| THE RELATIONSHIP OF STARTING TIME AND CUMULATIVE TIME TO SECOND LANGUAGE ACQUISITION AND PROFICIENCY
Elizabeth Murphy Faculty of Education
School boards and districts continue to establish policies and programs on these aforementioned assumptions and seek to improve French programs by increasing the number of instructional hours for French available to their students. Yet, as McNab (1976) reminds us: "time is not only the scarcest resource in the school system, it is also the most expensive" (p.265). In the 1976 report, French programs: Some major issues, authors Stern, Swain and McLean articulate some concerns regarding time and costs: Throughout the last decade, French programs in the Ottawa elementary schools were considerably expanded in the expectation that a small daily dose of French would be educationally and financially acceptable. When this expectation was not fulfilled, it was thought that the answer lay in increasing the amount of time devoted to the teaching of French. However, the demand for the extension of French gives rise to several considerations. An extension of time for French entails a rise in cost: is this additional expenditure commensurate with the increase in French achievement? (p. 43) Because time is at a premium and comes, not only at an economical cost, but a human one as well, it is important that we understand the relationship between time and proficiency. The purpose of this paper is to consider the relationship between the two through a synthesis of the various research findings concerning the relationship of time to second language proficiency. More specifically, this paper considers the following two questions:
Starting Time as a Predictor of Proficiency Levels There has been considerable research conducted regarding the relationship between age or grade at which learning begins, and the ultimate level of proficiency. In fact, not only has it been one of the most debated issues in language teaching theory, it is far from being resolved (Stern, 1983:361). The findings on this issue can be grouped into three categories; those that support the 'optimal' or 'critical age' hypothesis of a positive relationship between an early (child) starting age or grade and eventual achievement in French; those that refute the hypothesis of an 'optimum age' or 'critical period'; and those that support the hypothesis regarding a positive relationship between an older (adult) starting age or grade and level of achievement. Associated with the belief in the optimal age hypothesis is the neurophysiologist W.G. Penfield (1959) who observed that complete recovery of language ability after brain damage was possible in children but not in adults. Penfield maintained that the child is at an 'optimum' age for language learning because of the neural plasticity of the child's brain which makes it receptive and well adapted to the development of speech mechanisms. The age of the learner is the most important factor in the language-learning process and, if language learning is to be successful, it must take place before the age of ten: "The brain of a child is plastic. The brain of the adult, however effective it may be in other directions, is usually inferior to that of the child as far as languages are concerned" (Penfield and Roberts, 1959:240). This critical period hypothesis (CPH) was refined by Lenneberg (1967) who believed in a superiority of children over adults regarding language development and posited that the lack of specialization or lateralization of the pre-pubertal child's brain makes it more receptive to language development. Like Penfield, Lenneberg, based his observations on brain-damaged children. Although, Lenneberg's theories relating language ability to lateralization were questioned in later years, his basic hypothesis regarding the child's advantage in language development received a great deal of interest from researchers and the public alike. Schumann (1975) argued that, because of affective factors, children are more permeable to language influences than are adults. From a cognitive rather than an affective point of view, Krashen (1981) and Rosansky (1975) maintained that the critical period of language development is before the 'Piagetian' stage of 'formal operations' at adolescence and that in older language learners, second language learning is blocked by cognitive factors. Certain empirical studies in second language learning particularly in the area of pronunciation and global comprehension are consistent with the hypothesis that children are better second language learners than adults (e.g., Asher and Garcia, 1969; Oyama, 1973, 1976, 1978; Seliger, Krashan & Ladefoged, 1975; Patkowski, 1980; Strevens, 1972; Kirch, 1956). Many of these studies claiming the superiority of children as language learners were challenged on empirical and conceptual grounds. In relation to the studies of Penfield and Lenneberg, Genesee (1988) argues that their evidence is based on the ability of neurologically impaired adults to relearn first language skills and is therefore not relevant to the ability of adults with healthy neurological systems to learn a second language. Other research has directly or indirectly challenged or provided an alternative basis for the critical period hypothesis (see Krashen, 1974; Molfese & Molfese, 1979; Lamendella, 1977; Seliger, 1978; Walsh & Diller, 1981; Snow & Hoefnagel-Hohle, 1978). A number of studies have found empirical evidence contrary to the CPH. In a study by Olson and Samuels (1973) based on observations of immigrant children in natural settings, the authors concluded: "The general assumption is that younger children learn to produce foreign words with a more native like accent than older people. Not only is this assumption not supported by the test results but the trend is in a reverse direction favouring older students" (p. 267). Asher and Price (1969) compared the listening comprehension of children and adults using Russian as the L2 and found the adults to be superior to the children of any age group at p < .0005. Krashen, Long and Scarcella (1979) present evidence in favour of adult and older learners for age, rate of second language acquisition and generalize that: 1. adults proceed through the early stages of syntactic and morphological development faster than children. 