1. Introduction
English has is now being spoken by a great number of nonnative speakers all over the world. The spread of English raises the issue of mutual intelligibility among English speakers including both native speakers and nonnative speakers. As one method to deal with the issue of intelligibility, numerous studies have investigated the effects of accented English. Munro (1998) explained that foreign-accented speech could be defined as nonpathological speech produced by nonnative learners. Accented English consists of many various elements: segmentation, syllabification, intonation, speech rate or accentedness. In addition, some research broadened the definition of a dialect into the features of pronunciation and the regional background of speakers (Major, Fitzmaurice, Bunta, & Balasubramanian, 2005). The objective characteristics of accented English remain obscurely defined. However, even though the countless features of various types of English are changing sequentially, scaling the strength of accentedness has been devised to classify accented English.
   To measure the accentedness of speakers, the method of interval scaling has been used in numerous studies (Munro & Derwing, 1995a; Major et al, 2005). Even though the raters of accentedness are untrained native speakers, they could measure the accentedness sufficiently (Mackey, Finn, & Ingham, 1997; Bongaerts, van Summeren, Planken, & Schils, 1997). In fact, many experiments adopted native speakers as raters of accentedness, but even if inexperienced and untrained raters could measure the accentedness, it is unclear whether nonnative speakers can rate the accentedness. Some previous research consisted of mixed raters: both nonnative and native speakers (Olson & Samuels, 1973; Elliot, 1995), yet that factors influence the accentedness scaled by nonnative speakers was not clear, and how the accentedness scaled by nonnative raters influenced listening comprehension was ambiguous.
   Accents exist within both native speakers and nonnative speakers, and the diversity of accents sometimes disadvantages the understanding of listeners (Magen, 1998; Schimid & Yeni-Komshian, 1999). Moreover, strong accentedness caused listeners to underestimate the speakers’ intelligence and social status (Brennan & Brennan, 1981). Accent identification depends on the speaker’s role (Bresnahan, Ohashi, Nebashi, Liu, & Shearman, 2002).
   As stated above, speech with strong accentedness is judged negatively. However, should the accent be excluded from speech of nonnative speakers? Even if nonnative speakers learn model English, it is evident that their English pronunciation cannot avoid the influence of other aspects: mother tongue, amount of native language use, age of acquisition, and motivation as has been stated in previous research (Moyer, 1999; Piske, Mackay, & Flege, 2001). In addition, even if learners of English cannot avoid changing their own English and bumping into various types of English, how do listeners familiarize themselves with varieties of English including accented English?
   The familiarity with accented English has been scrutinized in several pieces of research (Gass & Varonis, 1984; Weil, 2001; Derwing, Rossiter, & Munro, 2002; Clarke, 2003; Clarke & Garrette, 2004;). Gass and Varonis (1984) found that topic familiarity was an important factor to understand accented English. However, with regard to training, interaction between tasks and adaptation occurred in several days (Clarke, 2000; Weil, 2001). The long-term training did not influence listeners’ comprehension but positive attitudes toward accented English (Derwing et al, 2002), although the findings suggested that native speakers could adapt to accented English immediately (Clarke, 2003; Clarke & Garrette, 2004). Familiarity possibly occurred in both short and long-term training. However, those experiments have been done by native speakers of English, so that the processing of accented English for nonnative learners of English was not clarified. In addition, given that the proficiency was based on high intelligibility, feasibly caused the positive mutual understanding (Bent & Bradlow, 2003), the nonnative speakers with high proficiency could familiarize themselves with accented English immediately.
   As for the practical contribution of familiarity with accented English, native speakers have encouraged nonnative speakers to decrease the strength of accentedness in pronunciation as judged by native speakers because accented English is unacceptable and avoidable. In addition, the investigation of attitudes for accented English illustrated that Japanese learners prefer the speech of native speakers to accented English spoken by nonnative speakers (Chiba, Matsuura, & Yamamoto, 1995; Matsuda, 2000). To summarize the above, it is necessary that not only the relationship between nonnative and native speakers, but also the issue of mutual intelligibility among nonnative speakers should be scrutinized in the context of English for nonnative speakers. The primary purpose was to investigate whether familiarity with accented English occurred immediately for nonnative speakers in an experiment based on a reaction time measurement. Secondly, if adaptation to accented English did occur within nonnative speakers of English, what factors caused the adaptation to accented English. To scrutinize whether the processing of accent perception for nonnative speakers is based on either language background knowledge or degree of accentedness, the experiment of familiarity with both shared language background and shared accentedness was examined.

