Individual Differences in L2 Learning
❝You can never understand one language until you understand at least two.❞
In this blog, we examine four studies focusing on how individual differences might affect one’s ability in second language acquisition (SLA). The purpose is to provide an overview on the progress in this field of study and to suggest possible research directions for further exploration. From the chosen studies, we have found that an individual’s ability to acquire language is linked to one of two factors i.e. either his cognitive capacity (e.g., how good his memory is etc) or external factors such as his environment (e.g., family background, social circles etc.), education levels and personal experiences. In short, we considered cognitive factors as well as social factors when accounting for individual differences in SLA. Ultimately, beyond identifying the factors that affect SLA, we asked how an individual’s ability to learn a second language could be enhanced or whether it was necessary in the first place.
At the heart of many SLA studies lies the intention to promote SLA. This is due to the many suggested advantages that multilingualism seems to pose for individuals. For the majority of this blog and in the studies we have chosen, we look at individual factors that affect SLA. However, we also aim to discuss whether or not SLA is truly relevant or beneficial to an individual, and whether it should be worth promoting at all.
The articles that we have selected for this review propose two kinds of factors that affect SLA – the first factor has to do with one’s cognitive abilities such as working memory or statistical learning ability. Working memory is sometimes referred to as short-term memory and it is the instantaneous processing and temporal storing of information in the brain. Statistical learning ability refers to the process of extracting structure and patterns from the world, like extracting grammatical rules/structures from a sequence. The second factor is concerned with external factors such as one’s personal experiences, his social environment and his level of education.
Each of the four studies aims to elucidate the importance of each factor and this review will attempt to synthesise the findings. Looking forward, these findings could be crafted into possible solutions for individuals who face difficulties in SLA, as well as for more specific groups of children who come from linguistically diverse backgrounds.
After all is said and done, the question remains if SLA is even necessary at all.
On one hand, multilingualism appears to complement the globalisation phenomenon, giving users of multiple languages a suggested advantage on international platforms that monolinguals may not enjoy. Furthermore, previous research on bilingualism have yielded results that suggest that bilingualism could delay the early onset of dementia and may even have positive effects on an individual’s creativity (Madhav, Anand, Swapna, & Sangeetha, 2012).
On the other hand, monolingualism seems to pave the way for the mastery of a single language, undisturbed by the interference of learning another language. Other research have also suggested that the acquisition of two or more languages could lead to the possibility of stuttering in one’s speech (Borson, Maes, & Foulon, 2001).
All in all, studying the individual factors that affect SLA bears not only theoretical import, but also practical implications for teaching, both at home and in the classroom setting. Greater insights into this field may be beneficial in helping teachers adjust their teaching methods and learning environments to better benefit individual learners in the acquisition of a new language.
Selected Studies for Review
The first article we have chosen by Yim & Rudoy (2012) studies the implicit statistical learning (ISL) and language skills of bilingual children. According to the researchers, ISL is an innate ability to “learn new information in which patterns and rules are embedded” in an incidental manner. They suggest that “the ability to learn sequential visual information is significantly correlated with language performance in children.” One of their hypotheses is that “bilinguals’ life experience with linguistic systemising will influence their statistical learning” (Yim & Rudoy, 2012).
They qualify that claim by introducing mediating factors that cause ISL abilities to vary among individuals. For example, they cite “early life experience” as a factor that determines ISL abilities, stating that the earlier infants are exposed to two or more languages, the more sophisticated their ISL skills are. In contrast, auditory deprivation, as in the cause for children with cochlear implants, may stunt the development of ISL development.
In their experiment, participants were presented with either non-namable visual shapes (Fig. 1) or nine pure-tone sounds i.e. sounds that could not be named. In the visual test, participants were shown two sets of three shapes. The former portion of the test was a ‘training phase’ whilst the latter portion was the ‘test’. Based on a training phase, participants had to decide which set looked more familiar during the test.
In the auditory test, three sounds were administered to the participants to form a continuous stream of sounds in a semi-random order. The former portion of the presentation was the ‘training phase’, and the latter was the ‘test session’. Participants were then asked to identify which sounds from the test session were similar to the training phase, as in the visual task.