2. Older children acquire meaning faster than younger children. Burstall, Jamieson, Cohen, and Hargreaves, (1974) in a major longitudinal study in British primary schools, found that children exposed to a basic program of French from age eight did not reach a higher level of achievement than the children who began at age eleven and concluded that the findings did not support the view that there is an optimal age for foreign language learning. In a comparative study of the teaching of French as a foreign language in eight countries, co-author J.B. Carroll (1975) affirmed that: There is no clear evidence that it makes any difference at what age or grade the student starts the study of French; if anything, students starting late in their educational careers make faster progress, other things being equal. That is, most of the evidence contradicted the commonly held assumption that students starting early have a special advantage.(p. 29) Similarly, a large number of studies conducted in Canada comparing the learning efficiency of Early and Late Immersion students confirm these results and support the hypothesis that older children are more efficient language learners. Research in Montreal conducted by Genesee (1981) showed that the achievement of students after 1,400 hours or two years in the Late Immersion program beginning at age twelve was comparable to that of students after eight years or 5,000 hours in the Early Immersion program beginning at age five. It was only in the area of listening comprehension that the earlier starters appeared to achieve/demonstrate an advantage. Genesee summarizes these findings as follows: That the two-year late immersion students achieved parity with the early immersion students despite the fact that the former had considerably less cumulative exposure to French than the latter at the time of evaluation implies relatively faster learning on the part of the older students and suggests, therefore, that beginning intensive second language instruction early in school is not necessarily advantageous, other things being equal (...) these results corroborate those from studies of short term second language learning in conventional instructional programs in demonstrating that younger children are not necessarily more effective than older children...and, in fact, may make slower progress. (p. 125) Swain (1987) provides an explanation for these results and argues that older learners are often more efficient than younger learners perhaps because of their cognitive maturity which may help them in the more formal aspects of language learning. Combined with this is the fact that older learners can transfer skills and knowledge already acquired from their first language to the second language context. Smythe et al. (1974), in reviewing the literature on the relationship between starting time and eventual achievement, conclude that: "it is a myth of contemporary folk-linguists that children are superior foreign language learners"(p.20). In an attempt to summarize the state of knowledge on this issue Stern (1983) concludes that Learning may occur at different maturity levels from the early years into adult life. No age or stage stands out as optimal or critical for all aspects of second language learning (...) There are differences in the acquisition of different aspects of language (phonology, vocabulary, syntax,etc.). (p. 366) In summary, although the debate on this issue is not yet over, we can affirm that the bulk of the empirical evidence does not support the hypothesis that an earlier starting time correlates with a higher level of proficiency than a later starting time. Recent evidence has refuted initial claims of a neural superiority of the brain of children with regard to language learning. In fact, a substantial amount of evidence suggests that, because older learners are more efficient learners, they may have an advantage over younger learners with regard to language learning. The results are not conclusive, however, and it is still not clear if this advantage (if it exists) holds true for all or only some aspects of language learning. It would seem then, that starting time, independent of cumulative time, is not an accurate predictor of the eventual level of proficiency in L2. It is quite possible that starting time could be an accurate predictor if there were more specific knowledge about this variable, and in particular about how it interacts with other variables. More research needs to be conducted in this area specifically in order to clear some of the confusion of the contradictory results of some studies and, in general, in order to further the understanding of the relationship of age to L2 acquisition. Cumulative Time as a Predictor of Proficiency Levels The second question posed in this paper seeks to identify the relationship between the amount of cumulative time and level of proficiency and to determine how well cumulative time correlates with level of proficiency. We can organize the findings pertaining to this question into four models; those which show a direct linear relationship between time and level of proficiency (an incremental model); those which suggest that cumulative time independent of starting time does not accurately predict the ultimate level of proficiency (older-age sensitivity model); those which suggest that there is not a strong correlation between cumulative time and level of proficiency (nonlinear model), and those which suggest that there are diminishing returns in the relationship of time to level of proficiency (diminishing returns or learning plateau model). Smythe et al. (1974) in reviewing the literature on the relationship between starting time and eventual attainment conclude that: "... there is strong support for the position that the more time spent studying a second language the greater the probability that the individual will achieve a high level of sophistication in the language" (p.21). Lazaruk (1977) reported that the principal finding of the Ottawa-Carlton experiments on the study of French as a second language is "... that the more instructional time that is available, the more learning is likely to occur"(p.8). Based on the findings of the longitudinal study in British primary schools, Burstall (1974) concluded