2. Experiment 1
The purpose of this experiment was to clarify whether adaptation to accented English could occur in sentence verification tasks (SVTs) and whether nonnative speakers of English could adapt to accented English. Some researchers illustrated that native listeners could rapidly adapt to foreign accented English in a word identification task (Clarke, 2003; Clarke & Garret, 2004). However, it was not clear whether this adaptation could happen for nonnative listeners. In addition, although accented English cost much more processing time than native English in SVT (Munro & Derwing, 1995b), the systematic effects of accented English were not scrutinized. To investigate whether nonnative listeners could familiarize themselves with accented English immediately, this experiment adopted a reaction time measurement because processing efficiency is a sensitive phenomenon.

2.1 Method
2.1.1 Subjects
Thirty-eight undergraduate students and graduate students were recruited from a local university for participation in this experiment. They are all Japanese learners of English and reported no hearing problems. Their English proficiency was based on self-reported TOEIC test scores, and was diverse including intermediate and advanced level.

2.1.2 Design
Experiment 1 used a mixed design with the four groups (accented English: Canadian English, Canadian-Japanese English, Japanese English, and Chinese English) for between-subject factors and four blocks (block: 1, 2, 3, and 4) for within-subject factors. The four blocks of twelve sentences were presented to each subject. Breaks between blocks were prepared if participants would like to take them. Each break depended on subjects’ fatigue. To investigate the adaptation to accented English, sentence verification tasks (SVTs) in this experiment were manipulated. Subjects in each group heard either forty-eight sentences recorded by the Canadian speaker (Canadian English condition), thirty-six sentences recorded by the Canadian speaker, followed by twelve sentences produced by the Japanese speaker (Canadian-Japanese English condition), forty-eight sentences recorded by the Japanese speaker (Japanese English condition), or forty-eight sentences recorded by the Chinese speaker (Chinese English condition).

2.1.3 Speech stimuli
The researcher manipulated sentence verification tasks, true or false judgments, to estimate processing time. The sentence verification tasks (SVTs) consisted of forty-eight sentences, some of which were created by Munro & Derwing (1995b). The researcher in this experiment constructed the rest of the speech stimuli in this present research. Those sentences were fundamentally universal or based on mathematical knowledge (Appendix 1). To produce the speech stimuli, the forty-eight sentences were recorded by the male Canadian speaker, by the female Chinese speaker, and the male Japanese speaker, who were rated in accentedness tasks.
   These speakers’ accentedness was 2.18 (Canadian), 5.70 (Japanese), and 7.70 (Chinese) on the rating scale follows labels: “9 = very strong accent” and “1 = no accent”. Prior to recording, the speakers familiarized with the sentences to remove errors of orthography to phonology translation. In some of the recording tasks in which audible and or visible hesitation lexical mistakes occurred, the researcher asked the speakers to produce the same sentences repeatedly and correctly. The speakers sat in a quiet room and read the list of SVTs at natural rate with the SONY ICD-SX20 recorder, and then it was transferred to a DOS operating system to be converted to .wav files (Channels: Stereo; Frequency = 44000 Hz). Each file was edited visually and aurally on a computer display and the experimenter removed extraneous noise to normalize the speech stimuli. After refining the sentences, the experimenter added silence to the end of the sentences.

2.1.4 Procedure
Subjects sat in a comfortable, quiet room in front of a computer screen. First of all, the researcher explained that this research was to investigate listening ability to relieve psychological stress and anxiety for tests. The sentence verification tasks (SVTs) required subjects to push either the true button if the meaning of sentence was correct, or the false button if the meaning of the sentence was not correct. Secondly, the researcher instructed subjects to respond as accurately and quickly as possible in the SVTs. Before the actual examination, subjects had six practice sentences, produced by a nonnative speaker who was different from the speakers in the experiment 1, to familiarize themselves the methodology of SVTs. After the practice tasks, subjects started the SVTs.
   In this experiment, the participants first saw three fixed crosses that remained visible for the duration before the listening trial. After 1000 milliseconds, listening materials produced by each speaker were presented to subjects. Immediately after the speech sentence then disappeared, the screen was white except for the words “True or False” until subjects pressed either the true or the false button on the computer keyboard. After listeners pushed one of the buttons, the next sentence of SVTs played without any milliseconds. These tasks continued repeatedly until listeners completed the forty-eight sentences in each condition. Figure 1 illustrates the sequence of the procedure in this experiment. According to auditory stimuli, the order of forty-eight sentences was randomized in each block during the SVTs between block 1 to block 4 in all conditions except the Canadian-Japanese condition. In the Canadian-Japanese condition, after block 3 ended, speech stimuli in Canadian-Japanese English condition changed the speech sentences produced by the Canadian English speaker into auditory sentences recorded by the Japanese English speaker. The speech stimuli did not include the same sentences in SVTs. In summary, all meanings of the sentences were different in the Canadian-Japanese condition similar to other conditions.

