Research

Cross-decoding of word production and comprehension in post-stroke aphasia

Thanks to an SNSF Postdoctoral Fellowship I was recently awarded, I am setting up an experiment to investigate the neural correlates of semantic and phonological processing across word production and comprehension using EEG. In this project, I aim to take advantage of multivariate pattern analysis to perform temporal decoding across production and comprehension in neurotypical speakers and speakers with aphasia.

Verb planning from a cross-linguistic perspective

In several EEG projects, I aim to understand what factors affect verb planning, using languages with different properties to address questions on morphological encoding, lexical retrieval and meaning composition. On-going work is focusing on Chintang, a Sino-Tibetan language spoken in Eastern Nepal (in collaboration with Tribhuvan University, Kathmandu, Nepal), Vietnamese (ULIS-VNU, Hanoi, Vietnam), Mandarin Chinese (thanks to the work of Master's student Runze Liang, now PhD student at the BCBL).

Discourse production in post-stroke aphasia

In many projects, both on neurotypical speakers and post-stroke aphasia, I apply methods from NLP and AI (e.g. Stanford parser, GPT-2) to the study of spontaneous speech. At the University of South Carolina, I developed a novel approach to the automatic quantification of verb argument structure in discourse using a dependency parser, which I used to compare verb argument structure in controls and speakers with aphasia (now published in JSLHR). I also presented this at SNL 2024 as part of a symposium on modern approaches to the characterization of grammatical deficits in aphasia, organized by Jeremy Yeaton and Danielle Fahey.

I also used CHAT-CLAN (AphasiaBank) to model speech fluency and complexity in spontaneous speech in post-stroke aphasia, which allowed me to understand which lesion characteristics are predictive of improvement in discourse following speech language treatment (Giglio et al., 2025, NOL).

Similarities and differences in the neural networks for sentence production and comprehension

During my PhD at the Max Planck Institute for Psycholinguistics, I ran several fMRI studies to understand whether speaking and listening recruit the same brain regions. We found substantial overlap in the networks, suggesting that higher-level linguistic processing is carried out by the same neural structures (Giglio et al., 2022, CerCor), with similar connectivity patterns (Giglio et al., 2024, NOL) and similar encoding of relational structures (Giglio et al., 2024, CerCor).

I also compared incremental syntactic processing in spontaneous production with comprehension, applying an fMRI analysis approach that had been used exclusively in comprehension to production (Giglio et al., 2024, PNAS). This line of work is now being extended in collaboration with Giulia Li Calzi, a PhD student at the University of Zurich, with the goal of developing novel ways of measuring incremental processing during production.

The neural correlates of (spontaneous) language comprehension

Language processing has traditionally been studied using constrained experiments or audiobook listening. In a recent project with Ellie Huizeling, we investigated the effect of disfluencies in spontaneous speech on the neural processing of word frequency and surprisal (Giglio et al., 2026). In another project with Cas Coopmans, we are asking how the type of input (spontaneous speech vs. audiobook) affects syntactic processing strategies using fMRI.

In collaboration with Muge Ozker, Peter Hagoort, and the Language in Interaction consortium (Donders Institute, The Netherlands), I am characterizing how brain networks underlying language processing are modulated by individual differences in cognitive skills. We have a first project currently under review (see Ozker et al., bioRxiv), where we show that cognitive skills differently engage neural networks not traditionally associated with language processing during task processing. In a parallel project, I am investigating this specifically during naturalistic listening, to complement previous task work. I recently presented posters on this at SNL and OHBM, and I gave a talk at the meeting of the Italian Association for Cognitive Science.