Dynamics of functional networks for syllable and word-level processing

Abstract

Speech comprehension requires the ability to temporally segment the acoustic input for higherlevel linguistic analysis. Oscillation-based approaches suggest that low-frequency auditory cortex oscillations track syllable-sized acoustic information and therefore emphazise the relevance of syllabic-level processing for speech segmentation. Most linguistic approaches, however, focus on mapping from acoustic-phonemic representations to the lexical level. How syllabic processing interacts with higher levels of speech processing, beyond segmentation, including the anatomical and neurophysiological characteristics of the networks involved, is debated. Here we investigate the effects of lexical processing and the interactions with syllable processing by examining MEG data recorded in a frequency-tagging paradigm. Participants listened to disyllabic words presented at a rate of 4 syllables/sec. Two hypotheses were evaluated: (i) lexical processing of words activates a 2 Hz network that interacts with syllable processing at 4 Hz; and (ii) syllable transitions contribute to word-level processing. We show that lexical content activated a left-lateralized frontal and superior and middle temporal network and increased the interaction between left middle temporal areas and auditory cortex at 2 Hz (phase-phase coupling). Mere syllable-transition information, in contrast, activated a bilateral superior-, middle temporal and inferior frontal network and increased the interaction between those areas at 2 Hz. Lexical and syllable processing interacted in superior, middle temporal, and inferior frontal areas (cross-frequency coupling), whereas syllable tracking decreased when lexical information was present. The data provide a new perspective on speech comprehension by demonstrating a contribution of an acoustic-syllabic to lexical processing route. Significance statementThe comprehension of speech requires integrating information at multiple time scales, including phonemic, syllabic, and word scales. Typically, we think of decoding speech in the service of recognizing words as a process that maps from phonemic units to words. Recent neurophysiological evidence, however, has highlighted the relevance of syllable-sized chunks for segmenting speech. Is there more to recognizing spoken language? We provide neural evidence for brain network dynamics that support an interaction of lexical with syllable-level processing. We identify cortical networks that differ depending on whether lexical-semantic information versus low-level syllable-transition information is processed. Word-and syllable-level processing interact within MTG and STG. The data enrich our understanding of comprehension by implicating a mapping from syllabic to lexical representations.

The Lateralization of Speech-Brain Coupling Is Differentially Modulated by Intrinsic Auditory and Top-Down Mechanisms

“…In Experiment C, as in the previously discussed study, speech tracking in both conditions presumably reflects both intrinsic auditory and externally driven contributions: on the one hand, the syllables are presented at a rate of 4 Hz – close to the natural frequency of the right auditory cortex – and on the other hand, the task requires phonological processing of the signal. During the condition in which German words were presented, additional lexical-semantic computations are necessarily performed (Rimmele et al, 2019).…”

“…Materials and methods of each experiment are briefly described below. For more detail see Assaneo and Poeppel (2018) for Experiment A, Assaneo et al (2019) for Experiment B, and Rimmele et al (2019) for Experiment C.…”

“…In total, 134 German disyllabic words were selected from the CELEX lexical database (Baayen et al, 1995) and 134 Turkish disyllabic words from the TELL database 1 [for details see Rimmele et al (2019)]. German and Turkish syllables produced by a female German/Turkish bilingual speaker were recorded.…”

“…However, under specific speech-listening conditions, or due to neuronal functional and structural individual differences, the externally driven mechanism can affect the neuronal activity, mostly in the left hemisphere, equalizing the strength of entrainment across hemispheres. We apply new analyses to three published magnetoencephalography (MEG) datasets (Assaneo and Poeppel, 2018; Assaneo et al, 2019; Rimmele et al, 2019) to present new evidence to support these claims.…”

Neural harmonics of syntactic structure

“…In a nutshell, speech is linear, while language is hierarchically organized. Temporal regularities linearly guide speech chunking in ways that are relevant for language production and comprehension [3, 5–13]): for example, entrainment to syllabic rhythm is relevant for word formation processes [3,6,14]. However, the core of language processing lies in the internal and tacit knowledge of syntactic rules, which determine how words combine into meaningful utterances [15].…”