Near-native , nativelike or native ? Some terminological and conceptual remarks on L 2 ultimate attainment research

The main objective of this paper is to draw attention to the problem of the inconsistent and potentially confusing use of the terms near-native, nativelike and native with reference to study participants and their level of L2 ultimate attainment in age-related research, as well as to highlight certain conceptual discrepancies pertaining to this issue. A basic analysis of some of the best-known publications in the relevant literature has demonstrated that the terms near-native and nativelike are not always defined in the same way, and furthermore, distinct criteria are applied to select highly advanced subjects for studies which set out to determine whether reaching a nativelike (native) level of L2 proficiency is feasible, thereby supporting or challenging the Critical Period Hypothesis. Such a state of affairs might have serious implications both for the methodology of L2 ultimate attainment research and the interpretation of the corresponding results. It is also argued that adopting a more bilingual-oriented approach to nativelikeness and reconsidering the role of the native speaker as the yardstick for L2 performance may significantly benefit the SLA theory by facilitating the interpretation of the findings and increasing the validity of studies investigating the age factor in L2 acquisition.


Introduction
The issue of the role of the age factor in second language acquisition has been the subject of intense debate for a long time.The question of maturational constraints in mastering an L2 is directly related to the Critical Period Hypothesis (henceforth CPH) originating in the work of Penfield and Roberts (1959) and later developed by Lenneberg (1964Lenneberg ( , 1967)).The CPH predicts that after puberty it is impossible to reach a level of L2 proficiency equal or comparable to native speakers due to the decreasing plasticity of the brain after a certain age and the fact that the process of cerebral lateralization is complete.Since its formulation the original, categorical version of the CPH has undergone a number of modifications and presently researchers postulate the existence of a sensitive period, multiple sensitive periods and even a gradual decline in language attainment from birth over a life span.The CPH, a source of academic controversy, has led to a considerable body of research on the age factor in second language acquisition (SLA), with opponents attempting to refute its core tenets.The main objective of a number of empirical studies was to find counterevidence to the existence of age limitations in L2 language acquisition.It was assumed that identifying late L2 learners -i.e., those with an age of onset after puberty -able to perform within the range of native speakers in several aspects of an L2, would suffice to challenge the CPH.Since the end of the 1980s certain studies have claimed to have found L2 speakers who appeared to defy maturational constraints by reaching a level of proficiency in their L2 comparable or presumably equal to native speakers, e. g.Birdsong (1992Birdsong ( , 2007)), Birdsong/Molis (2001), Bongaerts (1999), Colantoni/Steele (2006), Coppieters (1987), Ioup et al. (1994), Marinova-Todd (2003), Montrul/Slabakova (2003), Moyer (1999), White/Genesee (1996).
In scholarly literature these highly advanced individuals have been termed nativelike or more often near-native L2 speakers (henceforth NNSs).It is widely agreed that early beginners generally perform much better than late learners in terms of L2 mastery, and furthermore, the vast majority of L2 speakers with age of onset after puberty are incapable of achieving a level of proficiency approximating that of native speakers, in particular in the domain of L2 phonetics.This clearly explains why NNSs, some of whom are perceived as L1 speakers of their target language, and hence, pass for native speakers, are a rare phenomenon that has not surprisingly captivated the interest of SLA researchers.
Even though the proficiency of NNSs is frequently seen as compelling evidence against biological constraints in L2 acquisition, and consequently of the CPH, alternative explanations have also been proposed.It is claimed that the CPH does exist, but that in some exceptional cases it is possible to overcome neurological limitations and near-native speakers' levels of L2 ultimate attainment (UA) can be put down to so called compensatory factors of a social and psychological nature.These exogenous factors include extremely high language aptitude (e. g.DeKeyser 2000; Ioup et al. 1994), the quantity and quality of input (Bongaerts 1999;Bongaerts et al. 2000), motivation (Bongaerts 1999;Bongaerts et al. 2000;Moyer 1999) and the readiness to assume a new cultural identity (Schneiderman/Desmarais 1988).
So far no consensus has been reached among researchers regarding the actual role of maturational constraints in NNSs' ultimate attainment.Similarly, there is no agreement as to how psychological and social factors affect L2 acquisition in the case of highly advanced late learners, whether it is a question of a combination of such factors or if some of them are dominant.
