Time for words.

Szczesniak, Konrad

Time for words. Edited by Janusz Arabski. Frankfurt am Main: Peter Lang Press, 2002. Pp. 275.

As recently as a few decades ago there reigned the nonchalant notion that vocabulary was nothing but a disorderly welter of odds and ends. As the editor of the volume Time for words, Janusz Arabski observes, this view has changed dramatically and now the question is approached with the new-found conviction that "the lexicon is more highly structured than was thought before" (p. 7). Indeed, the structured lexicon has recently become a reverenced subject explored seriously from all conceivable angles. This resulted in an evolution which caused lexical studies to diverge into disparate approaches focusing on aspects as varied as lexical memory retention, psycho-analysis-style lexical associations, or the poetic side of word meanings--metaphor. The wealth of approaches and their original nature contribute to our understanding of how the lexicon works and how its potential can be tapped in foreign language teaching. It is in this spirit that the authors present the subject in the volume. Perhaps the only downside of this wealth is that the various topics discussed in Time for words are too many (the volume has 275 pages and 25 articles subsumed under three parts) to be comprehensively summarized in a review. Instead, I have selected the most symbolic examples of the current trends in lexical studies.

The volume opens with the article "Binomials, memes and the evolution of culture", by Hans W. Dechert, where he provides a new perspective on the question of form of expressions. Why do people say whiskey and soda, and not soda and whiskey? There is plenty of research on such questions and practically as many explanations as researchers. Among the most common explanations so far are the (unfortunately facile) suggestions of conventional agreement ("what people often repeat gains currency"), or (even more inane) de-gustibus-non-est-disputandum assertions of subjectivity ("what people like takes root"). There are more ambitious approaches, too. Steven Pinker (1995) points out that "the vowels for which the tongue is high and in the front come before the vowels for which the tongue is low and in the back"--hence we say whiskey first, and soda second.

In this cacophony of approaches, is there room for more theorizing? There certainly is room for intriguing thoughts. Dechert weaves a number of grand theories presenting an unusual way of thinking whose great advantage is that it handles the present problem whereas most previous theorizing does not. First of all, Dechert's proposal does not make one erroneous assumption (which the previous theories can be accused of): that the human mind is a stoically disciplined and predictable machine which will approach a linguistic problem according to one unchanging rule. Dechert is aware of the mind's capriciousness and proposes a theory that allows multiple forces at play, even contradicting ones, which results in more tolerance than the traditional approaches anticipate. He shows that it should not be a problem when some people say night and day, whereas others prefer day and night. Just how this is possible Dechert explains by pulling together anthropological evidence, philosophy of science, and the evolutionist Richard Dawkins's idea of the "meme" to cast new light on the old problem. This original approach comes as a genuine surprise in the midst of research activity traditionally performed in an unforgiving single fashion. Dechert's article is a reminder that any theorizing about this and similar questions is most fruitful when it is pluralistic and when it views the human mind as a compound of various (often conflicting) mechanisms. It does not dismiss the previous attempts to solve the problem. What it does is to demonstrate that one explanation does not necessarily exclude another, even if they turn out to be strange bed fellows.

Mental lexicons in multilinguals

The entire collection of articles talks about vocabulary in foreign language processing, but the question of the mental lexicon proper is discussed most thoroughly in Janusz Arabski's article "Learning strategies of L1, L2 and L3 lexis". Arabski approaches the lexicon not as a sum of words of a language, but as a mental store responsible for managing words. Implicit in this approach is the assumption that the lexicon is a highly complex mechanism. Its structure is probably so baroque that if it were to change only slightly, its functioning too would not be as we know it. For example, one remarkable fact about the lexicon is the unusually fast retrieval time needed to access words--something speakers owe to how the lexicon is organized. What about its structure in the second language? Is it the same as in the first? Which leads to yet another question: Does foreign vocabulary learning differ from that in the mother tongue?

