A Study of the Orthographic Influence of Learning Romaji Before English in Japanese Elementary Schools

A Study of the Orthographic Influence of Learning Romaji Before English in Japanese Elementary Schools

This thesis investigated the influence of learning romaji before English on the orthographic knowledge of Grade-5 Japanese elementary school students as measured prior to the Japanese education reform in 2020. While research has identified that orthographic depth can influence rate, strategy, and error types in reading, no research to date has investigated the orthographic influence from evidence in written responses of young children learning two orthographically contrastive alphabetic scripts.

One widely accepted assumption is that romaji facilitates English spelling. However, research surrounding orthographic influence indicates that shallow orthographies like romaji only partially facilitate orthographic knowledge transfer to deeper orthographies like English and this research considers the need to identify the place and direction of transfer. This thesis, therefore, contributes to the field of orthographic research by investigating the directional influence of learning romaji before English and by identifying what variables correlate with writing accuracy. As a result, three types of inhibitive orthographic influence were found, giving rise to the development of the Orthographic Gap Hypothesis (OGH).

To develop the OGH, a series of tests were administered to five groups of Japanese primary school students (N = 134) who had no formal English writing or reading skills. The primes/stimuli were designed according to developmental age word-frequency, word complexity, and orthographic complexity. Primes were ordered according to difficulty and grouped into categories. Each prime was presented for 300 milliseconds at three-second intervals. The series of tests included: a RAN test comprising native kanji and kana, colours, shapes and numbers, highly recognisable pictures, and aural primes of phonetically similar English alphabet names; a TACHiD test comprising 18 orthographically possible non-words; and two tests of implicit and explicit orthographic knowledge. Writing responses were collected on paper response forms and coded according to accuracy and error types for statistical and graphical analysis.

These tests provide new insights into the development of orthographic knowledge in preliterate learners of dual language alphabetic scripts of contrasting depth. Word frequency was related to word length as expected; however, word length had a non-linear, random relationship with errors, which contradicts previous research in more advanced learners, particularly monolinguals. In addition, the RAN test indicated that word complexity was significantly inversely proportional to writing accuracy (R2 = 0.98), which provides insights into unexplained variability in previous ODH research. Furthermore, RAN test results indicated that word frequency is a weak predictor of accuracy, but orthographic complexity significantly influenced writing accuracy, which accounts for the variation in previous research. Finally, orthographic complexity is influenced substantially by geminate obstruents, diphthongs with a mid-place 'y', and long vowels; a result that provides the answer to where orthographic influence is more significant during early literacy development.

The orthographic decision test (ODT) indicated that knowledge of English assisted the students' implicit understanding of word possibilities in each language, but poor English skills resulted in guessing. The orthographic recognition test (ORT) results, on the other hand, indicated that students with little to no English ability were better able to identify their native orthography based on the consistent V, CV, CVV patterns with a significance difference at p < 0.1, between the scores for romaji (M = 0.5, SD = 0.0829) and English (M = 0.46, SD = 0.0772), t(34) = 1.33, p = .0960. These results indicate that Japanese orthographic representations coincide with Japanese phonology as graphemic chunks, and therefore, can provide only negligible facilitation to English orthographic understanding at the letter level. These Japanese graphemic chunks bear little resemblance with English phonemic and graphemic patterns, which contradicts claims that romaji facilitates English.

These incidental results ushered the conception of the Orthographic Gap Hypothesis (OGH). The OGH recognises three types of inhibitory influences. A Type-0 gap is where there is no interference due to a one-to-one PG/GP correspondence between the two languages. A Type-1 gap is where there is no transference because the PG and GP correspondences are non-existent. Finally, a Type-2 gap is where the influence is confusing and requires mental processing that would usually take time and effort to correct (e.g., the vowels 'A' and 'U', and the phonemes for 'L' and 'R').

The development of the OGH in this thesis contributes to orthographic research because it identifies where and why some orthographic combinations are difficult to acquire and classifies the type of influence. Furthermore, the OGH is useful in identifying where students may struggle in reading and spelling new language scripts and also where learning a new language may interfere with existing orthographic knowledge even in other language settings. Finally, the OGH and methods developed in this thesis will be particularly useful in studying English earlier in Japanese schools.

File reproduced in accordance with the copyright policy of the publisher/author.