The study of the phenomenon of nativelikeness, even if not the key to falsifying the CPH, is still definitely a valuable contribution to ultimate attainment research.Unveiling the reasons for NNSs' success in L2 acquisition, irrespective of whether their proficiency actually equals that of native speakers, would have major implications for both the SLA theory and L2 teaching.
The present article examines terminology applied in literature on L2 ultimate attainment, with a focus on such terms as native, nativelike and near-native, as well as the theoretical assumptions underlying their use (see section 2).This is achieved by an overview of major publications in the age-related research.The following section analyzes the effect of the criterion of indistinguishability on the scale of L2 proficiency and corresponding terminology.Section 4 is devoted to the issue of the use of the native speaker benchmark in UA studies and the implications it may have for further research.Finally, some suggestions are made as to how to integrate several key concepts from the study of bilingualism with investigations into the age factor in L2 acquisition.

2
How far is it from near-native to nativelike?
Although the terms nativelike and near-native L2 speakers are today extensively employed in the literature on ultimate attainment, it seems that both of these notions are rarely explicitly defined or even appear to be somehow taken for granted.Furthermore, it seems that there exists no agreed-upon definition encompassing all the relevant characteristics of NNSs.This is evidenced by the manner in which scholars in the field of SLA refer to NNSs and how the terms nativelike and near-native are operationalized in research, in particular with respect to the criteria used for the selection of participants.It is often the methodological part of a study that actually makes it possible to infer what exactly is meant by the notions nativelike or near-native speakers.
A brief analysis of literature on the notion of nativelikeness indicates that in some cases the terms nativelike and near-native are, in fact, applied synonymously, albeit to a limited extent (e. g.Coppieters 1987;Ioup et al. 1994, Colantoni/Steele 2006).Even if one term is preferred by an author or used exclusively, a comparison of different studies shows that they both actually refer to the same phenomenon.As can be seen in excerpts provided below, the adjectives nearnative and nativelike seem to represent the same meaning.
"Do native and native-like non-native (i.e.near-native) speakers develop essentially identical underlying grammars of the same language?"(Coppieters 1987: 544) "(…) The native-like non-native speaker -hereafter, NEAR-NATIVE SPEAKER (NNS)offers an interesting testing ground for that question, and the opportunity to gain some insight into the problem" (ibd.)"(…) the fact that ultimate attainment can be fully native-like regardless of age of initial acquisition does not mean that it is always so, or that attainment of near-native speaker status is typical" (White 1996: 258-259) In the majority of publications, however, the term near-native constitutes part of various noun phrases denoting a specific category of L2 speakers, whereas the adjective nativelike frequently applies to the level of global language proficiency (or its particular domains) that can potentially be achieved by participants of studies aimed at investigating the upper limits of L2 acquisition (e. g. nativelike competence, nativelike pronunciation, nativelike ultimate attainment, etc.).This difference suggests that the two adjectives do not mean the same thing.In general terms, near-native describes L2 users whose mastery of L2 is close to that of native speakers, yet some deviations from the L1 norm can be identified.The term nativelike, on the other hand, defines a level of language proficiency (potentially) equal to that of native speakers.In other words, near-nativeness would refer to "incomplete" ultimate attainment according to the deficit model (cf.Bley-Vroman 1989), whereas nativelikeness would describe a perfect (complete) end state in L2 acquisition.
Determining if the term near-native L2 speaker is deliberately applied to refer to a specific, rare category of bilinguals who reach extremely high levels of L2 proficiency (yet never equal to that of native speakers) is not an easy task.This is due to the fact that often no explicit distinction is drawn between the terms near-native and nativelike.However, it seems that such a ten-dency has emerged in studies whose findings have revealed subtle differences between the performance of highly advanced late L2 speakers and native controls, even if near-natives' scores were within the range of L1 users (Hyltenstam 1992;Ioup et al. 1994, White/Genesee 1996).Leclercq (2009), for instance, examined the grammatical concept of ongoingness in French near-native speakers of English as an L2, and has found that their use of temporal adverbials and presentatives in narratives is strongly influenced by French (L1), even though the subjects were able to perform in a nativelike manner in certain discourse areas.These findings are congruent with the stance represented by Slobin (2003) andvon Stutterheim (2003), who postulate that grammatical encoding has an effect on speakers' selection of information necessary for verbalization, a process that occurs at the conceptual level.It is also seen as one of the reasons why some very advanced L2 learners are incapable of achieving native speaker competence.