that the longer the period of learning, the higher the educational level of achievement. In relation to the findings of a study of English teaching in the Philippines, Tucker (1977), reported that: "The results indicated that English proficiency was directly related to the number of years English had been used as the medium of instruction" (p.34). In a report of a two-year research study evaluating the effectiveness and costs of different ways of teaching French, co-author Halpern (1976) found that two groups of second grade students with similar aptitude but differing amounts of classroom exposure to French learned different amounts of French. Those with the greater number of hours learned more than the other group leading the author to conclude that "when more time is provided for learning, more learning occurs" (p. 165). These findings clearly indicate a positive correlation between the amount of cumulative time and level of second language proficiency. There would appear to be a linear relationship whereby incremental increases in time result in incremental increases in proficiency. In the Burstall (1974) study mentioned previously in this paper, it was reported that, with the exception of listening comprehension, regardless of whether or not the child learned French for five or eight years, comparable levels of proficiency were reached. The children exposed to French for five years from age eight reached equivalent levels of proficiency as the children exposed to French for eight years who began at age eleven. The Genesee (1981) study also referred to in the previous section of this paper reported results comparable to those of the Burstall study in that those with only 1,400 hours of study in French achieved comparable levels of proficiency with those students who had been immersed in French for more than 5,000 hours. Again the results favoured the older learners. Dunkel and Pillet (1962) used standardized tests of formal grammar to compare two groups of American school children and found that the older group with only one year of study outperformed the younger group with five years of study. Oller and Nagato (1974) found similar results in a study in Japan. These findings show that, although time is an important factor with regard to the level of proficiency, it is very often not an accurate predictor of level of proficiency except perhaps if it is considered in conjunction with the factor of age or starting time. We should note, however, that these studies are not conclusive about what ages produce what levels; they simply distinguish between older or adult learners and younger learners. Spilka (1976) conducted a study of the speech of a group of 20 early French immersion students' second language proficiency in grades five and six after six and seven years in immersion respectively. She taped their speech and compared it to the speech of classes of Francophone children of the same age. The immersion students had undergone the same evaluation procedure in the first four to five grades so that Spilka was able to report the findings relative to a period of six and seven years. Mean scores were computed for both the control and experimental group for each grade on such items as the percentage of incomplete and incorrect sentences, the percentage of gender, verb, preposition, pronoun, and reflexive pronoun errors. An analysis of the results show that, with the exception of the two items, verbs and pronouns, the experimental group made a higher percentage of errors in grade six than they did in grade one. In almost all cases, the number of errors made by the control group decreased from grades one through six. If the results of this study are significant and not a result of errors in the evaluation procedures or the design of the study then they clearly indicate that the level of proficiency did not improve with time. In a report on French programs in the Carleton and Ottawa School Boards, Stern et al. (1976b) concluded from the findings that "... the different time variations within the core programs have not led to any outstanding differences in language performance"(p.48). These findings and those of Splika would suggest a nonlinear relation between time and proficiency. The results of other studies indicate that, although there may initially be a linear relationship between time and level of proficiency, eventually, diminishing returns begin to set in whereby there cease to be incremental gains. Studies by Cummins (1981), Oyama (1976, 1978), and Patkowski (1980) have shown that, beyond a five-year term of exposure in a natural setting, the relationship of time to level of proficiency becomes minimal. Walberg, Hase and Rasher (1978) confirmed these results and postulated that the rate of acquisition of L2 diminishes with time. The sample for their study were 350 children of Japanese businessmen, university faculty and graduate students who had lived in the United States from zero to twelve years. The students completed self ratings and were also given ratings by their teachers. The first question which the results attempted to answer was whether or not the students gain in English language fluency in equal or diminishing increments with each additional USA month. An analysis of the results revealed that the gains in fluency and competency are made rapidly at first and more slowly as time goes on. The progression is equal for the first two months, five months, one year, the next two years and the next eight years. After this time, gains the same size as the initial ones take increasingly longer to reach. The acquisition rate is fast initially but the amounts of gain diminish with time. The authors depict this relationship between acquisition and time in a model which they refer to as the diminishing returns model and which they explain in the following way: It seems unreasonable that children can continue making equal language strides if adults can learn a foreign language reasonably well in 1,300 intensive hours. Whether or not one subscribes to the early age sensitivity hypothesis, there would seem to be diminishing returns for children of all ages to exposure or experience in natural language environments and possibly the contributed environment of the school. It is well known that a very small number