Construction of a Syntactic Processing Automaticity Measurement for Japanese EFL Learners

全文

(1)Construction of a Syntactic Processing Automaticity Measurement for Japanese EFL Learners Sawako HAMATANI Kansai University Abstract The reading rate of Japanese EFL learners’ is extremely low. Given the high complexity of the reading process, research on reading problems has widely relied on componential analysis, a method that divides reading into higher- and lower-level processes. In lower-level processes, text information is retrieved, while in higher-level processes, text information is processed for comprehension. According to previous research, lower-level processes must be automatized to enable fluent comprehension processes. Despite its importance, few empirical studies about the automatization of foreign language reading have been conducted, and only a handful of studies have focused on the automatization of syntactic processing (one of the lower-level processes). A reliable measure is necessary to estimate how syntactic processing is automatized. This study developed a test to measure the level of syntactic processing automaticity based on skill acquisition theory and implicit-explicit knowledge theory of second language acquisition. A pilot test was conducted to examine the reliability of the test, and 74 university students participated. The reliability analysis resulted in a Cronbach’s α of .685. Rasch measurement statistics for item difficulty and model fit suggested the need for further improvement in the test. After analyzing the reasons for the misfits, the items were modified or replaced with more appropriate ones. 1. Introduction Reading fluency has been investigated extensively in the L1 field over the past decades as it is considered a prerequisite for successful reading (e.g., Allington, 1983; Kuhn & Stall, 2003; Schreiber, 1980). However, reading fluency has received relatively less attention in the field of foreign language learning – attention that has been growing recently for several reasons. First, fluent reading allows readers to get a much greater amount of second language (L2) input. Second, reading fluency allows for enjoyable reading. Third, acquiring reading fluency is necessary for readers to process the overflow of information typical of the Internet era. Reading fluency is often defined as “the ability to read with ease and accuracy and to read with appropriate expression and phrasing” (Grabe, 2009, p.291). Since this study targets the silent 145.

(2) reading of Japanese EFL learners, reading fluency is defined here as quick and accurate reading, with no regard to appropriate expression and phrasing. Emphasizing accuracy, Japanese formal foreign language education has paid little attention to reading fluency. This characteristic, however, does not in itself justify the extremely low reading rate of Japanese EFL learners, which corresponds to half to one-third of an L1 student’s reading rate (Beglar, Hunt, & Kite, 2011; Iwahori, 2008; Taguchi, TakayasuMaass, & Gorsuch, 2004). There is no easy way to identify the main causes of the low reading rate of Japanese EFL learners, as reading involves complex processes. To address this issue, research on reading has widely adopted componential analysis, a method that divides reading into lower- and higher-level processes. In lower-level processes, orthographic, phonological, syntactic, and semantic information is perceived; in higher-level processes, words are processed for comprehension while syntactic and semantic information is preserved in working memory. Higher-level processes generally require attentional resources, whereas lower-level processes have strong potential to become automatized (Grabe, 2009). Fluent higher-level processes require the automatization of lower-level processes, because working memory, which is where reading comprehension processes are executed, is capacity-limited. Lowerlevel process automatization enables reduced workload processing memory, sufficient attention allocation in higher-level comprehension processes, and the probable occurrence of fluent comprehension (Koda, 2004; Stanovich, 1990, 2000). The purpose of this study is to develop a test to measure the level of automatization of syntactic processing, which is one of the most important lower-level processes. 2. Research Background 2.1 Automaticity and skill acquisition theory Automaticity has attracted substantial attention from cognitive psychologists over the past decades. While the term “automatization” (i.e., the process of making automaticity) may take on different interpretations, ranging from the mere acceleration of a task to a qualitative change in its execution (LaBerge & Samuels, 1974; Logan, 1997; Segalowitz, 2000), automatization is broadly defined and accepted as the phenomenon in which reaction time and error rates gradually drop off and interferences from and with simultaneous tasks are diminished (DeKeyser, 1997). Researchers disagree on how automaticity is brought about in cognitive psychology. Several competing models explain how automaticity occurs, among which Anderson’s adaptive control of thought (ATC*) model (Anderson, 1987, 1992) is the most widely endorsed. According to this model, a skill develops over three stages: declarative, procedural, and automatized (Anderson, 1987, 1992). These three stages are characterized by major differences in the nature of knowledge. Initially, learners acquire knowledge about a specific skill (i.e., 146.