Both tasks were designed to measure participants’ ability to pick out patterns amidst streams of information, whether visual or auditory.
The research question asked was “whether bilingual children would outperform monolingual children on implicit statistical learning” (Yim & Rudoy, 2012). From the tests, results suggested that no statistical difference between groups was found, either on the visual task or the auditory task. Thus, there was no group difference in ISL.
The second important finding was that bilinguals and monolinguals had different methods in predicting auditory and visual statistical learning skills. This finding helped to support the idea of age being an important variable in ISL and that even though there is no significant predictor for visual statistical learning in children, there seems to exist a significant predictor for auditory statistical learning in children. In general, the study has found that linguistic experience does not influence the learning of statistical regularity in children. However, auditory implicit learning is related to language performance, as mentioned above.
In the next study by Frost et al. (2013), the research question “what predicts successful literacy acquisition in a second language?” was asked. Similar to the findings in Yim & Rudoy’s article, this paper considers the acquisition of language to be the implicit assimilation of statistical properties of a linguistic environment. Throughout the course of the paper, Frost and his team answered the question of whether individual differences in statistical learning predict individual differences in SLA, bringing in factors that may affect the process of acquiring a second language (L2), as well as any relevant trends that may account for the variability in SLA in individuals. In general, it has been suggested that first language (L1) linguistic capacities appear to be reasonable predictors of the success in the acquisition of an L2. In addition, the level of literacy achieved in one’s L1 seems to also be a good determiner of the ability to acquire literacy in a foreign language.
Besides an individual’s statistical learning skills, an external factor that weighs in on SLA would be the degree of similarity and dissimilarity between statistical properties of the L1 and L2. Little else was mentioned in this paper, however, about why or how this is so.
In the experiment, it was hypothesised that statistical learning ability would correlate to learning to read in a new language that is characterized by a novel set of statistical regularities. They compared relative success in the Visual-Statistical Learning (VSL) task with relative success in Hebrew reading tasks, predicting that relative success in learning a series of random visual shapes would predict the speed and success of learning a new language.
American students were recruited for this experiment. They were administered a VSL task and this task examined participants’ ability to detect the patterns embedded in a continuous stream of visual shapes. Scores were collated for that task, then participants were administered three reading tasks in Hebrew to test for speed of decoding, naming of unpointed words and whether there would be a priming effect between modalities.
In the end, it was found that participants who scored well in the VSL task also scored well on two out of three of the Hebrew reading tasks, exclusive of morphological priming. From the findings, they conclude that VSL only correlates to learning the structural properties of Hebrew and not the semantic properties. In the end, they attribute any inconsistencies to factors like a small sample size and motivation among other factors.
Whilst children or younger individuals are often the main focus when studying language acquisition, we have also included a study that delves into the individual differences in adult SLA. The study by Kempe & Brooks (2009) provides a broad look at various factors of SLA. It introduces elements of psycholinguistics and brings in concepts on working memory, phonological short-term memory (PSTM), non-verbal intelligence, IQ and metalinguistic awareness. It also suggests the role of prior experience with other languages in SLA, stating that “individuals [consistently] try to transfer their knowledge from a previously learned language to the new language” (Kempe & Brooks, 2011).
Participants in this study were required to learn a small set of nouns from Russian, to listen and repeat short phrases, identify the referents, and then to produce short statements on their own, all without explicit teaching. Afterwards, participants were expected to make generalisations of the grammar, i.e. case-marking, as well as retention of vocabulary. Performance in the new L2 was then linked to various multiple cognitive abilities and it was said to predict performance in SLA as well.
While prior language experience can be beneficial, it has been shown that it benefits can only be applied under specific conditions. For example, an L1 speaker of Spanish or Italian, that is, languages with a fairly transparent gender-marking on nouns, would be able to apply this knowledge when acquiring a language like Russian (Kempe & Brooks, 2008). This has to do with the similarities between case markings which aids in the acquisition of another language.
