The role of electrophysiology in informing theories of word production: a critical standpoint

Abstract

Munding, Dubarry, and Alario (this issue) courageously and thoroughly summarise the MEG literature on word production. It is evident that their task was a real undertaking and word production researchers should applaud their efforts. In this commentary, I raise a few issues that are inspired by their report. These comments are not meant as criticism to Munding et al., but mainly reflect what I see as limitations of electrophysiological methods when it comes to making temporally specific claims. Three language production theoretical models are discussed by Munding et al. The hierarchical state feedback control model (Hickok, 2012) has an elaborated motor control aspect, but little specification regarding earlier stages of word planning. As such, it does not lend itself well as a model against which to evaluate the MEG evidence for serial vs. parallel processing. A second model discussed, Price’s model (2012), is solely based on positron emission tomography and functional magnetic resonance imaging (fMRI) data, which have no temporal resolution at the scale of single word production (i.e., often less than 1 second). Moreover, in fMRI the link between the blood-oxygen- level dependent (BOLD) signal and underlying neuronal activity, and hence electrophysiological measures, is still poorly understood (e.g., Ekstrom, 2010; Logothetis, 2008). As such, Price’s model may be difficult to assess on the basis of electrophysiological data. By contrast, Indefrey’s model comprises early and late stages of word production and is based on electrophysiological evidence along with haemodynamic evidence. Thus, Indefrey’s model possibly has the greatest potential to provide the framework for examining the evidence for either serial or parallel models, so my comments are mostly related to it.