Search Abstracts | Symposia | Slide Sessions | Poster Sessions
How can neural evidence constrain theories of language processing?
,
Organizers: Arka Mallela1, Brad Mahon2,3; 1Rush University Medical Center, 2Carnegie Mellon University, 3University of Pittsburgh
Presenters: Liina Pylkkanen, Nitin Tandon, Marina Laganaro, Jorge Gonzalez-Martinez
Increasingly sophisticated modalities can probe brain activity including functional MRI, lesion data, and extracranial/intracranial electrophysiology. These techniques have significant clinical utility in the neurosciences, and the spatiotemporal resolution of these modalities continues to grow. Causal manipulation is increasingly possible. Yet the role of these techniques in shaping theory remains unclear: do neural data meaningfully adjudicate between competing models of language representation and production, or primarily confirm frameworks derived from behavioral evidence? This symposium examines when and how neural evidence can meaningfully constrain theory, focusing on representational, temporal, and causal claims. Topics will include how these techniques can identify language representations in the brain, help preserve language in patients with neurological disorders, and whether they can generate theoretical insights beyond behavioral and cognitive data. This symposium aims to define if and how neural data can drive progress in theories of language processing.
Presentations
The brain does not “compose” structure
Liina Pylkkanen1; 1New York University, New York, NY, USA
The neurobiology of language has inherited the metaphor of composition from linguistics: the idea that the brain “builds” structure by combining words like Lego pieces. Yet neural effects of “composition” often occur faster than those of lexical access, suggesting that structure building precedes “picking up” the pieces that form the structure. Results from parallel visual presentation of multiword expressions clarify the picture. When the temporal contrast between single words and phrases is flattened, neural responses show identical electrophysiological component structure and timing. Both appear to move through the same stages. Analyses of experiential features show faster meaning access in phrases, suggesting that compatible meanings together increase the overall evidence for a specific meaning, resulting in faster comprehension. I argue it is time to let go of the metaphor of composition as a brain mechanism and instead articulate shared stages of mapping input to linguistic knowledge for words and phrases alike.
Dynamical language systems revealed by intracranial recordings
Nitin Tandon1; 1UT Health Houston, Houston, TX, USA
Theoretical models of language production debate a number of critical issues including the role of multilobar integration is language representations, the role of subcortical structures, and the parcellation of language areas such as the ventral temporal lobe. Intracranial electrophysiology can confirm or challenge such models with direct assessment of local activity and connectivity. These data, in conjunction with careful behavioral analyses, can explain language function as network states rather than focal activations. This enables us to reconcile localizationist and network models using transient stages as intermediate stages in language processing. Thus, this approach argues that experimentation is necessary for theoretical progress.
Parallel or serial processing in utterance production?
Marina Laganaro1; 1University of Geneva, Geneva, Switzerland
While it is acknowledged that the brain networks involved in language and speech production are not all active simultaneously, the issue of the amount of parallel versus sequentially occurring processing is still debated. Here we will seek and discuss evidence in favor or against sequential processing from EEG investigations on the dynamics of processes involved in utterance production focusing on different types of utterances (single words, phrases), different linguistic levels (semantic, lexical, phonological, motor speech) in neurotypical adults and in left hemisphere brain damaged speakers. I will discuss the extent to which EEG investigations can address this question and advance theories of parallel vs. serial language processing.
SEEG explorations of language processing and epilepsy
Jorge Gonzalez-Martinez1; 1University of Pittsburgh, Pittsburgh, PA, USA
The neural architecture supporting seizure semiology and language production often share considerable overlap. Manifestations of seizures can show abnormal language function. Stereo EEG is an invasive intracranial monitoring modality that utilizes anatomo-electrical clinical correlations to understand how seizures organize and propagate. This same methodology allows exploration of language processing with high spatial and temporal resolution. This is of not only strong scientific interest but also has immediate clinical relevance to understanding and preserving language function in patients with drug-resistant epilepsy. This talk will describe the use of stereo EEG in understanding language function, including stimulation to create causal evidence. With this data, we can both confirm or disconfirm models of language production but also provide evidence that can drive novel theoretical development.