Gibson/Fedorenko Young Scholar Award for work presented at the 2024 HSP conference. Maayan’s talk was titled ‘A transient binding model of interference in sentence processing‘ (along with Mandy Cartner, Aya Meltzer-Asscher, and Brian Dillon).

The Gibson/Fedorenko Young Scholar Award is meant to honor young scholars who deliver outstanding work in sentence processing at the HSP conference since 2022, continuing a tradition of the Jerrold J. Katz Young Scholar Award that was given to outstanding young scholars at the HSP (formerly CUNY) conference in honor and loving memory of Jerry Katz. As part of the Gibson/Fedorenko award, two young scholars are awarded $1000.

Here is Maayan receiving her award from Ted Gibson!

Shortly after the UMass Linguistics 50th anniversary reunion, Elan Dresher (PhD 1978) sent Gaja Jarosz and Joe Pater this history of his work on learning of parameter settings for stress.

It occurred to me that work on computational learning theories for metrical phonology has an even older connection to UMass Linguistics than anyone knows except me. So you might be interested in this story, which I pass on to you, as the keepers of UMass Linguistics history.

My first job after I left UMass was at Brown, and I remained in touch with Jean-Roger Vergnaud, who had been at UMass when I was a student there and was a mentor who helped me a lot in my time there. Jean-Roger suggested that we put in for an NSF grant to develop a computational learning model for metrical phonology, which we did sometime around 1980. We were unsuccessful, and I vividly remember the comments of one of the reviewers. The reviewer wrote that the project we were proposing was a logical next step that combined parametric metrical theory with thinking about learnability. Indeed, it was such an obvious thing to do that there was no need to fund us! It is inevitable, the reviewer went on, that someone, maybe several people, would come up with the exact same idea and would do it with or without a grant. 

Some years later, around 1983, I had moved to Ottawa and began working with Jonathan Kaye. He, too, was in touch with Jean-Roger (cf. the early work on Government Phonology), and J-R told him about our unsuccessful grant proposal. Jonathan suggested that we put in for the same project to SSHRC, and this application was successful. We both went out and bought Rainbow 100 computers, did a crash course in PROLOG with Peter Roosen-Runge, a computer scientist and friend of Jonathan who was a Co-PI on the grant, and the result was the project we reported on in our paper in Cognition (1990). This was, to my knowledge, the first computer program that attempted to learn metrical stress. Contrary to the prediction of the NSF reviewer, nobody else had come up with this obvious project in the meantime.

Alessa Farinella will present her research at the HFA Days Graduate Student Expo on March 27. The title of her talk is Inter- and intra- speaker variability in “BIN” utterances in African American English”. See the abstract below.

Inter- and intra- speaker variability in “BIN” utterances in African American English 
In African American English, one way of expressing the remote past is with a construction sometimes referred to as “stressed BIN” (Rickford 1973, 1975). For example, a sentence like “Mae been announcing the winners” could mean that Mae is announcing the winners, and has been doing so for a long time. It has been reported that in order for this construction to express the remote past meaning, there must be higher pitch and intensity on the word “been.” However, it has also been claimed that there is much variability across and within speakers (Green et al. 2022). In this project, I characterize the phonetic variability in these types of utterances using data-driven pitch measurements. I demonstrate that speakers vary their pronunciation in systematic ways across the entire contour, and not just on the word expressing the remote past (i.e., “been”). Better understanding this variation in pronunciation can help us understand the relationship between these phonetic characteristics and the factors that condition them.

The 38th annual conference on Human Sentence Processing took place in College Park, MD at the University of Maryland on March 27-29. UMass Scholars, past and present, were well represented in the program!

