The moment our eyes fall on a visual stimulus, a complicated set of processes— physiological, neurological, and cognitive—is set in motion, enabling us to decode the perceived information and convert print to meaning by comparing it to our mental lexicon. Specifically, nerve impulses from the eyes stimulate an area near the back of the brain—the visual cortex--that allows us to see the light and dark strokes that define each word or character (Caplan, 1991). After the stimulation of the visual cortex, the angular gyrus, located in the left temporal lobe of the brain, maintains connections to areas involved with speech comprehension and integrates the perceived alphabet letters with their corresponding sounds stored in our mental lexicon. Such a grapheme-tophoneme mapping process, also termed phonological recoding, is an automatic decoding process that helps beginning readers familiarize themselves with word structure and assists their comprehension (Swank, 1994). In this vein, Van Orden (1987) regards phonological recoding as a compensatory coding device in associating written symbols with underlying sound codes that directly connect to their corresponding semantic codes. (A) The first empirical evidence for the above stipulation comes from Conrad’s seminal study. (B) In contrast, beginners had more difficulties in identifying pictures of visually-similar items due to their holistic learning experience and reliance on a direct link between orthography (or pictures) and semantics. (C) Despite the fact that auditory word recognition precedes visual word recognition, empirical first-language (L1) cquisition studies have shown that beginning L1 learners seem to rely more on pictorial or orthographic representations as media in remembering newly learned lexical items. (D) The above evidence suggests that there appears to be a difference in how learners of different proficiency profiles and at different developmental stages process lexical items.

       In an orthography based on an alphabet such as English, words are still characterized by structural redundancy. Due to such redundancy, beginning readers of English are soon confronted with a large number of irregular words characterized by similar orthography but with different pronunciations (e.g., pint, hint). In this case, if readers only use a direct orthographic coding approach (i.e., whole-word reading) to word identification, a likely consequence is the failure to detect the grapho-phonemic anomalies in irregular words. On the other hand, total reliance on phonological recoding might also limit beginning alphabetic readers’ ability to acquire an adequate corpus of whole words that can be readily and rapidly identified on sight (Vellutino & Scanlon, 1987). Consequently, the development of automatic whole-word reading might be impaired due to reliance on phonological recoding as a compensatory coding device for lexical access. The above contention entails that training with whole word reading and practice with phonological recoding are both needed for beginning readers of an alphabetic language.

       With the improvement in reading skills, advanced readers of an alphabetic language have acquired a full inventory of word attributes, with each of them functioning as an alternative coding device to lexical meaning (Underwood & Batt, 1996). In this regard, phonological recoding functions more as a (postlexical) scaffold for higher-order thinking (e.g., integrating information from connected sentences and paragraphs) than as a prelexical lower-level device in accessing lexical meaning. In support of this view, Segalowitz and Hebert (1994) argued that such a phonological recoding is useful postlexically, helping advanced readers to better integrate sentence-, and paragraph-level information, without dampening word-level comprehension.

       The above observations of phonological recoding in advanced and beginning readers’ lexical processing were drawn from first language (L1) acquisition studies. Few studies have investigated the role of phonological recoding in advanced and beginning L2 readers—in particular those L2 readers of an ideographic language. It is possible that phonological recoding process may not be completely the same for readers of an alphabetic language (e.g., English) and for readers of an ideographic language (e.g., Chinese). Hitherto, this issue has not been empirically examined. Only a few scholars have discussed this issue in some position papers. Xu’s (1991) position paper is a case in point. For instance, Xu contends that in spite of the fundamental structural differences in alphabetic and ideographic languages, there exist some universal processing elements regarding how a language is decoded and how phonological recoding is performed by readers of different languages. According to Xu, readers of an alphabetic language access the phonological representation of printed words analytically because the writing system more or less directly reflects the phonological representation of its orthography. Similarly, the phonological code of a Chinese ideograph can also be analytically derived from its phonetic radical (if there is any)； this analytical process is especially evident in the decoding of unfamiliar characters (e.g., Feng, Miller, Shu, & Zhang, 2001) and semantically indeterminate lexical items (e.g., Tan, Hoosain, & Peng, 1995). Xu (1991) thus made an a-priori claim that the underlying phonological recoding process for alphabetic languages and for ideographic languages is similar. In light of the gap in existing phonological recoding research and to empirically establish Xu’s view, this study will be set out to investigate whether or not phonological recoding has parallel effects on L2 learners of different language systems (alphabet vs. ideograph) at different proficiency levels.