Introduction

Within reading comprehension research, much consideration has been devoted to the effect of reader characteristics on discourse processing (e.g., working memory, reading capacity, cognizance, background schemata, and so on.), elements which are critical in delineating the construct of reading. However, less research is conducted on the impact of the features of text. Therefore, to decipher the underlying mechanisms of L2 reading comprehension, various specialists have stressed the need to contemplate the differential contribution of content based attributes, for example, genre, text structure, rhetorical features, and textual markers (Geva 1992; Camiciottoli 2003; Carrell 1985; Marzban and Seifi 2013).

Texts are different inasmuch as they fulfill different functions (Cope and Kalantis 1993). They are many different text types two typical of which are expository and narrative. Expository texts are mainly aimed at informing or describing. Writers who compose expository writings investigate a subject or topic to gain information on it, while the essential target of a narrative is to recount a story. Narrative content consists of a start, center and end, characters, plot or conflict, and setting (Marzban and Seifi 2013). Narration “recounts events in sequence” whereas exposition “explains, classifies, makes clear ideas, terms, or propositions” (Smith 2003, p. 40). Based on a rational model of text type, i.e. a model which uses pre-determined textual features differing between text types and genres, text characteristics are hypothesized to differ between narrative and expository text types, and this can determine whether such factors influence readers’ comprehension of text or not (Landers 2010). The possible characteristics identified include content, word frequency, causality, and structure, or organization.

Besides text type, reading comprehension can also be envisaged from an assessment-oriented perspective. In his discussion on variables affecting reading and reading assessment, Alderson (2000) sorts them as (a) reader variable (background knowledge, inspiration, and language competency), (b) content variable (genre and text-type, lexical load, and subject matter), and (c) task variable (diction and question type). In spite of isolated investigations already carried out on these variables, it is their interactions and interdependencies that provide realistic insight into naturalistic text comprehension (Rapp and van den Broek 2005). The specific aim this study follows is the investigation of the literal and inferential levels of L2 reading comprehension of narrative and expository texts in Iranian EFL learners.

In the context of the present research, i.e. the Iranian EFL setting, systematic English learning and creating L2 reading capacity are constrained to the formal settings of private institutes and university classes because of the learners’ deprivation of real life exposure to authentic language use. Within this context, reading comprehension tests are usually adopted mostly from standard proficiency tests such as TOEFL, i.e. test of english as a foreign language. Keeping this in mind, this study used as instrument a selection of TOEFL reading comprehension narrative and expository passages, taken from ETS TOEFL practice tests, each followed by both literal and inferential multiple-choice items. Moreover, the extraneous variables of language proficiency, question type, and background knowledge relevant to the topic of the text were, to the extent possible, controlled to minimize their potential interfering effect.

Literature Review

Comprehension has been shown to vary between different text types (Best et al. 2008; Diakidoy et al. 2005; Weaver and Bryant 1995; Tun 1989). For example, Weaver and Bryant (1995) assigned participants to read either four expository passages or four narrative ones. The passages were screened on computer and reading was self–paced. After completing the passages, participants answered a series of questions designed to measure their comprehension. Results of analysis of variance test showed that the participants outperformed on narrative texts than on expository texts.

Previous research has generally identified content, word frequency, causality, and structure as the possible characteristics differing between expository and narrative text types (Landers 2010). For example, McCormick (2007) listed some factors that make expository texts difficult to read, namely, text structure, novel propositions, technical vocabulary, readability of the text, abstract concepts, and retention-based tasks assigned to the reader. Content can be defined as the topic and all information contained within a text, and has been assumed to vary between text types. Narrative texts are thought to focus more on topics that may be of general interest. In contrast, expository texts are thought to contain information that may appeal to only a subset of readers (Diakidoy et al. 2005; Gardner 2004). The content of narrative texts has also been described as more familiar than expository texts (Gardner 2004). Most learners find narratives more familiar as they are likely exposed to these texts beginning at a young age (Diakidoy et al. 2005; Hall et al. 2005). More specifically, young readers typically possess background knowledge required for narrative comprehension(Graesser et al. 2003).