2.2 Results
2.2.1 Accuracy
For this mixed design with four groups (accent English: Canadian English, Canadian-Japanese English, Japanese English, and Chinese English) for between-subject factors and four blocks (block: 1, 2, 3, and 4) as within-subject factor, a repeated measures Analysis of Variance (ANOVA) was conducted. Figure 1 shows the mean percentages of correct answers for the four groups of accents. No significant interaction between conditions and blocks, F (9, 102) = 1.61, ns, was calculated. However, there were significant main effect of both groups, F (3, 34) = 6.98, p< .01 and blocks, F (3, 102) = 4.76, p< .01.

Figure 1. Mean Percentage of Correct Responses Based on Twelve Stimuli per Block in Sentence Verification Tasks of Experiment 1.
  
   Post-hoc comparisons were performed using Bonferroni adjustment for multiple comparisons. The percentages of correct responses in the Canadian English group (M = 83.1%, SD = 2.4) were higher than the responses in the Chinese English group (M = 69.0%, SD = 2.5), p< .01. In addition, the percentages of correct responses in the Canadian-Japanese English group (M = 82.9%, SD = 2.4) were higher than the responses in the Chinese English group, p< .01. To compare the effect of blocks, the percentages of correct responses in block 4 (M = 82.8%, SD = 1.9) were higher than the responses both in block 1 (M = 72.2%, SD = 2.1) and in block 3 (M = 81.2%, SD = 2.1). In summary, correct answers increased during the task.

2.2.2 Fluency
Reaction time (RT) from correct responses was calculated. RT falling outside the range of ±2 standard deviations from the each subjects’ mean responses was excluded from this analysis in order to avoid other latency and error effects. Figure 2 shows the mean RT for four groups of accents. There were no significant effects for both groups, F (3, 34) = 2.15, ns and block, F (3, 102) = 0.50, ns. In addition, a significant interaction between groups and blocks, F (9, 102) = 3.50, p< .01, was calculated. The significant interaction between groups and blocks was explored using a simple main effects analysis. The simple main effects analysis yielded the significant effect of the groups within block 3, F (3, 34) = 4.20, p < .05. In block 3, RT in the Canadian English group (M = 979.16ms, SD = 229.46) was significantly quicker than RT in the Chinese English group (M = 2025.89ms, SD = 241.87), p < .05. Furthermore in block 3, RT in the Canadian-Japanese English group were significantly quicker (M = 1043.64ms, SD = 229.46) than RT in the Chinese English group, p < .05.
   In addition, the simple main effect of both the block within the Canadian English group, F (3, 32) = 3.44, p < .05, and the blocks within the Canadian-Japanese English group, F(3, 32) = 3.57, p < .05, was explored. In the Canadian English group, RT in block 3 significantly decreased more rapidly (M = 979.16ms, SD = 229.46) than RT in block 1 (M = 1382.87ms, SD = 234.46), p < .05. In the Canadian-Japanese English group, RT in block 4 (M = 1753.83ms, SD =341.48) decreased significantly more than RT in both block 2 (M = 1155.30ms, SD = 261.32), p < .05, and block 3 (M = 1043.64ms, SD = 229.46), p<.05.

Figure 2. Mean Reaction Time Based on Twelve Stimuli per Block in Sentence Verification Tasks in Experiment 1.

2.2.3 Trade-off relationship
In this experiment, additional trade-off effects between accuracy and fluency were considered, so the investigation of the trade-off relationship was analyzed using a standardized baseline measure. To compare both mean correct responses and reaction time in a one dimensional way, z-scores were adopted. There were significant main effects for blocks, F (3, 102) = 6.42, p< .01 and significant interaction between groups and blocks, F (9, 102) = 4.63, p< .01, and there was no significant main effect for groups, F (3, 34) = 0.01, ns. The simple main effects analysis indicated a significant effect of the groups within the block 4, F (3, 34) = 4.51, p < .01. In the block 4, z scores in the Canadian English group (z = 0.65, SD = 0.21) were significantly higher than z scores in the Canadian-Japanese English group (z = -0.413, SD = 0.21), p < .01.
   In addition, the simple main effects of both the block within the Canadian English group, F (3, 32) = 15.55, p < .01, and the block within the Canadian-Japanese English group, F (3, 32) = 4.55, p < .01, and the block within the Japanese English group, F (3, 32) = 3.79, p < .05, were explored. In the Canadian English group, z scores in block 4 were significantly higher (z = 0.65, SD = 0.21) than z scores in block 1 (z = -0.60, SD = 0.21), p < .01, and z scores in the block 2 (z = -0.19, SD = 0.23), p < .01. Furthermore, z scores in block 3 (z = 0.14, SD = 0.23) were significantly higher than z scores in block 1, p < .05. In the Canadian-Japanese English group, z scores in block 4 (z = -0.41, SD = 0.21) were significantly lower than z scores in both block 2 (z = 0.21, SD = 0.23), p < .05, and block 3 (z = 0.46, SD = 0.23), p < .01, and z scores in block 3 were higher than z scores in block 1 (z = -0.27, SD = 0.21), p < .05. In the Japanese English group, there was no significant relationship between the groups and blocks.

