Language Lunch: Dr. John Drury

The CRLMB Language Lunch is an informal, academic-community seminar on speech and language research. Please join us this week as Dr. John Drury (Communication Sciences and Disorders, CRLMB) discusses the use of ERP technology in language research, and in particular, the LAN/eLAN distinction in studies of syntactic processing.

Left anterior negativities in ERP studies of (morpho-)syntactic processing

In language ERP research, early negative-going deflections with left and/or anterior scalp distributions (LANs) and latencies ranging between 100-500 ms post-stimulus onset have been argued to be indices of early/fast/automatic syntactic and morpho-syntactic processing. In this talk I will consider data from two kinds of studies which bear on our understanding of these types of ERP-effects.

One important division that has figured prominently in the literature is that between very early (within the first ¼ second) LAN effects (“eLANs”) and later effects (roughly between 300-500 ms; aka “LANs”). eLANs have been argued to reflect the rapid action of syntactic category identification and structure assembly, while later LAN effects are suggested to index syntactic feature checking relevant to the connections between morpho-syntax and the establishment of thematic relationships. However, an examination of the ERP literature (here I will focus primarily on reading studies) shows that eLAN effects have not been consistently found across studies. Further, the experimental contrasts which have been argued to elicit these effects almost always have involved serious confounds.

One of the first paradigms to show both eLAN and LAN-type effects (though these labels were introduced later) is Neville et al. (1991), who contrasted well-formed control sentences like “He criticized Max’s proof OF the theorem” with word-order flips like “He criticized Max’s OF proof the theorem”). I present data from two separate studies which included these contrasts, and show that at least the very early LAN effect (Neville et al.’s “N125”) is likely an artifact driven by baseline differences associated with ERPs connected with the words immediately preceding the target words. The later LAN effect (in the 300-500 ms range), however, seems to not to be attributable to these confounds.

Finally, I will discuss another recent study examining the processing of regular versus irregular English verbs which suggests a fairly specific view of the nature of later (300-500 ms) LAN effects in terms of mismatches between lexically retrieved information and syntax-driven predictions about the categories/features and constituency of upcoming words. These combined results will be argued to: (i) cast some doubt on the eLAN/LAN distinction, and (ii) suggests a rather more subtle view than has been previously entertained about what exactly LAN effects index.