There were a number of platform presentations with UMass scholars:

• Shota Momma, Norvin Richards, and Victor Ferreira – Speakers encode silent structures: evidence from complementizer priming in English.
• Caroline Andrews, Sebastian Sauppe, Roberto Zariquiey, and Balthasar Bickel – Building a Cross-Linguistic Typology of Sentence Planning from Case Alignment
• Fernanda Ferreira, Julie Bannon, Madison Barker, Beverly Cotter, Casey Felton, Barbora Hlachova, and Adrian Zhou – Rethinking Prosodic Phrasing

Including a special demo:

• Lauren Salig, Erika Exton, Craig Thorburn, Alex Krauska – Conveying psycholinguistic concepts to general audiences: An interactive, problem-solving approach

There are many poster presentations from UMass Scholars as well:

• Yuhui Huang, Anthony Yacovone, and Jesse Snedeker – Switching meanings and forms: An ERP study on multilingual language processing in Mandarin-English bilinguals
• Beverly Cotter, Alberto Falcon, and Fernanda Ferreira – Flexibility in Bilingual Grammar: Judgments and Production of Noun-Adjective Sequences in Spanish-English Speakers
• Eva Neu, Maayan Keshev, and Brian Dillon – Modeling agreement attraction effects in vector space
• Jane Li and Grusha Prasad – Modeling morphological production with an algorithmically specified InflACT-R
• Satoru Ozaki and Shota Momma – Evaluating LLMs for abstract linguistic generalization using English parasitic gaps
• Briony Waite, Tatyana Levari, Anthony Yacovone, and Jesse Snedeker – Lexical access during naturalistic listening in middle childhood and early adolescence
• Thomas Hansen, Anthony Yacovone, Ivi Fung, and Gina Kuperberg – Changing the narrative: ERP markers of building and updating situation models during deep naturalistic comprehension
• Ashlyn Winship, Zander Lynch, John R. Starr, Yifan Wu, Lucas Li, and Marten van Schijndel – Experimentally extracting implicit instruments
• Özge Bakay, Faruk Akkuș, and Brian Dillon – Hierarchical relations in memory retrieval: Evidence from a local anaphor in Turkish
• Anzi Wang, Carolyn Anderson, and Grusha Prasad – To know what you might say, I will probably need to know the event type
• Mara Breen and Katerina Drakoulaki – Prosodic fluency in productions of The Cat in the Hat predicts reading comprehension skill in 6-10-year-olds
• Thomas Morton, Amber Jiang, and Victor Ferreira – Reaching for the unknown: sentence planning under message uncertainty and expectation violation
• Katerina Drakoulaki and Mara Breen – From lab to neighborhood: enhancing child language research through community-based collaborations
• Zander Lynch and Helena Aparicio – Failing Alternatives Lower the Acceptability of Definite Descriptions
• Adrian Zhou, Matthew Lowder and Fernanda Ferreira – Camping Tigers, Hiking Dragons: Dangling Modifiers Do Not Add Processing Difficulty
• Mandy Cartner, Brian Dillon, Aya Meltzer-Asscher, and Maayan Keshev – Rational inference does not predict agreement errors: Gender vs number attraction in Hebrew comprehension
• Suet-Ying Lam and Satoru Ozaki – Investigating the source of the passive ellipsis clause penalty in VP ellipsis
• Barbora Hlachova and Fernanda Ferreira – Garden-path dead-ends contextualized
• Beverly Cotter and Fernanda Ferreira – A Direct Comparison of RC and PP Attachment Preferences
• Shayne Sloggett and Quynh Chieu – Investigating island constraints in Vietnamese

A big congratulations to Lindsay Forauer (Linguistics, Theater ’25) and Evan Owens (Linguistics & Portuguese ’25) for being selected as 2024-2025 UMass Amherst Rising Researchers!

Lindsay says: “I was selected for my work as a dialect coach on UMass Theater Department and UMass Theatre Guild shows, as well as working professionally with high schoolers to professional actors. I have researched and coached a range of accents, from Polish to St Louis, Missouri and have also worked as a research assistant for my theater advisor, Professor Elisa Gonzales.” 

Evan says: “My research is on language ideologies surrounding the pluralization of Hawaiian loanwords in English.” Evan will also be giving a talk about this research at the 11th McGill Canadian Conference for Linguistics Undergraduates, to be held April 5-6, 2025!

Bruce Hayes will give a colloquium talk on March 28 at 3.30 in ILC S211. The title of the talk is: A wug-test of Catalan consonant alternations (joint work with Kevin Liang and Victoria Mateu). See abstract and details here.