The second factor, word frequency, refers to an estimate of how often a word appears in everyday usage. The ability to recognize words in a text is an essential part of the reading process, and word frequency is thought to be a key factor in word recognition (Gough 1984). It has been suggested that full understanding of a text can only occur when a reader is familiar with 95% of the words in the text (Gardner 2004). In light of this, then, expository texts are at a disadvantage as they are hypothesized to contain more low frequency, topic-specific vocabulary than narratives, which are thought to contain a greater number of high frequency words. This makes sense when we consider that most narratives are believed to concentrate on more common topics or authentic scenarios, which would likely use words that are frequently found in everyday life (Gardner 2004). They are often written in a language that is closer to oral language (Snow 2010).

Causality, as the third factor in differentiating text types, refers to the relationship between events in a text. Texts that are highly causal have a clear chain of events, making it easy to determine the relationship between different events or ideas in a text. It has been hypothesized that causality differs between different text types (Landers 2010). Narratives are thought to be highly causal. That is, it is not too difficult to determine what events influenced or caused other events in a narrative text (Zabrucky and Moore 1999). Many narrative texts are based on a goal or a problem that needs to be solved by the main character. Therefore, each event within the story can often be viewed as related to that initial goal or problem. This allows readers to easily create causal links between different events in the text, as it is apparent how one event relates to another within the context of the goal (Bower and Morrow 1990). In expository texts, though, these causal relationships are thought to be more difficult to determine (Zabrucky and Moore 1999). Expository texts have causal connections too, but they are more difficult for a reader to detect. One potential reason for this is because expository texts may present causal relationships that readers do not have enough background knowledge for (Tapiero et al. 2002).

The last factor is text structure, which can be defined as the elements within a text and their organization. An understanding of text structure can help identify the common elements of a text, how these elements go together, and the typical organization of different text types (Rumelhart 1980). More elaborate mental framework of a text is constructed for a reader provided s/he has knowledge of text (Van Dijk and Kintsch 1983). Narrative texts are often described as more predictable than expository texts, potentially due in part to the structural consistencies typically found in them (Landers 2010). This predictability may also be partly accounted by the fact that most people are exposed to narratives from a very early age, starting with fairy tales and story books and moving on to the narrative texts favored in early education, whereas expository texts are typically encountered much later through school years (Diakidoy et al. 2005; Hall et al. 2005). This repeated exposure to narratives may allow readers to develop a general schema for narratives more easily than for expository texts (Voss and Bisanz 1985). In contrast to narratives, expository texts may open with topic paragraphs that introduce the subject of the text through generalities; then proceed to explore the topic in greater depth with specific examples as the text continues (Berman and Katzenberger 2004; Voss and Bisanz 1985). Despite this generalization, however, many researchers have suggested that expository pieces have more varied structures that may change drastically from one text to the next, thus making their structures less predictable (Hall et al. 2005; Voss and Bisanz 1985). Moreover, as Stein and Trabasso (1981) maintain, in terms of structure, expository texts may bear more difficulty in interpretation due to their inclusion of abstract and logical relations.

Expository texts have proved to be more problematic for readers because they necessitate longer time to read and are more difficult to recall and apprehend compared with narrative texts (Saenz and Fuchs 2002; Zabrucky and Moore 1999). Moreover, research has proved that the strength of the relationship between reading comprehension and specific predictors, such as background knowledge and cognitive load, depends on text type. Specifically, decoding has been demonstrated to bear a stronger positive correlation with narrative reading comprehension than with the expository counterpart, whereas background knowledge has exhibited the reverse pattern (Best et al. 2008). Alvermann et al. (1995) found that short-answer items testing retention of text content yielded a higher performance on expository texts than on narratives. According to Roller and Schreiner (1985), however, there is no significant difference between narrative and expository texts regarding performance on a multiple-choice test or on the quality of post-comprehension written summaries.

