Content uploaded by David Birdsong
Author content
All content in this area was uploaded by David Birdsong on Aug 31, 2016
Content may be subject to copyright.
The Critical Period Hypothesis for Second
Language Acquisition: Tailoring the Coat
of Many Colors
David Birdsong
Abstract The present contribution represents an extension of David Singleton’s
(2005) IRAL chapter, ‘‘The Critical Period Hypothesis: A coat of many colours’’.
I suggest that the CPH in its application to L2 acquisition could benefit from
methodological and theoretical tailoring with respect to: the shape of the function
that relates age of acquisition to proficiency, the use of nativelikeness for falsifi-
cation of the CPH, and the framing of predictors of L2 attainment.
1 Introduction
David Singleton’s (2005) study, ‘‘The Critical Period Hypothesis: A coat of many
colors’’, is the second most-cited article ever to appear in International Review of
Applied Linguistics in Language Teaching. At its core, the piece is a critique of the
Critical Period Hypothesis (CPH) as it has been applied in the context of second
language acquisition (L2A). Singleton argues that, as an account of constraints on
L2A attainment, the CPH is underspecified in the literature. Crystallizing the
sometimes vague and decidedly diverse positions advanced by researchers in the
CPH tradition, Singleton (2005: 280) writes: ‘‘For some reason, the language
acquiring capacity, or some aspect or aspects thereof, is operative only for a
maturational period which ends some time between perinatality and puberty’’.
With respect to the notion of ‘period’, Singleton notes that various researchers
have pegged the end of the CP for phonetics/phonology at ages ranging from one
year to puberty. As for the affected language learning capacities, Singleton’s
review of the literature reveals that CP researchers have put forth accounts of
deficits in: general language learning ability, non-innate linguistic features, innate
D. Birdsong (&)
University of Texas at Austin, Texas, USA
e-mail: birdsong@austin.utexas.edu
M. Pawlak and L. Aronin (eds.), Essential Topics in Applied Linguistics
and Multilingualism, Second Language Learning and Teaching,
DOI: 10.1007/978-3-319-01414-2_3, Springer International Publishing Switzerland 2014
43
linguistic features, specific subparts of innate features, and implicitly acquired
linguistic features. As concerns the underlying sources of CP effects, Singleton’s
survey tallies six accounts of a neurobiological nature, four in terms of cognitive
development, and four relating to affect and motivation.
Singleton (2005: 280) characterizes with trademark pithiness his notion of ‘the
manifoldness’ of the CPH:
My conclusion from this exploration is that the CPH cannot plausibly be regarded as a
scientific hypothesis either in the strict Popperian sense of something which can be fal-
sified (see, e.g. Popper 1959) or indeed in the rather looser logical positivist sense of
something that can be clearly confirmed or supported (see, e.g. Ayer 1959). As it stands it
is like the mythical hydra, whose multiplicity of heads and capacity to produce new heads
rendered it impossible to deal with.
From Singleton’s perspectives on the CPH/L2A, ‘‘a coat of many colors’’ is indeed
an apt metaphor.
The present contribution piggybacks on Singleton’s work, taking complemen-
tary perspectives on mainstream research conducted in service of the CPH/L2A.
Adding a fitting metaphor to Singleton’s original title, I attempt to show that the
coat of many colors might warrant some methodological and theoretical tailoring to
accommodate the facts and phenomena associated with age and attainment in L2A.
2 What Critical Periods Look Like
To make a case for a CP in the L2 context, it does not suffice to demonstrate that
age of onset of L2 learning (often referred to as age of acquisition or AoA) and
ultimate L2 attainment are related. To qualify as a period, the geometry of the
function relating AoA to performance (usually characterized in terms of linguistic
proficiency or processing ability) should contain a slope that is bounded at some
points along the function.