Some interesting data on such questions comes from the empirical study conducted by Arabski. His work has the advantage of circumventing a common problem of this kind of study. Usually, the lexicons are compared in terms of one detail at a time. The L2 lexicon can be shown to perform very well in terms of retrieval time, but this leaves out the question of how lexical meaning is represented. Data on these separate questions must be collected separately in different tests, often with contradictory results. In Arabski's study these questions are integrated in an ingenious study, where the subjects were asked to memorize words in three languages (their native Polish, and English and German) and then to report the strategies they used to remember these words. An astonishing finding was that "the entire language experience is activated as a whole" (p. 211) in the memorization. When faced with the word equine, the subjects think (somewhat dadaistically) of egg wine, words they use to create a mnemonic association. It is important to stress that such mnemonic associations are based on words coming not only from their first language (Polish), but also from English, German, Silesian (Polish dialect), Latin, French, and even from Portuguese. What this means for the study of the lexicon is that, since words from many languages are called up for a given association with equal frequency as those from a person's mother tongue, the retrieval times must be comparable for all these languages. Secondly, since the meanings of these words are equally available for the mnemonic associations across all these languages, it could mean that the lexical representations in L2 or L3 do not differ dramatically from L1.

Of course, the exact structure of these lexicons as well as their mutual relationship remain an unknown quantity and require more research, preferably corroborated by scans of the neuronal aspect of the lexicons. But progress is under way already thanks to studies like the one above and the surprising clues they provide.

Metaphors

Foreign language learning, and vocabulary learning in particular is fraught with familiar difficulties--affective filter, insufficient input, poor fast-mapping, to name just a few. These limit the amount of vocabulary that students absorb, a problem further exacerbated by the students' diffidence in using vocabulary in new contexts. This problem is brought up in Danuta Gabrys's article "A universal or unique and amorphous feeling of anger". Although not really angry, Gabry's shows that the predicament is fairly serious, as even very proficient students are unnecessarily leery of perfectly correct English expressions especially when they are similar to their Polish equivalents. It is hardly a welcome status quo in foreign language business--in practice what it means is that, due to their diffidence, students get deprived of expressive power and most of their competence does not go beyond prior exposure, which leaves very little room for creativity.

Gabry's does not merely identify a problem, for this would hardly be news. An important point of her article is the solution she proposes to overcome the difficulty. She claims that a lot can be achieved by raising the student's awareness of the metaphorical nature of expressions both in their mother tongue and in the target language. She refers to the Theory of Metaphor by Lakoff and Johnson which has recently become extremely popular among linguists and psychologists, but which unfortunately has had little impact on language teaching. Gabry's sets forth to demonstrate the practical strengths of the Theory of Metaphor, arguing that if students were aware of the possibilities that metaphor offers, they would make better use of the English vocabulary. This sounds like a useful idea, considering the present disproportion of problems in the language classroom and the relevant working solutions. Gabrys's solution is promising in its robustness. There are reasons to believe that it could cut a huge swath of problems across the learner's lexicon at one fell swoop: It could stop students from relying so heavily on vocabulary uses witnessed in their experience; students familiar with Metaphor could recognize it in great quantities of new vocabulary which would thus become more salient and easier to remember; and finally Metaphor could also serve as a relatively risk-free method of inventing expressions by students themselves.

Really interactive programs

People are used to reports about computers beating chess champions. So why are there no computers beating language teachers? To be sure, the boon offered to language learners by computers is certainly no hype. In many respects, computers do hold some advantage over even the most dedicated teachers, if only in terms of superhuman patience. Moreover, thanks to the increasingly mind-boggling capacity offered by the hardware, computer-assisted learning is definitely an exciting prospect. Unfortunately, the progress in hardware is not matched by similar developments in software--language learning programs have shortcomings not addressed by their updates. The problem is that they are improved merely in quantity (more graphics and audio) but not in quality. For example, learners still cannot experiment and if they type in a novel sentence which the program does not recognize, too bad.

It is therefore an honor for the present volume to be graced by Brian Farrington's article describing the new software he has devised. Without any risk of sounding effusive, I feel confident to predict that Farrington's work will pioneer a new line of computer software--different from its precursors not in the number of audio-visual gimmicks, but because it represents a leap in quality. According to Farrington, the computer can cheek practically any sentence produced by the learner, and even finish incomplete sentences for him/her. One way to allow the learner to produce his/her own sentences (rather than rely on the program's examples to memorize and regurgitate on cue) is to build a rich database containing all the anticipated sentences a user might come up with. Of course, we are talking about a combinatorial grammar system here, so the number of possible correct sentences will exceed the number of bytes of the computer memory. This is one reason why no one before has ventured to "amass" as many sentences as possible in the computer memory.

Still, Farrington's idea seems perfectly feasible. After all, we are not interested in all the sentences that might strike the learner's fancy. The number of sentences is limited by the topic and vocabulary provided in exercises. For example, in exercises for salutation in business letters, the learner will no doubt stick to commonly predictable sentences whose number could not be multiplied too seriously even by Chomsky.