In a more recent study, Şahin (2013) investigated the effect of narrative and informative text types on reading comprehension levels of 4th and 5th grade students. The results of the study indicated that students understand narrative texts more easily and clearly than informative texts. Furthermore, it was found that there is a significant difference, in favor of female students, between the narrative texts reading comprehension marks of male and female students for all tests. In another research, Marzban and Seifi (2013) examined the effect of text structure instruction on reading comprehension of narrative and expository texts. The results indicated that the instruction of narrative text structure had a noticeable positive effect on the readers’ reading comprehension. However, knowledge of expository text structure did not significantly affect reading comprehension ability of the students.

Research on Levels of Reading Comprehension

The idea that there are different levels of reading comprehension, each requiring the reader to interact with the text in varying degrees, is not unprecedented (Herber 1970; Pettit and Cockriel 1974; Snider 1988; Pearson and Johnson 1978). Indeed, the theory of levels of comprehension underlies many recommendations in language teaching techniques and educational materials development procedures over the last five decades (Carnine et al. 2010; Herber 1970; Lapp and Flood 1983; Vacca et al. 2009). According to this theory, there is a continuum of reading comprehension skills, including literal, inferential, and evaluative levels, supposedly to be respectively ascending in difficulty and depth of processing, in which learners must engage to complete reading comprehension tasks (Herber 1970).

Literal comprehension, or reading on the lines, engages a student in the process of extracting information explicitly stated in a passage (Carnine et al. 2010). This level of understanding depends upon learners’ word-level processing capacity, or their ability to exactly identify individual words and apprehend the meaning created by the combination of words into longer strings including propositions and sentences (Perfetti et al. 2005). In line with this, Goff (2010) proposed that the components of literal comprehension include context, facts and sequence. A mental integration of these three building-block components is vital in attaining literal comprehension of a text: context can be envisaged as the whole mental image created by the interrelations and interdependencies of facts; facts as the core information put across in the reading text; and sequence as the chronological procession of events. According to Jude and Ajayi (2012), literal comprehension involves students’ ability to identify the exact meaning of the vocabulary utilized in the passage (reading for exact meaning at the word/sentence level), read for information (comprehending the gist of the text), as well as their capability to paraphrase or summarize what they understand from the text.

In inferential comprehension, or reading between the lines, on the other hand, readers transcend the literal meaning of the text to understand the implications of the text via knowledge-driven processes such as synthesizing, generalizing, summarizing, and extrapolating; therefore, inference making, by its very nature, involves reasoning beyond the text (Alptekin 2006). Whereas in literal comprehension readers pay attention to explicitly stated information and depend heavily on their linguistic resources, they tend to reduce their reliance on the text in inferential comprehension, and relate textual content mostly to their reasoning and pragmatic competence so as to form a coherent mental representation of the text subject. Hence, it can be said that rich and deep inference making relies less on the text itself and more on other mental processes (Alptekin 2006). At the inferential level, the reader is expected to engage in the process of manipulating information in the text to search for relationships among the main idea and details which is helpful in interpreting and drawing conclusions about the author’s intended meaning (Vacca et al. 2009), fill in deleted details, and/or elaborate upon what s/he has read (Dole et al. 1991).

Research studies have examined the relative degrees of difficulty of literal and inferential items in reading comprehension tests. Snider (1988), for example, conducted a research with “learning disabled” junior high school students to observe potential differences in their performance on items categorized based on the Pearson and Johnson (1978) taxonomy. Based on this taxonomy, questions were categorized as: (a) textually explicit, where the answer was directly stated in the passage (i.e. literal); (b) textually implicit, demanding readers to use information declared in the paragraph to answer a question (i.e. inferential); or (c) scriptally implicit, requiring readers to rely on and integrate their background knowledge with the information presented in the text, i.e. evaluative (Snider 1988). Students’ responses revealed that textually explicit (i.e. literal) questions were the least difficult for students to respond, followed by textually implicit (i.e. inferential) and scriptally implicit (i.e. evaluative) items; textually implicit questions were the most difficult for students to answer.

