Transcription

The Reading Comprehension Strategies of SecondLanguage Learners: A Spanish-English StudyKaren AcostaValdosta State UniversityAbstractThis study identified the reading comprehension strategies that English-speaking collegestudents enrolled in beginner, intermediate, and advanced Spanish language classes ata major Midwestern university in the United States used to comprehend a text in theirsecond language. The findings suggest that readers tended to use the same comprehension strategies when approaching a text in their second language, regardless of theirproficiency level. However, there was a qualitative difference in how these strategieswere used by readers of low, middle, and high proficiency in Spanish. Readers of alllevels used their first language when reading in their second language.Key words: reading comprehension, reading comprehension strategies, second language reading, Spanish language learnersBackgroundResearchers, educators, and foreign language program directors that work withstudents who are learning a second language need to understand what learners dowhen they approach language and literacy tasks in that language. The purpose ofthis study was to identify and describe the reading comprehension strategies usedby college students who were native speakers of English and who were enrolled inbeginning, intermediate, and advanced Spanish foreign language classes at a majorMidwestern university in the United States.Identifying what readers do when they encounter a text in a foreign languageand understanding their thought processes more thoroughly may provide relevantinformation to the development of curriculum and instruction, potentially guidingteacher training and informing curriculum planning decisions. Furthermore, educators who understand what readers of different proficiency levels do—and what theyneed to do in order to be successful in their reading efforts—will be more likely toattend to their students’ specific needs, helping them move toward achieving higherlevels of reading and language proficiency.Literature ReviewVocabularyVocabulary plays a major role in reading comprehension for both L1 and L2readers (Coady, 1997; NRP, 2000; NLP, 2006), since, without vocabulary, reading atext and understanding its meaning are not possible. According to Nation (1990,

58Dimension 20192001), to be successful readers, L2 learners need to know approximately 98% of thewords that are in the materials they read. This means that L2 readers need to haveand use their knowledge of L2 vocabulary in order to function in a second languagesuccessfully. While an essential vocabulary base of 2,000 words (Hinkel, 2006; Hirsh& Nation, 1992; Nation, 1990) is sufficient for daily interaction, that number increases to 5,000 if the goal is to comprehend written texts that are addressed to a generalaudience (Nation, 1990; Hirsh & Nation, 1992).L2 readers may benefit from using cognates by drawing from prior knowledgein their L1 when encountering new words in the second language (Tindall & Nisbet,2010). However, the transfer of cognates requires a certain degree of awareness onthe part of the reader, as not all words that look or sound alike are cognates. Falsecognates may be a source of misunderstanding and confusion for L2 readers. In addition, the transfer of L1 to L2 vocabulary does not occur when the writing systemsof the two languages are different (Birch, 2002; Koda, 1999, 2005; Hinkel, 2006), suchas Chinese and Hebrew.Vocabulary development is aided by extensive reading in the L2 (Coady, 1993;Constantino, Lee, Cho, & Krashen, 1997; Hinkel, 2006; Lervåg & Aukrust, 2010;Nation, 2001; Pitts, White, & Krashen, 1989) because extensive reading offers learners exposure to new and old vocabulary. However, the process is gradual and mayonly become evident after a certain level of L2 proficiency is achieved (Coady, 1993).Consequently, for less proficient language learners, graded or simplified texts withcontrolled vocabulary may be preferable to support decoding (Nation, 2001), butthey would offer fewer opportunities to learn new vocabulary.In a study that looked at depth of L2 vocabulary knowledge, Nassaji (2004)used think aloud protocols to identify the degree and types of strategies used by thereaders to derive word meaning from context. The study found that there was a significant relationship between depth of vocabulary knowledge, strategy use, and success. L2 readers who had stronger vocabulary knowledge utilized certain strategies,such as inferencing, more frequently compared to readers who had weaker vocabulary knowledge, and depth of vocabulary had a significant contribution to success.In addition to predicting the use of strategies and facilitating reading comprehension, vocabulary knowledge has been found to be strongly related to learners’ability to read and acquire new information from texts in both L1 and L2 (Nagy, 1997;Nation, 2001; Parry, 1997; Pulido, 2003; Qian, 1999, 2002; Read, 2000; Wesche &Paribakht, 1999). For example, Pulido (2003) found that vocabulary knowledge wascorrelated with incidental vocabulary gains from reading. Other studies related to L2reading vocabulary found that vocabulary makes a greater contribution to L2 readingcomprehension than grammar (Bossers, 1991; Brisbois, 1995; Taillefer, 1996).Reading ComprehensionReading comprehension is the process through which readers engage a text andextract meaning from it. Tindall and Nisbet (2010) call reading comprehension the“focus of all reading engagement” because readers need to be able to read text fluently,have sufficient prior knowledge and vocabulary, and be able to apply strategies whenreading. Some limitations to L2 reading comprehension include limited vocabularyknowledge, unfamiliar content, and limited knowledge of L2 language structures.

