The Bottom-up Processing Order Of Word Category Information And Word Meaning Information: Evidence From An ERP Study In Chinese

Distinguishing between category information and meaning information of input words in sentence comprehension is an important aspect of people's linguistic ability. In order to understand a spoken or written sentence, people have to integrate bottom-up information about category and meaning of each component word of the sentence into top-down expectation in the course of sentence formation, that is to say, two processes are involved—word category integration and word meaning integration. Since every input word of a sentence has to make both its category information and meaning information integrated into the preceding sentential frame for the sake of comprehension, would any intrinsic temporal relationship exist between these two bottom-up integration processes? To answer this question, we value Event-related Potentials (ERPs) as an effective approach, which is an advanced technique used to record cerebral electrical activity and with high temporal resolution. However, the previous experimental results acquired with the ERP technique did not provide an adequate answer to the concerned question. The ERP researches in literature, using Indo-European language stimuli, revealed that words violating the expectation of the preceding sentential contexts in terms of category elicited E(L)AN (100-200ms) or (L)AN (300-500ms); while words violating the expectation of the preceding sentential contexts in terms of meaning elicited N400 (300-500ms). Obviously, these findings can not be viewed as powerful evidence for any intrinsic temporal relationship between word category processing and word meaning processing from the perspective of bottom-up integration. We believe that on the basis of the three-phase sentence comprehension model proposed by Friederic(i2002), future ERP researches on the relevant issue should make temporal comparison between word category processing and word meaning processing under equivalent circumstances—that is, we should prepare a group of words that offer their category information and meaning information approximately at the same time in the lexical entries and set this group of words in different types of sentential contexts, so as to dig out whether an processing priority should be enjoyed by either kind of information (category or meaning) contained in these words.Therefore, for the present study, we carried out an ERP experiment using Chinese stimulus sentences strictly in accordance with the designing idea mentioned above, namely temporally comparing word category processing and word meaning processing. This experiment aims at finding out the ERPs related to word category violation and word meaning violation in Chinese and consequently using these two kinds of ERPs as hints about the temporal order of word category processing and word meaning processing. We used auditory-to-visual cross-modal stimulus presentation manner, for it contained the advantages of both visual and auditory linguistic stimuli: auditory stimulation simulates people's daily linguistic experience to a great extent, while visual stimulation is almost free of lexical ambiguity (which is very common for Chinese syllables) and can make a word's category information and meaning information input at the same time. Every stimulus sentence in our experiment was composed of two parts: preceding sentential context (auditorily-presented) and the critical word (visually-presented).Three experimental conditions were formed by establishing 3 kinds of relations between 40 sentence-final nouns and their preceding sentential contexts. In these conditions, the sentence-final nouns 1) met the top-down expectation in terms of category and meaning; 2) violated the top-down expectation in terms of category; or 3) violated the top-down expectation in terms of meaning. Subjects were asked to finish a sentence-correctness judgment task by pressing buttons on a response pad.The experimental results revealed that with category violation, the sentence-final words elicited a negative shift,most obvious at frontal sites during 150-250ms after stimulus onset, which resembled early AN; while with meaning violation, they elicited a negative shift,most obvious at central and frontal sites during 250-550ms after stimulus onset, which was similar to N400. These findings sufficiently suggest that the ERPs respectively related to word category violation and word meaning violation are to some extent of the same pattern across different languages and that word category processing intrinsically goes before word meaning processing for each individual word. This study not only provides evidence for the three-phase sentence comprehension model proposed by Friederici, but also demonstrates that distinguishing between word category information and word meaning information is a cross-language linguistic competence of mankind.