Prediction Of Segmental And Supra-Segmental Information In Constrained Context

The proposal that prediction is a fundamental aspect of human information processing has gained evidence from studies in various fields.Over recent years however it was extensively explored in the experimental study in the field of human language processing.Previous empirical studies on prediction have provided strong evidence that people can predict phonological information of up-coming words during language processing.These findings,however,are absolutely based on Indo-European languages which are non-tonal(e.g.English),while this aspect of tonal languages remains unexplored.In addition,previous studies on Mandarin Chinese(a tonal language)found that both segmental and suprasegmental information play critical roles in Chinese lexical recognition.More importantly,it is found that the constraint of context is crucial to the weight of supra-segmental information in lexical recognition,in which supra-segmental information plays a more critical role in a more constraint context.Taken together,whether phonological information is predictable in Mandarin Chinese remains unclear yet.The present study,using Visual World Paradigm(VWP),investigates the following questions: 1)Will Chinese native speakers predict phonological information of a highly predictable word in highly constraint sentence context? 2)Can both segmental and suprasegmental information be predictable? If this is the case,whether these two types of phonological information can be differently predicted in time course? Participants were presented four pictures on a screen while listening to auditory sentences which contained a predictable final word.The four pictures contained three unrelated distractor objects and one of the four critical objects: a target stimulus object whose English name was corresponded to the predictable word [? ying 1(cherry)],an Segmental-Tone competitor stimulus object whose phonemes and tones were exactly the same as the predictable word but differed in meaning [? ying 1(eagle)],a segmental competitor stimulus object whose phonemes were the same as the predictable word but differed in tone [? ying 2(fly)],and a baseline stimulus object that was unrelated to the predictable word [? liu 3(willow)].The names of the competitor conditions were termed according to what they have in common relative to the target.Fixations on each critical objects were analyzed.The collected data in the present study was initially analyzed as a whole,and then the whole items was divided into high prediction items and low prediction items from the median of the predictability of each items to explore whether the pattern of effects were different between high prediction items and low prediction items.Results showed that participants fixated on target objects more than unrelated objects and two competitor objects(Segmental-Tone objects and Segmental objects)before target word onset both in total data and in two split data.Additionally,objects in Segmental condition consistently gained more fixations than objects in baseline in the early phase before target word onset either in total data or in two split data,while in the latter phase significant fixation preference to Segmental objects has been detected in all sorts of data except for in high prediction data.As for tone information,the effect is intended to be measured by a direct comparison between Segmental-Tone condition and Segmental condition,and an indirect comparison to compare Segmental condition or Segmental-Tone condition with baseline respectively.No tone effect was found in direct comparison.Indirect comparison,however,showed significant tone effects in total data and high prediction data,but no such effect was found in low prediction data.These findings demonstrate: 1)Chinese native speakers make phonological predictions about the up-coming words.2)Both segmental and supra-segmental information are predictable,and segmental information is pre-activated earlier than supra-segmental information.In addition,the pre-activation of supra-segmental information requires a relatively high constraint of context.Taken together,the present study suggests that segmental information of up-coming words takes its part earlier than supra-segmental information in Chinese language prediction while the amount of contextual support is crucial to the pre-activation of supra-segmental information.This is compatible with the proposal that comprehenders employ their language production system to make representation in prediction.