2.3 Discussion
In this experiment, the purpose was to clarify whether adaptation to accented English could occur in sentence verification tasks (SVTs) and whether nonnative speakers of English could adapt to accented English immediately. The results showed that Japanese EFL learners of English could adapt to Canadian English during the SVTs, and showing that the processing time of Canadian English improved immediately. These findings were similar with previous research (Clarke, 2003; Clarke & Garette, 2004). In addition, the same result was replicated; the processing of accented English was more time-consuming than native accented English (Clarke, 2003; Clarke & Garette, 2004; Munro & Derwing, 1995b; Schimid & Yeni-Komshian, 1999). Some explanation of faster RT in the Canadian English group was the task specific strategy. However, on comparison the Canadian-Japanese English group with Japanese English group in block 4, such effects did not appear. This means that the subjects simply did not adopt a task-based strategy but a processing strategy for accented English. Furthermore, the percentages of correct responses increased during the tasks, and the scores of relatively stronger accented English groups were lower than other accented groups. Inconsistent with the experimental prediction, the adaptation to stronger accented English such as Japanese accented English and Chinese accented English did not occur in the SVTs. However, as the analysis of the combination of both accuracy and fluency, the processing of Japanese accented English might increase during the tasks. Exhaustively, to scrutinize the data, both that processing time in the Canadian accented group decreased in block 4 and sequential processing in Japanese accented English, it was possibly assumed that a greater amount of access to accented English was required to adapt to accentedness for nonnative speakers. In addition, given the high correct responses achieved in the Japanese accented group continuously from block 1 to block 4, abstract knowledge of accentedness could be structured for nonnative speakers of English.
   Conversely, if a retrieval system of abstract knowledge based on speech specific features improved over tasks, the existing storage of such knowledge could be readily adapted into processing. Otherwise, if both structuring and processing the knowledge simultaneously is operationalized, processing time would require much more time to adapt to new accented English. However, in this experiment, it could not clarify whether accentedness could contribute to the processing of knowledge based on abstract specific features.

3. Experiment 2
In the experiment 1, Japanese EFL learners could quickly adapt to Canadian English except for other accented English, although the percentage of correct responses increased in regard to the sequence of blocks. This result has suggested two possible paths of processing; 1) Listeners could acquire understanding of accentedness on line but not knowledge of features of L1 background. 2) Listeners could structure knowledge transferred from L1 specific features, and accentedness committed to construction of abstract knowledge of accented English, even if they are nonnative learners of English. To scrutinize this hypothesis, experiment 2 was conducted.

3.1 Method
3.1.1 Subjects
The subjects were twenty-eight undergraduate students and graduate students recruited from a local university for participation in experiment 2. . They are all Japanese learners of English and reported no hearing problems. Their English proficiency was based on self-reported TOEIC test scores, and was diverse including pre-advanced and advanced level.

3.1.2 Design
Experiment 2 used a mixed design with the three groups (conditions: Indonesian A-Indonesian B, Indonesian B-Japanese, and Indonesian B-Indonesian A) as between-subject factors and four blocks (block: 1, 2, 3, and 4) as within-subject factors. The four blocks of twelve sentences were presented to subjects. Breaks between blocks were prepared if participants would like to take them. Each break depended on subjects’ fatigue.
   To investigate generalization to accentedness, sentence verification tasks (SVTs) in this experiment were manipulated. In the Indonesian A-Indonesian B condition, the subjects in each group heard thirty-six sentences recorded by the Indonesian A speaker, followed by twelve sentences produced by the Indonesian B speaker. In the Indonesian B-Japanese condition, the subjects in each group heard thirty-six sentences recorded by the Indonesian B speaker, followed by twelve sentences produced by the Japanese speaker. Similarly in the Indonesian B-Indonesian A condition, in the first three blocks, subjects listened to thirty-six sentences produced by the Indonesian B speaker, and the twelve listening stimuli changed into sentences recorded by the Indonesian A speaker in the last block. Table 6 summarized the conditions in the experiment 2.