In a similar vein, Bartning (2012), who investigated last stages of acquisition in late oral French L2, argues that there exist certain typical morphosyntactic and discursive features that remain an obstacle in the way of achieving nativelikeness at the very advanced levels of L2 development.These seem to be language-specific and are to a large extent conditioned by the amount of input, which can be significantly increased by prolonged period(s) of residence in the target language community.
Results of studies like the ones mentioned above have led some scholars to the conclusion that nativelikeness is unattainable for L2 speakers with age of onset after puberty.Abrahamsson/ Hyltenstam (2009) assert that the presumed existence of cases of entirely successful L2 mastery reported so far can be attributed to methodological fallacies, in particular, to the lack of sufficiently sensitive research tools (for methodological problems in studies on nativelike L2 ultimate attainment see Long 2005).In other words, it is claimed that if more demanding, multifaceted linguistic tests were administered to scrutinize NNSs, the incidence of nativelikeness would be virtually nonexistent.
Furthermore, starting from the contention that nativelikeness studies are not entirely methodologically valid, and using the results of studies investigating the proficiency of adolescent and adult L2 learners of Swedish (Hyltenstam/Abrahamsson 2003;Abrahamsson/Hyltenstam 2009), Abrahamsson (2012) argues that "so far no study relying on a multivariate test design (including challenging tests and tasks, not just tests of very basic linguistic structures and trivial features) has been able to describe an adult L2 learner who, in every relevant respect, exhibits an L2 proficiency that is fully comparable to that of native speakers" (2012: 192).Consequently, he postulates that L2 learners achieving exceptionally high levels of proficiency should be termed near-native speakers rather than fully nativelike, a position shared by Gregg (1996) and Long/Robinson (1998).
Such an approach shows that a line has been drawn between terms near-native and nativelike, the latter indicating an unachievable end state of L2 development and simply fulfilling the function of a point of reference or a benchmark.Consequently, the notion of near-nativeness has been defined as ostensible nativelikeness, i.e., "levels of proficiency that cannot be distinguished from native levels in everyday spoken communication and only become apparent through detailed linguistic analyses" (Abrahamsson/Hyltenstam 2008: 484).This definition specifies a criterion that frequently appears in research on nativelikeness, namely the (in)distinguishability of NNSs, i.e., the ability to pass as native speakers in L1 conversational interaction.

Indistinguishability as a criterion
The fact that some bilinguals can be perceived as native speakers of their L2 deserves closer consideration for several reasons.First of all, it is a very rare phenomenon, one that proves that NNSs indeed have developed an extremely high level of L2 proficiency in a broad range of domains, varying from pronunciation and morphosyntactic features to pragmatic skills.The aspect of unaccented L2 pronunciation seems to be particularly intriguing and at the same time relevant in this respect.For most native speakers it might take only seconds to recognize nonnative features in their interlocutor's pronunciation, and even if not all L1 speakers are very adept at "carrying out" such an analysis, more often than not "foreignness" is recognized quickly and most probably subconsciously.Instances in which doubts could arise as to whether one is dealing with a nonnative interlocutor or not are not likely to be frequent given that the number of bilinguals who pass as native speakers is estimated to be very limited.
Acquisition of nativelike pronunciation is considered to be one of the central issues in agerelated research and one of the most demanding tasks for late starters (Leather/James 1996;Scovel 1981) as it has been claimed to be the first language domain affected by maturation (Long 1990).Not surprisingly, the overwhelming majority of L2 users will have an identifiable foreign accent although their ultimate attainment in this regard is subject to significant individual variation.Therefore, it seems reasonable to assume that NNSs (defined as individuals indistinguishable from native speakers) might play a special role in studies focusing on the limits of L2 ultimate attainment.Moreover, it is also probable that the level of their language skills is as high as their command of L2 phonology, or even higher.Undoubtedly, further research is needed to demonstrate to what extent such assumptions are justified.Hence highly advanced L2 speakers might indeed be excellent candidates for studies which set out to falsify the CPH and gain a better understanding of the nature of limitations in L2 acquisition.