Lastly, we reviewed Dabrowska & Street (2005)’s study on the comprehension of passive sentences by native and non-native English speakers. They wanted to see if a greater depth in experience with a particular construction would predict one’s proficiency at the language. It was suggested that proficiency of a language “is merely a function of the amount of exposure”, a proposal which contradicts claims in other studies. Such a claim presents us with the possibility that SLA could be enhanced with the intensive and diligent studying and practise of the language, given the proper motivation and instruction.
In the study of English passives, four groups of participants were involved in this study: graduates (native and non-native) who have had at least 15 years of English education, and non-graduates (native, with no formal English education after secondary school, and non-native, who are currently studying English). Participants were then asked to listen and identify the ‘do-er’ (i.e. agent) of the sentence.
Passives were chosen as for the experiment as they do not tax the working memory since it does not involve embedding. Furthermore, while passives are part of English’s ‘core’ grammar, experience with the structure differs significantly between speakers of the language, with passives found more often in formal written text. As such, academics and intellectuals are more likely, as compared to the layman, to have encountered and have been exposed to such sentence structures due to the very nature of their jobs.
Examples of a full passive sentence: (i) Passive and plausible: The man was bitten by the dog, (ii) Passive and implausible: The dog was bitten by the man.
Results showed that graduate speakers performed better than their non-graduate counterparts. Surprisingly, however, for the non-graduates, non-natives performed better than the natives.
The reasons why non-graduate non-natives performed better than the non-graduate natives could be that there may be a continuum of proficiency with passive sentences, where those with formal schooling do better than those without; and that perhaps the non-graduate natives might have misunderstood the task. But both explanations do not seem to be adequate in explaining this linguistic phenomenon, which will have to be substantiated with additional research adopting other methodology methods.
Further Research Directions and Future Applications
(i) The role of L1 in SLA
(ii) Development of non-verbal intelligence and statistical learning skills
(iii) Learning environments and the role of education
(iv) Early childhood opportunities and depth of experience
The role of L1 in SLA
The majority of the studies make reference to the speakers’ L1 and the ways in which it both aids with and hinders SLA. With that in consideration, it is therefore worthwhile to consider the role that the L1 of an individual might play in SLA and on ways to ensure that it becomes a vantage point for the acquisition of an L2. Thus far, there have been limited findings on whether two structural regularities in people’s lives help them to take advantage of the dual language context. In Frost et al. (2013), it has been shown that the linguistic capabilities of an individual in their L1 generally predict success in an individual’s acquiring of an L2. Although this might superficially imply that the L1 may provide a foothold for the acquisition of the second language, it is important to note that that may not necessarily be the case for all individuals. Whether or not the L1 aids in SLA depends largely on the degree of similarity and dissimilarity between statistical properties of the first and second language, for example, the language distance between the L1 and acquiring L2. In Kempe & Brookes (2009), it was found that individuals did attempt to transfer knowledge from a previously learned language to the learning of a new language(Kempe & Brooks, 2008).However, the success of that process was heavily dependent on how similar the L1 and L2, for example, in the case of Russian and Spanish/Italian gender-marking.
Naturally, this then prompts the question: To what extent can the L1 be useful in aiding SLA? In the case of Singapore, the population is faced with an environment of vast linguistic variety, with grammatical structures sometimes getting muddled between languages, with the mixing of (Fig. 2) and syntactic structures (Fig. 3).
Looking to the future, a possible area to explore is how do we, as linguists, then, ensure that individuals do learn the grammatical structures of the L1 and L2 if the grammatical structures differ (e.g., “Subject-Verb-Object“ as opposed to “Subject-Object-Verb”), without getting learners confused in the process, and thus achieving a good grasp of the L2.
Development of non-verbal intelligence and statistical learning skills
Discussed fervently in two of the studies we have selected, it is important that we take an even deeper look into the concept of Implicit Statistical Learning (ISL). ISL is said to be the ability to break down a complex language stream into regularities that can make learning a language more efficient. Conway et al. (2010) have identified a direct link between language skills and ISL. In a test that measured language ability based on word predictability in degraded listening conditions, subjects were asked to predict the last word of a sentence under one of two conditions – either under a semantically-predictable or non-predictable sentence. Results suggested that the correlation of ISL and language scores were significantly positive.