3.1.3 Speech stimuli
The script of the sentence verification tasks (SVTs) was the same as in experiment 1 (Appendix1). To produce the speech stimuli, forty-eight sentences were recorded by the male Japanese speaker, by the male Indonesian A speaker, and by the male Indonesian B speaker, who were rated in the previous scaling tasks. These speakers’ accentedness was 5.70 (Japanese), and 6.85 (Indonesian A), and 6.25 (Indonesian B) on the rating scale follows labels: “9 = very strong accent” and “1 = no accent”. The recording methodology was the same as experiment 1.

3.1.4 Procedure
Experimental procedures were identical to the previous experiment 1 except for speech stimuli.

3.2 Results
3.2.1 Accuracy
For this mixed design with three groups (conditions: Indonesian A-Indonesian B, Indonesian B-Japanese, and Indonesian B-Indonesian A) as between-subject factors and the four blocks (block: 1, 2, 3, and 4) as the within-subject factors, a repeated measures Analysis of Variance (ANOVA) was conducted. Figure 3 shows the mean percentages of correct responses for the three groups of accents. There were no significant main effects for both conditions F (2, 25) = 2.16, ns and blocks F (3, 75) = 1.15, ns. In addition, the significant interaction between conditions and blocks, F (6, 75) = 4.96, p < .01, was calculated.

Figure 3. Mean Percentage of Correct Responses Based on Twelve Stimuli per Block in Sentence Verification Tasks in Experiment 2.

   The simple main effects analysis yielded a significant effect of the conditions within both block 1, F (2, 25) = 4.12, p < .05 and block 4, F (2, 25) = 15.91, p < .01. In block 1, the percentages of correct responses in the Indonesian A-Indonesian B condition (M = 56.5%, SD = 4.9) were significantly lower than the responses in the Indonesian B-Japanese condition (M = 75.0%, SD = 4.6), p < .05. On the other hand, in block 4, the percentages of correct answers in the Indonesian B-Indonesian A condition (M = 57.4%, SD = 3.6) were significantly lower than the responses in the Indonesian B-Japanese condition (M = 83.3%, SD = 3.4), p < .01 and the Indonesian A-Indonesian B (M = 79.6%, SD = 3.6), p < .01.
   In addition, the simple main effects analysis indicated a significant effect on the blocks within the Indonesian B-Indonesian A condition, F (3, 23) = 3.44, p < .05, and within the Indonesian A-Indonesian B condition, F (3, 23) = 5.20, p < .01. In the Indonesian B-Indonesian A condition, the correct responses in block 4 (M = 57.4%, SD = 3.6) were lower than the responses in block 3 (M = 73.1%, SD = 4.3), p < .05. Furthermore, in Indonesian A-Indonesian B condition, the percentages of correct responses in block 4 (M = 79.6%, SD = 3.6) were significantly higher than in block 1 (M = 56.5%, SD = 4.9), p < .01.

3.2.2 Fluency
The analysis in this experiment was essentially the same as in experiment 1. Figure 4 shows the mean RT for four groups of accents. There were no significant effects on accent conditions, F (2, 25) = 0.86, ns. However, there were significant effects of blocks, F (3, 75) = 7.12, p< .01, and interaction between conditions and blocks, F (6, 75) = 2.97,p<.05.

Figure 4. Mean Reaction Time Based on Twelve Stimuli per Block in Sentence Verification Tasks in Experiment 2.

The simple main effects analysis yielded a significant effect of the block within the Indonesian B -Japanese condition, F (3, 23) = 5.71, p < .01. In the Indonesian B -Japanese condition, RT in block 4 was significantly quicker (M = 892.37ms, SD = 174.55) than RT in block 3 (M = 1509.71ms, SD = 275.11), p < .01. In addition, the simple main effects analysis indicated a significant effect of the block within the Indonesian A-Indonesian B condition, F (3, 23) = 3.14, p < .05. In this condition, RT in block 4 decreased significantly quicker (M = 1277.51ms, SD = 183.99) than RT in block 2 (M = 1752.33ms, SD = 267.52), p < .05.

