Literacy Across Cultures
September, 1997 1/2

Guessing Word Meaning from Context: Should We Encourage It?

David Dycus Aichi Shukutoku University

Introduction

Of all the reading strategies commonly recognized today in both L1 and L2 reading, arguably the most widely studied and encouraged is the guessing of the meaning of unknown words from context (hereafter referred to as the "guessing strategy"). It has a long history of research relative to L1 reading in English (Johnson and Bauman, 1984, cite studies on it from the 1940's, for example), with the great majority of studies demonstrating its value. Justification for applying it to L2 reading has come from cognitive science models of reading and schema theory, which are now widely accepted in ESL/EFL circles (see Jannuzi, this issue, for a discussion of schema theory and reading). This is especially true of models that emphasize top-down processing, with Goodman's (1967) famous characterization of "reading as a psycholinguistic guessing game" as probably the most influential.

The numerous studies which indicate that the strategy is effective provide validation for a strategy that is in itself intuitively appealing and appears to offer many advantages over laborious, time-consuming, methodical instruction in vocabulary and collocation. Another claim in support of the guessing strategy is that it involves generalizable skills of interpreting surrounding text, predicting, and testing predictions while reading, which enhance reading skills as a whole (Coady and Nation, 1988; Liu and Nation, 1985). In addition, guessing has been advocated instead of dictionary use because stopping to use a dictionary interrupts the flow of reading (Brown, 1972).

However, there is also a growing body of research that brings into question the value of encouraging the guessing strategy with L2 learners which, in turn, has implications for L2 reading instruction and for the psycholinguistic approach to reading. Since the guessing strategy is so widely encouraged, it is important to take evidence against it into consideration. Therefore, this essay centers on the question: Should we encourage adult L2 readers to use the guessing strategy?

Justification for the guessing strategy

The fact that the guessing strategy is often encouraged is not surprising considering the enormous number of words in the English language, the size of the average adult's working vocabulary, and the number of words one needs to know to recognize a reasonably high percentage of words on the average written page. Webster's Third New International Dictionary, for example, contains 460,000 words, and this number does not include plural forms of nouns, different present and past tenses of verbs, neologisms, and some technical terms (Denning and Leben, 1995, p. 3). Of course, the average person's actual vocabulary (both passive and active) is much smaller, but still considerable. Although estimates of the size of the working vocabulary of the average English-speaker vary widely, commonly accepted figures hover around 20,000 words (Nation, 1990, p. 11). Word frequency counts indicate that this number is more than sufficient for understanding the vocabulary of most non-technical texts, although estimates again vary. According to one, for example, the 25 most common words account for one-third of the words on a page; 135 words takes one up to 50%. After that, the number of words needed increases in lognormal distribution. So, while it takes 2500 words to cover 78% of the page, vocabulary size has to be doubled to 5000 to reach 86%, and doubled again to 10,000 to cover 92% of the text. One would need to know another 200,000 to cover the low frequency words that make up the remaining 8% (Diller 1978). However, Nation's (1990, p. 16) claim that the 2000 most frequently occurring words account for 87% of the average text, and that 2800 will account for 95%, is widely accepted today.

Regardless of the exact size of a native speaker's vocabulary, it is clear that the average second or foreign language learner faces a major challenge in trying to match it. Therefore, it is not surprising that the main reason given for encouraging use of the guessing strategy is the perception that it is the only reasonable way for L2 learners to learn enough words to form suitably large active and passive vocabularies.

Support also comes from experimental word recognition studies with L1 subjects, which have consistently shown that context plays a role in the identification of words in text (Gough, 1984; Underhill and Batt, 1996). Studies of context effects have established, among other things, that words are recognized better in context than out of context, and that simple word association enhances word recognition. For example, experiments show that lexical decision latency for a word is significantly reduced if it is preceded by a semantically related word (such as the word "wife" being displayed, then followed by "husband"). Appropriate sentential context has also been shown to improve the speed of lexical decision. Such results and their implications have been used to support the use of the guessing strategy for L2 readers.

Yes, but is it reading?