(3) declarative knowledge). Subsequently, they turn this knowledge into behavior (i.e., procedural knowledge). Finally, after extensive practice to decrease reaction time, error rate, and the amount of attention required, the automatization of knowledge occurs (DeKeyser, 2015). Automaticity is gradual and widely being accepted as a continuum of an affair, rather than as dichotomous (DeKeyser, 1997). 2.2 Studies on automaticity and lower-level processes Although L2 research on the automatization of lower-level processes is still in its infancy, several studies have explored the relationship between the automatization of lower-level processes and reading comprehension or reading fluency. Favreau and Segalowitz (1983) showed that word reading automaticity remained a powerful factor in the reading fluency of advanced L2 readers. Fukkink, Hulstijn, and Simis (2005) investigated the automatization of lexical access in L2 with computer-based training and verified that lexical access for some words was accelerated, but did not identify any transfer of acceleration of lexical access to reading speed. Nassaji and Geva (1999) investigated the role of phonological and orthographic processing skills in adult L2 reading and verified that efficiency in orthographic processing particularly contributed to reading measures. With the exception of Nassaji and Geva (1999), who explored syntactic processing in conjunction with word-level automaticity, all these studies mainly investigated word-level automaticity and its relationship to reading comprehension or fluency. However, to the best of our knowledge, few studies have investigated syntactic processing automaticity and its relationship to reading comprehension or fluency. Syntactic processing automaticity in reading is as important as word reading automaticity; therefore, it should not be ignored. Syntactic processing provides instructions for the construction of text comprehension by extracting information from determiners, word ordering, subordinate clauses, tense modality, and pronominal forms, among other information (Grabe, 2009). Furthermore, Alderson (1993) has produced evidence that grammar and reading are strongly connected, and Bernhardt (2000) emphasized the contribution of grammatical knowledge to reading performance. Therefore, there is a need to construct a measurement that evaluates syntactic processing automaticity. 2.3 The issue of explicit and implicit knowledge Influenced by the controversy of cognitive psychology, a discussion of implicit and explicit learning and knowledge figured in the field of L2 acquisition. Krashen (1981) distinguished “acquisition” (i.e., the implicit internalization of grammatical rules) from “learning” (i.e., the conscious formulation of explicit grammatical rules) and posited that implicit and explicit L2 knowledge involve different mechanisms of acquisition, rejecting the idea that explicit knowledge directly becomes implicit and vice versa. Such distinction led to the argument known as the “interface issue.” In contrast to Krashen’s concept, which is 147.

(4) described as “noninterface position,” the “strong interface position” claims that explicit knowledge can be derived from implicit knowledge, while explicit knowledge can be turned into implicit knowledge through practice (DeKeyser, 1998, 2007). This study adopts the “strong interface position” and posits that Japanese EFL learners usually obtain declarative knowledge through explicit learning in the first stage, develop it into procedural knowledge, and turn it (at least in part) into implicit knowledge. Although there is still controversy on their definition, explicit and implicit knowledge have not yet been properly addressed in empirical studies. Ellis (2005) argued that one reason for this is the lack of accepted instruments that distinctly measure explicit and implicit knowledge, and developed a battery of tests that could provide separate measures of implicit and explicit knowledge. The author postulated the key characteristics of implicit and explicit knowledge (see Table 1). According to Table 1, the characteristics of implicit knowledge include “procedural knowledge of rules and fragments” and “access to knowledge by means of automatic processing,” which can be compared to procedural and automatized knowledge, respectively. Therefore, from the perspective of skill acquisition, Ellis’s definition implies that implicit knowledge can be considered procedural knowledge that has automatized accessibility. According to DeKeyser (1997), some procedural knowledge may be in the process of automatization because automaticity is assumed not as dichotomous, but as a continuum. Ellis (2005) developed five tests: oral imitation test, oral narrative test, timed grammaticality judgment test (TGJT), untimed grammaticality judgment (GJT), and metalinguistic knowledge test. From the results of a principal component factor analysis of the test scores, he established that the imitation, oral narrative, and TGJT tests measure implicit knowledge, while the metalinguistic knowledge and GJT tests measure explicit knowledge. Table 1 Key characteristics of implicit and explicit knowledge Characteristics. Implicit knowledge. Explicit knowledge. Awareness. Intuitive awareness of. Conscious. linguistic norms. linguistic norms. Procedural knowledge of. Declarative knowledge of. rules and fragments. grammatical rules and. Type of knowledge. awareness. of. fragments Systematicity Accessibility Use of L2 knowledge. 148. Variable but systematic. Anomalous and inconsistent. knowledge. knowledge. Access to knowledge by. Access to knowledge by. means of automatic processing. means of controlled processing. Access to knowledge during. Access to knowledge during. fluent performance. planning difficulty.