Davey (1988) investigated the extent to which the location of response information and inference type could possibly account for the observed variance in students’ performance on a standardized reading comprehension test. The location of response information was set as a proxy for item type (i.e. literal, inferential, and evaluative) in that question information and the correct answers were found within the same sentence for textually explicit items, textually implicit items demanded readers to combine information across sentences, and scriptally implicit items required them to integrate information from the text with their own schemata (Davey 1988). Regression analysis results indicated that the location of response information was responsible for approximately 27% of the variance observed in the performance of struggling readers and 12% of the variance for proficient readers.

In another study conducted by Alptekin and Erçetin (2011), the impact of working memory capacity and content familiarity on literal and inferential comprehension in L2 reading among advanced Turkish university students was explored. The participants read either an original American short story or its “nativized” version. Then they responded literal and inferential multiple-choice comprehension items. The results showed that working memory capacity and content familiarity affected inferential understanding both independently (i.e. each factor at a time) and additively (i.e. both factors simultaneously). However, the effects of these two factors on literal comprehension were reported to be completely negligible.

Finally, Jude and Ajayi (2012) conducted a research study whose motivation was to examine students’ attainment in literal comprehension under reading for exact meaning, details, and gist in a text. The students’ performance in reading for exact meaning was higher than that in reading for details and gist.

As elucidated above, most of the studies working on comprehension have not considered different levels of comprehension, like literal and inferential, in their research to examine the difference between students’ performance in more details in different text types. Nor has the research on reading comprehension levels taken text typical characteristics into consideration in an L2 context (Saadatnia et al. 2016).

Purpose of the Study

Due to the inconsistency of the findings from different studies regarding the effect of text type on reading comprehension as well as the lack of elaboration on the concept of comprehension in such research, the authors are interested in examining different levels of comprehension of narrative and expository texts in Iranian EFL learners. Thus, the focus of the present study was on two levels of reading comprehension, namely literal and inferential, in four narrative texts as well as in four expository ones. The following null hypotheses, in line with the research questions of the study, were developed to be tested.

  1. 1.

    There is no significant difference between expository and narrative text type regarding the literal level of EFL reading comprehension.

  2. 2.

    There is no significant difference between expository and narrative text type regarding the inferential level of EFL reading comprehension.

Methods

Participants

The participants of the present study were a group of 180 EFL learners, both male and female, aged between 18 and 25, studying different majors in the University of Isfahan, Iran, selected through the oxford placement test (OPT), i.e. a standardized English proficiency test, to correspond to upper-intermediate level of English proficiency.

Instrumentation

The instruments used in this study were the OPT (2004), used for the selection of the participants, and four narrative and four expository texts, all having similar level of difficulty calculated by the readability formula including three indices of the Coh-Metrix 3.0 readability formula (2012), namely the Coh-Metrix L2 reading index, connectives incidence, and narrativity percentile.

The Coh-Metrix L2 reading index is calculated using three linguistic indices reported by the Coh-Metrix tool including CELEX word frequency (logarithm mean for content words) which is related to decoding, sentence syntax similarity (sentence to sentence adjacent mean) that is related to parsing, and content word overlap (proportional adjacent sentences unweighted) which is related to text cohesion and meaning construction. The index is reported as a percentile score varying from 0 to 100%, with higher scores meaning the text is likely to be more difficult to read than other texts. This index was 13.431, 11.41, 15.173, and 10.577 for the four narrative texts, and 12.613, 11.292, 14.535, and 10.372 for the four expository ones.

