Human sentence processing unaffected by sentence structure, eye-movement finds
- Date:
- February 16, 2011
- Source:
- Universiteit van Amsterdam (UVA)
- Summary:
- The hierarchical structure of sentences appears to be less important in human sentence processing than previously assumed, according to a new study of readers' eye movements. Readers seem to pay attention to simple word sequences above all, according to researchers in the Netherlands. Their findings, which run counter to the prevailing view, provide new insights into human language cognition.
- Share:
The hierarchical structure of sentences appears to be less important in human sentence processing than previously assumed, according to a new study of readers' eye movements. Readers seem to pay attention to simple word sequences above all. These are the conclusions of research conducted by Dr. Stefan Frank and Prof. Rens Bod from the Institute for Logic, Language and Computation (ILLC) of the University of Amsterdam (UvA).
Their findings, which run counter to the prevailing view, will soon be published in the journal Psychological Science. These results provide new insights into human language cognition: structures that have been considered extremely important in understanding language do not appear to have psychological relevance.
Seen superficially, sentences consist of a series of words. However, sentences also have a deeper hierarchical structure: they consist of phrases, which may themselves consist of phrases, and so on. Since the pioneering work of the linguist Noam Chomsky in the 1950s, most psycholinguists believe that this structure plays a crucial role in sentence processing. However, Frank and Bod show that the cognitive system of language users is especially sensitive to the superficial, serial structure of sentences, so the hierarchical structure does not really matter.
Expectations during reading
While reading text, expectations are continuously built up with regard to the words to follow. Infringing on these expectations slows down reading, which is detectable in the reader's eye movements. The UvA researchers took a dataset of these types of eye movement patterns and connected that to various statistical models of language. Based on certain assumptions about the sentence structure, these models calculate the degree to which each word was to be expected. It appeared that the eye movements were accurately predicted by models that look only at the superficial word sequences rather than hierarchical structures. This suggests that the reader largely ignores sentence structure and pays attention to the word sequences instead.
The study is part of the Vici-programme 'Integrating Cognition' funded by the Netherlands Organisation for Scientific Research (NWO) and headed by Rens Bod.
Story Source:
Materials provided by Universiteit van Amsterdam (UVA). Note: Content may be edited for style and length.
Cite This Page: