Role of Age on Second Language Acquisition

The role of age of acquisition on past tense generation in Spanish-English bilinguals: an fMRI study 

INTRODUCTION

This article explores how age of acquisition for a second language affects what areas of the brain are primarily utilized when Spanish-English bilinguals generate past tense verbs. The researchers hypothesized, according to the sensorimotor/emergentist model, that speakers with an earlier age of acquisition for their second language would rely upon sensory and motor processes of cognition to generate past tense verbs; whereas, later age of acquisition would require speakers to depend upon more executive cognitive control. The sensorimotor/emergentist (S/E) model (Hernandez, Li, & MacWhinney, 2005; MacWhinney, 2004) is a theory that aims to understand how age of acquisition (AoA) for a second language (L2) changes brain activity for late versus early bilinguals. It theorizes that when an individual establishes L2 early on (depending upon sensorimotor learning), there is less of an entrenched linguistic system for their first language (L1); thus, there is a lesser degree of competition and overlap between the two languages (Hernandez & Li, 2007). Comparatively, individuals that acquire L2 later on, use L1 as a reference for L2, which requires more cognitive control mechanisms (Hernandez et al., 2005; MacWhinney, 2004). Other theories that have been hypothesized for how early and late acquisition of L2 regulates brain activity are the declarative/procedural (D/P) model, which postulates that early bilinguals rely on procedural memory compared to late bilinguals who rely more on declarative memory (Ullman, 2001), and the cognitive control model, which theorizes that similar brain regions and systems are used with dependency on motoric and cognitive control for late and early bilinguals (Abutalebi, 2008). These models primarily examine the differences between early and late bilinguals in regards to proficiency, instead of age of acquisition like the S/E model. Previous studies have attempted to examine the effects of age of acquisition of L2 on neural activity, but have failed at controlling for the confounding of proficiency; therefore, this study controlled proficiency in order to highlight only the effects of age of acquisition.

METHODS

The participants in this study consisted of 22 individuals whose native language is Spanish. Their mean age is 23.18 with a 3.61 standard deviation. They were procured from the University of Houston, are right-handed, and psychologically and neurologically healthy. They were compensated with money or credit for courses. Eleven individuals are classified as early learners of English due to having an age of acquisition less than or equal to 6 years old (mean = 3.18, SD = 1.53, range = 1-6). The other 11 were classified as late learners of English with an AoA greater than or equal to 7 years old (mean = 11.00, SD = 3.33, range = 7-17). The stimuli for the experiment was made of 160 verbs; half of which were regular and half of which were irregular English verbs. These were taken from Bird, Lambon Ralph, Seidenberg, McClelland, and Patterson (2003) and Ullman (1999). Some irregular verbs in the experiment were: fly/flew, go/went, and lose/lost. A few of the regular verbs were: name/named, crush/crushed, and seem/seemed. Once inside the scanner, a rear projection video screen and a series of mirrors were used to show the participants English verbs that they were then to covertly generate the past tense form of each of them as successfully as possible. Each verb was shown for an average of 425 ms, but the presentations were at irregular intervals in order to avoid expectation effects. After the study was completed in the scanner, the researchers utilized a behavioral study on the computer where they could determine if the participant had accurately generated the past tense form of each verb during the scanning portion of the experiment. During the scanning portion, brain activation to specific regions was measured using a 3.0 Tesla Siemens Allegra scanner at Baylor College of Medicine. Behavioral performance was measured based on the percentage of errors for past tense generation.