According to Gough (1984) we need to be cautious in interpreting the results of such experimental studies. While acknowledging that semantic priming (word association, etc.) is an established context effect that supports a top-down view of processing, he points out that semantically related words rarely occur together in such a way in authentic texts, casting doubt on the applicability of these findings to true reading. He also grants that the influence of sentential context on word identification and the guess- ability of word meanings is also firmly established, which implies that reading is a top-down process because, if it were bottom-up, words would have to be recognized before contextual factors could come in to play. However, while context has an effect on word recognition, Gough points out that the effect is not at all constant. Studies show "larger effects with younger and poorer readers than with older and better ones" (Gough, 1984, p. 245), possibly because poor readers, because of deficiencies in language ability, resort to context as a way to compensate for problems in recognizing words.

Finally, he points out that the context used in the studies he surveyed does not correspond well with real-world reading conditions at all. In these studies, target words were always nouns in the final position in a sentence, making them highly predictable. In addition, nouns make up only a small part of all content words, and content words themselves only make up only about half of the words in a running text. He concludes that context may play almost no role at all in skilled reading, which he concludes is probably a bottom-up, language driven process most of the time (Gough, 1984).

Just what is context in real reading?

In challenging seemingly hard and fast findings, Gough forces us to consider the question of exactly what context is. Clearly, defining it is not as easy as it may seem. At a basic level, it can be seen as information. Information, in turn, is that which reduces uncertainty. In reading, context can be defined as information that reduces uncertainty about the elements of a text, their meanings, and the meaning of the text as a whole.

Traditionally, context (as well as meaning) was seen as a given, existing fully and completely in any properly written text, and the key to using it was linguistic knowledge. Today's cognitive theories claim this view places too much emphasis on linear, bottom-up processing. In response, various definitions of context have been proposed that include language knowledge but emphasize the role played by high-level knowledge sources and personal experiences. Still, the commonly used general distinction between local context (provided by intrasentential and sentential information) or global context (provided by intersetential to discourse level information and world knowledge) is useful to this discussion, especially regarding the guessing strategy and L2 readers. As will be shown below, successful use of the guessing strategy often depends on which of these contexts is available and how it is used, if at all.

Context as a reader construct

Of the many theoretical descriptions of the elements and nature of context, Bialystok's (1983; cited in Barnett, 1989) has important implications for any discussion of context and the guessing strategy. She proposes that context exists in relation and proportion to the reader's implicit knowledge (intuitive and unanalyzed knowledge of the L2), other knowledge (knowledge of other languages and world knowledge), and context (linguistic and physical aspects (in this case, of a text) which provide clues to meaning). From this perspective, context is not an absolute presence in a text, but is instead created by the reader, and is therefore influenced by the reader's linguistic and world knowledge. The implications of this constructivist view of context will be returned to later.

L2 studies: Second-guessing guessing

The number of studies indicating the value of the guessing strategy is huge. Still, it is important to note that the vast majority are of L1 readers, and the subjects are often children, not adults. Although it is tempting to make generalization from these studies to L2 reading, there is no reason to simply assume that the results automatically apply to adult L2 readers (i.e. Coady, 1996), as a growing of body of L2 reading research points out. For example, Bensoussan and Laufer (1984) found that their subjects could successfully guess only 25% of the unknown words in a text used in their study. Haynes' (1984) subjects did not do well using the guessing strategy either, and Schatz and Baldwin (1986) got such poor results in their own study that they concluded that the guessing strategy is so unproductive that it should not be taught at all. Clearly, something different is happening with L2 readers.

Factors affecting L2 readers' use of the guessing strategy

More and more studies show that a key factor affecting L2 readers' ability to make use of context is vocabulary knowledge. Laufer's (1996, p. 20-22) summary of L2 research on this topic provides some interesting conclusions regarding the importance of vocabulary in reading comprehension and strategy use:

L2 learners tend to rely heavily on words as landmarks of meaning in text, less so on background knowledge, and to virtually ignore syntax.
Vocabulary knowledge has been consistently shown to be more strongly related to reading comprehension than other components of reading.
Even if a reader has and uses good metacognitive strategies in L1, they will not be of use in the L2 until the reader develops a solid language base.