(5) Self-report. Nonverbalizable. Verbalizable. Learnability. Potentially only within. Any age. critical period Note. Adapted from “Measuring implicit and explicit knowledge of a second language: A psychometric study,” by R. Ellis, 2005, Studies in Second Language Acquisition, 27, p. 151.. According to the definitions in Table 1, explicit knowledge can be accessed by controlled processing (i.e., not automatic). However, DeKeyser (2015) argues that explicit knowledge can also be automatized (or near automatized) and that automatized explicit knowledge and implicit knowledge are distinct constructs. As DeKeyser posits, the difference is that implicit knowledge occurs without awareness (with intuitive awareness in Table 1), while automatized explicit knowledge is accompanied by awareness. In the TGJT, which is presumed to measure implicit knowledge according to Ellis (2005), although grammaticality must be judged automatically or fairly quickly, the test taker’s attention is directed to forms (with awareness); “therefore, timed grammaticality judgment tasks are considered to tap into automatized explicit knowledge” (Suzuki & DeKeyser, 2017, p. 749). The purpose of this study was to develop a test to measure automatized syntactic knowledge, as there is no established instrument to measure the automatization of syntactic processing in L2 reading. To this end, a TGJT for Japanese EFL learners was developed based on Ellis (2005). Although Ellis claims that TGJTs measure implicit grammar knowledge, this study relies on DeKeyser’s (2015) assumption that a TGJT can measure automatized explicit knowledge; this is because, as all participants were educated primarily in the Japanese formal education system, they will be aware of the grammatical forms of the sentences tested. 3. Method 3.1 Participants Participants were 74 native Japanese students (33 women, 41 men) enrolled in undergraduate courses at a university in western Japan. Twenty-six majored in commerce, 23 majored in foreign language studies, 18 majored in chemistry and bioengineering, five majored in economics, and two majored in policy studies. Students in general had received formal education for approximately six to nine years. Participants who majored in foreign language studies had studied abroad in an English-speaking country for about one year. 3.2 Instrument Specifications regarding the test context and purpose were established based on Carr (2011), as shown in Table 2. The purpose of the TGJT was to measure the extent to which the syntactic processing of Japanese EFL learners was automatized. The construct to be assessed 149.

(6) and their definition is automatized syntactic knowledge. A norm-referenced interpretive framework has been adopted, as this test was not designed to be used in a specific situation; however, a criterion-referenced framework could be adopted in the case of a specific situation, such as the assessment of the level of achievement of grammar learning in school education. In other words, the interpretive framework depends on how the test results will be used. The target language use is the academic domain: formal education, cram school, among others. The task type is yes/no questions. Test takers were assumed to be Japanese EFL learners who have been educated in the Japanese formal school system. Finally, as for the minimum acceptable levels for each of the usefulness qualities, practicality was emphasized because, as this test had to be conducted along with other tests in a single session, it was necessary to determine an appropriate number of items. Reliability and construct validity should be confirmed for the test to be considered a reliable measure. Table 2 Specifications regarding the test context and purpose Component Purpose of the TGJT. Explanation To measure the extent to which Japanese EFL learners’ syntactic processing is automatized. Construct to be assessed and its definitions. Automatized syntactic knowledge. Interpretive framework. Norm-referenced. Target language use (TLU) domain and. Academic domain: formal education, cram. common/important task types. school, etc. Task type: yes/no. Characteristics of test takers. Japanese EFL learners who have been educated in Japanese schools. Minimum acceptable levels for each of the. Reliability, construct validity, practicality. qualities of usefulness. Grammaticality Judgment Test (GJT). GJTs have been extensively used in the field of second language acquisition (SLA) since the mid-1970s. In GJTs, learners are required to make judgments about individual sentences and decide whether the sentences are grammatical or ungrammatical. Although there has been some criticism about their validity and reliability, GJTs are still widely used because they are assumed to provide data that represent a second language learner’s competence to a high degree and allow researchers to collect specific types of data about particular grammatical structures (Davies & Kaplan, 1998). The timed GJT (time-constrained GJT) for Japanese EFL learners in this study was developed based on Ellis (2005). Ellis’s GJT includes 17 grammatical structures that were considered universally problematic. Ellis’s TGJT and untimed GJT adopted the same questions. 150.

