You are viewing the SNL 2017 Archive Website. For the latest information, see the Current Website.

Poster A10, Wednesday, November 8, 10:30 – 11:45 am, Harborview and Loch Raven Ballrooms

Semantic context reverses the polarity of P200 effects during word planning

Daniel Kleinman1, Kara Federmeier1;1University of Illinois at Urbana-Champaign

How long does it take to retrieve a word’s representation in long-term semantic memory when preparing to speak? Recent EEG studies using language production tasks have found that the P200 component has a greater amplitude when lexical selection is more difficult. This relationship holds across several different manipulations: Larger P200s are evoked when bilinguals name pictures in their L2 (vs. in their L1), or when speakers name pictures with lower-frequency (vs. higher-frequency) names or pictures with recently named semantic competitors (Costa, Strijkers, Martin, & Thierry, 2009; Strijkers, Costa, & Thierry, 2009). Strikingly, P200 amplitude in these tasks positively correlates with picture naming latencies on a trial-to-trial basis. Collectively, these findings have led researchers to suggest that speakers initiate lexical access within 200 ms, and that the P200 component (particularly in occipital regions) indexes the difficulty of this retrieval process. However, to our knowledge, all such reports have used a picture naming task, which affords only a single correct response that must be produced in the absence of a broader context – factors that may limit the generality of these findings. Here, we investigated the relationship between P200 amplitude and lexical selection difficulty using a sentence completion task, which affords many acceptable responses that are generated within a semantically rich context. METHOD: We recorded participants’ EEG (n=38) as they read RSVP sentences that varied in constraint (how strongly the context predicted a particular final word). On 50% of trials, the last word of the sentence was omitted and participants instead saw a blank, prompting them to overtly produce their own completions. To permit comparison with standard effects of context on comprehension, the other 50% of sentences were completed with a final word that was either expected or unexpected. RESULTS: On production trials, single-trial analyses revealed a significant relationship between response latencies and P200 amplitudes in occipital regions. However, in contrast to previous studies, this relationship was negative – that is, a larger P200 was associated with a faster RT (which was in turn associated with higher sentence constraint and higher cloze probability of participants’ productions). CONCLUSIONS: The relationship between P200 amplitude and word production difficulty (as indexed by production latencies) is task-dependent. As there are many differences between picture naming and our sentence completion task that could account for the polarity of this effect, this inconsistency may help to reveal what exactly P200 amplitude signifies. If P200 amplitude scales with the amount of activation injected into the production system, for instance, then that extra activation could lead to different RT effects when spread among a set of close competitors (as during picture naming) vs. when concentrated on a single response (as during sentence completion for a strongly constraining sentence). Alternatively, P200 amplitude during production may reflect attention-related and pre-lexical effects that differ between tasks. Until further research reveals how (and not just when) P200 effects reflect word production processes, using that component as a barometer of lexical selection difficulty may be premature.

Topic Area: Control, Selection, and Executive Processes

Back to Poster Schedule