E-texts or p-texts? Evidence from reading comprehension tasks for Polish teenage learners of English

Research investigating differences between reading comprehension of electronic and paper texts has so far provided conflicting evidence. Thus, the present paper aims to examine the comprehension of paper and electronic texts by Polish intermediate learners of English as well as to present their attitudes towards the two types of texts in question. In a pre-test, the participants representing a similar level of this subskill were selected and divided into two groups. In the study proper, the control group read a paper version of the text while the experimental group worked with a text displayed on a computer screen. Although the computer group scored higher than the paper group, these differences were not statistically significant.


Introduction
Due to the unprecedented omnipresence of computers, smartphones or, broadly speaking, hand-held devices, it is common for people to substitute paper books with their electronic counterparts.Consequently, more and more reading that is taking place nowadays is not done in a traditional, that is paper way, but rather via electronic devices (e.g., Myrberg & Wiberg, 2015, p. 49;Walsch, 2016, p. 160).This is why since the arrival of the first personal computers, the differences between reading of paper and electronic texts have been thoroughly explored by reading comprehension specialists.
The present paper introduces a theoretical background to the problem of reading comprehension of e-and p-texts by discussing selected empirical studies devoted to that subject matter.Since the research has accumulated plentiful, yet conflicting evidence, three trends were identified: 1) traditional reading is more successful than digital reading in terms of reading comprehension; 2) digital reading is said to be more effective than its traditional counterpart; 3) the subjects, who read paper and electronic texts, achieve similar scores in reading comprehension tests.Having studied varied data obtained from the research into reading comprehension of e-and p-texts, we designed and conducted the experimental study with an intention of discussing traditional and digital reading comprehension from the perspective of Polish students of English.
Thus, the objectives of the present paper are threefold: (1) to provide a review of the relevant literature in the area of reading comprehension of e-and p-texts, (2) to describe the experimental study undertaken with Polish intermediate students of English and, finally, (3) to draw some conclusions with regard to further research on reading comprehension.

An overview of research on media-related reading comprehension
The first analyses concerning the differences between reading of electronic and paper texts emerged in 1980s.A variety of factors have been considered, including the effect of age (e.g., Meyer et al., 1998;Ball & Hourcade, 2011), attention span (e.g., Schneps, Thomson, Chen, Sonnert, & Pomplun, 2013;Gudinavicius, 2016) and gender (e.g., Sun, Shieh, & Huang, 2013) on reading processes.For the purpose of the present paper, however, we will limit ourselves to an overview of research into the comprehension of texts in traditional and computer-supported reading environments.
The late 1990s research reported consistent results which demonstrated the major advantage of traditional reading over its digital equivalent (e.g., Meyer et 129 al., 1998;Mayes et al., 2001).The studies proved that people were not willing to read on computer screens since such differences between p-books and e-books as fonts, spacing or the act of scrolling up and down instead of page turning, proved to be the reasons for readers' tendency to disregard digital reading.Their general reluctance to read on computer screens was reflected in their lower scores achieved during reading comprehension tests.
Even though the technological development revolutionised the digital world, the results of 1990s research were confirmed by more recent studies (e.g., Jeong, 2010;Ackerman & Lauterman, 2012;Mangen et al., 2013;Chen et al., 2014;Hosseini et al., 2014).The data have not pointed to the superiority of electronic texts over their paper counterparts, refuting the theory that the low quality of computer screens was the reason for the readers' problems with reading e-texts.
Some researchers maintained, however, that thanks to higher quality display systems, readers were provided with the format of texts resembling the ones printed on paper (e.g., Margolin et al., 2013, p. 513).As a result, it was concluded that reading e-texts is either equivalent to reading p-texts (e.g., Dundar & Akcayir, 2012;Margolin et al., 2013;Mojarrad et al., 2013) or even more effective than traditional reading in terms of comprehension (Porion et al., 2016).Moreover, the findings revealed that not only was it the kind of medium used, but also the kind of question posed and the readers' familiarity with the device that influenced the results of comprehension tests.For instance, Chen et al. (2014) conducted an experiment with a group of 90 college students with a view to investigating the differences between reading comprehension tests performed on paper, tablet and computer screens.To check the participants' shallow and deep level 1 comprehension of a text, two sets of questions were given.It was demonstrated that as far as shallow level reading and multiple choice questions were concerned, paper group got a higher score.Nevertheless, it was also argued that the participants' degree of familiarity with the medium used was a significant factor since those subjects with high-level tablet familiarity outperformed other readers while dealing with deep level questions.
The comparison of traditional and digital reading does not only produce conflicting results with respect to reading comprehension tests, but also differing levels of attention span among readers who process a text written in their native versus foreign language.In his pilot study, Gudinavicius (2016) attempted to measure the readers' concentration while working with p-and e-texts in their L1 and FL.He proved that the kind of media used by readers affected their levels 1 Shallow comprehension enables readers to grasp an explicit meaning of the text.It is possible thanks to the surface code and the textbase.Deeper comprehension, on the other hand, can be developed by identifying causes of events or inferring messages of the text which are then related to the readers' background knowledge (Davoudi, 2005, p. 108). of attention span.Reading p-books in L1, for instance, turned out to be the least cognitively demanding activity whereas processing a FL text proved to be, as the scholar defined it, "neutral" (Gudinavicius, 2016, p. 182).On the other hand, the presence of any electronic device required the participants increased levels of attention and, hence, was described as the most engaging.
As already mentioned, no successful conclusion has been reached as to the effectiveness of reading comprehension of paper and electronic texts.By no means exhaustive, Table 1 lists the selected examples of research into the subject matter.There was no statistically significant difference in the reading speed or comprehension between the control and experimental groups.