L2 Reading Comprehension Strategies59In addition, cultural and social elements related to language might also be alimitation to reading comprehension because values, experiences, beliefs, and concepts can vary across languages and cultures. Different studies suggest that L2 readers may benefit from working with culturally familiar texts (Johnson, 1981, 1982;Pritchard, 1990; August, 2003) because reading comprehension is enhanced in children and adult readers when they read culturally familiar content.Reading Comprehension StrategiesReading comprehension strategies are “the conscious actions readers use to repair breakdowns in comprehension (cognitive strategies) or the deliberate actionsreaders use to monitor and oversee those attempts at repair (metacognitive strategies)” (McNeil, 2011, p. 885) and they are important to both L1 and L2 reading.L2 reading comprehension is also impacted by L1 reading ability and L2 languageknowledge (Bernhardt & Kamil, 1995; Lee & Schallert, 1997; Perry, 2013; Song, 1998).Research shows that more proficient L2 readers, those with high reading comprehension and/or a high knowledge of the L2, are different from less proficientL2 readers in how they use strategies (Anderson, 1991; Block, 1986, 1992; Ikeda& Takeuchi, 2006; Jimenez, Garcia, & Pearson, 1996; Oxford, Cho, Leung, & Kim,2004; Wang, 2016; Yang, 2006). The differences in strategy use by less proficient L2readers compared to more proficient L2 readers are due to deficits in lexical knowledge, decoding skills, and syntactical knowledge (Alderson, 1984; Clarke, 1979;Koda, 2007; Nassaji, 2007; Wang, 2016). In addition, less proficient L2 readers havefewer resources to apply to higher-level cognitive or metacognitive strategies.Jiménez, García, and Pearson (1996) found that successful bilingual readersunderstood the relationship between the L1 and L2, were aware of the similarity between the languages, and explicitly transferred information or strategies learned inone language to the other language as they thought aloud. They also knew EnglishSpanish cognate relationships and substituted words from their other language whenthey encountered unknown vocabulary. However, less successful L2 readers wereunable to identify strategies to help their comprehension of the text and tended toview their L1 and L2 as two separate, unrelated languages. Perhaps the most compelling finding from the Jiménez et al. (1996) study is, however, that successful L2learners used strategies that were unique to their bilingual status. These findingsindicated that students reprocessed L2 words into their L1 while reading L2 texts.The strategies that L2 learners used were cognate knowledge, information transferbetween languages, and mental translation.Similarly, Upton and Lee-Thompson (2001) explored the way students usedtheir L1 and L2 while they read and found that L2 readers accessed and used theirL1 in the comprehension strategies they employed. They found that mental translation was a common way for adult learners to “reprocess” L2 words into L1 words asthey read a text in their L2. However, the degree to which learners relied on their L1declined as their proficiency in the L2 increased.Research has found that reading strategies can be transferred between languages, and that there is a correlation between reading performance in the L1 andL2, especially for more proficient readers (Bernhardt & Kamil, 1995; Brisbois, 1995;Perales Escudero & Reyes Cruz, 2014; Taillefer, 1996; Yamashita, 2002a). However,

