The contribution of free reading to students ’ writing proficiency in EFL

Many factors contribute to students’ writing proficiency. This paper intends to find out whether free reading is one of them. A group of Romanian students of English as a foreign language has been tested for their amount of free reading and writing proficiency. Students’ scores on the two tasks have been compared by calculating the Pearson’s product-moment correlation coefficient. The degree of covariance obtained is quite weak, showing no significant relationship between free reading and writing proficiency. However, the coefficients of all the variables observed are higher than zero, which can imply that reading is important in acquiring a (foreign) language. The paper concludes by suggesting possible ways for EFL teachers to stimulate students’ interest in free reading.


Introduction
Writing in a foreign language is a difficult task and requires the simultaneous use of various skills.Teachers try hard to teach their students the art of writing and, unfortunately, they succeed completely in very few cases.This is due to the fact that writing is a very complex process which can be influenced by many factors, such as the piece of writing's purpose and generic structure, and/or students' transfer of techniques from their mother language/first language, their motivation or their knowledge of the topic.Another important factor affecting writing is free reading, i.e., reading for pleasure.

Theoretical framework
Free reading has been the object of study for many research papers conducted in the field of linguistics.The majority have proved that the reading for pleasure improves students' writing performance.It seems that when reading, students have an example of the language of writing and of the techniques writers use to transmit their messages.Thus, they can observe and learn to write by themselves.Moreover, free reading appears to provide additional knowledge of literary language and, as such, develop students' confidence in writing.According to Krashen and Lee (2004, p. 1), this Reading Hypothesis is 'a corollary of the more general Comprehension Hypothesis', hypothesis which postulates that 'we acquire language when we understand messages, when we understand what people tell us and when we understand what we read' (Krashen, 2004a, p. 1).The hypothesis is further related to literacy, since 'our reading ability, our ability to write in an acceptable writing style, our spelling ability, vocabulary knowledge and our ability to handle complex syntax is the result of reading' (Krashen, 2004a, p. 1).In other words, it can be said that those who read more acquire more of the written language.This fact has been proved by several first and second language studies.For example, in one of his studies, Krashen (2004c), while investigating the relation between writing instruction and learning, pointed out that frequent reading leads to an improved writing quality, followed by the discovery of one's own style.In another study (Krashen, 1984), the same linguist concluded that formal instruction of sentence-level rules can help improvement in writing, but is only complementary to receiving comprehensible input through reading.Another linguist, Lee (2005), considered free voluntary reading, together with out-of-school writing practice and students' beliefs in formal instruction on reading and writing, to be facilitative factors affecting writing performance.One of his hypotheses was that students who report doing more free reading would have higher writing performances; hypothesis turned to truth by the results of his study.Lee also showed that 'free voluntary reading was the only significant predictor of writing performance' (p.358).
The findings described previously are consistent with other research studies demonstrating the positive influence of reading in a second language on that second language' writing proficiency.For example, Gradman and Hanania (1991) suggested that two background factors, in this case extensive outside reading and high teacher proficiency in English, have a significant positive effect on TOEFL performance.Elley and Mangubhai (1983) and Elley (1991) considered oral language, reading comprehension and writing to be measures of language use.The findings of the Fiji 'Book Flood' studies showed that an 'enriched diet of regular reading' has a powerful effect on L2 learners (Mangubhai, 2001, p. 147).The free reading students outperformed their traditionally taught peers on tests of reading and listening comprehension, vocabulary, oral language, grammar and writing.Even more, an interesting finding in some of the studies was that the improvement made in writing appeared most dramatically in the tests given 2 years after the beginning of the book flood.
All these studies prove Krashen's (2006, para. 2) belief that 'free reading is the source of our reading prowess and much of our vocabulary and spelling development, as well as our ability to understand sophisticated phrases and write coherent prose'.Nonetheless, the Reading Hypothesis has also some counterpoints.Even if free reading is not done at home, during students' leisure time, but at school, in the form of sustained silent reading, self-selected reading or extensive reading, the difficult issue for teachers is to control the amount and type of reading.Krashen (2004b, para. 20) agrees with the fact that 'light reading is not enough' and that 'in order to master the academic language one needs to read truly challenging texts'.Unfortunately, 'our usual practice in foreign language education is to force students to jump from elementary level courses where they read only simplified texts to very demanding texts ' (para. 20).Having this in mind, he proposes light reading as 'a bridge to heavier reading', since 'it supplies the competence necessary to understand challenging texts' (para.20).Furthermore, it seems that the benefits of extensive reading do not come in the short term (Nation, 1997).This is also consistent with other studies, such as Krashen (2006), which highlights the most negative research result, namely, that some sustained silent reading and comparison groups make the same gains.The studies that show no difference between the two groups are short-term ones, some lasting as little as 8 to 10 weeks.
Finally, another problematic issue that needs to be pointed out is represented by students' motivation to read in a second language.As Prowse (n.d.) claims, students do not read in their own languages; it is even more difficult to make them read in a foreign language.Here is where teachers and parents have an important role.The challenge is 'to get this keyboard obsessed, video-game playing generation to start reading' (para.15) and convince them of its benefits.
Considering all this, the present paper intends to test the Reading Hypothesis.Namely, its aim is to find out whether there is any significant relationship between free reading and writing proficiency, as far as students of English as a foreign language are concerned.For this, it compares students' results on a test measuring free reading and, respectively, their writing performance on a composition.In the end, as a practical implication, the paper intends to provide also some ideas on how to enhance reading motivation in English.

