Chunking Language: Phase-Locked Oscillations or Evoked Responses?

Poster Session A - Saturday, April 13, 2024, 2:30 – 4:30 pm EDT, Sheraton Hall ABC

Lena Henke¹ (henke@cbs.mpg.de), Burkhard Maess¹, Lars Meyer^1,2; ¹Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany, ²Clinic for Phoniatrics and Pedaudiology, University Hospital Münster, Germany

Our ability to cognitively integrate immediate sensory information may be confined to a time window of 2–3 seconds. In language, this window may constrain the grouping of individual words into larger multi-word chunks, which is required for comprehension. Chunk boundaries are accompanied by a characteristic event-related potential, the Closure Positive Shift (CPS). More recently, frequency-domain analyses have found neural oscillations in the delta band (< 4 Hz) to predict chunking; we have argued that their preferred wavelength could explain the endogenous time limit. Hence, we here tested whether the CPS might consist of underlying neural oscillations, with increases in evoked amplitude resulting from increased phase-locking of oscillations across trials. We recorded participants’ magnetoencephalogram while they listened to globally ambiguous sentences in which the choice between two possible interpretations would be biased by a single chunking decision. Chunking was not externally imposed, but we hypothesized that temporal limits of integration windows would influence chunk termination and consequently, interpretation. Accordingly, we manipulated the sentence duration up to the potential chunking decision point to range between ~2–4 seconds in order to increase the likelihood of chunk termination, a CPS, and delta-band phase locking. Preliminary results indicate an evoked response at the chunk boundary that is influenced by sentence duration. This evoked response correlates with delta-band phase-locking. In conclusion, we tentatively suggest that the CPS may reflect the time-domain equivalent of phase-locked delta-band oscillation. We are currently working on localizing the neural sources of this endogenous, time–/wavelength-driven chunking mechanism.

Topic Area: LANGUAGE: Other