To this list, we can add points from Barnett's (1988) discussion of research on the guessing strategy:

Usable context varies from rich to poor, and is affected by the proportion of known to unknown words.
Readers with larger active vocabularies can use available context better than those with smaller vocabularies.
Beginning readers and advanced readers have been shown to use guessing strategies more than middle level readers.

These findings have important some implications. First, they support Bialystock's proposition that context is created by the L2 reader in proportion to preexisting knowledge, and show that vocabulary is an important part of that knowledge. Second, they make it clear that a critical level of vocabulary and general language mastery is essential, not only for successful use of the guessing strategy, but also for the transfer of L1 strategies to L2 reading (Laufer, 1996). Third, the seemingly paradoxical fact that low- and high-level L2 readers use the guessing strategy more than middle-level readers is, in fact, another indication that level of linguistic development plays an important part in guessing. All of these points have direct implications for L2 reading instruction.

The threshold vocabulary

We have seen that an insufficient vocabulary can easily prevent the L2 reader from constructing enough context to guess unknown words, regardless of how much effort is expended in top-down processing strategies, and that this deficit can prevent L1 strategies from being transferred to L2 reading. An important question, then, is how many words the L2 reader needs to automatically recognize to reach the threshold level. Laufer (1996, p. 23-24), using results from her own studies, concludes that transfer of L1 strategies to L2 reading occurs at about the 3000 word family⁽¹⁾

level, which translates to roughly 5000 lexical items.

The "beginner's paradox"

Beginning readers and advanced readers have been shown to use the guessing strategy more than readers in the middle levels (Barnett, 1988). This is probably because beginners don't know much language and have to guess. Advanced readers are likely to guess for the opposite reason; they know enough L2 vocabulary to successfully apply the strategy to unknown words. This levels effect presents us with Coady's (1996) "beginners paradox." Extensive reading is commonly proposed as a way for L2 readers to expand their vocabulary. However, as we have seen, readers who don't know enough words can't read well or guess well. The paradox, then, is how can they learn enough words through reading when they don't know enough words to read well?

A big part of the answer to the paradox is methodical instruction in vocabulary, but this has been in disfavor for many years. It is claimed that directly teaching the large number of words needed for fluent reading is too time-consuming. Extensive reading (which involves intensive use of the guessing strategy) has been strongly advocated as the only reasonable means of building a suitably large vocabulary (Krashen, 1989). But clearly, the beginner's paradox brings this into question, as does the fact that claims for the effectiveness of extensive reading are largely based on results of L1 studies of children which, as we have seen, do not necessarily apply to adult L2 readers. This is not to say that extensive reading has no value in helping readers learn new words. Indeed, with proper attention to materials and assessment methods, learners can benefit very much from extensive reading (Coady, 1996). Still, it may not be nearly as useful for vocabulary learning as we have been led to believe.

Should we teach the guessing strategy?

This discussion has led us back to the question of whether or not we should teach and encourage L2 readers to use the guessing strategy. The answer is "Yes," but not at all times, not with all learners, not with all contexts, and certainly not as the main means of learning vocabulary. We have seen that readers use the strategy for different reasons and in different ways, partly because differences in levels of vocabulary knowledge affect their ability, need and willingness to construct context. This implies that we need to be selective about who we encourage to use the strategy, and that in the early stages extensive language learning should not be expected to take place through reading. In Eskey's (1988) words, the emphasis would be on "learning to read" as opposed to "reading to learn." It also indicates we need to learn much more about how readers make the transition from the intermediate level to the advanced level, why they are hesitant to guess, and when and why that hesitancy declines.

We have also seen that a vocabulary threshold of about 3000 word families or 5000 words is essential to effectively transfer L1 strategies to L2 reading. Indeed, a large sight vocabulary has been shown to enhance guessing form context (Laufer, 1996). Clearly, this evidence supports active teaching of vocabulary. Although direct vocabulary teaching has been out of favor recently, there is definitely reason to reassess arguments against it and to look for effective ways to balance vocabulary learning through direct instruction and incidental exposure.