3.2.3 Trade-off relationship
In the experiment 2 as well as in the experiment 1, additional trade-off effects between accuracy and fluency were analyzed using a standardized baseline measure. To compare both mean correct responses and reaction time in a one dimensional way, z-scores were adopted. There were significant main effects on blocks, F (3, 75) = 4.10, p< .01 and significant interaction between conditions and blocks, F (6, 75) = 7.03, p< .01, was calculated. In addition, there was no significant main effect on conditions, F (2, 25) = 0.01, ns.
   Significant interaction between groups and blocks was explored using a simple main effects analysis. The simple main effects analysis indicated a significant effect on the conditions within block 4, F (2, 25) = 10.91, p < .01. In block 4, z scores in the Indonesian B- Indonesian A condition (z = -0.37, SD = 0.16) were significantly lower than z scores in the Indonesian B-Japanese condition (z = 0.55, SD = 0.16), p < .01, and z scores in the Indonesian A-Indonesian B condition (z = 0.58, SD = 0.17), p < .01.
   In addition, the simple main effects analysis of both the block within the Indonesian B-Japanese condition, F (3, 23) = 8.09, p < .01, and the block within the Indonesian A-Indonesian B condition, F (3, 23) = 6.95, p < .01, were explored.
   In the Indonesian B-Japanese condition, z scores in block 4 (z = 0.55, SD = 0.16) were significantly higher than z scores in block 2 (z = -0.10, SD = 0.23), p < .05, and z scores in the block 3 (z = -0.41, SD = 0.25), p < .01. In the Indonesian A-Indonesian B condition, z scores in block 4 (z = 0.58, SD = 0.16) were significantly higher than z scores in block 1 (z = -0.44, SD = 0.27), p < .01, block 2 (z = -0.04, SD = 0.25), p < .05, and block 3 (z = -0.10, SD = 0.26), p < .01.

3.3 Discussion
In this experiment, the results showed processing time was influenced by accentedness more than knowledge of accented English based on language background. In both the Indonesian A-Indonesian B condition and the Indonesian B-Japanese condition, the processing time decreased in block 4. This finding suggested that the perception of accentedness was more sensitive than the understanding of the accent of specific features. However, in the Indonesian B-Indonesian A condition, there was no influence of accentedness on reaction time. Even if the accentedness in block 4 increased more than the one between block 1 and block 3, processing time was the same as in previous tasks. This result presumably provided the more minute effects of knowledge based on language background on the processing time because the reaction times in block 4 would cost more than from block 1 to block 3 due to the outcome of the Canadian-Japanese condition in experiment 1.
   With regard to the correct responses, accentedness was the most important factor on listening judgment. In block 1, the percentages of correct responses in speech stimuli produced by Indonesian B were easier than the speech sentences recorded by Indonesian A. In addition, the Japanese accented English was easier than Indonesian accented English in block 4, and the correct responses in Indonesian B accented English was better than the scores in Indonesian A accented English. Furthermore, in the Indonesian B-Indonesian A condition, the responses in block 4 were lower than those in block 3. The achievement of SVTs depended on the accentedness. Even if the processing time in block 4 was not different from block 3 in the Indonesian B-Indonesian A condition, the percentages of correct responses in block 4 were lower than in block 3. With the investigation the integration of accuracy and fluency, in Indonesian B-Indonesian A condition, the integrated processing for accented English suggested the stronger effect of accentedness over knowledge based on speakers’L1 background. Those findings suggested generalization to accented English consisted of both accentedness and knowledge based on the language background. However, theses two factors indirectly influence on processing and semantic identifications relatively.

