Exploring the Transfer Relationship of Summarizing Skills in L 1 and L 2

The purpose of the present study is to examine whether the summarizing skills in the first language (L1) of learners of English as a Foreign Language (EFL) affect their summary performances in a second language (L2). To examine the transferring of L1 summarizing skills to L2 summary performance it is necessary to figure out which L1 knowledge and skills EFL learners already possess. A total of 47 Japanese university students with low intermediate English proficiency were asked to write a summary in their L2 (i.e., English) and L1 (i.e., Japanese) of a text written in each of the same languages after they received a quick lecture on how to write a summary. The relationship between their L1 and L2 summarizing skills was examined by using the scores from their L1 and L2 summary performances. The results showed that a small variance of L1 summarizing skill affected the overall summary performances in L2, which supports the Cummins’s (1976) Linguistics Threshold Hypothesis. This study concludes by offering several suggestions for teachers of summary writing, and implications for future research.


Literature Review
Summarizing a text is a highly complex cognitive skill (Kirkland & Saunders, 1991), and summary production involves a complex interplay of cognitive and metacognitive activities (Hosseinpur, 2015).The production of a summary is assumed to follow three stages, including understanding the text, identifying the main ideas of the text and integrating the main ideas into a text (Kato, 2018).
Though producing a summary is considered a complex cognitive skill, some researchers highlight how the experience of writing summaries influences a writer's summary performance.For example, Johns (1985) examined the difference in quality of English summaries produced by novice first language (L1) summary writers and expert L1 summary writers (e.g., eighth-grade students vs. adults) and Winograd (1984) compared "underprepared" U.S. university students and academically "prepared" U.S. university students.Moreover, researchers argue that the skill of summary writing improves after summary writers receive instructions in summary writing (e.g., Choy & Lee, 2012;Keck, 2006;Wichadee, 2013).However, many students, especially students from China, Japan, and Korea, have different backgrounds of writing education to students from Western countries (e.g., Liebman, 1992;Pennycook, 1996;Rinnert & Kobayashi, 2005;Shi, 2006).As Liebman (1992) mentions, the L2 rhetorical instruction in the Japanese school system tends to focus on grammar, and Japanese high school students even receive little L1 writing training.Moreover, according to Kobayashi andRinnert (2001, 2002), Japanese high school students experience very little writing of any kind, and their knowledge of writing can be called self-taught because they do not receive sufficient L1 composing instruction throughout their academic contexts (Casanove, 1998).Therefore, when students from East Asian countries study in U.S. universities, they require explicit instruction in integrated reading and writing skills (Leki & Carson, 1997).
Researchers (e.g., Johns & Mayes, 1990;Kim, 2009) have suggested another cause for the difficulties with summary writings in L2, the students' L2 skills and writing proficiency.Hirvela (2004) also argued that the limitations of the second language (L2) writers' English language skills lead to difficulties not only with writing in English, but also with reading in English, and that is a very important difference between the first language (L1) and L2 writing instruction contexts.
In studies on second language acquisition, it is considered that L1 linguistic knowledge of L2 learners transfer to the L2 (e.g., Cook, 2000;Gass & Selinker, 2008;Garcia-Mayo, 2009;Goad & White, 2009;Snape, 2008).Several studies examining the relation between L1 and L2 have adopted Cummins's (1979) Linguistic Interdependence Hypothesis, suggesting that the level of L2 competence learners accomplish is significantly influenced by their L1 competence.Focusing on writing studies, Cumming (1989) and Sasaki and Hirose (1996) examined whether L2 writing performance is influenced by L1 writing competences, and found that proficient writers in L1 tend to also be proficient writers in L2.In addition, given the relation between L1 and L2 reading skills, some studies found significant positive correlations between the interlanguage reading skills (Carrell, 1991;López & Greenfield, 2004;van Gelderen et al., 2007), even in the case of the combination of a non-alphabetical language as L1 (e.g., Japanese) and English as L2 (Yamashita, 2002).Cummins (1976) proposed the Linguistic Threshold Hypothesis before he suggested the Linguistic Interdependence Hypothesis (Cummins, 1979), and some researchers (e.g., Alderson, 1989;Bernhardt & Kamil, 1995;Cummins, 1979) were skeptical about the Linguistic Interdependence Hypothesis and they reported that a proficient transfer of L1 ability to L2 could not occur.In other words, L2 learners require a threshold of L2 proficiency which refers to L2 grammar and vocabulary knowledge, and the short-circuit of L1 knowledge transfer only occurs if they do not achieve said L2 proficiency threshold (Alderson, 1984;Carrell, 1998;Clarke, 1979;Cziko, 1980).
The studies on the relation between L1 reading and L2 reading (e.g., Bernhardt & Kamil, 1995;Bossers, 1991;Brisbois, 1992;Carrell, 1991) support the Linguistic Threshold Hypothesis, and these studies showed that L2 proficiency explains more variance in L2 reading than L1 reading does.Also, Ito (2009) investigated the existence of the threshold level in L2 writing, and he tentatively confirmed the existence of the threshold level in L2 writing.
However, in recent years, evidence both for and against the Linguistic Threshold Hypothesis has been shown (e.g., August, 2006;Jiang, 2011;Pae, 2018;Park, 2013).In any case, there are few studies exploring the relationship between the integrated skill of reading and writing in L1 and L2, e.g., summarizing skill in L1 and in L2.

