Thursday, June 3, 2010

Back to the future on syntax and Broca's area

Thirty-five years ago Caramazza and Zurif (1976) made a startling claim that literally changed the way the field thought about Broca's aphasia, Broca's area, and the neurology of syntax. Up until that time Broca's area was thought to be basically a motor speech area. Even the agrammatic speech output of Broca's aphasics was thought, by prominent researchers, to reflect not a syntactic deficit but an economy of effort induced by the difficulty of articulating speech. In this context, Caramazza and Zurif showed that Broca's aphasics failed to comprehend sentences that required syntactic analysis. (Footnote: the same was true of conduction aphasics, but no one remembers that fact.)

Based on their findings C&Z made the following claim:

...for the Broca’s aphasics, brain damage affects a general language processing mechanism that subserves the syntactic component of both comprehension and production. The implication that follows is that the anterior language area of the brain is necessary for syntactic-like cognitive operations. (p. 581)

State of the art, 1976: Broca's area is the seat of syntax.

This, of course, was a beautiful study that has since been replicated repeatedly. Their conclusions were perfectly reasonable, except they turned out to be wrong. Subsequent work (e.g., Linebarger, Schwartz, & Saffran, 1983) showed that Broca's aphasics had not entirely lost their syntax -- they could still make grammaticality judgments pretty darn well.

State of the art, 1983: Broca's area is not the seat of syntax.

The field reacted with a variety of new ideas about Broca's area: It supports only a restricted component of syntax (Grodzinsky), it supports the mapping between syntax and meaning (Schwartz and Saffran), it supports fast access to lexical information (Zurif and others).

State of the art, after 1983: Who knows what Broca's area is doing, but we all agree: it is not the seat of (all of) syntax.

Now fast forward to today and scan the literature on the role of Broca's area in syntax. You might come across a paper by Fadiga et al. (2009) which states:

we propose that Broca's area might be a center of a brain network encoding hierarchical structure regardless of their use in action, language, or music. (p. 455)

Similar claims have been made by Friederici and colleagues who have suggested a role for Broca's area in hierarchical structure processing and phrase structure building.

State of the art [?!], 2009: Broca's area is the seat of syntax via its more general role [?!] in hierarchical processing of any kind.

What happened between 1983 and 2009 to cause the regression back to the interesting, but ultimately incorrect claim of C&Z? Functional imaging happened. It seems that when functional imaging became a widespread tool, with the development of fMRI in particular in the 1990s, the field either forgot about the decades of good research that came before, or just decided to start over. This is a mistake.

New rule: If you want to claim Broca's area is the seat of syntax, please (i) cite Caramazza & Zurif, and (ii) provide an explanation for the Linebarger et al. results.


Caramazza A, Zurif EB. 1976. Dissociation of algorithmic and heuristic processes in sentence comprehension: Evidence from aphasia. Brain and Language. 3:572-582.

Fadiga, L., Craighero, L., & D’Ausilio, A. (2009). Broca's Area in Language, Action, and Music Annals of the New York Academy of Sciences, 1169 (1), 448-458 DOI: 10.1111/j.1749-6632.2009.04582.x

Friederici AD. 2009. Pathways to language: fiber tracts in the human brain. Trends Cogn Sci. 13:175-181.

Friederici AD, Bahlmann J, Heim S, Schubotz RI, Anwander A. 2006. The brain differentiates human and non-human grammars: functional localization and structural connectivity. Proc Natl Acad Sci U S A. 103:2458-2463.

Linebarger MC, Schwartz M, Saffran E. 1983. Sensitivity to grammatical structure in so-called agrammatic aphasics. Cognition. 13:361-393.


Nina Kazanina said...

Hi Greg,

A comment from someone who is not following the issue closely but:

I think you would agree (or at least that's the impression I get from how your state-of-the-art bullets are phrased) that there is an implicit difference between 1976 and 2009. In CZ’s 1976-view Broca’s area was a seat for the speaker’s language-specific representational capacity, whereas in 2009 Fadiga, Friederici and others it carries out computations performed during structure building. So it’s ‘syntax’ in both cases only because the term ‘syntax’ is fuzzy to cover both usages.

I don’t think that Linebarger’s et al’s finding that Broca’s aphasics can still make grammaticality judgments is a fatal argument against the 2009-view. However, it should serve as a trigger for Fadiga, Friederici, etc. to look into why some structures are more difficult to compute than others (which will probably overlap to a large degree with what some of the post-1983 approaches have done already). All of this obviously links to the discussion below the previous post and reinstates the need for explicit linking between grammaticality judgments and computational processes that underlie them.

Greg Hickok said...

Hi Nina,
If my memory serves, CZ's claim is not necessarily language-specific. There quote, "syntactic like cognitive operations" hints at this. Edgar Zurif (my former Ph.D. adviser) is retired now and probably doesn't lurk on this blog, but Harvard Prof., Alfonso Caramazza may hear about this discussion. If so, maybe he can tell us what the concept was at the time.

But it doesn't matter. The current views are suggesting a very general hierarchical processing function for Broca's region. Because the linguistic knowledge underlying the ability to make most grammaticality judgments is highly dependent on hierarchical representations of sentences, damage to this system should obliterate the ability to make such judgments. It does not. This fact *has* to be discussed if one wants to claim such a general function for Broca's area, but it never is.