72
The students who read a printed version of the texts scored better on the reading comprehension test than those who read the electronic text.

The experimental study
Below there is the description of the goals, participants, instruments and procedure adopted in the study.

Goals
Unlike the abundance of studies investigating reading comprehension of p-and e-texts done with L1 students, there is a dearth of available research conducted in EFL instructed settings.Due to the scarcity of empirical research concerning the relationship between reading comprehension assessment and the use of technology in Poland, this study aimed to determine whether the use of computer screens affected the students' overall EFL reading comprehension performance.
Two research questions guided this study: RQ1.Which of the two groups, control or experimental, performed better in a reading comprehension test?
RQ2. Did reading on computer screens have positive, negative or neutral effect on the students' reading comprehension?
RQ3.What is the significance of paper and electronic texts as perceived by the participants of the study?

Participants
The study was conducted at Kazimierz Wielki Secondary School in Lublin, Poland during the 2018/2019 academic year.Forty-two pre-intermediate learners of English participated in the first stage of an experiment.

Instruments
Two methods, quantitative and qualitative, were adopted for the investigation of reading comprehension of e-and p-texts.Quantitative approach was taken to investigate the subjects' reading comprehension skills during the pre-test stage as well as the study proper.For this purpose two reading comprehension tests in English consisting of ten multiple choice questions were selected2 .The difficulty of the texts could be compared to reading comprehension tests approximating B1 level.In order to check the students' shallow and deep understanding of the text, three types of questions were posed: surface, inferential and semantic ones 3 .
The qualitative analysis was used to gain insights into the participants' reading experiences in their L1 and FL, that is Polish and English, respectively.The questionnaire aimed to find out whether the amount of the participants' exposure to a given medium affected their performance in a reading comprehension test.Three closed-item and two open-ended questions were asked to elicit the students' responses.

Procedure
In a pre-test, the participants representing a similar level of reading comprehension were selected.Students were given 15 minutes to read an approximately 800 word The two texts used in the study as well as the comprehension questions can be found in the Appendices below.
E-texts or p-texts?Evidence from reading comprehension tasks...

133
long narrative and choose one correct option out of four possible from a multiple choice list.Those who provided five, six or seven correct answers were to participate in the second stage of an experiment.Such a threshold, 50%, 60% and 70%, allowed to select a group of participants on a similar level of EFL reading competence.
Thirty students achieved the intended score and were divided into two random groups.The control group, or the paper group, read a paper version of the text while the experimental group, or the computer group, worked with electronic texts displayed on a computer screen.The format of the reading comprehension test proper was similar to the one from the pre-test stage.The readers did not have to switch between the text format as the participants from the paper group 4 provided the answers on paper and the readers from computer group 5 underlined their answers in a Microsoft Word file.They were given 15 minutes to take the test.
To supplement the data, a post-study questionnaire was administered to examine the participants' attitudes towards traditional and digital reading in a foreign language.It took the students approximately five minutes to complete the questionnaires.

Results
The paper and computer tests were marked and the results were compared to assess which of the two groups performed better with regard to EFL reading comprehension.The data were analysed by means of the programme STATISTICA.Two types of operations were performed, including descriptive statistics, that is the mean, median and SD, and inferential statistics, the student t-test and the nonparametric Mann-Whitney U test.As shown in Table 2, the mean reading comprehension levels and other statistics were calculated for the two groups.There was a slight difference between the mean and the median in the control group, that is ±1.06.The test results were balanced and the points that the subjects scored ranged from five to nine.In the control group, however, correct answers were more varied.To be more precise, 4 The control group read the paper text The Carpet Fitter which was characterised by the following features: the format, A4; the font type, Times New Roman and the font size, 12.