Purpose of the Study
As shown in Figure 1, when summary writers produce a summary in L1 from a source text written in L1, they need to manipulate summarizing skills which are defined as the central executive system, making it possible for them to produce a summary.Summarizing skills imply the summarizing rules, i.e., finding the main ideas from the source text, rearranging the order of the statements logically with examples of integration and connectives, and expressing accurate information from the source text in their own words and structures (Hedgcock & Ferris, 2009;Kintsch & van Dijk, 1978;van Dijk & Kintsch, 1983).On the other hand, when they produce a summary in L2 from a source text written in L2, they need to have the required L2 ability in addition to summarizing skills.

Figure 1. Model of summary production skill in L2
In this paper, Japanese EFL learners' L1 summary performances will be compared with their English summary performances to examine if their L1 summarizing skills transfer to their performances of English summary writing.

Research Question
Previous studies (e.g., Cumming, 1989;Sasaki & Hirose, 1996;Carrell, 1991;López & Greenfield, 2004;van Gelderen et al., 2007;Yamashita, 2002) suggest that L1 proficiency skills have the important role of influencing both L2 reading and writing performances, and predict that L2 summary performance is influenced to no small extent by L1 summarizing skills (i.e., components of summarizing skills in Figure 1).Therefore, the present research addressed the following research question: Do L1 summarizing skills transfer to L2 summary production?

Participants
A total of 82 participants in this study were all enrolled in freshman English courses offered at a university in Japan.Their major was engineering, and they had two English classes a week; one provided by a native English teacher, and the other provided by the present author who is a Japanese English teacher.
According to Nation, (2006, 2011), Sato (2017), and Schmitt and Schmitt (2012), vocabulary size is strongly related to second language proficiency, and to select the appropriate text to be used for producing a summary in English, all the students took a vocabulary size test (Mochizuki, Aizawa & Tono, 2003).The result of the test showed that the students were at the level of 3,529.60 (SD =1,253.13)words out of 7,000.Based on that, it was determined that the students' proficiency in English was lower intermediate.All the students were unfamiliar with writing a summary in both English and Japanese.

English and Japanese Texts
The English text materials were adapted from a section of the reading comprehension of the EIKEN Test, Japan's leading English proficiency assessment (See Appendix A).The readability of the English text to be used for producing a summary in English was measured using Microsoft Word (see Table 1 in detail).Flesch Reading Ease is one of the simple approaches to clarify the grade-level of readers (Flesch, 1949): A score of around 80 reflects a fairly easy level which is equivalent to the 7th grade school level for English native speakers.In linguistic typology, the Japanese language belongs to agglutinative languages, while English belongs to inflectional languages.Hence, I divided the Japanese vocabulary by morpheme and counted the number of morphemes to determine the word count.The Japanese text was translated from the English text by the present author (See Appendix B).I paid close attention to not changing the meaning of words or sentences when translating.As Table 1 shows, the count of characters, paragraphs, and sentences was counted by hand.

Data Collection Procedures
Before starting data collection, all students received a lecture on how to write a summary in English by the author.They were mainly instructed to write a summary which is short but detailed, to allow readers who have not read the source text to understand the main idea, that they do not copy any sentences directly from the source text but paraphrase them, and that they place the main idea of the source text in the first sentence of the summary as a topic sentence.Subsequently, the teacher distributed a worksheet (See Appendix A) to the students, and asked them to write a summary in English in for 30 minutes.They were allowed to look at the source text while writing the summary.In the next class, the teacher distributed a worksheet (See Appendix B) to students, and asked them to write a summary in Japanese in 30 minutes.They were again allowed to look at the source text while writing the summary.