Method
This section presents the participants, methods and procedures selected to fulfil the aim of the paper.

Participants
Foreign language students have been chosen as participants, having in mind the fact that evidence in the foreign language situation is more convincing than evidence in the first or second language situation, since input is available from fewer sources in the foreign language situation (Krashen & Lee, 2004).The target group has 13 members, which are characterised as follows: Romanian, 19year-old, 11 years of studying English as a foreign language and first year students of Translation and Interpretation.

Materials
Since the study aims to compare free reading and writing proficiency, particular materials are required to measure them.For free reading, the example of other research studies (Lee, 2005;Takase, 2007) has been followed and students have been asked to complete a questionnaire in 15 minutes.Lee's (2005) literacy questionnaire, already tested for validity and reliability by different researchers, has been used.This contains items referring to various aspects, such as reading and writing students do at leisure or their attitudes to reading and writing class activities.The focus has been only on the items probing students' involvement with reading, i.e., from 5 to 10.However, students were administered the complete set of questions, in order to distract them from the goal of the study.The questionnaire is included in Appendix A. For clarity reasons, the questions under discussion in this paper have been marked with italics.
As far as the measure of writing proficiency is concerned, participants were required to write a composition on a given topic in 25 minutes.They did not know the topic beforehand.The instructions are mentioned in Appendix B. The narrative composition has been chosen as a measure for students' writing performance because this type of text may be the most frequently encountered throughout their readings.

Procedures
Both the questionnaire and the composition were completed in the classroom during the same class in this particular order.Furthermore, the students were not allowed to speak to each other, to consult any source or to ask the teacher any question.The literacy questionnaire was given in its initial form, i.e., in English.As far as the questions relevant to this study are concerned, the students had to choose one of the following answers: almost always, often, sometimes, occasionally or almost never.The answers have been scored from 5 to 1, as opposite to Lee (2005), who scored them from 1 to 5. His procedure has been changed in order to balance students' scores on the two tasks, so that the one who reads more obtains more points than the one who reads less, much the same as the one who has a better composition scores higher than the one who has a worse composition.
The compositions have been corrected by an English teacher.The teacher did know neither the students nor the purpose of the research.She also could not score the students freely but she has been provided with an essay rating profile (Martin Uriz & Whittaker, 2004).The profile is included in Appendix C and focuses on several aspects of written texts, i.e., content, rhetorical structure, grammatical form, diction and tone and mechanics.For each aspect, she had to choose from excellent, good, satisfactory or unacceptable and justify why.The teacher did not know the scores for each of these aspects.In this way, subjectivity has been controlled as much as possible.

Analysis
The independent variable (X), i.e., the amount of free reading done by students, measured by the literacy questionnaire, and the dependent variable (Y), i.e., students' writing performance, measured by the composition task, have been compared.Namely, the relation between the scores on the questionnaire and those on the compositions have been analysed by calculating the Pearson product-moment correlation coefficient.According to Brown (1988), this correlation coefficient is a common measure of covariance between two variables.If the coefficient is zero, there is a null or no significant relationship between the two variables.If, instead, it is higher or lower than zero, there is a significant relationship between them.In order to have a precise calculation of Pearson's r, Excel Spread sheet has been used (Excel Easy, n.d.).A more liberal alpha level has been chosen and settled at <0.05, with a degree of freedom of n − 2. The study is non-directional.In addition, for a more detailed analysis, the correlation coefficients between the scores on the questionnaire and each aspect of the essay rating profile mentioned above have been calculated.For each Pearson's r calculated, its required assumptions have been checked and met.

Results
The findings related to the literacy questionnaire are presented below.Table 1 contains students' scores on each question (Q), as well as, their final scores.In order to have a more general picture of students' answers, Table 2 points out the central tendency, i.e., the mean (X ), and dispersion, i.e., standard deviation (SD), of the scores for each question.For the calculation of SD, N − 1 has been used, as the sample is quite small.The answers of the targeted students indicate that they read in English for pleasure, the majority often or sometimes and, unfortunately, some almost never.All the students go to the library for outside reading, the majority occasionally and none often or almost always.They all visit bookstores, but the majority almost never, occasionally or sometimes.As far as reading on the Internet is concerned, the majority of the students answered often or almost always.Quite many students do not read English newspapers frequently, but almost never or occasionally.Finally, students usually read English magazines sometimes or occasionally.Thus, having all this in mind, this group of students does not seem to do much free voluntary reading.The only exception is the amount of reading on the Internet.
As far as the writing performance is concerned, Table 3 shows students' scores on the composition.It presents each of the aspect looked at by the rater provided to the correcting teacher (namely, content, rhetorical structure, grammatical form, diction and tone and mechanics) and the final scores.Table 4 summarises the results and presents the descriptive statistics of the present study.It indicates the number of items analysed (N); the mean (X ) and SD; and the lowest and highest scores observed, as well as their range, both for the literacy questionnaire and the composition.Taking into consideration all the results mentioned above, the Pearson product-moment correlation coefficient between students' free reading and writing proficiency has been calculated.With an alpha decision level (α) higher than 0.05 and a degree of freedom (df) equal to 11, Pearson coefficient (r) came up to be equal to 0.02.The same coefficient has also been computed between each aspect of the essay rating profile and free reading, respecting the same values (α > 0.05; df = 11) for the calculations of all Pearson coefficients (r).Table 5 presents the means and standard deviations for each rating aspect and the obtained coefficients.