of different words comprise most of the oral and written discourse of all languages that have been studied. The word the comprises about 7% of the 5 million words in a large sample of school textbooks passages and shows this nearly constant percentage across grades 3 through 9; (...) Repeated exposure in such texts and even greater repetition in oral discourse would eventually lead to learning plateaus with respect to vocabulary and probably syntax. Rarer words by definition are repeated less often; and it is difficult to imagine how steady, constant rates of vocabulary acquisition would be maintained in natural environments of learning. (p. 429) These studies by Walberg et al. estimate that it is after eight years that diminishing returns begin to occur; however, this conclusion is on the basis of teacher and self ratings, and we must question whether or not ratings are accurate estimates of level of proficiency. Furthermore, these results are based on minority language learners in a majority language environment which leads us to question whether or not the results would be similar for English majority language learners studying French, the language of the minority. This is not to suggest that the results are not valid; rather is points to the need for more studies to see if the results can be replicated and to determine if they can be generalized to the study of French as a second language. There is also a need for further definition of the point at which the diminishing returns actually occur. Is it after five years as studies of Cummins (1981), Oyama (1976,1978), Patkowski (1980) suggest or is it, according to Walberg et al. (1978), after eight years? These four models present somewhat contradictory representations of the relationship between time and L2 acquisition. It would be imprudent to choose one more than any other as being more representative of the relationship since no one category of findings is more conclusive than the other. The linear, incremental model suggests that, by computing the amount of time that a student spends studying a language, we will be able to predict the level of proficiency which he or she might eventually achieve. If this is true, then, why did the amount of time not predict level of proficiency for those students in the studies of Brustall (1974), Genesee (1981), Spilka (1976) ? Obviously, this model alone does not account for the differences in achievement in relation to time for the groups studied. If we combined this linear model with the age sensitivity model we could account for the differences in the findings of the first two groups of studies. By considering the interaction of age at which the person began the study of French combined with the amount of time of exposure, we could predict the eventual level of proficiency that the student might be expected to achieve. However, combining these two models would still not account for the differences between the first two sets of studies and the third set which found diminishing returns. If we were to combine the three models, then we could suggest that there is a linear relationship between starting time plus cumulative time and the student's eventual level of proficiency up to the pint where there is a plateau or diminishing returns. However, even by combining the three models, we are not accounting for the results of the studies of Spilka (1976) and those reported by Stern (1976) which depicted a nonlinear relationship. We are forced to conclude, therefore, that based on the review of the studies cited in this paper, we cannot clearly or definitively conclude that there is a linear relationship between the amount of cumulative time and level of proficiency or whether the amount of cumulative time can predict the level of proficiency. Conclusion The relationship of time to achievement is an essential issue which needs to be clearly understood in order to establish a foundation on which to base policy, programs and curriculum. Research has provided many answers to questions on this topic and has shown that starting age is not in itself an accurate predictor of level of achievement. It now seems that the optimum or critical age hypothesis favouring young children is not valid and that older children require less cumulative learning time because they are more efficient learners. Cumulative time does not appear to be an accurate predictor of achievement and, depending on the studies considered, there may be a linear or incremental relationship, a diminishing returns relationship, a nonlinear relationship, or a linear relationship which includes the variable of starting time as well as cumulative time. Research might provide some answers to the remaining questions. Concerning the relationship between starting time and level of proficiency, research could attempt to define systematically the effect of different ages (not only younger/older, child/adult, or pre/post pubertal learners) on rate of acquisition and level of proficiency. Studies could be conducted on the effect of starting time and cumulative time on the different aspects of language in isolation and together (i.e., grammar, vocabulary, listening skills, speaking skills, writing skills, pronunciation, etc.). Longitudinal studies (at for example grades 1, 6, 9, 12) which estimate for each grade the rates of acquisition and the level of proficiency of the different skills might indicate how learning progresses and whether or not a plateau is actually reached and if so at what grades. Certainly, one of the most important questions needing to be answered is how time interacts with other factors such as cognitive maturity, attitude, motivation, teaching methods, aptitude, class size, etc. It would be beneficial to conduct research on the other aspects of time, such as time on task and time spent learning the L2 outside the classroom. This research could take into consideration general theories relating time to learning such as those elaborated by Carroll (1963); Bloom (1974, 1981); Frederick and Walberg (1980). Understanding the entire issue of time and its relationship to second language acquisition is central to understanding the entire process of language acquisition in general. It is a complex relationship which we are only beginning to understand. References
|