Cognates and other clues: strategies for comprehension. (Language Teaching & Learning).

Abstract

The act of reading consists of several hierarchal processes that play a key role in the construction of a situation model, that is, a mental representation of a literary text. However, a breakdown in the processing of incoming information can impair this construction and affect overall comprehension. The present study looked at the comprehension strategies of two third-semester French students when confronted with unfamiliar vocabulary. An analysis of the transcripts using a model of reading strategies (de Bot et al.,1997) revealed that the strategies employed can influence comprehension and thus an awareness of these strategies can be used to assist early language readers.

Introduction

The act of reading consists of several processes that occur at many levels. Comprehension is said to be reached by creating inferences through the linking of incoming with previous information. Thus, the processes that take place during this action are believed to affect the construction of the text, known as a situation model (Trabasso, Suh, Payton, & Jain, 1995, p. 219). According to Graesser, Millis & Zwaan (1997) the situation model is the "microworld" represented by the information about people, space, time, and events within the text.

The ability to draw inferences and to construct meaning is attributed to the schema theory. Schemata are used to interpret data, retrieve information, organize it, and generally guide the operation of reading comprehension (Rumelhart, 1980, p. 34). The schemata not only contain knowledge but also information about how this knowledge is to be used. Schema-directed processing is hierarchical in nature, thus the activation of lower-level processes is needed in order to proceed to the higher-level processes (Rumelhart, 1980, p. 42).

However, sometimes a breakdown within the schema activation or at other point in the processing of information can block comprehension. This may occur at the encoding stage, where words are not recognized in the first place, or during lexical processing. Or, this breakdown may occur when the appropriate schemata are not there in the first place or if there is not enough printed information to process properly. Another possible block to comprehension may occur when the appropriate schemata have been activated and processed but not in the manner intended. That is, "the reader may find a consistent interpretation of the text but may not find the one intended by the author" (Rumelhart, 1980, p. 48).

While these breakdowns in the processing of information can happen in one's first language (L1), they are more likely to happen in in a second language (L2). According to the linguistic threshold hypothesis, a minimum level of L2 proficiency must be attained before readers can achieve comprehension when reading a L2 text. Without this minimum level, readers will spend more time on the lower-level processes than on the higher-level ones. A reader who devotes more resources to the lower-level processes will not be able to move on to the higher-level processes necessary for the construction of a situation model. This is supported by Just & Carpenter's capacity hypothesis (1992) which states that lower-level processing will be prioritized over higher-level processing and that this is more likely to occur during L2 comprehension than during L1 comprehension.

Relevant Research

Zwaan & Brown (1996) examined the influence of language proficiency and comprehension skill on situation model construction during narrative comprehension.

Zwaan & Brown had twelve third-semester French students participate in a think-aloud study that compared their comprehension of English and French stories. Scores on these two tasks indicated that the students generated more explanatory inferences in English than in French and constructed stronger situation models in English than in French. In addition, those considered "skilled comprehenders" (based on high verbal SAT scores) generated more explanatory inferences and stronger situation models than those considered less-skilled.

Zwaan & Brown's findings support the theory that lexical access and word-and-sentence level processing are more resource consuming in L2 than in L1 comprehension and that the efficiency of L2 syntactic processing is dependent upon L2 fluency. In addition, Zwaan & Brown's findings support the hypothesis that skilled comprehenders will generate more explanatory inferences and will construct stronger situation models than less-skilled comprehenders. In this manner, language proficiency appears to have a direct bearing on the reader's ability to construct a situation model.

The role of language proficiency was further investigated by Horiba (1996) in a study on comprehension processes in L2 reading. This particular study examined four groups of readers: two groups of native speakers of English learning Japanese (L2-Intermediate and L2-Advanced), a control group (L1-English), and a native group (L1Japanese).

Horiba found that L1 readers devoted their attention to higher-level processes and recalled more events from the text than did less competent readers. L2 readers, however, paid more attention to lower-level processes. This study supports the premise that limited language competence causes slower and more inefficient processing, leading to the construction of an underdeveloped representation of a text.

A Discussion of the Study

Given the role of language proficiency in L2 reading, a think-aloud experiment was conducted at a large, state university with two third-semester French students in order to determine their strategies for comprehension of a L2 text, particularly when confronted with unfamiliar vocabulary. These students, one male and one female, volunteered to participate in the study. Both were in their final semester of their senior year and were of traditional college age. Both had studied two years of French in high school and two semesters in college. They were considered to be of "average" ability as determined by their SAT scores and previous grades; however, they considered themselves to be "below average" in their reading ability in French.

The stimulus was the first chapter from Candide, by Voltaire. This text was selected for several reasons: first, it is a common entry in the canon of French literature and this story is language-appropriate for undergraduate students. Secondly, it is a popular story with colorful characters and action, enhancing the potential for students to remember the story. Also, there are subtle literary elements of sarcasm and implied action that supersede language alone. Finally, the students indicated they had no prior knowledge of the story.