Many studies have found AoA effects over the full span of AoA’s, suggesting
unbounded functions (Birdsong 2005). Conversely, non-linearities or inflections in
the AoA-attainment function have been interpreted as suggestive of a period, in the
sense that changes in slope would mean that AoA-related effects are bounded
(Hakuta et al. 2003; Stevens 2004). The logic here is that a significant slope
change would be consistent with a qualitative change in sensitivity of the learning
mechanism. To suggest that maturational effects are at play, the changes in slope
should line up with recognized developmental milestones that are uncontrover-
sially maturational in nature.
In this context, Birdsong and Molis (2001) reanalyzed the L2 proficiency data
from Johnson and Newport’s (1989) study of Korean and Chinese learners of L2
English. Using a piecewise linear regression model, the reanalysis placed the
breakpoint in Johnson and Newport’s AoA-proficiency slope at 18 years, i.e. at an
AoA beyond puberty. Similarly, Vanhove (2013) applied piecewise regression
44 D. Birdsong
analyses to DeKeyser et al.’s (2010) data from Russian immigrants learning L2
English in North America and L2 Hebrew in Israel. Vanhove’s reanalysis of
DeKeyser et al.’s Hebrew grammaticality judgment results revealed that including
an inflection point in the AoA-attainment function did not result in a better fit than
a simple linear regression model. In other words, AoA effects were best modeled
as a straight-line function, across the full range of AoA. The reanalysis of the
English grammaticality judgment results revealed that a model with a breakpoint
at around AoA =16 was a marginal improvement over a simple linear model.
However, like the Hebrew data, the slope of the function after AoA =16 did not
flatten, i.e. a decline in performance continued throughout adult AoA.
Vanhove’s study suggests that piecewise regression models, which have been
used only infrequently in L2 attainment studies, are appropriate for determining
whether the timing and geometry of the AoA-attainment function conform to
assumptions of what a CP should look like.
1
Made-to-measure analytical methods
may be required to suitably fit the coat to the function.
3 Nativelikeness and the CPH/L2A
Long (1990) stipulates that the way to falsify the CPH in the L2A context would
be to find a single late learner who is indistinguishable from an adult monolingual
native. The operational logic goes something like this: the absence of observed
nativelikeness is due to maturational factors, and nativelikeness can disconfirm the
CPH/L2A.
On a complementary view of non-nativelikeness, many researchers point out that
non-monolingual-likeness in both the L1 and the L2 is a defining characteristic of
bilingualism (early and late) (for a review, see Ortega 2009: 26–27). For example,
VOT values of the L1 may extend toward those of the L2, just as VOT values of the
L2 may extend toward those of the L1 (see e.g. Fowler et al. 2008). Among bil-
inguals, effects of maturation (in the sense of biologically determined declines in
learning ability) cannot straightforwardly explain the fact that syntax, lexicon, and
phonology of the L1 are altered in bilingualism, and have features reflecting contact
with and use of the L2 (see e.g. Cook 2003). Non-monolingual-nativelikeness in the
1
Granena and Long (2013) applied multiple linear regression analyses to the relationship of
Chinese natives’ AoA to their attainment in L2 Spanish morphosyntax, phonology, and lexis and
collocation. For each of these three linguistic domains, including breakpoints in the model
revealed a small (5 %) but statistically significant increase in variance accounted for, as
compared to the variance accounted for in a model with no breakpoints. According to the authors,
the fact that the improvement was so small ‘‘could mean that the less complex (i.e. more
parsimonious) model with no breakpoints is already a good enough fit to the data or, alternatively,
that a larger sample size is needed to compensate for the loss of degrees of freedom and to
minimize the risk of overfitting’’ (2013: 326–327).
The Critical Period Hypothesis for Second Language Acquisition 45
L1 of bilinguals cannot be due to maturationally induced impairment of a presumed
language learning mechanism, inasmuch as the L1 has been fully acquired, before
the end of maturation.
Arguably, the fact that the L1 can be influenced by the L2 in adulthood is
evidence for maturationally conditioned representational plasticity. In other words,
non-monolingual nativelikeness in the L1 is suggestive of a capacity to learn
language in adulthood. For example, ‘speaking with an accent in the native lan-
guage’ is common among immigrants returning to their homeland for visits, as are
noticeable changes of accent among individuals who move across dialect
boundaries within a single country. Such permeability of the L1 would not be
possible if the neural systems underlying phonetic perception and production were
not plastic. To fully clothe the big-picture facts about late L2 and late L1 learning,
the CPH/L2A coat might benefit from some broadening through the shoulders.