(7) Five of the 17 structures were replaced with others that were considered to be more appropriate for the purpose of TGJT for Japanese EFL learners (Table 3). Among the five structures, question tags and “yes/no” questions were considered to be too colloquial and therefore not appropriate, since the purpose of the TGJT for Japanese EFL learners was to measure the grammatical knowledge used during reading. In addition, indefinite articles are extremely difficult for Japanese EFL learners to master (Yamada & Matsuura, 1982; Butler, 2002). Ergative verbs are not explicitly taught in the Japanese formal education, which makes it difficult for Japanese EFL learners to use them correctly (Otaki & Shirahata, 2017). On the other hand, the use of possessive -s is relatively easy for Japanese EFL learners, in part because the system resembles that of the Japanese language (Shirahata, 1988). As shown in Table 3, these structures were replaced by five other structures that are considered problematic for Japanese EFL learners. The test had 68 sentences (34 grammatical and 34 ungrammatical), comprising 17 grammatical structures (see Table 4), which were extracted from and based on EIKEN Grade Pre-1 to Grade 4 (Eiken Foundation of Japan, 2017), Grammar in Use Intermediate (Murphy & Smalzer, 2009), Practical English Usage (Swan, 2005), and New College English-Japanese Dictionary (Takebayashi, Higashi, Ichikawa, & Suwabe, 2003). Table 3 The replaced experimental grammatical structures Ellis (2005) Question tags. colloquial. S. GJT for Japanese EFL learners Active and passive. S. Yes/no questions. colloquial. M. Infinitive modifier. S. Indefinite article. too difficult. M. Tense agreement. S. Ergative verbs. too difficult. M. Present perfect. S. Possessive -s. easy. M. Past perfect. S. Note. S = syntactic; M = morphological. The test was programed using SuperLab version 5 software and run on a laptop computer. Question sentences appeared on the computer screen one by one and participants should press the “y” key if they considered the sentence to be grammatical or the “n” key if they considered it ungrammatical within time limit. In a preliminary session, the response time data for each sentence were obtained from eight native English speakers; then, four graduate and five undergraduate Japanese students took the test with 50 percent and 60 percent more than the average time of the native speakers, respectively. Considering the results, 70 percent of the native speakers’ average response time was added to establish time constraints. Participants answered several practice questions and then completed the test individually in a classroom or 151.

(8) office. The test lasted about 10 to 15 minutes in total. Table 4 Experimental grammatical structures Structure. Error example. Type. Verb complements. Jim wanted his decisions independent of his parents' advice.. S. Regular past tense. Ken finds some money in front of the library yesterday.. M. Modal verbs. My grandfather could spoke five languages fluently.. M. Unreal conditionals. If you are in my position, what would you do?. S. “Since” and “for”. They have known each other for they were in high school.. S. Active and passive. Japanese are depend on fish for half of their animal protein.. S. Plural -s. Our life would be difficult without waters.. M. Third person -s. Julie don't drink coffee very often.. M. Relative clauses. This weekend, Brian will go to the town which he grew up.. S. Embedded questions. I don't know why didn't Kelly come to the party.. S. Dative alteration. They donated the museum some money.. S. Comparatives. The band was the worse of the three bands at the concert.. S. Adverb placement. Kate has every week a new hair style.. S. Infinitive modifier. I'd like something stop my toothache.. S. Tense agreement. It was the first time that I hear her sing.. S. Present perfect. Susan really loves that movie, so she sees it eight times.. S. Past perfect. Amy has just gotten home when I called.. S. Note. S = syntactic; M = morphological. Analysis. Test results were analyzed using the one-parameter IRT dichotomous model (Rasch). Data were processed using WINSTEPS 3.81 software. The Rasch model was adopted because it provides stable estimates of item difficulty using a true interval scale presumed to underlie the traits. Furthermore, Rasch fit statistics provide useful data for examining the quality of test items (Bond & Fox, 2015). Fit statistics shows the degree of match between the pattern of observed responses and the modeled expectation, which can exhibit the pattern for each item in each person. On the other hand, WINSTEPS calculates the amount of distortion between the modeled expectation and the observed results as mean squares. Item difficulty estimates are expressed in logits, in which a logit value of zero is set as the mean. Items with positive logit estimates are more difficult, while items with negative logit estimates are easier. Person ability estimates are also indicated in logits. Persons who have positive logit values have higher ability, while persons with negative logit values have lower ability.. 152.

