Clahsen and Felser (CF) deserve praise for their superlative synthesis of literature relating to grammatical processing, as well as for their original contributions to this area of research. CF “explore the idea that there might be fundamental differences between child L1 and adult L2 processing.” The researchers present evidence that adult second language (L2) processing is often less automatic and less efficient than first language (L1) processing. Qualitative differences are suggested as well. Adult L2 processing may be restricted to shallow computations, whereas L1 processing typically involves detailed representations. These conclusions are reached in large part by comparing highly proficient L2 learners with natives on various neurological and behavioral dimensions of processing. I propose that additional comparisons might be carried out that involve an understudied population: learners whose L2 is their dominant language.

Although dominance has been operationalized in terms of the relative personal benefits of L1 versus L2 use (Cutler, Mehler, Norris, & Segui, 1989), more germane to the present discussion, and less controversial, are psycholinguistic metrics of dominance. Psycholinguistically, an individual's L2 is dominant if, compared to this person's L1, performance on a battery of language tasks is characterized by greater speed, fluency, automaticity, or efficiency (accuracy) in processing. These tasks include reading aloud, speeded picture and number naming, and recognition and recall of words under noise (see Flege, MacKay, & Piske, 2002; Golato, 1998; Grosjean, 1998, for discussion of various operationalizations of dominance). Dominance is understood in relative and continuous terms. That is, L2 dominants differ across individuals in the degree to which their L2 is superior to their L1 along the behavioral dimensions just mentioned.

Operationalized psycholinguistically, dominance reflects quantitative differences between processing in the L1 and processing in the L2. Importantly, CF suggest that quantitative differences in processing are not associated with qualitative differences in processing behaviors, a point to which we will return shortly.

Proficiency and dominance are overlapping and confusable constructs, as levels of proficiency and degrees of dominance tend to correlate. However, it is important to maintain logical and functional distinctions between L2 dominance and L2 proficiency. As we have seen, the former is defined in processing terms, whereas the latter is typically viewed in terms of attainment in areas of grammar, lexis, pronunciation, and so forth. One can imagine, for example, an L2-dominant individual whose L2 performance is peppered with “fossilized” nonnativelike morphosyntax, or who scores one SD below the mean of native controls on a given L2 attainment measure. By the same token, one does not have to be L2 dominant to be a highly proficient speaker or writer in the L2.

The incomplete convergence of proficiency and dominance highlights a key methodological concern. To determine the upper limits of L2 processing, we should take a look at L2 learners who do well on processing measures, not just those who do well on proficiency measures. It would therefore be reasonable to target L2 dominants, who have greater speed, accuracy, and automaticity in their L2 than in their L1, and who might be found to outstrip high-L2 proficients in these respects. One could further envision comparing the L2 grammatical processing of L2 dominants and L2-high proficients who are matched for age of acquisition.

As previously noted, CF and others have found both quantitative and qualitative processing differences between high-L2 proficients and natives. Will quantitative differences (e.g., in reading aloud and word recall tasks) be found between L2 dominants and natives? Will qualitative differences be found? For example, will L2 dominants display nonnativelikeness in ambiguity resolution, processing filler-gap dependencies, and ERP negativities in sentence processing? Will CF's shallow-structure hypothesis be a good fit for data from L2 dominants?

These are open questions, but they are not idle questions. Recent studies suggest that data from high-L2 proficients do not tell the whole story about what L2 learners are capable of. For example, Flege et al.'s (2002) study of Italian L1/English L2 bilinguals showed that the pronunciation of L2-dominant bilinguals was indistinguishable from that of natives, whereas both L1-dominant and balanced bilinguals (i.e., individuals who were highly proficient in both languages) displayed detectable accents. This finding led the authors to suggest that interlingual interference effects in pronunciation may be absent among L2 dominants.

A more relevant example comes from Golato's studies (1998, 2002) of word processing among late-learning French–English bilinguals who were proficient in both languages. Golato found that word parsing routines for French and English words varied as a function of the participants' dominant language. English-dominant subjects were able to switch between segmentation strategies and parse both French and English words in the manner typical of monolingual speakers of these languages. French dominants, on the other hand, were not flexible in their routines, applying French-based open-syllable parsing biases to both English and French words. (The study by Cutler et al., 1989, of French–English bilinguals from birth found French dominants, but not English dominants, to be flexible in this respect.) In Golato (1998), processing patterns were shown to be independent of the participants' native language. That is, no matter what their L1, the performance of English dominants differed from that of French dominants. Late L2 learners can become L2 dominant, although as both Golato (1998) and Flege et al. (2002) point out, this outcome is not typical. Late age of acquisition notwithstanding, the results of Golato's research suggest that bilinguals' processing performance in both the L1 and the L2 depends on which of their languages is dominant.

If L2 dominants were included in the study of grammatical processing, what kinds of results might we expect? By the psycholinguistic definition of L2 dominance, we know that processing in the L1 is relatively slow and less efficient than in the L2 across a range of tasks. It would not be surprising if L2 dominants' processing of their L1 also turns out to be slower and less efficient than that of monolingual speakers of that language. In qualitative terms, can we expect L2 dominants' processing of the L2 to resemble that of monolingual native speakers of that language? Given the asymmetries found by Cutler et al. (1989) and Golato (1998, 2002), the answer would appear to be: in some cases, yes; and in some cases, no.

As long as the L1 is actively used, it is reasonable to hypothesize at least some degree or some type of non-monolingual-likeness in grammatical processing among L2 dominants. It is in the nature of bilingualism to find L1 effects in the L2, and indeed to find L2 effects in the L1 (Cook, 2003; Flege, 2002; Grosjean, 1998). However, in the case of L2 dominants where the L1 is unused or infrequently used, one could conjecture that their L2 grammatical processing might turn out to be indistinguishable from that of L1 monolinguals. Monolingual-like processing of the L2 might be more likely still in those extraordinary instances where the L1 is completely forgotten, as suggested by behavioral and imaging data from Korean adoptees moved to Paris and deprived of contact with the Korean language (Pallier et al., 2003). For these individuals, the L2 is not merely dominant, it has effectively supplanted the L1.

Because there are degrees of L1 and L2 dominance, in all the comparisons suggested it would be important to look at data from individuals as well as from groups. Finally, it goes without saying that future work with L2 dominants should involve not only behavioral measures of processing, but should include results from neurofunctional imaging and electrophysiological techniques as well.