Raters and Rubric
The scoring rubric used for evaluating the summaries in English was adopted from Li (2014) (Appendix C), and it consisted of four components addressing different aspects of summarization: Main idea coverage; Integration; Language use; and Source use.These categories, except for Language use, refer to components of the summarizing skills shown in Figure 1.Can-do lists of each category were established, and scores from zero to five were categorized as shown in Appendix C. All the raters used the same rubric.
A total of 47 English summaries produced by the participants were evaluated by three raters who taught English at a senior high school and a university in Japan.The internal-consistency reliability of raters measured by the Cronbach alpha was as follows: α = 0.90 for Main idea coverage; α = 0.84 for Integration; α = 0.99 for Language use; and α = 0.78 for Source use.Meanwhile, a total of 47 Japanese summaries produced by the same participants were evaluated by three different raters.They were all Japanese native speakers who were elementary school teachers in Japan, and they evaluated the summaries using the same rubric as the English raters.The internal-consistency reliability was as follows: α = 0.94 for Main idea coverage; α = 0.98 for Integration; α = 1.00 for Language use; and α = 0.98 for Source use.

Data Analysis Procedures
All the summaries were analyzed using IBM SPSS statistics version 23.0.For the sake of consistency, a consensus among three raters was set to exclude students who had not written anything or written just one sentence since they were asked to write a summary which is short but detailed.Also, students who just copied whole sentences from the source text were excluded.As a result, 47 students' samples were investigated.
Firstly, to compare the Japanese and English summaries statistically, a t-test was conducted.Secondly, to examine the influence of the Japanese summaries on those in English, a regression analysis was conducted.

Results
The results of the analysis are presented in Table 2.In both languages, students obtained a higher score for Main idea coverage than any of the other categories: Mean = 3.27, SD = 0.77 in Japanese; and Mean = 3.26, SD = 1.17 in English.Writing a summary in their own words while writing in English was hard for them because the average score for Source use was the lowest of the four: Mean = 3.13, SD = 0.65 in Japanese; and Mean = 2.05, SD = 1.27 in English.
To compare the score of each evaluation category for both summaries (i.e., English summary and Japanese summary), a Welch's t-test was conducted.As shown in Table 2, though a significant difference could not be seen between the scores of Main idea coverage in English and Japanese (t(79.72)= 0.03, n.s.), there were significant differences between the scores of other categories: t(61.88)= 3.04, p < .01 for Integration; t(46.00)= 12.70, p < .001for Language use; and t(68.56)= 5.18, p < .001for Source use.Consequently, a significant difference was seen between the total scores for the Japanese summary and the English summary, (t(61.56)= 6.34, p < .01).To examine which category performances (i.e., Main idea coverage, Integration, Language use, and Source use) in the Japanese summary were significantly predicted to influence overall English summary performance, multiple regression analysis was conducted.As shown in  Note.x = Independent variable, y = dependent variable, MI(J) = Main idea coverage in Japanese, Int(J) = Integration in Japanese, LU(J) = Language use in Japanese, SU(J) = Source use in Japanese Though it seemed that the performance in Main idea coverage in L1 influenced the overall performance in English summary, to examine how the performance of each category in L1 influences the performance of each category in L2, a multiple regression analysis was conducted again.Needless to say, summary writers are required to obtain the skill/ability of all four categories in English (i.e., Main idea coverage, Integration, Language use, and Source use) because these four categories are essential components of producing a summary in English.The results in Table 4 show that the performance in Main idea coverage in Japanese is significantly predicted to influence both the performance in Main idea coverage in English (β =.51, p <.05), and the performance of Integration in English (β =.55, p <.05), while Source use in Japanese is significantly predicted to influence the performance in Source use in English (β =.62, p <.01).