4 Scrutiny Across the Board
According to Long (1990) and Hyltenstam and colleagues (e.g. Hyltenstam and
Abrahamsson 2003; Abrahamsson and Hyltenstam 2009), there are two key ele-
ments of the linkage of nativelikeness to the CPH/L2A. One is the requirement that
the nativelikeness in the L2 must be observed ‘across the board’, that is, with
respect to all L2 linguistic features and processes, for it to be sufficient to falsify
the CPH. The other is that the evidence for (non)-nativelikeness (be it, presumably,
behavioral or brain-based) should be uncovered from close scientific scrutiny, lest
some evidence be overlooked. Thus, on this view, an individual who appears
nativelike to the casual observer or on coarse or too-easy performance measures is
insufficient evidence for rejecting the CPH. In sum, falsification of the CPH/L2A
would require ‘scrutinized nativelikeness’ (Abrahamsson and Hyltenstam 2009)on
a comprehensive set of linguistic measures.
There is a sensible rationale for psycholinguists to look beyond what is noticed
by the untrained ear. With sensitive measures, our understanding of linguistic
behaviors—especially inter-group and inter-individual differences—is enhanced.
In the L2 context, as in scientific inquiry generally, the precision of information
available from granular observation is valuable and welcome. From this per-
spective, there is no argument with scrutiny. The concern is with the application of
evidence for non-nativelikeness—be it obtained by scrutiny or by any other
methodological orientation—to theory. Monolingual-bilingual differences are
inevitable, and more differences are sure to emerge from challenging tasks and
fine-grained analyses than from simple tasks and coarse analyses. But it is not clear
that non-monolingual-like behaviors and brain functions are decisive for CPH/
L2A theory. Given what is known about reciprocal L1-L2 influences in bilinguals’
behaviors, evidence for non-nativelikeness—be it detected on the street or under
microscopic examination, be it present in outer patches or inner pockets, in bolts of
cloth or in buttonholes—does not compel, uniquely, a maturational explanation.
46 D. Birdsong
And so it is with across the board nativelikeness. Since bilinguals are not like
monolinguals in either of their languages, it is hard to argue that comprehensive
nativelikeness, scrutinized or not, should be held up as the gold standard for
falsifying the CPH/L2A.
If the idea is to look around for non-nativelikeness in bilingualism, then non-
nativelikeness will eventually be found. If the follow-on idea is to stipulate that
across-the-board nativelikeness is what is required to disconfirm the CPH, then the
CPH is invulnerable to falsification. This being the case, the coat would need some
letting out in the chest to accommodate the Kevlar vest underneath.
5 Framing the Issues
A study by DeKeyser (2000), entitled ‘‘The robustness of critical period effects in
second language acquisition’’, investigates the roles of factors such as AoA, lan-
guage learning aptitude, and years of schooling in predicting L2 English gram-
maticality judgment (GJ) accuracy by 57 Hungarian immigrants to the US. A look
at each of these factors in turn is revealing.
•AoA. For all participants, AoA was predictive (r=-0.63, p\0.001). On the
other hand, breakout correlations with groups divided by early arrivals
(AoA \16; n =15; r=-0.26 ns), and late arrivals (AoA 17–40; n =42;
r=-0.04 ns), revealed no significant declines at either pre- maturational or
post-maturational AoA epochs. Thus, definitional evidence for a critical period,
in the form of pre-maturational declines in proficiency, is not found. DeKeyser
acknowledges this failure to replicate the pre-maturational AoA effects observed
by Johnson and Newport (1989) (the items used in DeKeyser’s grammaticality
judgment task were a slightly modified subset of those used by Johnson and
Newport). DeKeyser considers this discrepancy ‘‘hard to interpret’’ (2000: 513),
and goes on to develop an explanation based on putative artifacts of sampling
(2000: 514).
