This blog is a forum for the discussion of speech related papers and research, especially as it relates to the brain. I'm going to kick things off with a discussion of Phonological Repetition-Suppression in Bilateral Superior Temporal Sulci. Vaden KI, Muftuler LT, Hickok G Neuroimage 2009 Jul 31. This paper uses non words of varying phonological similarity and a repetition suppression design, and finds that bilateral superior temporal lobe areas are sensitive to repetition of more similar stimuli. This is interpreted as showing that both the left and right STS is sensitive to phonological structure. However, the design contained no control for the acoustic changes that correlate with these phonological properties, so they can't rule out an acoustic basis for these effects. Also, they don't show any plots of the effects, and it would be interesting to know if there is a consistent effect across conditions, or whether one condition is driving all the effects. It would also be interesting to know what would be seen using pseudo-words, since in addition to the phonological and acoustic repetition, in the low and medium conditions, four different words are heard in each condition, and in the high repetition condition, the same word is heard four times.
Is the prelexical or phonological processing of speech bilateral? Clinically, left temporal lobe lesions are the ones that result in aphasic syndromes, while right temporal lesions are often 'silent' (though patients can show problems with aspects of auditory processing when tested e.g. music perception). While these aphasic problems frequently improve, presumably through recovery mechanisms, there isn't really any debate clinically that speech is linguistically processed in the left temporal lobe - there are very few examples of people having a receptive aphasia following damage restricted to the right temporal/parietal lobe.
Many papers (not cited in Vaden et al, 2009) have indeed shown left lateralised responses to intelligibility in speech (e.g. scott et al, 2000; 2009, narain et al, 2003; uppenkamp et al, 2006, leibenthal et al, 2005). Indeed, the results in Vaden et al (2009) show a greater response in terms of the size of the effect (t=6.25 on the left STS peak, t=4.88 on the right STS). They argue that this could in fact reflect better phonological processing on the right than on the left (due to more efficient processing). I suspect that had this data been collected by our group (though we would probably not run the exact same design) then we would argue the opposite - the result indicate a dominance for 'phonological' processing in the left STS.
A final and an important point is what, exactly, do we mean by phonological? Does it refer to some abstract property of the stimuli or can it entail more complex acoustic-phonetic changes? If we use, like Muftuler et al, four CVC strings, and varying the final C of each word, does this mean that the CV component of each string is the same? Phonetically and acoustically, it most likely does not, and it means we need to think about issues of co-articulation and fine phonetic detail. For example, the two words 'let' and 'led' are in theory the same phonetically, except for the voicing of the final alveolar plosive. However this word final voicing difference also leads to differences in the initial phoneme: the /l/ sound is darker when the word finished with the voiced /d/ then when it finished with the unvoiced /t/, and listeners are sensitive to these effects (Sarah Hawkins has written at lot about this). There isn't a list of stimuli for the Muftuler et al paper, but it would be interesting to know who they controlled for this. This point also raises the issue that it is hard to distinguish phonetic and acoustic factors in speech, and it is really important therefore to have acoustic baselines to control out truly phonetic effects from acoustic effects.