60Dimension 2019without explicit strategy instruction, readers may continue using only those strategies instead of developing new strategies for the L2. This practice may be detrimentalbecause L1 strategies are not always fully successful in helping readers comprehendL2 text (Yamashita, 2002b).Think Aloud ProtocolsThink aloud protocols have been used in language research to identify and studythe ways in which learners notice and process language. L1 reading research has employed think alouds (Bereiter & Bird, 1985; Fox, 2009; Kucan & Beck, 1997; Kuusela& Paul, 2000; Strømsø, Bråten, & Samuelstuen, 2003) to investigate reading strategiesused by young and adult learners to determine differences between the thought processes of less and more successful readers, to provide explicit instruction to improvelearners’ reading skills, and to explore students’ writing processes in their L1.In second language acquisition research, think alouds have been used to gaininsight into the cognitive processes and strategies learners use when they read intheir L2 (Pressley & Afflerbach, 1995; Yang, 2006); to examine the role of mentaltranslation as a strategy that L2 readers use when they encounter a text (Kern, 1994);and to compare the reading strategies that readers use in their L1 and L2 (Davis &Bistodeau, 1993; Jiménez, García, & Pearson, 1996; Upton and Lee-Thompson, 2001;Wang, 2016).Many L2 readers spend much of their time thinking about L2 texts in their firstlanguage. Research that examined other L1s and a range of language proficiencies(Kern, 1994; Lee, 1986a, 1986b; Perry, 2013; Upton, 1997) found that L2 readersuse their L1 as they try to comprehend an L2 text. This may be a way for learners toconfirm their understanding of the text or to store what they comprehend in a moreefficient way. Other studies suggest that this may simply be the readers’ “language ofthought.” Lee (1986a, 1986b) found that college students taking Spanish as a foreignlanguage were able to express their understanding in a more complete way whenthey were allowed to write in their L1. Similarly, Moll (1988) found that the readers’reports in their L1 provided a better picture of their reading comprehension. Thus,allowing readers to think aloud in their L1 when reading in their L2 may result in abetter understanding of the reading process.Limitations of Think AloudsAlthough think aloud protocols have been successfully used to explore different reading processes in L2, there are also limitations to using them as a tool forresearching reading. Block (1986) states that think alouds are most useful when theyprovide information about the learners’ reading processes as they have trouble understanding what they are reading; however, processes that are already automatic orcannot be easily verbalized by learners are more challenging to study. Pressley andAfflerbach (1995) write that fully automatic processes are difficult to self-report because “they occur very quickly, so much so that intermediate products of processingare not heeded in short-term memory and, thus, not available for self-report” (p. 9).Therefore, think alouds are better for studying processes “that have not been automatized, ones that are still under conscious control” (p. 9).Even though researchers have frequently used think alouds to study language

L2 Reading Comprehension Strategies61and reading, their use has been at times controversial. Rossomondo (2007) explainsthat, “concerns have been raised as to the validity of employing think aloud protocolsas a means of data collection because of the possibility that the act of thinking aloudactually adds an additional task that might affect processing” (p. 44).In order to determine whether verbalization affected the subjects’ task performance, several studies have used separate groups, with one group completing thetask silently and the other groups completing the task while doing a think aloud(Bowles & Leow, 2005; Leow & Morgan-Short, 2004; Rossomondo, 2007), and foundno significant difference between the groups, concluding that “thinking aloud is notreactive; that is, thinking aloud did not add an additional attentional burden” (Rossomondo, 2007, p. 60).Ericsson and Simon (1993) found that in groups that were asked to completethe think aloud non-metacognitively; that is, without justifying or hypothesizingabout the process, the subjects’ performance was usually not significantly differentfrom the subjects who completed the same task silently. However, if subjects wereasked to complete the task by thinking aloud metacognitively; that is, providing reasons, hypotheses, or conjectures about the process, their performance was significantly different from the performance of the silent subjects, sometimes underperforming and sometimes outperforming the silent group.Non-metacognitive verbalizations do not seem to have an impact on cognitiveprocesses when compared to silent control groups. Therefore, this type of concurrentverbal protocol appears to be a valid way of exploring learners’ cognitive processes asthey read and complete tasks. Leow and Morgan-Short (2004) recommend that thistype of verbalization be collected because this allows learners to focus on the taskwithout having to look for an explanation as to why they are thinking what they arethinking, instead simply voicing their thoughts as they read.Research QuestionsThe purpose of this study was to identify and describe the reading comprehension strategies used by college students who are native speakers of English and whowere enrolled in beginning, intermediate, and advanced Spanish foreign languagecourses. Specifically, this study addressed the following research questions:1. What are the reading comprehension strategies used by native English languagecollege students who are beginner, intermediate, and advanced learners of Spanish as a second/foreign language when approaching a text in their L2?2. How do these college students use their first language (English) when they encounter reading or comprehension difficulties in a Spanish text?MethodsParticipantsThe study was conducted in the Spanish foreign language program of a majoruniversity in the Midwest United States. The participants of this study were studentswhose first language is English, who were enrolled in intensive beginner courses,upper intermediate courses, and advanced level courses in the Spanish program. Inorder to identify such students, participants filled out a background questionnaire