4. General discussion
This research investigated the notion that the adaptation to accented English consists of three dimensions: processing efficiency, accentedness and knowledge based on language background. In addition, the two experiments demonstrated four findings: 1) Adaptation to accented English could occur for nonnative speakers immediately, 2) nonnative speakers of English adapted to both knowledge of transferred L1 language background and accentedness, 3) the adaptation to accentedness was more important to process accented English than knowledge based on features of speech, and 4) the learned abstract knowledge could improve processing of accented English.
   First, for nonnative speakers, the adaptation to low accented English occurred immediately. In experiment 1, nonnative Japanese speakers of English could improve processing sequentially, although Canadian English was not native-like accentedness. However, nonnative listeners did not adapt to English with a stronger degree of accentedness rapidly. Similarly, this result was confirmed in experiment 2. In these two experiments the accentedness of English was more than a middle strength accent, so that processing efficiency required much time to familiarize itself with accentedness. This means that the stronger the accentedness is, the more processing time listeners require (Clarke, 2003; Clarke & Garette, 2004; Munro & Derwing, 1995b; Schimid & Yeni-Komshian, 1999).
   Secondly, adaptation to accented English consisted of both accentedness and knowledge based on language background. Attempting to hypothesize on flexible familiarity with accented English, rapid adaptation to accented English includes familiarity with accentedness and abstract knowledge fundamental to the first language background. In experiment 2, the listeners recognized the accentedness to preserve the features of accent and immediate processing would occur in a processing device. However, if the input has stronger accentedness, much processing time is required to become familiar with accented English. In addition, simultaneously, unconscious features based on a speakers’ mother tongue were scrutinized in the processing device, but this knowledge is not defined completely. In summary, abstract knowledge is similarly processed to become familiar with accented English. The effects of this instable knowledge were small but limited quantity, so while adaptation is progressing, the extent of the effects of knowledge is limitedly affective.
   Inconsistent with the prediction, both accentedness and knowledge of the first language could manipulate the processing of accented English. In experiment 2, the generalization to accented English based on the same language background did not occur. This illustrated that relatively speaking accentedness was the more important factor to adaptation to accented English than abstract knowledge in speech. The findings of Clarke (2003) and Clarke and Garette (2004) showed that generalization to the Spanish accented English including shared features occur immediately for native speakers of English. However, in experiment 2, generalization to Indonesian English did not occur for Japanese speakers of English. With regard to those results, even if two accented Englishes consists of the same specific features, the perception of accentedness is the most important factor to process speech for listeners, and the reception of accentedness is very sensitive for nonnative speakers.
   In incorporating the results of experiment 1 and 2, the small effect of stored knowledge might be considered in processing accented English. The integrated processing of Japanese accented English was systematically similar to that of the Indonesian A-Indonesian B condition in experiment 2. This illustrated that the empirical knowledge of Japanese accented English could be retrieved from the storage system because Indonesian A-Indonesian B condition shared the same condition, relative effect of accentedness, but shared knowledge of accented English in block 4 was different. In addition, because accuracy of responses with Japanese accented English could improve over tasks in experiment 1, seemingly the knowledge could support the processing of accented English less positively.
   Other data elucidated the mechanism of accent perception. Some subjects did not identify what nationality speakers had. Furthermore, several subjects conjectured that the Indonesian B was from an English community and he was a native speaker of English. This meant that it was difficult to identify the nationality of speakers of English from speech stimuli for nonnative speakers, if the listeners had no experience of accented English. In these experiments, even if subjects were exposed to sequential accented speech, the striking knowledge of accented English was accumulated incompletely as stated in retrospection.
   Pedagogically, this research provided one insight into effects of accented English. The present research included two experiments and suggested previous results showed that mental maintenance for accented English required processing time to store vague concepts, and to access the assumable meaning in speech stimuli. Consequently listeners’ listening comprehension deteriorated. However, as suggested in the experiment by Bent & Bradlow (2003), even if speech materials were accented, have English proficiency with high intelligibility for example near-native standard, nonnative speakers could understand such speech stimuli to be similar to speech recorded by native speakers. Furthermore, by storing this feature of accented English learners can adopt the knowledge and generalize it to other accented English. To summarize, expanding the experience of accented English improved nonnative speakers’ knowledge and flexibility of speech perception. Secondly, the results in the two experiments suggested that accented English has an attraction as effective teaching materials. For instance, qualitative data from the questionnaires also showed the positive attitudes toward the varieties of English. Some listeners reported that listening to other accented English was really fun and we should notice varieties of English as real languages in the world. Exposure to accented English could change attitudes positively (Derwing, Rossiter, & Munro, 2002; Matsuura, Chiba, & Fujieda, 1999). From the questionnaires the researcher found that short-term exposure to accented English could motivate subjects to listen to obscure speech even if it was not easy for nonnative speakers to adapt to accented English because the very strong accentedness confused and sometimes disregards the speech.
   Furthermore, to strictly control the influences of accentedness, the researcher provided new methods to scrutinize both the adaptation and generalization. First, English proficiency should be controlled. If the English proficiency with high intelligibility was near-native, nonnative speakers could understand such speech stimuli (Bent & Bradlow, 2003), so presumably the direction of influence of accentedness would be different between high proficiency and low proficiency English language learners. A threshold of adaptation to accentedness possibly exists. Secondly, adaptation and generalization to other accented English can be scrutinized. Scaling accentedness can apply to other types of accented English. The accentedness scaling was verified by much research including this research (Major et al., 2005; Munro & Derwing, 1995a; Piske, MacKay, & Flege, 2001). To comprise and investigate various factors of both adaptation and generalization, an integrated database of accentedness could be useful as well as a bank of sophisticated speech materials.

References

Bent, T. & Bradlow, A. R. (2003). The interlanguage speech intelligibility benefit. Journal of the Acoustical Society of America, 114, 1600-1610.

Bongaerts, T., van Summeren, C., Planken, B., & Schils, E. (1997). Age and ultimate attainment in the pronunciation of a foreign language. SSLA, 19, 447-465.

Brennan, E. M., & Brennan, J. S. (1981). Accent scaling and language attitudes: reactions to Mexican American English speech. Language and Speech, 24, 207-221.

Bresnahan, M. J., Ohashi, R., Nebashi, R., Liu, W. Y., & Shearman, S. M. (2002). Attitudinal and affective response toward accented English. Language and Communication, 22, 171-185.

Chiba, R., Matsuura, H., Yamamoto, A. (1995). Japanese attitudes toward English accents. World Englishes, 14, 77-86.

Clarke, C.M. (2000). Perceptual learning of foreign accented English. Unpublished Master’s thesis, The University of Arizona, Tucson, AZ.