Connectives incidence is reported by Coh-Metrix as an incidence score for all connectives as well as different types of connectives. Indices are provided on five general types of connectives: causal (because, so), additive (and, moreover), temporal (first, until), logical (and, or), and adversative/contrastive (although, whereas). In addition, there is a distinction between positive connectives (also, moreover) and negative connectives (however, but). It is believed that the more connectives a text enjoys, the less difficult it is to comprehend (McNamara and Graesser 2011). The index is a cardinal number of all the connectives incidence within the text. Connectives incidence was 117.949, 122.093, 115.928, and 120 for the four narrative texts, and 89.457, 84.507, 80.214, and 88.112 for the four expository ones, respectively.

A narrative text tells a story, with characters, events, places, and things that are familiar to the reader, and is closely affiliated with everyday oral conversation. This robust component is highly affiliated with word familiarity, world knowledge, and oral language (McNamara and Graesser 2011). Narrativity percentile, as a percentile measure, is the index reported as a figure ranging from 0 to 100, with higher percentiles indicating more narrative features present in the text. This index was 54.38, 51.6, 68.08, and 51.6 for the four narrative texts, and 11.9, 7.35, 18.77, and 6.43 for the four expository ones, respectively.

The texts were selected from ETS TOEFL practice tests. Each text was followed by five literal and three inferential multiple-choice items. The literal items were composed of questions on gist, detailed information, and exact meaning of words. It is noteworthy that prior to the actual data collection, the elicitation instrument was validated in a pilot phase through administration to a group of 30 similar Iranian upper-intermediate EFL learners.

Procedure

The OPT test was employed to select the participants according to their proficiency in English. 180 male and female students who had taken the test and attained between 60 and 80score were chosen to participate in the study. They were asked to read the eight texts, do the multiple-choice items on the texts, and answer one last item demanding them to indicate how much topic-related background knowledge they had already had on each text. They had to specify whether they had little, moderate, or much background knowledge on each text. Accordingly, those who ticked much background knowledge were eliminated from the study to minimize the possible interference of this extraneous variable. Subsequently, the performance of the participants on the multiple-choice items were statistically analyzed through paired-samples t test to provide answers to the research questions.

Results

The descriptive and comparative results regarding the research questions of the study are presented below.

Table 1 Descriptive statistics of literal items in expository and narrative texts
Full size table
Table 2 Paired-samples t test in comparison of literal items between expository and narrative texts
Full size table

The Comparison of Expository and Narrative Texts Regarding Literal Comprehension

According to Table 2, the mean difference of correct answers in the expository (M = 11.507, SD = 4.309) and narrative texts (M = 8.771, SD = 3.970; Table 1) was 2.738 with a SD of 3.403. In order to compare literal comprehension in the expository texts with that in the narrative ones, a paired-samples t test was run. The results of the test indicated that there was a significant difference between these two text types (t = 9.513, df = 179, p \(\,<\,\).05). The participants significantly outperformed in the expository texts.

Table 3 Descriptive statistics of inferential items in expository and narrative texts
Full size table
Table 4 Paired-samples t test in comparison of inferential items between expository and narrative texts
Full size table

The Comparison of Expository and Narrative Texts Regarding Inferential Comprehension

Table 4 reveals that the mean difference of correct answers in the expository (M = 5.221, SD = 2.289) and narrative texts (M = 5.386, SD = 2.365; Table 3) was −.164, and the SD was 2.28. To compare expository and narrative texts regarding inferential comprehension, a paired-samples t test was run. The results of the t test revealed no significant difference between the two text types (t = −.852, df = 179, p \(\,>\,\).05).

The Comparison of Literal and Inferential Comprehension in Expository Texts

Due to the difference in the number of literal and inferential items, 10 versus 6, to investigate the issue the percentage of the correct answers at each comprehension level was calculated, and the comparison was made based on these percentages.