The model used to analyze the strategies used is based on a previous study conducted by de Bot, Paribakht, & Wesche (1997) that examined strategies of lexical processing, which are listed in Table 1. See <http://rapidintellect.com/AEQweb/fallp.htm>.

For the present study, the students participated in the task separately and were instructed to first read the text silently and to underline any words or phrases they did not recognize. After completing this portion of the task, the students were then asked to explain, in general terms, their understanding of the story. Their thoughts were recorded and later transcribed. Here is a portion of what the first student, known here as Mary, reported:

Mary: The only thing that I'm confident with is that it introduces you by describing a house, I know there are some young people, authority figures..... Cunegonde is young, it describes the action that's going on, and the character Pangloss, I'm not sure about him but he seems kind of snotty.... I think I could understaand it a lot better if I knew the words that were underlined.

Mary speculates that familiarity with the vocabulary in question would assist her comprehension of the text. When in turn the second student, known here as Ross, was asked to explain his understanding of the story, he replied:

Ross: We are in Westphalia, at the castle of Thunder-ten-tronckh's house ... I don't know who Candide is, but I know he was chosen...... there's the daughter and she's very beautiful. It looks like she went into her mother's room for a lesson from Pangloss ... she was agitated and pensive ... maybe it's another kind of lesson! (laugh)

Although Ross made no direct mention of vocabulary, he too seems to be trying to articulate his thoughts and comprehension of the text. Both Mary and Ross mentioned what they "thought" might be happening and began to make some progress towards a base of understanding. When asked about their general understanding of the story, both Mary and Ross had focused on character names and descriptions within the text.

Of the two, however, only Ross seemed to grasp that some of the idea units within the text could be interpreted either literally or figuratively. The passage in question read "Docteur Pangloss gave a lesson in experimental physics to a chambermaid while in the woods." While this is the literal representation of the sentence, the figurative meaning is actually "Docteur Pangloss was involved in illicit relations with a chambermaid while in the woods." Ross was able to deduce this because of the words "pensive" and "agitated," which have cognates in French. These words clued him into the fact that there was more than a mere lesson going on. According to Graesser et al. (1997), this hints at a deeper level of discourse analysis. The literal meaning of the text usually corresponds with comprehension at the textbase level, while the figurative meaning of the text usually corresponds with comprehension at the communication level, more closely related to the situation model.

According to de Bot et al., inferencing seems to be the most often used strategy for vocabulary comprehension. The data provided by Mary and Ross seem

to support this viewpoint. Mary seemed to prefer major linguistic sources in her strategies whereas Ross seemed to use mostly extra linguistic sources. Upon further examination of the data supplied by Mary, she employs many of the major linguistic sources cited in the model. For example, she looks at relationships among speech parts in a sentence in the following examples:

Mary: That's a verb because it's after this noun ... I'm not sure what that is [the meaning] without looking it up.

She also looks at orthographic cues to guide her in her attempts to comprehend the vocabulary at hand:

Mary: Let's see ... I know these are verbs and I know these are endings we studied ... I know when there is an "ait" ending it's a verb; like here it's teacher, um, teaching.

Mary not only recognized adjectives from the orthographic cues, but also from their punctuation:

Mary: I look at sentence structure a lot; here where there are commas, clearly these are adjectives.

Mary, more than Ross, seemed to focus on the structure of the passage; she was very aware of paragraph transitions, punctuation, and orthographic cues. Both Mary and Ross used extralinguistic inferencing in their reading of the text. Ross in particular used word knowledge, that is, knowledge of the theme and topic. For example, previously, Ross had noted that Candide was of sound judgment and good character. He used this previous knowledge towards another passage by speculating that the Baron, too, must have similar qualities:

Ross: I think that's an adjective, my guess is that this is honest; he's a gentleman; this means royalty-this man is a good and honest gentleman and of royalty.

Mary and Ross both used information from other parts of the text in guessing the meaning of the target word. For example, Mary said:

Mary: ... I'm not sure what it means, but it must be something positive because it's here with some other positive words ...

Ross, who first thought of Pangloss as a servant, used subsequent information in the text to correct this.

Ross: Well, he's talking about metaphysics and giving lessons, so he must be a teacher or tutor of the house.

Mary was the only one who employed a form of homonymy, that is, knowledge of sound relationships to guess the meaning of an unknown word. She found herself "sounding out" the words in an oral attempt at word retrieval. She explained that it helped her comprehension to "hear" the word in question:

Mary: Um, I can't help but read in my mind; but I can hear it being spoken and I can tell a different tone.

