Birdsong and Speech

In a session on animal communication at the Science of Aphasia meeting (Potsdam), Constance Scharff (Freie Universität Berlin) and Erich Jarvis (Duke University) gave two good presentations about their research programs on different aspects of vocal communication. Constance focuses more on the potential of the different FoxP2/FOXP2 preps and related approaches (her talk was about the ‘hope versus hype’ in foxp2 approaches). Erich’s work is oriented towards comparative work, looking in detail at the neural circuitry underlying vocal learning. Both focus on vocalization and communication -- and the toolboxes they have available are, to those of us working on the ‘awake behaving human prep,’ jealousy-inducing, of course. The research exemplifies a beautiful piece of biology. I am a big fan of this work.

There is no but … But I do want to insert a cautionary note -- and a suggestion for a new collaborative research effort. The laudable thing, obviously, is that Constance and Erich attended a meeting on linguistics and aphasia to begin with. They are among the very open-minded investigators, genuinely wanting to understand what’s going on in neurolinguistics, psycholinguistics, biolingustics, linguistic theory – or whatever x-lingustics comes to mind. (For the record, I wish that other scientists interested in speech and language had the same drive; too often, studies about some aspect of language in other domains proceed without acknowledging that a rich and detailed literature exists. Imagine doing that in physics, say … that would be really weird.)

The really tricky part – for Constance and Erich, on the one hand, and for cognitive scientists, on the other – has to do with the potential alignments between the research programs. There is an (often explicit) desire to identify analogies between birdsong and speech/language. For example, some argue that a phoneme is perhaps like a note/syllable in birdsong; that a word is like a motif; that a sentence is like a bout … Presuppositions are made about how such representational units line up. But the key word here is ‘representation.’ When language researchers talk about syllabic structure, lexical entries, or phrasal combinations, these are technically specified notions that serve the purpose of investigating how language works in the human mind/brain. Implying that a bout is like a sentence comes with the assumptions and machinery (at least for cognitive scientists) that words/motifs are being combined to yield meaning that is arrived at by combining the pieces.

In my view, two things are noteworthy. First, it’s absolutely essential to separate speech perception and language comprehension when even discussing such analogies. The casual mixing of ideas from speech perception (e.g. do syllables matter) versus language comprehension (e.g. how do combinatorics work) is a recipe for confusion. There is much to be gained for cognitive neuroscience of speech perception by attending to the insights derived from birdsong. There is, however, much less gained for cognitive neuroscience of language comprehension. Second, and closely related, the significant insights we gain about sensorimotor transformation from birdsong research are very relevant to processing and learning speech perception and production, but that does not extend to language processing more generally. Again, it seems like it’s extremely important to be a “splitter,” conceptually speaking, and respect distinctions that have yielded useful descriptions of phenomena at different granularities.

It is my impression that the research being pursued in that domain (e.g. by Constance and Erich, but also by Todd Troyer, and Tchernikovski, and Woolley, and others) is making terrific progress in its own right, yielding genuine biological insight about the neurophysiological foundations of sensorimotor learning, development, neural coding, etc. Potential analogies to language are perhaps sometimes useful as heuristic devices (or useful as rhetoric in the context of grant applications), but potentially misleading if we take the parallelism too seriously. In my view, we should jump on this as a real opportunity for collaboration, injecting what cognitive science can contribute to the analysis of that vibrant area of research without making assumptions about weak (or dis-) analogies.