Discussion
As shown in Figure 2, three summarizing skills (i.e., Main idea coverage, Integration, and Source use) are required when writing a summary in L1 (e.g., Japanese), while an ability in L2 in addition to the summarizing skills are required when writing a summary in L2 (e.g., English).
It is hypothesized that L1 summarizing skills affect the performances of summarizing skills in L2 (See Figure 2 in detail), and indeed, the results show that parts of the summarizing skills such as finding main ideas (i.e., Main idea coverage) and expressing accurate information from the source text in their own words (i.e., Source use) seemed to influence students' performances in English.However, in fact, their English ability was not good enough to express appropriate grammar and vocabulary, and Language form in English was not influenced by any skills in Japanese as shown in Figure 2, which resulted in only 18.6% of the variance being explained by the influence of L1 summarizing skills on the L2 summary performance.According to a previous study which examined the effectiveness of the provision of L1 clues to produce a summary in L2 (Kato, 2018), even though the EFL summary writers completely understood the English text when helped by L1 clues such as L1 translation and L1 glossary, they needed to possess enough knowledge of English grammar and vocabulary when they produced a summary in English.The same can be deducted from this study.To select a suitable text for the participants in this study, their vocabulary size and readability of the English text were measured, so it was predicted that the selected English text was not too difficult for them to comprehend.
When the raters evaluated the participants' English summaries, they paid attention to their grammatical and lexical errors in terms of Language form, but they focused on the content rather than their grammatical and lexical errors in terms of other evaluation categories.Nevertheless, only 18.6% of their summarizing skills in Japanese affected their summary performances in English, and the result of multiple regression analysis showed that only the score of Main idea coverage which referred to the skill of reading comprehension was significantly predicted to have an influence on the performance of L2 summary, and the scores of other categories (i.e., Integration, Source use, and Language use) were not significantly predicted to have an influence.
In other words, summary writers are required to possess sufficient proficiency in English exposing the mental representation which they build in their mind while reading the text, and they especially need knowledge of English grammar and vocabulary as an absolute condition.If they had possessed higher proficiency of L2, their English summary would have been more sophisticated in line with their Japanese summaries.Accordingly, as Cummins (1976) proposed, the linguistic threshold is expected to exist in the case of producing a summary as well.
Considering the findings from this present study and applying them to a future related study, the participants' L2 proficiency such as vocabulary and grammar needs to be measured to examine if the threshold exists in the case of summary writing.Also, qualitative protocols such as think-aloud, immediate recall, and interview should be adopted to identify the process and strategies EFL summary writers employ to produce a summary in L1 and L2.
Furthermore, some educational implications were raised through the findings of the present study.First, the teachers teaching summary writing to non-proficient students or students who are unfamiliar with summary writing should instruct English grammar and vocabulary separately from the instruction of summary writing in English.Also, as a co-operational instruction with their first language education (i.e., Japanese), both language teachers teaching Japanese and English should create more opportunities for EFL learners to produce summaries in each language.

Conclusion
The purpose of the present study was to investigate the influence of L1 summarizing skills consisting of the skills referring to the three components in the rubric (i.e., Main idea coverage, Integration, and Source use) on summary performance in L2, and an attempt was made to clarify the complex skills by exploring the role of summarizing skills in L1 in producing a summary in L2.Though the present author used the participants' vocabulary size and the readability of the English text to select suitable materials for producing a summary, the participants could not express the mental representation in their own words using appropriate English grammar and vocabulary.In fact, their performances in Japanese summary writing were quite sophisticated, but the role of L1 summarizing skills was not as large as that of English proficiency.
The finding from this study is that L2 proficiency may play a great role in producing a summary in L2, but it is predicted that the role of L2 proficiency may be different depending on the L2 proficiency of summary writers.For example, L2 proficiency is vital for students with poor L2 proficiency, but it may not be for proficient L2 learners.Considering the influence of L1 summarizing skills, as Cummins (1976) proposed, sufficient L2 proficiency causes the successful transferring of L1 summarizing skills to L2 summary performances.Therefore, a threshold level in L2 proficiency to transfer summarizing skills from L1 to L2 is expected to exist there much like in the case of reading and writing.
Also, to understand the complexity of summary writing, future studies need to be carefully planned to observe the process that the learner uses and the strategies they employ to produce a summary, in order to deepen our understanding of the nature of summary production and to offer many useful suggestions for EFL teachers to better teach summary writing.

Figure 2 .
Figure 2. Transferring the summarizing sub-skills from L1 to L2 Note.MI = Main idea coverage, Int = Integration, SU = Source use, LU = Language use

Table 1 .
Readability and features of the text Note.N.A. = Not applicable.

Table 2 .
Comparison of the participants' Japanese summaries and English summaries

Table 3 .
Table 3, the results indicate that three predictors explained only 18.6% of the variance (R 2 = 0.19, F(3, 46) = 4.51, p <.01), and among them, only Main idea coverage (β = .52,p <.05) is significantly predicted to have an influence.Result of multiple regression analysis of overall summary performance in English

Table 4 .
Result of Multiple Regression Analysis of Each Performance in English