Word association was another strategy used by Mary and Ross. For example, Mary knew the French word "rouge" meaning "red" and applied this to her comprehension of the unfamiliar verb, "rougit:"

Mary: This one looks like rouge, like red, so maybe it has to do with red. Maybe he's getting red.... he's blushing.

Ross also used this strategy in trying to determine the definition of the verb "baisa" meaning "kissed:"

Ross: OK ... something ... um ... he does something to her hand ... hold it? Kiss it? Yeah, kiss it because that [baisa] looks like kiss [bisous].

Ross, for instance, utilized his recognition of the French cognates vivacite, sensibilite, grace, s'enflamment, tremblerent) for comprehension more than Mary:

Ross: ... vivacity, sensibility, grace ... now, you've got mouth ... reflexive; the eyes.... it looks like they're inflaming ... they're trembling.

Note Mary's comments about the same passage:

Mary: Listen, I could tell that, even though I really couldn't describe what was going on I could tell from the paragraphs what was important; I know this one is important and I really want to know what it means because it seems juicy, it seems important.

When questioned about how she could tell the passage was important, she replied:

Mary: You can just tell that something different is happening here because of all the verbs ... the other paragraph had a bunch of adjectives.

Conclusion

While both Ross and Mary employed several of the strategies listed in the table of knowledge sources used in inferencing, their dominant strategies can be easily identified. Mary focused on grammatical structure and Ross recognized cognates. While both were able to comprehend more of the text with these strategies, it was Ross who was more articulate about the story; he was the one who seemed to grasp not only the primary action of the story, but even some of its more figurative connotations. This can be explained by the fact that while grammatical structures are useful in the comprehension of text, they tend to be based on surface meanings. Cognates, on the other hand, carry deeper meaning for the reader because of the ability to quickly infer links between L1 and L2 towards the construction of meaning associated with the representation of the text. According to Kintsch (1988):

In contrast to expectation-based, predictive views of discourse comprehen- sion, a model is developed in which the initial processing is strictly bottom- up. Word meanings are activated, propositions are formed, and inferences and elaborations are produced without regard to the discourse context. However, a network of interrelated items is created in this manner, which can be integrated into a coherent structure through a spreading activation process (163).

As Kintsch (1988) explains, general knowledge about words constrains the construction of discourse representations at all levels. Recognition of certain key cognates from the passage, such as honnete (honest), gentilhomme (gentleman), and caractere (character) lend themselves to certain themes associated with the text, which in turn assists the reader in his/her comprehension of the text as a whole. From here, a coherent situation model is likely to emerge.

As exemplified through the data collected from Mary and Ross, it is evident that reading strategies have an impact on the comprehension of an L2 text. Deliberate instruction on strengthening vocabulary strategies, particularly in the area of cognates, can assist readers who might be limited lexically, syntactically, or semantically. This awareness, and the employment of specific strategies towards comprehension, can enable readers of L2 to construct adequate situation models that will enable them to not only understand, but enjoy, the literature they read.

References

De Bot, K., Paribakht, TS & Wesche, MB (1997). Toward a lexical processing model for the study of second language vocabulary acquisition: evidence from ESL reading. Studies in Second Language Acquistion, 19, 309-329.

Graesser, AC, Millis, K & Zwaan, RA (1997). Discourse comprehension. Annual Review of Psychology, 48, 163-189.

Haberlandt, K. (1994). Methods in reading research. In Handbook of Psycholinguistics ed. MA Gernsbacher (pp. 1-31). NY: Academic Press.

Horiba, Y. (1996). Comprehension processes in L2 reading. Second Language Acquisition: 18 (4), 433-473.

Just, MA and Carpenter, PA (1992). A Capacity Theory of Comprehension: Individual Differences in Working Memory, Psychological Review 99(1), 122-149.

Kintsch, W. (1988). The role of knowledge in discourse comprehension: a construction-integration model. Psychological Review, 95, 163-182.

Rumelhart, D. (1980). Schemata: the building blocks of cognition. In Theoretical Issues in Reading Comprehension: Perspectives from Cognitive Psychology, Linguistics, Artificial Intelligence, and Education. eds. RJ Spiro, BC Bruce, & W. Brewer (pp. 33-58). Hillsdale, NJ: Erlbaum.

Trabasso, T., Suh, S., Payton, P. & Jain, R. (1995). Explanatory inferences and other strategies during comprehension and their effect on recall. In Sources of Coherence in Reading. eds. RF Lorch, Jr. & EJ O'Brien (pp. 219-239). Hillsdale, NJ: Erlbaum.

Voltaire (1970). Candide. Paris: Librairie Larousse. (original work published 1759).

Voltaire (1961). Candide, Zadig, and Selected Stories. trans. DM Frame, Bloomington: IU Press. (original work published 1759).

Zwaan, RA & Brown, CM (1996). The influence of language proficiency and comprehension skill on situation model construction. Discourse Process 21, 2189-327.