•Aptitude. DeKeyser administered to all participants a Hungarian-language
adaptation of Carroll and Sapon’s (Carroll and Sapon 1959)Modern Language
Aptitude Test. The average aptitude score of all participants was a low 4.7 out of
a possible 20. DeKeyser divided the 57 participants into a high aptitude group
(n =15) whose aptitude scores were 6 or higher, and an average- or low-
aptitude group consisting of 42 participants. To clarify, the 15- and 42-partic-
ipant breakouts for high aptitude and average/low aptitude, respectively, were
not the same participants as the groups of 15 early arrivals and 42 late arrivals.
Across all 57 participants, aptitude was not predictive of GJ scores
(r=0.13 ns). The reported correlation of aptitude with GJ scores for early
arrivals was not significant either (r=0.07 ns). However, for late arrivals, a
significant positive correlation of aptitude and GJ scores was observed
(r=0.33, p\0.05). DeKeyser had predicted that adult learners would not
The Critical Period Hypothesis for Second Language Acquisition 47
score within the range of early AoA participants unless they had high language
learning aptitude. The combination of: a significant positive correlation of
aptitude and performance among late arrivals, a non-significant correlation
of aptitude and performance for early learners, the performance near ceiling of
early learners, and an examination of 5 (of 6) higher-aptitude late learners whose
GJ scores were within the range of early learners, leads DeKeyser to the fol-
lowing generalization: ‘‘Whereas the younger acquirers in the present study all
reached a native or near-native level regardless of aptitude, only the adults with
above average aptitude eventually became near native’’ (2000: 515). ‘‘Aptitude
plays a role for adult learners’’ (2000: 515) in the sense that, on L2 proficiency
measures, high aptitude trumps, or compensates for, high AoA. Thus, the
basting that sews together the AoA variable and proficiency is the interaction of
AoA and an additional learner variable, language learning aptitude: aptitude
conditions performance among late learners, but not among early learners. This
is a notable finding, to the extent that its interpretation allows for rationalization
of high GJ scores among late learners. However, what is also notable, and what
the DeKeyser study does not adequately investigate in its data, is a clear-cut set
of relationships involving the education variable.
•Years of schooling. With the data provided in Appendix A of the DeKeyser
chapter, I conducted correlations of years of schooling with performance on the
grammaticality judgment task. I found that, over all AoA (n =57), years of
schooling significantly correlate with grammatical proficiency (r=0.45,
p\0.001). Education also predicts GJ scores among late learners (n =42;
r=0.51, p\0.01) as well as among early arrivals (n =15; r=0.78,
p\0.001).
2
With learners separated into aptitude groups, my analysis reveals
that education is again predictive of proficiency. For the 15 high aptitude par-
ticipants, years of schooling correlate significantly with GJ scores (r=0.564,
p\0.05). Likewise, for the 42 low- to average-aptitude participants, education
predicts proficiency (r=0.43, p\0.01). Meanwhile, education and aptitude
are not correlated over all AoA (r=0.03 ns), nor among early arrivals
(r=0.006 ns), nor among late arrivals (r=0.08 ns), suggesting the indepen-
dent contributions of education and aptitude. To summarize, years of schooling
predict GJ results across all relevant correlations. Importantly, unlike AoA and
unlike aptitude, the ‘education effect’ is systematic: significant correlations are
not restricted to certain AoA spans or certain aptitude levels.
The DeKeyser (2000) narrative is about finding a connection between AoA and
L2 proficiency that is consistent with the CPH/L2A. But by framing the study
around the ‘robustness of critical period effects’, the most robustly predictive
factor in proficiency—education—is neglected (see Hakuta et al. 2003 on the role
of education in L2 proficiency over AoA).
2
DeKeyser (2000: 515) erroneously reports that the correlation of years of schooling and GJ
scores is r=0.006 ns, for early arrivals, and r=0.08 ns, for late arrivals. In fact, these reported
coefficients reflect correlations of years of schooling with aptitude; see discussion to follow.