(9) 4. Results and Discussion Table 5 presents the mean and standard deviation for the measure. The internal consistency of the test (Cronbach’s alpha) was .684. Table 6 shows the summary of item statistics. “Infit Mnsq” and “Outfit Mnsq” show the amount of distortion between the modeled expectation and the observed results. “Infit” is more sensitive to the pattern of responses of those persons located closer to the item’s difficulty value, and “outfit” is more sensitive to responses to items with difficulty far from the person. “Zstd” comprises the standardized mean squares (z-scores). Z-scores are expressed in terms of standard deviations from their means, and have a distribution with a mean of 0 and a standard deviation of 1. The productive measurements considered for the mean square and standardized values were 0.5–1.5 and -2.0– 2.0, respectively. Table 6 indicates that the maximum standardized infit and outfit mean square values (both 2.7) and the minimum standardized infit mean square value (-2.2) are out of range. Figure 1 shows item difficulty and person ability on the same logit scale. The numbers from 2 to -2 on the left side show logits. The left side of the broken line shows the distribution of participants indicated by x. The right side of the broken line shows the distribution of items, which are indicated by the numbers. The higher on the scale, the more difficult the items and the higher the persons’ ability. According to Figure 1, item 22 is the most difficult and item 16 is the easiest. The letters M, S, and T, common to both items and persons, correspond to mean, standard deviation, and two standard deviations, respectively. Comparison of item and person distribution shows that many persons are positioned around the mean, but a fair number of items are distributed under M, meaning that the TGJT was relatively easy for participants. Table 5 Descriptive statistics for the test M SD GJT for JEFLLs. 36.3. 6.8. N. α. 74. .684. Table 6 Item measure statistics Measure Mean SD Max Min. 0.00 .85 1.69 -2.18. Infit Mnsq 1.00 .06 1.17 .89. Outfit Zstd .0 .8 2.7 -2.2. Mnsq 1.00 .10 1.35 .84. Zstd .1 .9 2.7 -1.8. 153.

(10) Figure 1. Item-person map. 154.

(11) Table 7 shows item statistics for the eight worst misfit in order of degree of misfit. “Measure” shows the difficulty with logits. The infit and outfit mean squares for the eight items are within the productive range, but the infit zstd for item 67 and the outfit zstd for items 12 and 67 are out of range. The point-measure correlations for items 48, 12, and 67 show negative values. Table 8 shows the item category, option, and distractor frequencies of the eight worst misfit items in order of degree of misfit. The average ability of participants who correctly answered item 48 was .06*, which is much lower than those who gave incorrect answers (.20). Ten low-ability students correctly answered item 48. Similarly, the average ability of participants who correctly answered item 12 was .14*, against .21 of those who gave incorrect answers, and one very high-ability student failed. For item 67, the average ability of those who answered correctly was .10* against .22 of those who gave incorrect answers, and one lowability student correctly answered unexpectedly. These figures indicate that items 48, 12, and 67 need some revision. Item 2 was ranked as the third worst misfit item because seven relatively high-ability students incorrectly answered it unexpectedly, although it is an easy item (-1.18 in measure in Table 7). Conversely, with regard to item 47, five low-ability students correctly answered it unexpectedly, although it is quite difficult (1.36 in Table 7). Although items 34, 32, and 29 are relatively easy (-.87, -.29, -.18 in measure, respectively, in Table 7), five relatively high-ability students failed in item 34, and one very high-ability student failed in items 32 and 29. The actual sentences of misfit items are shown in Table 9. Table 7 Misfit items in order of degree of misfit Item 48. Total Score 14. Total Count 74. Model S.E. .30. Infit Mnsq 1.10. Zstd .6. Outfit Mnsq 1.35. Zstd 1.6. Point-measure Corr. Exp. -.11 .19. 12. 44. 74. -.24. .24. 1.13. 1.9. 1.20. 2.5. -.08. .22. 2. 58. 74. -1.18. .29. 1.05. .4. 1.19. 1.0. .02. .18. 67. 32. 74. .45. .24. 1.17. 2.7. 1.19. 2.7. -.12. .23. 47. 18. 74. 1.36. .28. 1.05. .4. 1.16. 1.0.. .06. .20. 34. 54. 74. -.87. .27. 1.06. .6. 1.15. 1.0. .03. .20. 32. 45. 74. -.29. .24. 1.09. 1.3. 1.14. 1.6. .02. .22. 29. 43. 74. -.18. .24. 1.08. 1.3. 1.12. 1.6. .04. .22. Measure 1.69. 155.

