We (me, Ed Stabler, John Hale et al.) have a new paper in Brain and Language in which we compare various grammatical models (Markov, context-free, minimalist) in terms of their fit against fMRi signals related to sentence processing. The key take-away is that the most abstract (Minimalist) representations improve the statistical fit against left temporal lobe signals even after taking in to account lower-order syntax models and various control covariates. I think this is exciting stuff for at least two reasons:
First, debate about abstract hierarchical structure tends to get bogged down in broad disagreements about the definition of language which leave neither side satisfied. We are able to test a very concrete hypothesis about whether or not of abstract hierarchy is used during a specific every-day task: listening to a story. I suspect (but don’t know!) that if such representations are used here, then there are very few instances where less rich representations would be deployed.
Second, the method we use is generalizable. Any incremental parsing model can be applied and a variety of different ways of counting complexity can be tested (we compare expectancy, or Surprisal, and tree size, or Node Count). We have a how-to-guide at Language & Linguistics Compass. So, please pick favorite grammar/parsing model and try to do something similar: The stimuli and ROI time-courses that we used are available online!
Brennan, J., Stabler, E. P. Jr., Van Wagenen, S. E., Luh, W.-M., & Hale, J. (2016). Abstract linguistic structure correlates with temporal activity during naturalistic comprehension. Brain and Language, 157–158: 81–94
doi: 10.1016/j.bandl.2016.04.008 [Download Stimulus & Data]
Abstract Neurolinguistic accounts of sentence comprehension identify a network of relevant brain regions, but do not detail the information flowing through them. We investigate syntactic information. Does brain activity implicate a computation over hierarchical grammars or does it simply reflect linear order, as in a Markov chain? To address this question, we quantify the cognitive states implied by alternative parsing models. We compare processing-complexity predictions from these states against fMRI timecourses from regions that have been implicated in sentence comprehension. We find that hierarchical grammars independently predict timecourses from left anterior and posterior temporal lobe. Markov models are predictive in these regions and across a broader network that includes the inferior frontal gyrus. These results suggest that while linear effects are wide-spread across the language network, certain areas in the left temporal lobe deal with abstract, hierarchical syntactic representations.