62Dimension 2019during the data collection session. Fifteen think alouds from each level were selectedfor analysis for a total of 45 think alouds across the three levels of proficiency.Materials, Procedures, and Data CollectionParticipants completed a Spanish placement exam, which was a version ofthe University of Wisconsin System Spanish Language Usage and Reading Exam,modified due to the time constraints of the data collection session. It was possibleto establish the modified test’s own reliability measure and confirm that it was anaccurate measure of Spanish proficiency, with an overall Cronbach’s α coefficient αof 0.835, and individual test items ranging between 0.821 and 0.838. This placementexam served to establish the participants’ level of proficiency in the L2, independently from the class in which they were enrolled and from their self-reported Spanishlevel. The scores also determined which students’ think alouds were to be includedin the data analysis.Participants were provided with instructions in English explaining thinkalouds and their procedure, a sample think aloud transcript, and a warm-up activity before they recorded their own protocol. Students were asked to start readingand thinking aloud non-metacognitively, that is without justifying or hypothesizingabout the process, as they worked through the text passage (Bowles & Leow, 2005;Leow & Morgan-Short’s, 2004; Rossomondo, 2007). The language of verbalizationwas English (Bowles, 2010).An expository text from a world news source in Spanish about a culturally unfamiliar topic was used for the study. To determine their actual familiarity with thetopic, participants completed a familiarity questionnaire during the data collectionsession (Block, 1986; Davis & Bistodeau, 1993). Participants also completed a written recall protocol to assess reading comprehension (Lee, 1986b) without being ableto look back to complete the task, and a background information questionnaire thatfocused on students’ language knowledge, experience, and reading. In order to assessthe participants’ comprehension of the text, participants also completed a multiplechoice comprehension test that addressed (a) low level/in text information, (b) highlevel/go beyond the text information, and (c) vocabulary related questions.In order to participate in the study, students signed up electronically and theirinformation was kept confidential. Data collection sessions lasted 50 minutes andwere conducted in a computer language lab using software that allowed control ofparticipants’ access to the text and their computer screens, as well as the ability tostart and stop their think aloud audio recordings. All materials were presented using software that made it possible to lock the students’ work stations and limit theirInternet access.To ensure confidentiality, each student’s data were identified by a number, thusmaking it impossible to tell which students recorded which think alouds. During thedata collection session, participants wore headsets with microphones, which prevented them from listening to other people’s recordings.Data AnalysisInclusion of Participants in the Data Analysis. In order to determine which participants would be included in the data analysis, three groups of 15 participants eachwere formed. The information gathered during the data collection sessions was or-