In studies to date, the criterion of distinguishability has not been approached in a uniform manner -this probably stems from terminological inconsistencies concerning definitions of nearnativeness.However, it seems that an increasing number of scholars have gradually decided to adopt stricter selection criteria for participants in their studies; this is with a view to improving the validity of their research.One such prerequisite was the condition that subjects should not be immediately distinguished from L1 speakers on a variety of measures.According to Montrul/Slabakova, the application of this criterion has shown that a considerable number of NNSs (20-30% or more, depending on the study) perform in the range of native speakers in different domains (2003: 353).The reliability of such estimates might, of course, be affected by both conceptual aspects and methodological factors, such as the requirement of residency in the target community, age of onset or issues related to the selection of control groups (cf.Birdsong 1992; Long 2005).
Even when applied, the indistinguishability criterion can be differently operationalized in various study designs.For instance, Coppieters (1987) selected near-native subjects based on their own judgments and those of L1-speaking friends of participants.Proficiency, on the other hand, was determined by an interview conducted on the basis of the ACTFL 1 scale.These screening procedures were criticized by Birdsong (1992), who argued that such informal ratings would not ensure nativelike performance over a broader range of tasks and could be considered be the result of examination in limited contexts.In a replication (1992) of Coppieters' study, which focused on biographical variables, Birdsong selected participants based on their length of residence in a target community and their self-assessment of L2 proficiency without employing any formal testing tools.Furthermore, it was reported that some of the subjects had detectable foreign accents.Yet White/Genesee (1996) mention Birdsong's study among those meeting the criterion of participants passing as native speakers.
In their own research, White and Genesee set out to examine whether late starters have access to Universal Grammar, and hence, are able to achieve nativelike linguistic competence.The language samples elicited from both L1 and L2 speakers were evaluated by two native-speaking judges on phonological, morphosyntactic, lexical and speaking fluency measures.As a result of this procedure, 45 participants were classified as NNSs.White/Genesee (1996) emphasize that the remaining subjects were highly proficient and if less stringent selection requirements, similar to those from earlier studies, had been applied, these participants might have been categorized as near-native L2 speakers.This clearly suggests that the term near-native has been used by different scholars to refer to distinctive groups of L2 speakers; this obviously has important implications for the comparability and generalizability of results obtained in this line of research.
Screening for nativelikeness by means of native speaker perceptual evaluation was applied in a number of subsequent studies investigating the possibility of attaining nativelike L2 command (e. g.Bongaerts et al. 2000;Colantoni/Steele 2006;Montrul/Slabakova 2003;Moyer 1999).Some researchers, however, advocate neither the application of independent judgments nor participant self-perception.Abrahamsson/Hyltenstam (2008, 2009) postulate examining the actual L2 proficiency of subjects with the application of well-controlled testing procedures and instruments sufficiently sensitive to detect even very subtle deviations from the native point of reference.
This approach is based on the claim described above, according to which late bilinguals who pass as native speakers only appear to be nativelike on the surface, but are actually NNSs -a fact that can be verified by a comprehensive and detailed linguistic analysis.Thus, a distinction has been made between L2 speakers who are perceptually indistinguishable and those who are linguistically indistinguishable.The other category refers "exclusively to L2 speakers whose L2 linguistic competence after in-depth analyses appears to be completely indistinguishable from native-speaker behavior" (Stölten/Abrahamsson/Hyltenstam 2015: 23).This definition implies that linguistic indistinguishability characterizes solely nativelike speakers, and since such individuals do not exist it follows that nativelikeness is a theoretical construct that is the benchmark for unattainable levels in L2 acquisition.A question that arises at this point is whether the terms perceptual and linguistic indistinguishability correspond to the distinction 1 American Council on the Teaching of Foreign Languages.
between performance and competence, which is suggested by the definition provided above.It appears that the actual objective of many UA researchers is to determine if the performance of NNS is indicative of native speaker competence.However, the upper limits of L2 achievement are referred to as e. g. "native-speaker performance", "nativelike competence", "native mastery" or "proficiency" (see excerpts in 2.2), which is another potential source of confusion in NNS-related research.
