| |
Abstract:
One of the best described brain potentials related to
language processing is the N400 component. The N400 reflects a
word's expectancy and degree of association with its context.
Presentation of any word will generally elicit an N400, but less
associated and less expected words tend to elicit N400s with larger
amplitudes (e.g. Kutas, 1993; Kutas & Hillyard, 1984). Some
authors have proposed that N400 can be elicited by factors other
than semantic expectation. For example, presentation of the second
word of a non-rhyming pair (in contrast with a rhyming word)
elicits a late negative component termed N450 (Rugg, 1984). Rugg
concludes that the N450 belongs to the same family as N400 but
instead reflects processing at the phonological level. Importantly,
deviations in familiar melodies, and physical or grammatical
anomalies fail to produce the N400 effect (Besson, Maacar, &
Pynte, 1984; Kutas, Neville & Holcomb,1987).
In Mandarin Chinese, the lexical tones with which words are
pronounced influence word meaning: changing the tone changes the
meaning. The present research investigates the processing of tonal
information by observing the brain responses elicited by the
presentation of erroneous tones in otherwise perfectly grammatical
Chinese sentences. ERPs were obtained from twenty-five speakers of
Mandarin Chinese during auditory presentation of sentences with
normal and anomalous sentence endings. Three different types of
semantic anomalies were created by manipulating the tone, syllable,
or both tone and syllable of the sentence final words. We
hypothesized an N400 effect for all three types of semantic
anomalies, and that changing both the tone and the syllable of the
expected ending word would elicit the largest negativity.
Consistent with previous findings, we found an N400 effect elicited
by the three types of semantically anomalous words. The earliest
effect had an onset of 200ms post-stimulus and continued through
1000ms in some areas. An analysis of the large 150-1000ms latency
window revealed a significant area by condition interaction.
Particularly interesting were the effects over Broca's area:
contrary to our expectations, the double-anomaly condition elicited
responses comparable to the negativity elicited by the tone
condition and only slightly more negative than the control
condition (p= .0509). Furthermore, this negativity was
significantly smaller than that elicited by the syllable anomaly.
Smaller N400 effects for the tone condition are not alarming,
considering that tone errors are not always noticed (see Cutler,
1997). Although the smaller N400 effect for the double anomaly was
somewhat surprising, a possible account for this finding is that
the N400 effect may be a reflection of a differential expectancy
developed for each type of anomaly based on the amount of
word-initial phonological and semantic information. More
specifically, in the tone and the syllable conditions, the final
(anomalous) word shared at least one feature (tone or syllable)
with the expected word. In contrast, in the double-anomaly
condition, the final (anomalous) word had no information in common
with the expected ending. If semantic integration takes place in an
incremental manner (Van Petten et. al. 1999), then the initial
acoustic information provided by the tone and the syllable
conditions may have increased the semantic expectation for the
correct ending. The double-anomaly condition in contrast would not
benefit from any of this shared acoustic information. The net
result would be a smaller N400 effect. This proposal based on
phonological activation is discussed in terms of a
spreading-activation model such as the one outlined in Dell
(1986).
|