5
The experimental group read the Microsoft Word version of the paper text The Carpet Fitter on a computer screen, the size 21.higher dispersion of the test results around the mean 7.87 (±2.20) as well as more extreme values, ranging from 3 to 10 points, concerned the group reading a text on a computer screen.The results indicated that the group reading an an article -had higher mean value of reading comprehension.The mean score comparisons did not render any statistical differences between the control and treatment group.Although t-test performed with two independent samples (t = -1.16;df = 28; p = 0.25) was not relevant, the box plots presented in the figure below show that the results in the computer group were slightly higher than the ones from the paper group.Three types of questions were asked to check the students' understanding of the p-and e-texts.They included surface questions (Q1, Q3, Q4, Q6, E-texts or p-texts?Evidence from reading comprehension tasks... 135 Q7), inferential questions (Q2, Q5, Q8) and semantic questions (Q9, Q10).The analysis of the answers provided by the participants from the paper and computer group indicated that the students reading on a computer screen gave more correct answers.
(Q1, Q3, Q4, Q6, Q7), inferential questions (Q2, Q5, Q8) and semantic questions (Q9, Q10).The analysis of the answers provided by the participants from the paper and computer group indicated that the students reading on a computer screen gave more correct answers.As shown in Figure 3 and Table 3, the paper group outperformed the computer group with regard to Q6 and Q10.What proved to be the most problematic question for the subjects reading the text on paper  As shown in Figure 3 and Table 3, the paper group outperformed the computer group with regard to Q6 and Q10.What proved to be the most problematic question for the subjects reading the text on paper was Q8 (33%) and Q9 (20%).At the same time, Q6 (53%) and Q10 (53%) got the lowest number of correct answers in the experimental group.For both groups, Q3 turned out to be the easiest question since all of the students answered it correctly.Based on the table above, it is evident that the participants from the control and experimental groups were more successful with respect to the surface questions (Q1, Q3, Q4, Q6 and Q7).Therefore, one can ask whether they were less difficult for the subjects.The answer is not straightforward.Having assumed the statistical significance of differences at 0.05, one can refute the hypothesis about the difference in reading comprehension results both in the holistic approach (Z=1.88;p=0.06) and between groups -paper group (Z=1.77;p=0.07); computer group (Z=1.15;p=0.25).In order to assess that, the nonparametric Mann-Whitney U test was used.
was Q8 (33%) and Q9 (20%).At the same time, Q6 (53%) and Q10 (53%) got the lowest number of correct answers in the experimental group.For both groups, Q3 turned out to be the easiest question since all of the students answered it correctly.Based on the table above, it is evident that the participants from the control and experimental groups were more successful with respect to the surface questions (Q1, Q3, Q4, Q6 and Q7).Therefore, one can ask whether they were less difficult for the subjects.The answer is not straightforward.Having assumed the statistical significance of differences at 0.05, one can refute the hypothesis about the difference in reading comprehension results both in the holistic approach (Z=1.88;p=0.06) and between groupspaper group (Z=1.77;p=0.07); computer group (Z=1.15;p=0.25).In order to assess that, the nonparametric Mann-Whitney U test was used.E-texts or p-texts?Evidence from reading comprehension tasks...

137
Interestingly, a subtle difference in the difficulty of the questions can be noticed in the figure above, which suggests that further research could potentially prove the accuracy of the statement mentioned above.
The subjects' answers provided to the post-study questionnaire confirmed that more and more young people tended to opt for electronic versions of texts.Question 1 asked the participants to mark on a scale the frequency with which they read paper and electronic texts.Even though no significant differences in the number of reading p-and e-texts on a weekly and monthly basis (or less often) were found, the figure below illustrates a significant advantage of e-texts over their paper counterparts with regard to day-to-day reading.The subjects were also unanimous in the choice of the electronic device that they used while reading e-texts.Smartphones got the largest number of students' responses (n=23), superseding laptops (n=11), tablets (n=3) and e-readers (2).
In Question 2 the students (n=1) from the control group expressed complete indifference towards reading paper texts in English.The remaining subjects either preferred reading electronic texts in English (n=7) or it made no difference to them what kind of a text they dealt with (n=7).As far as the treatment group was concerned, the three options got an even number of points from the participants, that is five each.
In the following question (Question 3), the readers were to state whether they were satisfied with the versions of the text that they had received during the experiment under discussion.The control group was to answer if they were content with a p-text whereas the experimental group was to express their attitudes towards the e-text.The students' answers proved that eight participants from the paper group would prefer an electronic version of a text because, according to them, it was more comfortable to read on the screen.The remaining students (n=7) liked to work with a paper version of the text.They stated that it was comprehensible and legible, enabling them to find easily the key extracts without unnecessary disruptions.On the other hand, the majority of the readers from the experimental group were satisfied with electronic versions of the text (n=11).To justify their choices, they stated that they spent most of their time in front of computers and hence, they were accustomed to working with e-texts.
In Question 4 both groups demonstrated similar preferences towards the reading comprehension practice during English lessons.While ten participants from both the control (n=5) and experimental group (n=5) preferred reading paper texts, 12 students opted for electronic versions of texts (the control group = 6; the treatment group = 6).The remaining subjects (n=8)would prefer to include these two kinds of instruction in their EFL classes.
On being asked (Question 5) whether the medium of a text could affect one's reading comprehension, the participants were in a difference of opinion.While nine readers from the control group were convinced that the type of medium used did have influence on reading comprehension, nine participants from the treatment group stated that there was no connection between the two.