48 D. Birdsong
Researchers in SLA have an interest in knowing what factors account for L2
proficiency in a sampled population. This interest is not limited to explanations of
high-aptitude late learners’ proficiency as a function of assumptions of the CPH/
L2A. A more fundamental concern is accounting for L2 proficiency globally, over
all AoA and over all aptitudes. Perhaps the coat’s palette might include a few
neutral tones alongside the many bespoke hues.
6 Conclusion
The CPH coat of many colors, pointedly so named by David Singleton, has a
history going back to Penfield and Roberts (1959) and Lenneberg (1967). Over the
ensuing years the garment has graced the torso of many a modish scholar. The
present contribution has suggested that a gusset here, a gather there, might mean
the difference between a well-worn coat and one that is worn well.
References
Abrahamsson, N. and K. Hyltenstam. 2009. Age of onset and nativelikeness in a second
language: Listener perception versus linguistic scrutiny. Language Learning 59: 249–306.
Ayer, A. J. 1959. History of the Logical Positivist movement. In Logical Positivism, ed. A. J.
Ayer, 3–28. New York: Free Press.
Birdsong, D. 2005. Interpreting age effects in second language acquisition. In Handbook of
bilingualism, eds. J. Kroll and A. DeGroot, 109–127. Oxford: Oxford University Press.
Birdsong, D. and M. Molis. 2001. On the evidence for maturational constraints in second-
language acquisition. Journal of Memory and Language 44: 235–249.
Carroll, J. B. and S. M. Sapon. 1959. Modern Language Aptitude Test: Manual. New York:
Psychological Corporation.
Cook, V. 2003. Effects of the second language on the first. Clevedon, UK: Multilingual Matters.
DeKeyser, R. M. 2000. The robustness of critical period effects in second language acquisition.
Studies in Second Language Acquisition 22: 499–533.
DeKeyser, R., I. Alfi-Shabtay and D. Ravid. 2010. Cross-linguistic evidence for the nature of age
effects in second language acquisition. Applied Psycholinguistics 31: 413–438.
Fowler, C. A., V. Sramko, D. J. Ostry, S. A. Rowland and P. Hallé. 2008. Cross language
phonetic influences on the speech of French-English bilinguals. Journal of Phonetics 36:
649–663.
Granena, G. and M. H. Long. 2013. Age of onset, length of residence, language aptitude, and
ultimate L2 attainment in three linguistic domains. Second Language Research 29: 311–343.
Hakuta, K., E. Bialystok and E. Wiley. 2003. Critical evidence: A test of the Critical-Period
Hypothesis for second-language acquisition. Psychological Science 14: 31–38.
Hyltenstam, K. and N. Abrahamsson. 2003. Maturational constraints in SLA. The handbook of
second language acquisition, eds. M. H. Long and C. J. Doughty, 539–588. Malden, MA:
Blackwell.
Johnson, J. S. and E. L. Newport. 1989. Critical period effects in second language learning: The
influence of maturational state on the acquisition of English as a second language. Cognitive
Psychology 21: 60–99.
The Critical Period Hypothesis for Second Language Acquisition 49
Lenneberg, E. H. 1967. Biological foundations of language. New York: Wiley.
Long, M. H. 1990. Maturational constraints on language development. Studies in Second
Language Acquisition 12: 251–285.
Ortega, L. 2009. Understanding second language acquisition. London: Hodder Education.
Penfield, W. and L. Roberts. 1959. Speech and brain mechanisms. Princeton, NJ: Princeton
University Press.
Popper, K. 1959. The logic of scientific discovery. New York: Basic Books.
Singleton, D. 2005. The Critical Period Hypothesis: A coat of many colours. International
Review of Applied Linguistics in Language Teaching 43: 269–285.
Stevens, G. 2004. Using census data to test the critical-period hypothesis for second-language
acquisition. Psychological Science 15: 215–216.
Vanhove, J. 2013. The critical period hypothesis in second language acquisition: A statistical
critique and a reanalysis. PLoS ONE. 8(7): e69172. doi:10.137/journal.pone.0069172
50 D. Birdsong