Clarke, C.M. (2003). Processing time effects of short-term exposure to foreign-accented English. Unpublished doctoral thesis, The University of Arizona, Tucson, AZ.

Clarke, C. M., & Garret, M. F. (2004). Rapid adaptation to foreign-accented English. Journal of the Acoustical Society of America, 116, 3647-3658.

Derwing, T. M., Rossiter, M. J., & Munro, M. J. (2002). Teaching native speakers to listen to foreign-accented speech. Journal of Multilingual and Multicultural Development, 23, 245-259.

Elliott, A. R. (1995). Field independence/dependence, hemispheric specialization, and attitude in relation to pronunciation accuracy in Spanish as a foreign language. The Modern Language Journal, 79, 356-371.

Gass, S. & Varonis, E. M. (1984). The effect of familiarity on the comprehensibility of nonnative speech. Language Learning, 34, 65-89.

Mackey, L. S., Finn, P., & Ingham, R. J. (1997). Effects of speech dialect on speech naturalness ratings: a systematic replication of Martin, Haroldoson, and Triden (1984). Journal of Speech, Language, and Hearing Research, 40, 349-360.

Macwhinney, B., Cohen, J., & Provost, J. (1997). The PsyScope experiment-building system. Spatial Vision, 11, 99–101.

Magen, H. S. (1998). The perception of foreign-accented speech. Journal of Phonetics, 26, 381-400.

Major, R. C., Fitzmaurice, S. M., Bunta, F., & Balasubramanian, C. (2005). Testing the effects of regional, ethnic, and international dialects of English on listening comprehension. Language Learning, 55, 37-69.

Matsuda, A. (2000). Japanese Attitudes Toward English: A Case Study of High School Students. Unpublished PhD dissertation, Purdue University.

Matsuura, H., Chiba, R., & Fujieda, M. (1999). Intelligibility and comprehensibility of American and Irish Englishes in Japan. World Englishes, 18, 49-62.

Moyer, A. (1999). Ultimate attainment in L2 phonology; the critical factors of age, motivation, and instruction. SSLA, 21, 81-108.

Munro, M. J. (1998). The effects of noise on the intelligibility of foreign-accented speech. SSLA, 20, 139-154.

Munro, M. J., & Derwing, M. T. (1995a). Foreign accent, comprehensibility, and intelligibility in the speech of second language. Language Learning, 45, 285-310.

Munro, M. J., & Derwing, T. M. (1995b). Processing time, accent, and comprehensibility in the perception of native and foreign-accented speech. Language and Speech, 38, 289-306.

Olson, L. L. & Samuels, S. J. (1973). The relation ship between age and accuracy of foreign language pronunciation. The Journal of Educational Research, 66, 263-268.

Piske, T., MacKay, I. R.A., & Flege, J. E. (2001). Factors affecting degree of foreign accent in an L2: a review. Journal of Phonetics, 29, 191-215.

Schimid, P. M., & Yeni-Komshian, G. H. (1999). The effects of speaker accent and target predictability on perception of mispronunciations. Journal of Speech, Language, and Hearing Research, 42, 56-64.

Weil, S. A. (2001). Foreign Accented Speech: Adaptation and Generalization. Unpublished Master's Thesis. The Ohio State University.

APPENDIX1
Sentence Verification Tasks
Gasoline is an excellent drink.
Elephants are big animals.
Japan is an island.
Spaghetti grows on tall trees.
Hot and cold are opposites.
The sun always sets in the north.
The inside of an egg is blue.
An airplane can fly in the sky.
It always snows in Singapore.
March has thirty-eight days.
Most people wear hats on their feet.
Dolphins swim in the desert.
Exercise is good for your health.
All men can have babies.
All dogs have fifteen legs.
Pork is a kind of vegetable.
Vegetarians eat both fish and beef.
Babies can drink alcohol.
1000 yen is less than 100 yen.
The earth is round.
There are many cities on the moon.
Korea is a country in Asia.
People eat through their noses.
Red and green are colors.
Gold is a metal.
Milk comes from chickens.
A monkey is a kind of bird.
Ships travel on water.
A grandfather is a man.
A grandmother is a man.
An orange is a kind of fruit.
Dinosaurs are an extinct species.
Wine is made from grapes.
Penguins can fly in the sky.
Butter is made from milk.
Germany is a country in Europe.
A hamburger is a kind of metal.
Ants are bigger than crocodiles.
400 is twice as many as 200.
A lemon is a kind of fruit.
Thirty three is smaller than thirty nine.
An uncle is an older woman.
There are many rabbits on the moon.
Christmas is on December 25th.
Wednesday is the first day of the week.
A soccer ball is bigger than a baseball.
A developing country is a poor country.
English is a foreign language in Japan.