L2 Reading Comprehension Strategies63ganized, collapsing the files to make (a) a single file that contained the data of allstudents who completed a session, (b) separate files for each course, and (c) separatefiles for each level; that is, combining the participants who were beginner students butenrolled in different courses. Every participant’s level check test was scored and theirresults were sorted along with enrollment information. Groups were based on theparticipants’ level of proficiency as evidenced by the level check, as opposed to the levels in which students were enrolled. This entailed mixing students enrolled in differentlevels to make groups of participants that scored similarly on the proficiency test.The SPSS Statistics software was used to separate the participants’ scores onthe test into three separate groups. Students who reported a first language other thanEnglish and/or a primary language spoken at home other than English were eliminated from the group. Of the 82 remaining students, participants of each level whoreceived the same or similar scores on the placement test and whose first languagewas English were considered for inclusion in the data analysis. Additional criteriafor making the three 15-participant groups were (a) excluding participants who didnot complete all the tasks, (b) excluding participants whose recordings were difficultto hear/poorly articulated or that suffered technical difficulties, (c) excluding thinkalouds in which the student was often quiet, and (d) when possible including participants whose proficiency based on the level check matched the course in whichthey were enrolled, in order to keep participants who were enrolled in courses trueto their proficiency level together.In order to avoid confusion, an alternate set of labels for the groups in the studywas created based on their level as evidenced by the proficiency test. When discussing groups formed for analysis for the purposes of this study, the labels low-proficiency, middle-proficiency, and high-proficiency are used. When discussing groupsbased on enrollment, the labels beginner, intermediate, and advanced are used.Qualitative Analysis.A strict transcription of the think alouds was done, including participants’pauses, sighs, and yawns. The think aloud transcriptions were coded qualitatively,according to the strategies identified in each paragraph, in order to keep the original context of the participants’ think alouds (LaPelle, 2004). Reading comprehension strategies found by other studies using think alouds (Jiménez et al, 1996; Kamhi-Stein, 2003, Upton and Lee-Thompson, 2001; Wang, 2016; Yang, 2006) withL2 learners served as a guide during the collection and transcription of data andbecame the basis for the qualitative codebook. The following reading comprehension strategies were used in the coding of the think aloud protocols: focusing onvocabulary, summarizing, restating/rereading the text, paraphrasing, using contextclues, decoding, inferencing, questioning, predicting, confirming/disconfirming, integrating information, invoking prior knowledge, monitoring, visualizing, evaluating, noticing novelty, demonstrating awareness, searching for cognates, translating,code-switching, and transferring.The coded transcriptions of the think alouds were used to address the researchquestions. The coded transcriptions provided information about participants’ specific strategies. They also provided information about how participants used the sameor different strategies when they came across difficulties in the text. In addition, the

64Dimension 2019think alouds made it possible to draw connections between different participants ofthe same level who struggled with the same sections of the text similarly. Further, itwas possible to make comparisons of certain strategies used for specific sections ofthe text by participants across levels.During the qualitative coding process it became evident that participantstended to use the same strategies and that some strategies were used more widelythan others. Consequently, rather than addressing all nineteen strategies that wereoriginally described in the codebook, the most commonly observed strategies became the focus of the analysis. In order to determine which strategies were the mostcommonly used, the coded transcriptions were reviewed and counted to determinehow many strategies were used by each of the participants throughout the readingand how many times each strategy was used by each participant. The frequency ofstrategies participants used was determined and compared across proficiency levels,and the qualitative data was then quantified using the data transformation approach(Creswell, 2003).Quantitative Analysis.Following the concurrent model, the qualitative data was quantified. According to Creswell (2003), the data transformation approach involves “creating codesand themes qualitatively, then counting the number of times they occur in the textdata” (p. 221). Creswell (2003) argues that this quantification of qualitative data “enables a researcher to compare quantitative results with the qualitative data” (p. 221).This approach made it possible to identify and describe the reading comprehensionstrategies qualitatively by using the data that emerged from the think alouds, andthen to quantify the frequency of their use.After the qualitative coding process was completed, the SPSS statistical analysissoftware was used to run (a) descriptive tests, (b) analysis of variance (ANOVA),and (c) post hoc tests, such as the Tukey HSD, in order to determine the number ofstrategies used by each participant and the frequency with which each strategy wasused by each participant.FindingsStrategy UseThe following tables show how many strategies, in all, participants used whenreading the Spanish text, and whether there was a difference in frequency of strategyuse between groups and within groups. In addition, the tables shown below providean itemization of which specific strategies were used by participants in each level,and how frequently they used them throughout the reading passage.The descriptive statistics for the number of strategies used are presented belowin Table 1.