The aforementioned disambiguation between perceptual and linguistic indistinguishability is directly connected with the model of proficiency levels in L1 and L2 proposed by Hyltenstam/ Abrahamsson (2003).The underlying assumption of this framework is that the level of ultimate attainment declines linearly with the increasing age of onset, without a clear cut-off point marking the offset of a critical or sensitive period.This refers both to early and late starters, the latter being able in some cases to compensate for their deteriorating language-learning ability with non-maturational factors.The bilinguals identified as L1 speakers were classified as NNSs (or seemingly nativelike speakers) -i.e., those L2 learners whose non-nativeness is not discernible in everyday interactions and whose lower levels of attainment indicate the upper limits of perceivable non-nativeness.Only by carrying out in-depth analyses can it be ascertained that the proficiency of NNSs is not nativelike, in other words not identical with the native norm.
The model developed by Hyltenstam/Abrahamsson in 2003 is definitely a valuable attempt to reconcile conflicting observations and the results of studies on nativelikeness due to the fact that it incorporates both maturational and socio-psychological factors in L2 acquisition.Furthermore, it clearly specifies what criteria need to be met to categorize an L2 user as an NNS, which is not a common practice in previously conducted research within this line of enquiry.Even if the model described above assumes that nativelike speakers do not actually exist, a claim which calls into question the results of some UA studies, any proposals aimed at resolving conceptual and terminological inconsistencies in SLA research are bound to benefit this field of study.
Although subjecting potentially nativelike speakers to rigorous scrutiny has its advocates, not all researchers agree that meticulous examination is necessary to define the uppermost limits of L2 attainment, thereby verifying the validity of the CPH.Birdsong argues that "the falsification process should not require data from every imaginable nook and cranny of linguistic behavior" (2005: 322) "until a betraying shibboleth is found" (2005: 322) as not all deviations from the native norm are attributable to compromised learning mechanisms.He also emphasizes that the (non-)nativelikeness standard should be applied in a reasonable manner, and consequently, a line must be drawn to determine what is considered nativelike and what is not.Establishing even more stringent criteria for the close investigation of late L2 speakers might prove detrimental to age-related L2 research.
Conflicting perspectives on nativelikeness are undoubtedly one of the reasons for terminological discrepancies, which in turn must have affected the interpretation of empirical data acquired in this field of research.As can be seen above, it would be difficult to find a consistent definition of nativelikeness that explains what counts as nativelike behavior.Consequently, delineating the definition of an NNS seems problematic, at least to the extent that would allow valid comparisons of study results.It is important to point out that terminological incongruence in UA theory might be related to another issue, namely that of the application of the native speaker norm in L2 research.

The native speaker benchmark
One of the questions addressed by a number of studies investigating maturational constraints in language acquisition is whether NNSs are able to reach nativelike L2 mastery.This implies that the term nativelike is used to describe a level of L2 proficiency identical to that of native speakers, which means that "an entirely nativelike ultimate attainment" (Abrahamsson/Hyltenstam 2008: 484) would result in L2 = L1.The assumption of potential "overlap" shows that what is in fact examined is whether in some rare cases late L2 speakers can "become" native speakers in each and every aspect of their linguistic knowledge and skills.
The fact that non-native proficiency is often seen as potentially equal to that of native speakers is reflected, for instance, by the scale of proficiency levels in L1 and L2 proposed by Hyltenstam (2016: 6).The L2 scale has been divided into two parts -one covering the non-native proficiency (from low to advanced) and the other near-native and nativelike proficiency.The last category describes "proficiency levels among L2 users who are, in fact, equal to those of native speakers" (Hyltenstam 2016: 5).
The concept of near-nativeness as non-perceivable non-nativeness has been adopted by Bartning (2012), who agrees that in-depth examination can identify linguistic domains that distinguish NNS from native speakers.In contrast to Hyltenstam (2016), however, an additional category of L2 learners is proposed as part of an acquisitional continuum (Bartning/Schlyter 2004), namely that of functional bilinguals.Such an L2 user has been defined as "a speaker who has a very rich, elaborated L2 -e.g. native-like quantity of formulaic language, constituents of PAs, complex syntax/discourse, (…) acquired during important LORs in the TL country, but who is still perceived as non-native" (Bartning 2012: 182).Thus, functional bilingualism would refer to an intermediate stage of language development prior to reaching near-nativeness.
