| |
Abstract:
This paper investigates the role of canonical word order in
sentence processing. Previous psycholinguistic studies on English
using the cross-modal lexical priming paradigm have shown that the
parser actively recovers the underlying SVO word order of English
during ongoing comprehension (Love & Swinney 1998). Similar
results were reported from Spanish, while no such evidence was
observed in Bulgarian. Clearly, more work is needed to determine
the role of underlying word order, in particular with respect to
different canonical word orders in different languages. In the
present study, we will investigate the processing of verb placement
in German, where SVO is used in main clauses and SOV in subordinate
clauses. In a self-paced reading study and in two ERP experiments
the following types of sentences were visually presented to
subjects:
a. SVO in main clauses, grammatical
b. SOV in main clauses, ungrammatical
c. SOV in subordinate clauses, grammatical
d. SVO in subordinate clauses, ungrammatical
Experiment I: Self-paced reading
We found that sentences with ungrammatical word order were
associated with longer reading times compared to grammatical word
order, but only for sentence type (b), i.e. for SOV instead of the
required SVO. For the reverse, i.e. for sentence type (d), no
differences in reading time were observed. These results suggest a
preference for SVO irrespective of (un)grammaticality.
Experiment II: ERPs on SVO vs. SOV
We found evidence for an SVO preference in this experiment. An
anterior negativity (possibly reflecting working memory load,
Weckerly & Kutas 1999) was associated with SOV compared to SVO
word order, even for grammatical SOV sentences. Effects of
ungrammaticality occurred later, between 700 and 1000 ms
poststimulus. Here, we found a large P600 for ungrammatical SOV
compared to grammatical SOV. For ungrammatical versus grammatical
SVO however, no reliable P600 effect was found. Given that the P600
amplitude varies as a function of the "severity" of the syntactic
anomaly (Osterhout 1997), the observed results are compatible with
the hypothesis that SVO is the word order which is easiest to
process; hence, changing (grammatical) SVO to (ungrammatical) SOV
represents a more severe violation than changing SOV to SVO.
Experiment III: The role of finiteness
In this ERP experiment we examined whether the SVO preference
found in experiments I and II is due to lexical-semantic properties
of verbs or due to the verb's morpho-syntactic properties (in
particular finiteness). ERPs were elicited to SVO and SOV sentences
containing auxiliaries (instead of lexical verbs as in exp. II).
The results are similar to those of experiment II and provide
evidence for an SVO preference caused by the position of the finite
verbal element, rather than by the lexical-semantic content of
verbs. Taken together, the results of the present study indicate
that the mind/brain prefers SVO word order during ongoing
comprehension, even in cases in which the grammar of German
requires SOV. This finding is compatible with previous results on
English and Spanish. Moreover, we found that morpho-syntactic
properties seem to be crucial for the observed SVO preference, in
particular the position of the verbal element bearing the feature
[+ finite].
References:
Love, T.E. & D.A. Swinney (1998). The influence of canonical
word order on structural processing. Syntax and Semantics, Volume
31: 153-166.
Osterhout, L. (1997). On the brain response to syntactic
anomalies: manipulations of word position and word class reveal
individual differences. Brain and Language 59, 494-522.
Weckerly, J. & M. Kutas (1999). An electrophysiological
analysis of animacy effects in the processing of object relative
sentences. Psychophysiology 36: 559-570.
|