The ideas presented by Farrington are so exciting and thought-provoking that one feels tempted to recount them all here, which would be supererogatory. They are all to be found elegantly compressed in a short deliciously readable and funny article in the volume.

Objective indicators of competence

To stay with computers and computer-aided language learning for a moment, let us look at an article discussing students' essays. Romantic views of written composition as a kind of inexplicable art aside, students' essays can be evaluated with precision based on a number of reliable indicators of competence. To take a banal example, mistakes are one indicator--an essay riddled with mistakes is visibly less effective than a clean passage. But research on students' writing competence needs to go beyond banalities.

Andrzej Lyda does just that in his article "Disjuncts and conjuncts in discourse management". He suggests one useful indicator by showing that the long-term improvement of language skills is naturally accompanied by the increased use of link words like of course, however, on the other hand, etc. The use of such words more than doubled in the written production (some words were used a few times more often than at the beginning), according to Lyda's longitudinal study of a group of Polish college students, which started in their first year, and ended two years later. Lyda shows that an increased use of these words is a reliable signal of language development. Because they serve to link pieces of information into larger logical structures, to contrast facts, or give examples, etc., their use by a student shows that his/her language competence allows him/her to handle complex ideas. This is also confirmed by the length of sentences (which doubled over the two years of the study). These discoveries are intuitively correct. By confirming their validity, Lyda has pointed out a conveniently straightforward indicator of students' writing competence.

To appreciate its value, it is helpful to look at ways in which such data routinely finds its practical application. Here is another contribution to computer-aided language learning. At present, new software is being developed that serves to evaluate students' written compositions (the most sophisticated is the E-rater produced by ETS's team led by Jill Burstein and Karen Kukich). Unable as computers are to truly comprehend essays, they are still nevertheless very good at high-precision assessment of the written text. To grade an essay, computers use indicators like how many relevant words are used in it. The problem with relevant vocabulary is that it is topic-specific, so for each topic, a new batch of such word indicators has to be fed into the computers. Disjuncts and conjuncts are definitely a universal indicator of an essay's quality, independent of its topic. I am not aware whether E-rater uses disjuncts and conjuncts, but they certainly would be a strong ingredient.

It is interesting that research like Lyda's often mines the first two years of university studies for data on language development. As Maria Wysocka noted in her article "On the learning and use of foreign vocabulary in advanced learners of English", "the third year is that period during the five year cycle when students achieve the best results in language learning" (p. 202).

Maria Wysocka takes this fact as a premise of her study aiming to collect specific data on which words during this period of time are best consolidated, and which remain elusive to the learner. Because answering such questions requires more than just a single study conducted at one sitting, Wysocka proposes a global effort by many practical English university teachers. She then goes on to propose a set of rules that should be observed in the course of the study. They include specific procedures for discovering the most effective techniques of introducing and practicing new vocabulary.

This is a promising plan. Although methodology already makes practical use of working ideas about such questions, there are still a lot of unknowns that the present study can cast light on. At present, the best metaphor for the functioning of the student's vocabulary is still Jean Aitchison's idea that it is like looking at a subway network from the outside--trains come and go, that we know for sure, but where they originate and what routes they take has to be discovered by a lot of ingenious guesswork. In the case of a learner of foreign vocabulary, the subway enigma is complicated by an extra twist--the fact that that the subway is under construction and interested in constant purchase of new trains. If Wysocka's research can provide clues about at least one sure route, that would be progress. A massive study coordinated by a single clear procedure warrants optimism.

Wysocka's idea is perhaps the most symbolic of the entire volume. The nether regions of the mind, its lexical subway can best be probed in a cumulative effort, preferably by specialists interested in different aspects, willing to come together and compare their findings the way they do in Time for words.

REFERENCES

Aitchison, Jean

1989 Words in the mind. An introduction to the mental lexicon. Blackwell: Oxford.

Arabski, Janusz (ed.)

2002 Time for words. Studies in foreign language vocabulary acquisition. Peter Lang: Frankfurt.

Pinker, Stephen

1995 The language instinct: How the mind creates language. New York: Harper.

ELECTRONIC SOURCES

Burstein, Jill--Karen Kukich

1998 "Automated scoring using a hybrid feature identification technique", at: http://www.ets.org/research/dload/ac198.pdf

Reviewed by Konrad Szczesniak, Uniwersytet Slaski, Sosnowiec