Finally, it is important to remember that not all contexts are equal. Haynes (1984) found that guessing which only required reference to immediate sentence context was more effective than guessing which depended on textual elements farther away from the target word. In other words, guessing using local context is superior to guessing using global context. Because of this, she believes we should only encourage guessing if clues are in the immediate context, but that we should also teach when not to guess. Accordingly, if guessing requires global context, the guessing strategy should be abandoned and the dictionary or other resource should be used instead.

Conclusion

In this essay I have questioned the usefulness of the strategy of guessing word meaning from context with L2 readers. There is ample evidence in support of its effectiveness in L1 reading, especially with children, but research and experience show that these findings do not apply well to adult L2 readers.

Because there is so much written in support of the guessing strategy, I have focussed mostly on evidence against it to present my case, and much of it supports the importance of bottom-up processing. I do not wish to imply that top-down factors are unimportant. Indeed, background knowledge and schema activation can play an important part in the guessing strategy (Barnett, 1989). Still, top-down strategies can also lead readers to ignore words, and improper use of a schema or of background knowledge can lead readers to wrong conclusions about context and word meaning (Laufer, 1996). All things considered, the case for direct vocabulary teaching and against the guessing strategy is strong, at least for low- and intermediate level L2 readers.

Some guessing may be useful to teach because it encourages readers to make and test predictions, which is a useful generalized reading skill (Liu and Nation, 1985). But the evidence discussed above dictates a selective approach, by the reader and the teacher. Instruction should include training in what contexts provide the best opportunities for successful guessing, and must avoid urging use of the guessing strategy in all cases where readers encounter unknown words. Otherwise, guessing can easily become a strategy for frustration and demotivation instead of for improved reading and learning.

References

Barnett, M. (1989). More than meets the eye. Englewood Cliffs, N.J.: Prentice Hall Regents.

Bensoussan M., & Laufer, B. (1984). Lexical guessing in context in EFL reading comprehension. Journal of Research in Reading, 7(1), 15-32.

Brown, H. D. (1972). Cognitive pruning and second language acquisition. The Modern Language Journal, 56(4), 218-227.

Coady, J. (1979). A psycholinguistic model of the ESL reader. In R. MacKay, B. Barkman, & R. Jordan (Eds.), Reading in a second language (pp. 5-12). Rowley, Mass.: Newbury House.

Coady, J. (1997). L2 vocabulary acquisition through extensive reading. In J. Coady & T. Huckin (Eds.), Second Language Vocabulary Acquisition (pp. 225-237). Cambridge: Cambridge University Press.

Denning, K, & Leben, W. (1995). English vocabulary elements. Oxford: Oxford University Press.

Diller, K. (1978). The language teaching controversy. Rowley, Mass.: Newbury House.

Goodman, K. (1967) Reading: A psycholinguistic guessing game. Journal of the Reading Specialist., 6 (1), 126-135.

Gough, P. (1984). Word recognition. In P. D. Pearson (Ed.), Handbook of reading research (pp. 225-253). New York and London: Longman.

Haynes, M. (1984). Patterns and perils of guessing in second language reading. In J. Handscome, R. Orem, & B. Taylor (Eds.), On TESOL '83: The question of control (pp. 163-177). Washington D.C.: TESOL.

Johnson, D., & Bauman, J. (1984). Word identification. In P.D. Pearson (Ed.), Handbook of reading research (pp. 583-608). New York and London: Longman.

Laufer, B. (1997). The lexical plight in second language reading. In J. Coady & T. Huckin (Eds.), Second language vocabulary acquisition (pp. 20-34). Cambridge: Cambridge University Press.

Liu, N. & Nation, I.S.P. (1985) Factors affecting guessing vocabulary in context. RELC Journal, 16(1).

Nation, P. & Coady, J. (1988) Vocabulary and reading. In R. Carter & M. McCarthy (Eds.), Vocabulary and language teaching (pp. 97-110). London and New York: Longman.

David Dycus can be contacted at: Aichi Shukutoku University, 9 Katahira, Nagakute, Nagakute-cho, Aichi-gun 480-111, JAPAN. By e-mail, contact: <dcdycus@asu.aasa.ac.jp>

Notes

1. A word family includes all the derivations of a word (i.e. 'live, lived, life, living').

Back to Table of Contents Back to FLL home page Back to LAC main page Back to JALT home page