(12) Table 8 Misfit items category, option, and distractor frequencies in order of degree of misfit No. 48. 12. 2. 67. 47. 34. 32. 29. . Data Code 0. Score Value 0. Data Count % 60 81. Av. Ability .20. S.E. Mean .06. Outf. Mnsq 1.1. Ptma Corr. .11. 1. 1. 14. 19. .06*. .17. 1.4. -.11. 0. 0. 30. 41. .21. .11. 1.3. .08. 1. 1. 44. 59. .14*. .06. 1.1. -.08. 0. 0. 16. 22. .16. .15. 1.2. -.02. 1. 1. 58. 78. .17. .06. 1.0. .02. 0. 0. 42. 57. .22. .07. 1.2. .12. 1. 1. 32. 43. .10*. .09. 1.2. -.12. 0. 0.. 56. 76. .15. .06. 1.0. -.06. 1. 1. 18. 24. .22. .14. 1.2. .06. 0. 0. 20. 27. .15. .13. 1.2. -.03. 1. 1. 54. 73. .18. .06. 1.0. .03. 0. 0. 29. 39. .16. .10. 1.2. -.02. 1. 1. 45. 61. .18. .07. 1.1. .02. 0. 0. 31. 42. .15. .10. 1.2. -.04. 1. 1. 43. 58. .19. .07. 1.1. .04. Item 48 asked whether the plural of “electricity” was correct or not. Most participants knew that “electricity,” ranked 924 in JACET 8000 (committee of Revising the JACET Basic Words, 2016), is a material noun and has no plural form; however, under the time-pressure condition, this item could have been misleading because it contained relatively complicated spelling. In the revised item, “water,” which is ranked 174 in JACET 8000, was the material noun that determines the correctness of the sentence. Item 12 questioned the usage of “occur to,” which means “(of a thought or idea) come to the mind of.” This item may have been less familiar to participants compared to the usage of “happen,” which is placed in the first definition in most dictionaries. The former usage was not found in the International Corpus Network of Asian Learners of English (ICNALE) (Ishikawa, 2013). Moreover, generative linguists theorize verbs such as “occur” as “unaccusative verbs,” which do not have a semantic agent as grammatical subject and are difficult for Japanese EFL learners to use (Saito, 2005). Therefore, the item was changed to a sentence that included “concentrate,” which has both 156.

(13) transitive and intransitive usages but only one main meaning. Table 9 Misfit items and replaced ones No. 48. Item sentences Our lives would be difficult without electricities.. Grammatical structures plural -s. W. active/passive. C. verb complements. C. past perfect. W. plural-s. W. past perfect. C. present perfect. C. tense agreement. C. → Humans can’t live without waters. 12. Just then a bright idea occurred to me. →Ken needs to concentrate on his new job.. 2. The ancient Greeks believed Mount Olympics to be sacred. →The Japanese believe Mt. Fuji to be sacred.. 67. We are not hungry, as we had just had lunch. →I didn’t know the woman, as I have never seen her before.. 47. Do you speak any foreign language? →Would you like some more teas?. 34. Ken was very nervous on the plane because he had not flown before. →The house was dirty because they hadn’t cleaned it for weeks.. 32. Kelly lives in Toronto, and she has lived there all her life. →Kelly has lived in Tokyo all her life.. 29. The princess decided that she didn’t like staying at home all day. →Jane said that she didn’t like staying at home all day.. Note. C = correct; W = wrong. In item 2, unfamiliar words (“Greeks” and “Mount Olympics”) may have consumed extra time. They were changed to much familiar words to participants: “Japanese” and “Mount Fuji.” Items 67, 34, 32 have perfective aspects. Sentences that have perfective aspects may have been difficult even for some of the most advanced participants. Item 67 may have been problematic, because it presented two “had,” one being an auxiliary verb and the other a verb. Item 67 was changed to a sentence whose tense distortion was assumed to be easier to find. Items 34 and 32 were changed to sentences that were shorter or had familiar topics. Item 47 may have taken considerable time for participants to remember the plural form that follows “any.” This item was changed to a sentence that had a mistake, “teas,” which was assumed to be easier to find. Finally, item 29 was shortened to make it easier to find the correctness of tense agreement. 157.

