Distance effects or similarity-based interference?: A model comparison perspective

Shravan Vasishth
Universitaet des Saarlandes, Germany

Hawkins' Early Immediate Constituents (EIC) and Gibson's Discourse Locality Theory (DLT) predict that greater processing difficulty results from increasing distance between heads and dependents. By contrast, Lewis' Retrieval Interference Theory (RIT), which defines processing complexity in terms of similarity-based interference, predicts that increasing distance will result in easier processing if syntactically similar arguments are no longer adjacent. I show that
similarity-based interference, if broadly construed, can account for certain distance effects, and that increasing head-dependent distance can, surprisingly, facilitate processing.

Four studies (one acceptability rating task, and three self-paced reading studies) involving Hindi center embeddings were performed to investigate the above predictions. Distance was manipulated in two distinct ways: (a) by fronting indirect objects, and by fronting direct objects (Experiment 1, 2, 3); and (b) by inserting an adverb between the final NP and the innermost verb. Examples are shown in (1) below: (1a) shows canonical order, (1b) indirect- and (1c) direct-object fronting, and (1d) adverb insertion.

(1) a. siitaa-ne hari-ko kitaab-ko khariid-neko kahaa
Sita-erg Hari-ko book-the buy-inf told
`Sita told Hari to buy the book.'
b. hari-ko siitaa-ne kitaab-ko khariid-neko kahaa
c. kitaab-ko siitaa-ne hari-ko khariid-neko kahaa
d. siitaa-ne hari-ko kitaab-ko jaldi-se khariid-neko kahaa
quickly

There were two main results. First, when noun phrases were fronted, processing difficulty increased at the innermost verb, as predicted by Hawkins' and Gibson's models, and contra Lewis' model ((1a) vs. (1b,c); Expts. 1-3). Second, adverb insertion resulted, surprisingly, in easier processing at the innermost verb ((1a) vs. (1d); Expt. 4); this goes against Hawkins' and Gibson's models, but is predicted by Lewis' model.

These mutually contradictory results regarding the three models can be explained as follows. Fronting indirect or direct objects renders them more similar to subjects (since fronted objects are in the prototypical subject position), causing increased similarity-based interference between the actual subject and the fronted object. The easier processing due to adverb insertion occurs because the adverb strengthens the activation level of the current hypothesis (in working
memory) regarding the sentence completion.

In sum, I argue that: (a) although distance (or decay) in general does play a role in sentence processing, other factors, such as the degree of confidence in the predicted verb, play a greater role in certain contexts; and (b) similarity-based interference is not restricted to syntactic similarity but may extend to similarity in other dimensions, and this explain the putative distance effects such as those found in Expts. 1-3.

References

Gibson, E. 2000. Dependency Locality Theory: A Distance-based theory of linguistic compexity. In A. Marantz, Y. Miyashita, & W. O'Neil (Eds.). Image, Language, Brain: Papers from the first mind articulation project symposium. Cambridge, MA, MIT Press.

Hawkins, J. A. 1994. A Performance Theory of Order and Constituency, Cambridge University Press, New York.

Lewis, R. L & M. Nakayama. 2001. Determinants of processing complexity in Japanese Embeddings: New theory and data. In Sentence Processing in East Asian Languages, ed. by M. Nakayama. CSLI: Stanford.