Discussion
As indicated by the analysis, students' scores on the literacy questionnaire are quite low.This means that they do not do large amounts of free reading, excepting the case of the Internet.In fact, very few students answered almost, always or often; the majority read outside school sometimes or occasionally.However, they do read since very few of them answered almost never.As far as the composition task is concerned, the group seems quite homogeneous or has a similar English level since the majority of the scores range from 5 to 7.Only few students have scored lower than 5 or higher than 7.
The calculation of correlations resulted into quite low Pearson product-moment correlation coefficients.All the coefficients, namely, total r = 0.02, content r = 0.06, rhetorical structure r = 0.11, grammatical form r = 0.01, diction and tone r = 0.12, mechanics r = −0.38,are smaller than the critical r for the present study.According to Fisher and Yates (1963), a two-tailed test with α > 0.05 and df = 11, as the one here, is expected to have a critical value, i.e., 'the value that the researcher might expect to observe simply because of chance' (Brown, 1988, p. 119), equal to 0.55.Considering that the observed value is lower than the critical value, the null hypothesis must be accepted.This means that the data found show there is no significant relationship between free reading and writing proficiency.Nonetheless, it is important to mention that the Pearson product-moment correlation coefficient does not equal to zero, which indicates that a certain relation, even if weak and insignificant, between free reading and writing proficiency does exist.There are even some coefficients that are not as weak as the others, such as rhetorical structure r, diction and tone r and mechanics r.This can reveal that certain aspects, as the ones just mentioned, are influenced by reading.

Conclusion
To sum up, the present study aimed to observe whether there is a significant relationship between the amount of free reading students do and their writing proficiency.For this, it has calculated the Pearson product-moment correlation coefficient between students' scores on a literacy questionnaire and on a composition.The results have indicated that there is no significant relationship between the two variables.However, this conclusion is limited to those aspects of free voluntary reading covered in the questionnaire.
The findings may be influenced by the fact that the students do not report to do large amounts of reading or any heavy reading, fact consistent with the controversy related to the influence of reading on language acquisition (Krashen, 2004b).This is also supported by the fact that the Pearson product-moment correlation coefficients observed are not equal to zero.Thus, correlations exist, but yet very insignificant, indicating a weak relationship between free reading and writing proficiency.
The limitations of the study may also have influenced the results.The sample of participants is quite small and not randomly chosen.They were not involved in a reading programme, as it was the case of several studies mentioned previously (Elley, 1991;Elley & Mangubhai, 1983), implying that their amount and type of reading did at home could not be totally known and controlled.Moreover, self-reports can be problematic, since students usually want to please and, thereby, provide desirable answers.Then, the items related to free reading in the literacy questionnaire proposed by Lee (2005) are quite few.The answers may be interpretable as students can understand whatever they want from words such as often or sometimes.As far as the composition is concerned, a single task may not be enough to reflect students' proficiency and level of writing.
To conclude, despite the results and limitations of the present study, there is a certain relation between free reading and writing proficiency, implying that reading can influence writing up to a certain point.As such, free reading should be emphasised as part of developing students' writing ability, since 'reading anything at all will help all writing, to at least some extent' (Krashen & Lee, 2004, p. 1).Teachers should encourage students to read for pleasure and provide them with interesting reading materials even if this may not guarantee that every child will become a dedicated, highly literate reader (Krashen, 2006).Moreover, when students read for pleasure, they can continue to improve their language 'without classes, without teachers, without study and even without people to converse with' (Krashen, 2004c, p. 147).
In the end, it is necessary to emphasise one of the most important practical implication of the study.The questionnaire item with the most favourable answer was the one related to reading on the Internet.The targeted students, and most probably all students, read more online materials than other types of materials.This practice should be encouraged by teachers and parents.Many studies (Akanda, Hoq & Hasan, 2013;Liu, 2005;Noor, 2011) confirm the necessity to adapt education to technology and make its participants digitally aware.This is true also for reading activities.Teachers should use the Internet to promote reading and enhance students' motivation, as this provides access to all types of materials and can be accessed from a PC, tablet or phone.Students can be part of social networks, blogs or book sharing sites.Teachers can create online groups to share books and opinions.Another idea to encourage free reading can be to organise different events related to reading at school, such as the presentations of books, contests or workshops to create book trailers.