1. Introduction

The importance of prediction in language communication was introduced, and it was pointed out that predictive ability plays a crucial role in both native and foreign language communication. Research suggests that difficulties in foreign language comprehension may be related to lower predictive efficiency. The research in psycholinguistics indicates the existence of semantic and syntactic prediction, while speech prediction is a relatively new field of study. This study will use the eye movement visual contextual paradigm to explore the speech prediction effect in a foreign language environment and analyze the impact of foreign language proficiency on the prediction effect.

2. Literature review

Research on speech prediction effects in native language environments shows that native speakers are able to predict the phonetic features of upcoming words. DeLong et al.'s ERP experiment found that native English speakers start predicting the phonetic features of articles before hearing nouns, and articles that do not meet expectations can trigger the N400 effect. Subsequent studies have shown that the effect of speech prediction is influenced by the degree of contextual constraints and the duration of word intervals, while semantic prediction is not affected by time intervals, and speech prediction only occurs when the time interval is long and the degree of constraint is extremely high. Research on visual contextual paradigms shows that readers begin to focus on speech interference maps approximately 1000 milliseconds before hearing the target word, indicating that speech information has been activated in advance. But other studies have found delayed speech prediction effects or have not observed any speech prediction effects. The study of Chinese visual context paradigm has also found the effect of speech prediction, but the role of tone information in speech prediction is relatively limited.

3. Research methods

This chapter provides a detailed description of the research methods for pre activation of native language speech information in a second language context. Firstly, it was estimated through G * Power software that 30 participants were needed, and ultimately 32 high-level English learners from China were recruited. They were all English major students who passed the TEM-8 exam. The English proficiency of the subjects was evaluated through the LexTALE test, and the scores were generally high. The experimental materials include 32 English experimental sentences, each sentence ending with a highly constrained English word as the target word, accompanied by four versions of four grid image stimuli. The experiment adopts a Latin square design to ensure the balance of experimental conditions. The experimental process is divided into two parts: image familiarization task and eye tracking experiment, using Eyelink 1000 eye tracker for data collection. In terms of data analysis, a mixed effects logistic regression model is used to analyze the temporal trend of gaze data, with a focus on eye movement gaze data before and after the appearance of the target word, and the temporal process of speech prediction effect is explored through time window analysis. In addition, exploratory analysis was conducted to divide participants into high-level and low-level groups based on LexTALE scores, in order to investigate the impact of foreign language proficiency on speech prediction effects.

4. Experimental results

Eye tracking experiments showed that the accuracy rate of task completion for the subjects was 95.41%, indicating that the task can be effectively executed. The experiment observed a change in the fixation ratio before and after the appearance of the target word, and the target image and interference image attracted more fixation 500ms before the appearance of the target word. By merging the significant effect windows, it was found that within the -500 ms to -200 ms and 0 to 200 ms windows, both the target image and the interference image caused more fixation. The degree of contextual constraints does not affect the effectiveness of speech prediction. The proficiency level of foreign languages has an impact on the speech prediction effect, with the high-level group showing significant effects before and after the appearance of the target word, while the low-level group only shows effects after the appearance of the target word.

5. Discussion

This study found through the eye movement visual contextual paradigm that Chinese English learners can activate Chinese phonetic information corresponding to English vocabulary in advance when understanding English sentences. 500 milliseconds before the appearance of the target word, the subjects were able to predict the segmental information of the translated Chinese word, but the predictive effect of tone information was limited. The effect of speech prediction is influenced by foreign language proficiency. High level learners have more attention to interference words with segmental tone overlap and segmental overlap within a specific time window, while low-level learners have no significant difference. This indicates that pre activation of native language speech information exists in foreign language contexts, but the predictive effect of tone information is limited, and the strength of speech prediction effect is influenced by foreign language proficiency.

6. Conclusion

The study revealed the pre activation phenomenon of native language speech information in foreign language context through eye movement visual contextual paradigm, and found that the prediction effect of sound segments is stable, while the prediction effect of tone information is limited.