ERP Evidence for Rapid Meaning Access to Newly Learned Words

• In Event: 1-208 - The Neural Basis of Word Learning

Integrative Statement

For well-known words, meanings are accessed rapidly and automatically from semantic memory. However, such fast meaning access is not achieved immediately following learning. According to the account of complementary learning systems (CLS, Davis & Gaskell, 2009), a novel word is initially represented as part of episodic memories about learning event. Overnight consolidation is usually needed for such memories to be transformed into semantic memory and integrated into the mental lexicon, allowing for rapid meaning access. Existing research has examined such fast and automatic meaning access using semantic priming or interference paradigms. For example, the emergence of semantic priming between novel words and existing words indicates the integration of novel words into semantic memory (e.g., Tamminen & Gaskell, 2013).

Another approach to the time course of meaning activation is to take advantage of the domain specificity of meanings, including their sensorimotor components (Binder & Desai, 2011; Kiefer & Pulvermuller, 2012). Previous studies find that words with different sensorimotor meanings evoke different brain activation patterns within 200 ms of word presentation (e.g., Moseley, Pulvermuller, & Shtyrov, 2013; Pulvermuller, Lutzenberger, & Preissl, 1999). The current study aimed to provide more direct evidence for meaning activation by observing ERPs evoked by novel words that were assigned sensorimotor meanings.

Across three days, participants learned novel spoken words associated with either action or non-action (i.e., static visual) meanings presented as verbal definitions. On the fourth day, they performed a semantic judgment task on novel and existing words presented in separate blocks while EEG were recorded. In the ERP task, participants were presented with word stimuli. Following approximately 20% of the words, they saw a phrase (e.g., “leg movement”, “dark color”) on the screen and were asked to judge whether the phrase was semantically related to the preceding word.

Behavioral data showed that participants learned both types of novel words very well; their performance in cued-recall and recognition tests reached ceiling. ERP data showed that novel action words evoked a larger negativity at the central clusters than novel non-action words within N1 (82-122 ms) and P2 (136-236 ms) time windows. For existing words, the difference between action and non-action words first emerged within the N400 (300-500 ms) and LPC (500-700 ms) time windows. While the latency of semantic category effect was different between novel and existing words, the effect emerged around the recognition point of both novel words (~103 ms) and existing words (~408 ms). In addition, novel words overall evoked a larger frontal negativity and parietal positivity than existing words, especially after 500 ms.

The results indicate rapid and (arguably) automatic meaning access to newly learned words three days following learning. This semantic domain effects emerges even though the meaning was presented as a verbal definition, with no nonlinguistic sensorimotor input. Nevertheless, novel words may not be fully integrated, as suggested by the persistent difference in ERPs between novel and existing words after 500 ms. The findings have implications for the examination of word knowledge integration and embodied language.

Authors

• Xiaoping Fang, University of Pittsburgh
  Presenting Author

• Charles A Perfetti, University of Pittsburgh
  Non-Presenting Author