📌 VFFs from Gahl et al. (2004)'s manually annotated (i.e. gold-standard) VFFs
📌 against preferences for competing frames (the dative alternation and NP/SC ambiguity) 🧵6/8

We benchmarked it thoroughly. The LLM consistently outperformed benepar & the Stanford Parser:
📌 300 human-annotated sentences (LLM accuracy = 79%, vs. 69% for benepar and 59% for Stanford) 🧵5/8

We took a closer look at what was going on in prefrontal cortex. This revealed that these sustained representations traced back to different regions depending on a word's sentence position: when it was a subject, it was encoded in IFG, while MFG encoded objects. 🧵6/9

In passive sentences like "Frankenstein was hit by Dracula", we observed sustained neural activity encoding BOTH nouns simultaneously throughout the entire utterance. This was particularly true in prefrontal cortex. 🧵5/9

For straightforward active sentences ("Dracula hit Frankenstein"), the brain activated words sequentially, matching their spoken order. But things changed dramatically for more complex sentences... 🧵4/9

We trained machine learning classifiers to identify each word's specific neural pattern. 🔑We ONLY used data from picture naming (single word production) to train the models. We then used the models to predict what word patients were saying in real time as they said sentences.🧵3

We recorded brain activity directly from cortex in neurosurgical patients (ECoG) while they used 6 words in two tasks: picture naming ("Dracula") and scene description ("Dracula hit Frankenstein"). 🧵2/9

For folx at #HSP2025, tune in at 2:15 for our talk on the processing of Switch-Reference Marking in Nungon, a language spoken by ~1000 ppl that requires speakers to inflect the verb not just for features of its subject, but also for the UPCOMING subject!

hsp2025.github.io/abstracts/15...

Just presented our work using #ECoG to decode words during sentence production at #HSP2025. Really grateful for all the great feedback. I got more clever ideas for future directions than I can possibly follow up on. Love this conference!

doi.org/10.1101/2024...

Little late here but this talk at #hsp2025 yesterday was SO neat. Literacy effects disappear when you control for differences in SES. Work by Jessica Vélez Avilés and Paola (Giuli) Dussias

hsp2025.github.io/abstracts/26...

But in prefrontal cortex, both the subject and the object were active throughout the whole passive sentence. Inferior frontal gyrus (IFG) sustained a representation of the subject, while middle frontal gyrus (MFG) sustained the object.🧵6/9

What about sentences with non-canonical word orders? We looked at how the brain processes passive sentences, and found a division of labor between brain regions. In sensorimotor cortex, where articulatory information is stored, each noun was activated in the order that it was said.🧵5/9

During the production of active sentences like "Dracula hit Frankenstein,” the model predicted each word as it was being said - first "Dracula" (blue = significant prediction of subject), then "Frankenstein" (green = object significant). But here's where things get interesting -> 🧵4/9

First, using data from a picture naming task, we trained classifiers to learn the patterns of cortical activity for each of 6 words - Dracula, Frankenstein, etc. 🧵2/9

New preprint! We know a lot about how the brain produces single words. But how does word production work in sentences? We tracked 6 words in the brain while ECoG patients used them to produce sentences like "Dracula hit Frankenstein". 🧵 1/9
doi.org/10.1101/2024...