The findings in Frost et al. (2013) also suggest the correlation between ISL and language proficiency. They further hypothesised that ISL was the basis of learning to read in a new language that is characterised by a novel set of statistical regularities. In the case of English L1 speakers learning to read Hebrew, this would mean that they if they had relative success in learning the transitional probabilities of random visual shapes (liken to the Hebrew script), then they would be able to predict the speed and success of learning to read Hebrew. From the results obtained from the study conducted, there seemed to be a correlation between statistical learning scores and language learning scores.
Kempe & Brooks’ (2009) study also noted that their non-verbal intelligence tests prompted subjects to tap into their ability to notice and identify patterns in complex stimuli. They, too, reported the ability of individuals to detect complex patterns and that it seems to be a powerful predictor for learning grammar over and above the effects of other cognitive predictors like IQ, working memory and PSTM. As such, the weight of this particular variable has to be considered and thereby, further our research on how this ability can be honed to aid SLA.
Given that ISL seems to be important to the adequate acquisition of the L2, the next question we have to ask is how then we can cultivate the skill, taking into consideration limitations such as individuals’ cognitive capacities, ageing, and perhaps a more challenging circumstance of disabilities like hearing or speech disabilities. The study done by Yim & Rudoy (2012) found that kids who had been deprived of auditory input had significantly weaker ISL abilities, which put them at a disadvantage when it came to the learning of a language. Therefore, it is pertinent to explore other various types of input that one can receive in order to be in equal standing as those without disabilities.
Learning environments and the role of education
It is important to note the vital role that education plays in the ultimate attainment of SLA proficiency, expounded on in the two studies of Dabrowska & Street (2005), and Kempe & Brooks (2009). Both studies have shown that there is a strong correlation between education and performance on sentence types. That implies that the more highly educated produce higher language scores. Furthermore, there seems to be an indication of a relationship between IQ and the command of passive constructions. The role of education and one’s learning environment, which can include family members or caretakers, has the potential to play a highly substantive role in SLA and we would be remiss not to take a closer look at them. Therefore, it is important to consider SLA through instructional methods, that is, how lessons are conducted in schools, and how teachers interact with the children.
Learning at home also needs to come under greater scrutiny. The amount of emphasis placed on language proficiency in the home may influence how well children eventually pick up the language. Ensuring that there is a conducive environment for language use and growth begins with the adults of the household, where the language they use and the way they use it (whether formally or colloquially) may have a bearing on how the children use language in their later years. Little can be done about how parents choose to speak to their children at home, but in school, teachers could act as norm-makers, using only exonormative standards as their method of instruction.
The Singaporean government has done just that by implementing language policies in schools (Low & Brown, 2005; Pakir, 1994). In general, it is about consistency – the more children listen to and are exposed to a standard, the more likely they will be to adopt that standard.
Early childhood opportunities and depth of experience
In Yim & Rudoy (2012), the question is posed as to whether children who are exposed to two languages simultaneously develop more robust statistical learning mechanisms due to receiving dual-language input. They hypothesised that the greater the depth and amount of experience children had with tracking statistical probabilities for multiple languages, the more robust their statistical learning abilities.
In Lany & Gomez’s (2008) study, it was found that infants as young as 12-months old, who were passively exposed to artificial languages, were better able to predict the last word of a semantically predictable sentence.
Furthermore, additional studies by Dabrowska & Street (2005) demonstrated the crucial role experience plays through the use of the full passive. A full passive is a paraphrase of an active clause where the object precedes the subject, and where the agent of the action is identified (e.g., the rat was chased by the cat). Considering that the full passive is a structure that is usually found in formal written texts, it is less common and therefore, would help to separate people into two relatively distinct groups – those who are familiar with this kind of discourse (academics or intellectuals) and those who find the discourse unfamiliar (e.g. the layman).
On the basis that the more highly educated should be more familiar with this type of structure (more experience with the structure), the researchers were able to prove that they would outperform the less educated on tasks tapping knowledge about the passive construction.