(14) 5. Conclusion The purpose of this study was to develop a test to measure the syntactic processing automaticity of Japanese EFL learners. A TGJT was constructed based on skill acquisition theory and explicit-implicit knowledge theory of SLA. The reliability score was not high but satisfied the criteria (α= .684). The eight worst misfit items were revised in accordance with the Rasch analysis. The revised items would need further analysis in the next pilot study. To identify the causes of students’ slow reading speed, a word recognition test must be conducted simultaneously, because the TGJT does not certify whether the reason for the wrong answers or the lack of answers is slow syntactic processing or word recognition. References Alderson, J. C. (1993). The relationship between grammar and reading in an English for academic purposes test battery. In Douglas, D. and Chapelle, C. (Eds.), A new decade of language testing research: selected papers from the1990 Language Testing Research Colloquium: dedicated in memory of Michael Canale (203-219). Alexandria, VA: Teachers of English to Speakers of Other Languages. Allington, L. (1983). The reading instruction provided readers of differing reading abilities. The Elementary School Journal, 83, 548–559. Anderson, J. R. (1987). Skill acquisition: Compilation of weak-method problem solutions. Psychological Review, 94, 192–210. Anderson, J. R. (1992). Automaticity and the ACT*theory. American Journal of psychology, 105, 165–180. Beglar, D., Hunt, A., & Kite, Y. (2011). The effect of pleasure reading on Japanese university EFL learners’ reading rates. Language Learning, 62, 665–703. Bernhardt, E. B. (2000). Second language reading as case study of reading scholarship in the 20th century. In M. L. Kamil, P. D. Pearson, and R. Barr (Eds.), Handbook of reading research. Vol. lll (pp.793–811). Mahwah, NJ: Lawrence Erlbaum. Bond, T. G., & Fox, C. M. (2015). Applying the Rasch Model: Fundamental Measurement in the Human Sciences. New York: Routledge. Butler, Y. G. (2002). Second language learners’ theories on the use of English articles. Studies in Second Language Acquisition, 24, 451–480. Carr, N. T. (2011). Designing and Analyzing Language Tests. Oxford University Press. Committee of Revising the JACET Basic Words (Ed.). (2016). The new JACET list of 8000 basic words. Tokyo: Kirihara-shoten. Davies, W. D., & Kaplan T. I. (1998). Native speaker vs. L2 learner grammaticality judgements. Applied Linguistics, 19, 183–203. 158.