Based on Table 6, the mean difference of literal (M = .575, SD = .215) and inferential comprehension (M = .435, SD = .191; Table 5) in the expository texts was .14 bearing a SD of .15. A paired-samples t test was run to compare literal and inferential comprehension in the expository texts. The results of the test demonstrated that there was a significant difference between the two comprehension levels (t = 11.0 55, df = 179, p \(\,>\,\).05). Literal comprehension was significantly superior to inferential comprehension in expository texts.

Table 5 Descriptive statistics of literal and inferential items in expository texts
Full size table
Table 6 Paired-samples t test in comparison of literal and inferential items in expository texts
Full size table
Table 7 Descriptive statistics of literal and inferential items in narrative texts
Full size table
Table 8 Paired-samples t test in comparison of literal and inferential items in narrative texts
Full size table

The Comparison of Literal and Inferential Comprehension in Narrative Texts

According to Table 8, the mean difference of literal (M = .439, SD = .198) and inferential comprehension (M = .449, SD = .197; Table 7) in the narrative texts was −.01 with a SD of .19. The results of the paired-samples t test, run to make a comparison of literal and inferential comprehension in the narrative texts, indicated that there was no significant difference between the two comprehension levels in the narrative texts (t = −.639, df = 179, p \(\,>\,\).05).

Discussion and Conclusion

The study attempted to delve into the relationship between text type and levels of reading comprehension in Iranian upper-intermediate EFL learners. The first research question was to uncover if there is any significant difference in literal comprehension between expository and narrative texts. The data from this study indicated that the participants significantly outperformed on the expository texts at this level of comprehension. This finding may be discussed in light of Goff ’s (2010) stipulation on the concept of literal comprehension: the building-block components of literal comprehension are context, facts and sequence, and to achieve literal comprehension of a text, the reader has to combine these three components. It seems logical that combining these elements should be more facilitated for narratives since context and chronological sequence are typically more readily conceived with narrative discourse.

However, it may be argued that less proficient readers are constrained during the reading process due to their reading deficiency. Although the cues and textual information may help them recognize the employed text structure of the narrative discourse, they are likely to commit errors in making interactions to extract the desired information from the text in order to relate it with their schemata. Consequently, they may not be able to construct their own meaning from the provided resources. In addition to reading deficiency, lack of native-like knowledge in a language is another reason which constrains the less proficient readers to process information which leads to poorer performance as compared to proficient readers. In the context of the present study, the participants were assumed to be upper-intermediate in English proficiency level and this cannot be a true representative of proficient native-like readers, especially in an EFL context, in which, by its very nature, learners are deprived of free real-life opportunities for authentic, day-by-day exposure to native discourse.

The finding of the first research question may also seem objectionable considering the word frequency difference between narrative and expository texts. According to Gardner (2004), expository texts typically contain a greater number of low-frequency, topic-specific vocabulary than narratives, which are thought to contain more high frequency words. The probable reason for this is most narratives are thought to focus around more common topics or real life scenarios, which would likely use words that are frequently found in everyday life (Gardner 2004). This could debilitate expository text literal comprehension. Nevertheless, one may argue that the word frequency advantage of narratives can fully work in a context where the readers have had natural, authentic exposure to everyday language from an early age so that they have had the opportunities to build up a repertoire of high frequency lexical items in their mind. The EFL participants of the present study have lived, grown up, and developed their language learning habits in a situation far away from this.

As a response to the second research question, the study also compared inferential comprehension between expository and narrative texts. The data from this study indicated that, in contrast with literal comprehension, there was no significant difference between the two text types regarding inferential comprehension. This may seem in partial contradiction with Basaraba et al. (2012), who suggested that without a firm knowledge of the lexicon and structures used in the text, i.e. a literal understanding of the text, it is difficult, if not impossible, to infer implicit meanings. This may imply that a strong literal comprehension of a text can potentially contribute to a better inference making of that text. Considering this, one may expect that, in the context of the present research, outperformance on literal items in expository texts should lead to outperformance on inferential items in these texts, too.