Through this study, they demonstrated the effects of linguistic experience in relation to familiarity (and hence, acquisition) of language. Future developments in this area should, therefore, concentrate on a child’s growing years. Considering that the amount of experience in navigating through various sets of grammar can help to aid a child’s statistical learning ability. A more robust statistical learning ability would then enable the individual to acquire languages with greater ease and proficiency.
Whilst the main focus of this review would be how individual differences can affect the acquisition of a second language, a possible question to consider based on what has been discussed would be that: is acquiring a second language, and being proficient at it, really that important?
The studies that have been discussed seem to have the common agreement that bilinguals and multilinguals seem to be at more of an advantage when acquiring a new language, as compared to monolinguals. This is as monolinguals are only familiar with one set of linguistic structures, and also taking into account the ‘more advanced’ individual differences that have been discussed in detail of bilinguals and multilinguals, In addition, there are arguments to this effect that are based both on the advantages of being bilingual and the disadvantages of having only one language in one’s back pocket. Studies have shown that bilinguals tend to be more creative and learning two languages could even delay the onset of dementia (Madhav, Anand, Swapna, & Sangeetha, 2012).
However, there are opposing arguments proposing that the mastery of one language enables individuals to be that much more focused, and hence, being more proficient at the acquisition of that language.
As the world becomes smaller with globalisation, there exists a motivation to be able to communicate in more than one’s own language, perhaps in multiple new languages, therefore calling out the importance of learning a new language. Furthermore, the degree of mastery of a language conveys competence and credibility to others, and therefore serves as an incentive in itself as well.
- Borsel, J, V, Maes, E., Foulon, S. (2014). “Stuttering and Bilingualism – A Review.” Journal of Fluency Disorders 26 (2001): 179-205. Elsevier. Web. 15 Jan. 2014. <http://www.cckm.ca/CPSLPR/pdf/van%20Borsel2001.pdf>.
- Brooks, P, J., Kempe, V., (2009). “Individual Differences in Adult Foreign Language Learning: The Mediating Effect of Metalinguistic Awareness.” Psychonomic Society, Inc. 2012. 8 Oct. 2014
- Dabrowska, E., Street, J., (2005). “Individual Differences In Language Attainment: Comprehension of Passive Sentences by Native and Non-native English Speakers.” Language Sciences 28 (2006): 604-615. 11 Oct 2014
- Finn, A, S., Lee, T., Kraus, A., Hudson Kam, C, L., (2014). “When It Hurts (and Helps) to Try: The Role of Effort in Language Learning.” PLOSone. 6 Nov 2014
- Frost, R., Siegelman, N., Narkiss, A., Afek, L., (2013). “What Predicts Successful Literacy Acquisition in a Second Language?” Psychological Science. 8 Oct 2014
- Madhav, A., Anand, A., Swapna, N., Sangeetha, G, S. (2012). “Effect of Bilingualism on Creativity – An Exploratory Study.” Journal of All India Institute of Speech and Hearing 31 (2012): n. pag. Web. 14 Jan. 2014.
- Snow, C, E., Hoefnagel-Hohle, M., (1978). “The Critical period for Language Acquisition: Evidence from Second Language Learning.” Child Development, Vol 49, No. 4:1114-1128. 25 Oct 2014
- Yim, D., Rudoy, J., (2013). “Implicit Statistical Learning and Language Skills in Bilingual Children.” Journal of Speech, Language & Hearing Research (2013) 56, 1: 310. 23 Sep. 2014
Exonormative standards: standards of language based on the metropolitan norms of the ‘mother’ country and outside of the place in question
Full passive: a paraphrase of an active clause where the object precedes the subject, and where the agent of the action is identified
Implicit Statistical Learning (ISL): innate ability to incidentally learn new information in which patterns and rules are embedded
Phonological Short-Term Memory (PSTM):a phonological store and an articulatory rehearsal mechanism which maintains decaying representations, that has the ability to retain sequences of sounds in memory over short periods of time.
Linguistic systemising: involves the extraction of structures and statistical regularities in a language; synonymous to statistical learning ability
Second Language Acquisition (SLA): the acquisition of a second language
Working memory: usually synonymous to ‘short term memory’, where information is instantaneously processed and stored temporally
First created by Natalie Goh, Kerensa Chew, AY2014/15 Semester 1