The Effects Of Word Frequency And Semantic Transparency On Word Segmentation In Chinese Reading

Words are regarded as the primary semantic units for lexical segmentation and recognition in reading processing for both alphabetic languages and Chinese. Inserting inter-word spaces into Chinese texts has been proved to have a facilitating effect on the word processing in Chinese reading for native Chinese as well as non-native Chinese learners; while inserting spaces among characters are found to have an inhibitory effect on word processing. However, according to the IIC (Inter/Intra Connection) Model, compounds with different linguistic features are represented and stored differently in mental lexicon. Compounds of high frequency or opaque meaning are represented holistically; but the constituents of compounds with low frequency or high semantic transparency are stored separately. So the potential modulating effects from the linguistic features on Chinese word segmentation might be neglected by the previous studies. We predict that bimorphemic words of different frequency and semantic transparency might be segmented and recognized differently in Chinese reading, and that inserting spaces among characters would facilitate the processing of words with low frequency and high semantic transparency. Therefore we inserted spaces into Chinese sentences to form four presenting conditions (ie. normally unspaced condition, word spaced condition, character spaced condition and non-word spaced condition), and manipulated the word frequency and semantic transparency of the target words to form four treatments (ie. high frequency transparent words, high frequency opaque words, low frequency transparent word and low frequency opaque words). With repeated measures ANOVAs, we conducted both global analyses based on the sentences reading and local analyses based on the regions of target words processing. The results showed reliable spacing effects. The local analyses showed a robust facilitating effect of word spaced presenting condition, which was even much greater in the later processing stage; more interestingly, there was also a facilitating effect for character spaced condition in the early processing stage. The findings suggested that words were the primary processing units in Chinese reading, and inserting spaces among characters also facilitated the early stage of word processing. Besides, sentences with high frequency target words were processed faster than those with low frequency ones; while semantic transparency had an inhibitory effect on the early and overall word processing course. There was not interaction between the sentence presenting condition and the word frequency or semantic transparency, indicating that word frequency and semantic transparency did not influence the word segmentation in Chinese reading in an integrative way.