DISCUSSION

Congruent with the researchers' hypotheses, they discovered that second language age of acquisition does affect brain activity during past tense verb generation. The activation that occurred for late versus early bilinguals supports the sensorimotor/emergentist model in that early Spanish-English bilinguals rely on sensorimotor control and motor planning networks; whereas late bilinguals utilize more executive cognitive control and working memory. According to the S/E model, the activation that occurred for early bilinguals is due to a higher degree of competition and overlap between the neural networks of the first and second languages. Comparatively, the hippocampus activation seen for late bilinguals during the task may be attributed to them relying upon associations, like a picture, to generate the past tense verb since the hippocampus plays a large role in lexical retrieval (Heckers, Weiss, Alpert, & Schacter, 2002). Similarly, late bilinguals' utilization of executive control and working memory leads to activation of areas such as the middle frontal gyrus and inferior parietal lobule. Previous studies have shown that early bilinguals rely on sensorimotor structures due to the involvement of activation within the superior temporal gyrus, which has lead researchers to conclude that early bilinguals maintain a different lexical retrieval path (Graves, Grabowski, Mehta, & Gordon, 2007). Both of the groups from this study, employed brain regions that play a role in cognitive control, congruent with the cognitive control model of Abutalebi & Green (2007); however, the type of control (motor or executive) varied depending upon the individual's age of acquisition for L2 because the cognitive capacities and neural systems that are already in use during the time of acquisition influence the type of control. This study was able to further advance support for the S/E model and examine the effects of age of acquisition while controlling for proficiency, which is particularly difficult as it tends to be a confounding variable. It also displayed many of the brain regions that are employed differently for late and early bilinguals, which qualifies the way development of a second language changes brain activation depending upon AoA. This study specifically examined a small group Spanish-English bilinguals; thus, further research with different populations must be done to prove their results as well as lead to a broader understanding of the effects of AoA on brain activity for bilinguals. There have been quite a few studies done since 2013 that have based their research off of this particular study (including a meta-analysis of neuroimaging for L1 and L2 processing in the bilingual brain). The importance of a study like this is that it serves as a foundation for other studies to come along and build off of in order for a greater understanding of bilingualism to be reached.

REFERENCES

Abutalebi, J. (2008). Neural aspects of second language representation and language control. Acta Psychologica, 128(3), 466-478.

Bird, H., Lambon Ralph, M. A., Seidenberg, M. S., McClelland, J. L., & Patterson, K. (2003). Deficits in phonology and past-tense morphology: What's the connection? Journal of Memory and Language, 48, 502-526.

Graves, W. W., Grabowski, T. J., Mehta, S., & Gordon, J. K. (2007). A neural signature of phonological access: Distinguishing the effects of word frequency from familiarity and length in overt picture naming. Journal of Cognitive Neuroscience, 19, 617-631.

Heckers, S., Weiss, A. P., Alpert, N. M., & Schacter, D. L. (2002). Hippocampal and brain stem activation during word retrieval after repeated and semantic encoding. Cerebral Cortex, 12, 900-907.

Hernandez, A. E., & Li, P. (2007). Age of acquisition: Its neural and computational mechanisms. Psychological Bulletin, 133(4), 638-650.

Hernandez, A. E., Li, P., & MacWhinney, B. (2005). The emergence of competing modules in bilingualism. Trends in Cognitive Sciences, 9(5), 220-225.

MacWhinney, B. (2004). A unified model of language acquisition. In J. K. A. D. Groot (Ed.), Handbook of bilingualism: Psycholinguistic approaches. New York: Oxford University Press.

Ullman, M. T. (1999). Acceptability ratings of regular and irregular past-tense forms: Evidence for a dual-system model of language from the word frequency and phonological neighbourhood effects. Language and Cognitive Processes, 14(1), 47-67.

Ullman, M. T. (2001a). The neural basis of lexicon and grammar in first and second language: The declarative/procedural model. Bilingualism: Language and Cognition, 4(1), 105-122.

Ullman, M. T. (2001b). A neurocognitive perspective on language: The declarative/ procedural model. Nature Reviews Neuroscience, 2, 717-726.     

Table 1

Demographic and behavioral data for early and late bilinguals.

Figure 1

Error percentage in postscan behavioral task for early and late bilinguals.

Table 2

Activation regions by condition: group analyses.

Figure 2

Cross group comparison. Verb Generation, early bilinguals > late bilinguals. Activations are significant at p < .001 uncorrected. Areas of activity noted in bilateral sensory and motor regions. Please see Table 3 for a complete listing of activations.

Table 3

Activation regions by condition: main effect L2 AoA analyses.

Figure 3

Cross group comparison. Verb Generation, late bilinguals > early bilinguals. Activations are significant at p < .001 uncorrected. Areas of activity noted in regions involved in direct executive cognitive control, and lexical retrieval. Please see Table 1 for a complete listing of activations.

Table 4

Activation areas by condition: L2 AoA and verb regularity.