Following the logic that NNSs are not capable of becoming nativelike due to minor differences in their performance compared to native speakers (as can be proven by rigorous scrutiny), it would seem that the highest level in the L2 scales described above is actually non-existent, and as such is only a theoretical construct.Furthermore, the nativelike level is equated with the native level, which raises the question of whether such a point of reference for L2 is needed at all given that only apparent nativelikeness exists in real language use.Such a way of reasoning leads to one more question, namely whether the term nativelike actually means native and is applied only to show that we are dealing with a potential native level of L2 proficiency in the case of speakers born with a different L1, or whether it is used based on the underlying assumption that NNSs are doomed to "failure", unable to reach a command of their L2 equal to that of native speakers; this would consequently mean that labeling their upper limits of attainment as native would be inadequate or simply pointless.In other words, an L2 speaker might be nativelike because they are by definition not a native speaker but achieves native levels of proficiency, or it may be that they are nativelike because it is possible only to approximate the native standard, even to the extent of not being indistinguishable from L1 speakers in their L2 performance.
The issue of the precise level of proficiency specified by the term nativelike, i.e., examining whether L2 proficiency of NNSs equals native speakers' L1, or only assessing to what extent it deviates from the native speaker norm, can be observed in the way in which the end of the proficiency continuum is defined.The examples presented below, taken from some of the bestknown studies in the line of research discussed in this paper, show that the L2 end state of an NNS is referred to as either native or nativelike (in some cases both are used): "The results show that it is possible to attain a native level of proficiency 2 for learners who start acquiring an L2 after puberty, even for learners with a typologically distant L1" (van Boxtel et al. 2005: 355) "One of the central questions in research on second language (L2) acquisition is whether it is possible for late L2 learners to reach native-like ultimate attainment in the L2.Early studies addressed this question by investigating whether L2 speakers could reach native-speaker performance in general L2 proficiency" (Hopp 2009: 463).
"(…) after which adult L2 learners are not likely to attain complete nativelike competence in the L2 phonology or morphosyntax" (Montrul/Slabakova 2003: 352) "In this paper, we provide further evidence that L2 learners may achieve native-like mastery of target language phonetic properties" (Colantoni/Steele, 2006: 59) "(…) the putative inability of late learners to attain native-like competence in an L2" (White/Genesee 1996: 235) For the reasons mentioned above it is not always clear whether this inconsistency in defining the potential upper limits of ultimate attainment is the result of divergent theoretical assumptions in various studies (the potential outcome L2 = L1 vs. L2 ≈ L1), or is instead random in nature, evincing a lack of unified terminology.For instance, in an oft-cited study Ioup et. al (1994) refer to the participants' level of global proficiency and particular domains either as native or nativelike.
"The study presents the results of extensive testing of an adult who has apparently acquired native proficiency in Egyptian Arabic (EA) in an untutored setting" (Ioup et al. 1994: 73).
"This task provided good examples of Julie's ability to manipulate EA discourse markers and pause fillers and to adapt them to the situation.For example, the recorded sample of her speech exhibited nativelike use of conventionalized language such as malh wi filfil wi kammuun" (Ioup et al. 1994: 80).
Ioup et al. concede that although Julie, one of the participants, achieved a level of proficiency within the native speaker range on a number of measures, there were still areas (tasks dealing with discourse semantics) in which departures from the control group attainment were observed.However, nativelike is not applied here to refer to a performance below the native 2 Emphasis is mine in all quotations.
benchmark.It appears to be used in a similar way to the term native.One could speculate, however, that nativelike in the extract below means that Julie was perceived as native, but since Arabic was not her L1, she could only receive the label nativelike; obviously, this is not how she was identified by the assessors.It seems to have instead been used as a synonym; replacing the adjective nativelike in the context presented below with native would not affect the meaning of the sentence whatsoever, other than in purely stylistic terms."Julie and Laura were rated as native speakers by 8 of the 13 judges (62%), with 6 judges considering them both to be nativelike and 2 rating one but not the other as a native speaker" (Ioup et al. 1994: 80).