Discussion
The purpose of the present paper was to provide some background information to the experimental study which investigated the impact of traditional, that is paperbased, and electronic, that is computer-displayed, texts on EFL teenage students' reading comprehension.A considerable body of evidence was accumulated which proved that it was e-texts that were associated with a poorer reading comprehension performance as opposed to traditional reading (e.g., Ackerman & Lauterman, 2012, Mangen et al., 2013;Hosseini et al., 2014).Nevertheless, the current study showed that the Polish students performed slightly better with e-texts than p-texts.As opposed to the findings from early 1990s, the present study proved that the experimental group achieved slightly better results than the control group, at the same time supporting the research carried out by Dundar and Akcayir (2012) or Porion et al. (2016).The scores achieved by the group of 30 Polish students of English allowed one to claim that the use of technology had a positive effect on the subjects' reading comprehension.These differences, however, were not statistically significant.
Three limitations to this study need to be acknowledged.The first of them concerned the problem of self-reported data that came from the post-study questionnaire.In spite of the fact that the majority of the answers were carefully marked by the students, there were some instances of open-ended questions being left blank.The second limitation referred to the measure used to collect the data.There were only three types of multiple choice questions asked in a reading comprehension test.These questions were to check the students' shallow and deep understanding of the text.Nevertheless, no open-ended questions in which students were to write their own answers were assigned.The third limitation was connected with the introduction of semantic questions.Since they were to check the students' knowledge of lexis used in the narratives, the provision of a correct answer was not strictly connected only with the comprehension of the text, but also readers' general linguistic competence.
Taking into account the design of the study in question and the answers provided to the questionnaire, we are fully convinced that the research on reading comprehension of paper and electronic texts ought to be further pursued in Polish EFL classroom surroundings.The students' answers showed that, in the majority of cases, whenever they were willing to read a text in English, they opted for its electronic version.More importantly, it was not computers or tablets that they relied on while reading e-texts in English, but smartphones were their most frequent choices.Therefore, it is justified to propose that the future area of investigation could target the analysis of reading comprehension of paper and electronic texts displayed on the screens of smartphones.

Conclusion
The results of our experimental study cannot be generalized.Nevertheless, not only do they give a very interesting account of how the conceptualization of literacy has been changing over the years, but also they represent a gradual cultural shift from a traditional to a more modern, and hence digital view on the ability to read.In our opinion, such inconsistencies which have characterised the last twenty years of research into traditional versus digital reading comprehension seem natural since they reflect humans' increasing familiarity with and reliance on technology which, with the omnipresence of electronic gadgets from their infancy onwards, will probably become even more considerable in the foreseeable future.

Figure 1 .
Figure 1.Histogram of results from paper and computer tests conducted in the control and treatment groups.

Figure 2 .
Figure 2. Box plots for the two groups (p=paper group; c=computer group).

Figure 3 .
Figure 3. Number of correct responses to Q1-10 given by the paper and computer group.

Figure 3 .
Figure 3. Number of correct responses to Q1-10 given by the paper and computer group.

Figure 5 .
Figure 5. Box plots presenting the correct responses provided to the surface questions (s) and inferential/semantic questions (e).

Figure 6 .
Figure 6.Percentage of the students from the control and treatment group reading L1 and FL paper and computer versions of texts every day, once a week, once a month, a few times a year and once a year.

Table 2 .
Means, medians, SD and between-group comparisons of reading comprehension.

Table 3 .
Percentage of correct answers provided by the paper and computer group.

Table 3 .
Percentage of correct answers provided by the paper and computer group.

Table 4 .
Means, medians, SD and between-group comparisons of reading comprehension of surface, semantic and inferential questions.

Table 4 .
Means, medians, SD and between-group comparisons of reading comprehension of surface, semantic and inferential questions.