However, the aforementioned finding may earn partial justification based on Alptekin (2006), according to whose explanation readers in inferential comprehension, as opposed to the literal counterpart, tend to relate textual content to their reasoning and pragmatic competence to build a coherent mental representation of the text content. In other words, richer and deeper inference may quite possibly result in less dependence on the text itself. This reduction in reliance on linguistic resources neutralizes the literal outperformance of the participants on the expository texts and, conceivably, no meaningful difference is observed between expository and narrative texts in inference making. Put another way, inferential comprehension, envisaged from this perspective, is not as much pertinent to text typical features as may be expected. Moreover, the facility in inference making for narratives, due to an advantageous dependence on a generally more familiar background knowledge, may quite possibly have further counteracted the apparent outperformance of the participants on the literal items in the expository texts.

Finally, the study included a comparison of literal and inferential comprehension in each of both expository and narrative texts. This comparison in the expository texts revealed that, in line with mainstream research, literal comprehension meaningfully outweighed inferential comprehension. The reason for this observation could be, as Perfetti (1999) asserts, while making inferences, the reader constructs a situation model of the text in mind which is more complex than the text base, which is a simple representation of the propositions of the text, or a literal understanding of each word as it relates to those around it in the text. This results in more cognitive load when processing the text inferentially and, hence, inference making becomes more difficult for the reader (Alptekin and Erçetin 2011). This seems more evident for EFL learners, who typically, due to lack of sufficient exposure to authentic language use, have relatively limited linguistic resources and, hence, become cognitively overburdened while processing the text at inferential level.

However, in the narrative texts, there was no significant difference between literal and inferential comprehension. This finding is in contrast with the mainstream research, which posit different degrees of difficulty in items designed to target literal, inferential, and evaluative comprehension, resulting in different comprehension performances by readers on these levels. This apparent conflict could be partly compromised as one, in line with Britton and Pelligrini (1990), takes into consideration the fact that the knowledge about all the actions, goals, events, and emotions of a narrative are deeply embedded in our perceptual and social experience. Furthermore, as elucidated by Graesser et al. (2003), narratives require background knowledge that even young readers typically possess. Aligned with such considerations, one can argue that the life-long earned facility in constructing the situation model of the narrative text, with the help of preexisting world knowledge structures in the reader’s mind, blurs the difference between reading and understanding on the lines and that beyond the lines. In such a context, literal and inferential comprehension bear more resemblance regarding difficulty. In a nutshell, in spite of the generally suggested outperformance of literal over inferential comprehension, in the case of the present study, the typically more familiar background knowledge on narrative texts, may have compensated for the commonly poorer performance on inferential items.

Regarding the findings of the study, a number of implications can be made. First, due to the relative difficulty of inferential comprehension, more work seems necessary to be devoted and spent on inference making skills in language classrooms by teachers. Besides, more exploratory work could be done on the mechanisms and processes of comprehension at different levels on expository and narrative texts so as to provide a more realistic picture of the mental processes readers go through while comprehending a piece of text. Also, from a material development viewpoint, examining text difficulty based on such features as readability or interest could, and should, not be the sole criterion in developing and selecting reading materials for different levels of instruction; rather, other factors such as the type and structure of the text should also be taken into account. In this regard, material developers can include reading tasks that are more focused on text typical features and have activities that deal with sentence relationships, discourse markers, main ideas, and supporting sentences in both text types of narration and exposition without engaging students in finding the meaning of all and every difficult word. Furthermore, most of the reading practice time in class should be allocated to tasks and activities which develop an awareness of text typical characteristics in students and, thereby, help them enhance their reading skill.

As with all other research, the present study suffers from insufficiencies and limitations. The researchers did their best to check and control all the possible extraneous variables; however, parameters such as participants’ learning styles and test taking strategies might have found their way to interfere with the results. Furthermore, due to the descriptive-empirical nature of the study, the findings could become more robust and, hence, generalizable to a wider variety of L2 learners if a more representative sample of English learners could be provided to take part in such research.