L2 Reading Comprehension Strategies65Table 1Descriptive Statistics for Number of Strategies e mean scores for the participants in the low-proficiency group (7.80) werelower than the mean scores for the middle-proficiency (8.40) and high-proficiency(8.67) groups. However, the results indicated that the mean scores for the middleproficiency (8.40) and high-proficiency (8.67) groups were almost identical. Thedescriptive statistics also revealed that the minimum and maximum number ofstrategies used by readers in each group were similar. The mean scores were thensubmitted to a one-way ANOVA, which is presented below in Table 2.Table 2ANOVA for Number of Strategies UsedTaskStrategies 956.290.74942427.333 10.17544433.244The results of the ANOVA revealed no significant difference in the numberof strategies used between the groups [F(2,42) 0.290, p 0.749]. Taken together,these results suggest that the number of strategies that participants used in this studywhen reading a text in Spanish was not significantly different from the readers ofother proficiency levels. The descriptive statistics for the nineteen strategies used byreaders in all groups revealed the most frequently used strategies to be (1) focusingon vocabulary, (2) decoding, (3) monitoring, (4) inferencing, (5) paraphrasing, (6)searching for cognates, and (7) translating (See Appendix A for an itemized viewof the frequency with which each strategy was used by readers across proficiencygroups). The mean scores were then submitted to a one-way ANOVA that revealedno significant difference for the frequency with which each strategy was used by thethree proficiency groups, in most cases (See Appendix B for details).These findings suggest that, in most cases, the frequency with which participants in this study used the reading comprehension strategies was not significantlydifferent from readers of other proficiency levels. The readers in the low-proficiency group decoded much more frequently than the readers in the high-proficiencygroup. Although there was no significant difference found in the frequency of use ofthis strategy between either the low-proficiency and middle-proficiency groups, or

66Dimension 2019between the middle-proficiency and high-proficiency groups, these last two groupsapproached significance in a way that aligned with the expected trend. That is, moreproficient readers decoded words less frequently than less proficient readers, whichmay indicate that readers who had more vocabulary knowledge also read more fluently, thus needing to use the decoding strategy less when reading.Further, these results also suggest that even though participants of all groupsused the searching for cognates strategy, more proficient readers used this strategyless frequently, which may indicate that because more proficient readers had a largervocabulary, they did not need to rely on cognates as often as the less proficient readers.There was a statistically significant difference in how frequently this strategy was usedby both the low-proficiency and high-proficiency groups, and the middle-proficiencyand high-proficiency groups. The frequency with which readers in the low-proficiency and middle-proficiency groups used this strategy was not significantly different.Reading Comprehension StrategiesNineteen observable strategies were coded to analyze the think-aloud transcripts. Briefly, the comprehension strategies were: focusing on vocabulary, summarizing, restating/rereading the text, paraphrasing, using context clues, decoding,inferencing, questioning, predicting, confirming/disconfirming, integrating information, invoking prior knowledge, monitoring, visualizing, evaluating, noticingnovelty, demonstrating awareness, searching for cognates, translating, code-switching, and transferring. The last four strategies required participants to use their firstlanguage, and therefore were defined as bilingual comprehension strategies and willbe discussed in a separate section.Five of the non-bilingual reading comprehension strategies were found to becommon and widely used by participants in all three proficiency groups. These fivestrategies were (a) focusing on vocabulary, (b) decoding, (c) monitoring, (d) inferencing, and (e) paraphrasing. Although the same strategies were commonly used bythe readers in this study, within as well as across groups, there were, at times, qualit

the part of the reader, as not all words that look or sound alike are cognates. False cognates may be a source of misunderstanding and confusion for L2 readers. In ad-dition, the transfer of L1 to L2 vocabulary does not occur when the writing systems of the two languages are diff