(15) DeKeyser, R. (1997). Beyond explicit rule learning: Automatizing second language morphosyntax. Studies in Second Language Acquisition, 19, 195–221. DeKeyser, R. (1998). Beyond focus on form: Cognitive perspectives on learning and practicing second language grammar. In C. Doughty and J. Williams (Eds.), Focus on Form in Classroom Second Language Acquisition (pp. 42–63). New York: Cambridge University Press. DeKeyser, R. (2003). Implicit and explicit learning. In C. Doughty, & M. Long (Eds.), The handbook of second language acquisition (pp.313–348). Malden, MA: Blackwell. DeKeyser, R. (2007) Introduction: Situating the concept of practice. In R. DeKeyser (Ed.), Practice in a Second Language: Perspectives from Applied Linguistics and Cognitive Psychology (pp. 1–18). New York: Cambridge University Press. DeKeyser, R. (2015). Skill acquisition theory. In B. VanPatten, & J. Williams (Eds.), Theories in second Language Acquisition (2nd ed., pp.94–112). New York: Routledge. Eiken Foundation of Japan (2017). Shiken-naiyou, kako-mon [the contents tested, the questions made in the past]. Retrieved June, 2017, from https://www.eiken.or.jp/eiken/exam/. Ellis, R., (2005). Measuring implicit and explicit knowledge of a second language: A psychometric study. Studies in Second Language Acquisition, 27, 141–172. Favreau, M., & Segalowitz, N. (1983). Automatic and controlled processes in the first-and second language reading of fluent bilinguals. Memory and Cognition, 11, 563–574. Fukkink, R., Hulstijin, J., & Simis, A. (2005). Does training in second-language word recognition skills affect reading comprehension? An experimental study. The Modern Language Journal, 89, 54–75. Grabe, W. (2009). Reading in a second Language: Moving from Theory to Practice. New York: Cambridge University Press. Ishikawa, S. (2013). The ICNALE and sophisticated contrastive interlanguage analysis of Asian learners of English. Learner Corpus Studies in Asia and the World, 1, 91–118. Iwahori, Y. (2008). Developing reading fluency: A study of extensive reading in EFL. Reading in a Foreign Language, 20, 70–91. Koda, K. (2004). Insights into second language reading. New York: Cambridge University Press. Krashen, S. (1981). Second Language Acquisition and Second Language Learning. New York: Pergamon Press. Kuhn, M. R., & Stall, S. A. (2003). Fluency: A review of developmental and remedial practices. Journal of Educational Psychology, 95, 3–21. LaBerge, D., & Samuels, S. J. (1974). Toward a theory of automatic information processing in reading. Cognitive Psychology, 6, 293– 323. Logan, G. D. (1997). Automaticity and reading: perspectives from the instance theory of automatization. Reading & Writing Quarterly, 13, 123–146. 159.

(16) Murphy, R., & Smalzer, W. R. (2009). Grammar in use intermediate. New York: Cambridge University Press. Nassaji, H., & Geva, E. (1999). The contribution of phonological and orthographic processing skills to adult ESL reading: Evidence from native speakers of Farsi. Applied Psycholinguistics, 20, 241– 267. Otaki, A., & Shirahata, T. (2017). The role of animacy in the acquisition of ergative verbs by Japanese learners of English. Annual Review of English Language Education, 28, 177– 192. Saito, S. (2005). Nihonjin eigo gakushusha no hitaikakudoshi no shutoku ni tuite [On the Japanese English learners’ acquisition of unaccusative verbs]. Arutesu riberatesu (Iwate university faculty of humanities and social sciences bulletin). 76, 39–50. Samuels, S. J. (1994). Toward a theory of automatic information processing in reading, revisited. In R. Ruddell, M. Ruddell, & H. Singer (Eds.), Theoretical models and processes of reading (4th ed., pp. 816–837). Newark, DE: International Reading Association. Schreiber, P., A. (1980). On the acquisition of reading fluency. Journal of literacy behavior, 12, 177–186. Segalowitz, N. (2000). Automaticity and attentional skill in fluent performance. In H. Riggenbach (Ed.), Perspectives on fluency (pp. 200–19). University of Michigan Press. Shirahata, T. (1988). The learning order of English grammatical morphemes by Japanese high school students. JACET Bulletin, 19, 83–102. Stanovich, K. (1990). Concepts in developmental theories of reading skill: Cognitive resources, automaticity, and modularity. Developmental Review 10, 72–100. Stanovich, K. (2000). Progress in understanding reading: Scientific foundations and new frontiers. New York: Guilford Press. Suzuki, Y., & DeKeyser, R. (2017). The interface of explicit and implicit knowledge in a second language: insights from individual differences in cognitive aptitudes. Language Learning, 67, 747–790. Swan, M. (2009). Practical English usage. New York: Oxford University Press. Taguchi, E., Takayasu-Maass, M., & Gorsuch, G. (2004). Developing reading fluency in EFL: How assisted repeated reading and extensive reading affect fluency development. Reading in a Foreign Language, 16, 70–96. Takebayashi, S., Higashi, N., Ichikawa, Y., & Suwabe, H. (2003). New college EnglishJapanese dictionary 7th edition. Tokyo: Kenkyusha. Yamada, J., & Matsuura. N. (1982). The use of the English article among Japanese students. RELC Journal, 13, 50–63.. 160.

(17)