Irrespective of terminological and theoretical differences, it is evident that the point of reference applied in ultimate attainment studies is that of native speaker proficiency.On the one hand, this issue could be considered irrelevant -whatever the theoretical model, L1 control groups are still needed to determine the position of NNSs on the proficiency scale in relation to native speakers, thereby indicating the upper limits of L2 attainment.On the other hand, application of the native speaker criterion might have certain consequences of a terminological and conceptual nature for L2 research.
There is no doubt that both terms near-native and nativelike were coined in relation to nativelevel proficiency as the yardstick for an L2 end state.As stated above, the question is whether the underlying hypothesis is that exceptional late L2 users can reach at best a nativelike level (assuming the difference between nativelike speakers and NNSs) or actually a native level (can become native speakers).I would argue that in both cases the objective is to verify whether L2 proficiency equals that of native speakers.Such an approach might, however, lead to certain contradictions.
First of all, non-native speakers by definition, whatever their level of proficiency, differ from native speakers due to a constant interaction between their languages and the effect that each of them exerts on the other (see 3 below).Therefore, using the term native to describe the point of reference in UA research could be misleading as it presupposes that we are comparing two possibly identical types of users of a given language.Obviously, non-native and native speakers are not and cannot be the same, which means that differences (deviations from the native norm) are inevitable.Expecting that the research results will reveal a 1:1 correspondence might be a misconception in many cases.
Secondly, assuming that NNSs are incapable of achieving a native-equal level of proficiency, it would be justified to use a term defining their upper level of attainment (below the native benchmark), such as nativelike.The suffix -like, used to form adjectives from nouns, might suggest a feature that is typical or characteristic of a given object, phenomenon, person, etc.However, it might also imply that certain features are shared but that we are still dealing with something that is not the referent, i.e., the morpheme on the basis of which the -like adjective was created.In such a case, the very use of the term nativelike speakers may -by definitionbe interpreted as actually being not native.Therefore, using the term nativelike interchangeably with the adjective native to indicate the potential level of ultimate attainment of L2 speakers would be tantamount to applying two different points of reference.
Depending on the position held by the researcher as to whether nativelikeness is attainable or not, employing the term nativelike might have its advantages, which is why discarding it entirely might be unwarranted.However, it can also be problematic, inasmuch as this notion is semantically based on the concept of the native speaker as the benchmark.Thus, in a sense, L2 and L1 speakers are assigned to one category, as a result of which bilinguals might not be seen as independent entities with essentially distinct characteristics.Perhaps it is only a matter of adjusting the angle of the approach to nativelikeness.For instance, the objective of a number of studies could be to identify and explore differences instead of attempting to prove that for various reasons NNSs are not native speakers.
Obviously, this is by no means to say that native control groups should not be used in this line of research.What is suggested is rather a change in perception that could benefit studies focusing on the age factor in L2 acquisition by providing a better insight into the nature of differences between near-native/nativelike and native speakers found in research to date.Furthermore, adopting a more bilingual-oriented perspective could also have an effect on the basic assumptions underlying the conceptual framework of studies on nativelikeness.

A bilingual-oriented perspective
With regard to the terminology denoting the subjects of UA studies, it is interesting to note that the term bilingual does not appear frequently in the relevant literature.Definitions of bilingualism may differ, especially in terms of the criterion of the level of L2 proficiency, however, today most researchers specializing in bilingualism would qualify NNSs as bilinguals.This fact most likely indicates that the approach developed in the field of bilingualism, inextricably linked to issues of L2 ultimate attainment, is not always taken into consideration.I would argue that CPH-related research could greatly benefit from incorporating certain theoretical assumptions proposed by scholars of bilingualism, which may be a valuable contribution to the debate on nativelikeness.
In contrast to lay perception of bilingualism, the academic approach to the phenomenon has significantly changed in recent decades (Zubrzycki 2018).Initially it was based on the idealized model of L1 speaker, which goes back to the Chomskyan notion of the monolingual native speaker demonstrating perfect knowledge of his/her language (Chomsky 1965) .This paradigm was widely applied in testing and measuring bilinguals' language skills.Gradually the concept of bilingualism has become much broader (cf.Mackey 2007) and the most recent definitions focus on the use and not any more on the proficiency of bilinguals (cf. Hakuta 1992, Grosjean 2008).
The above-mentioned changes can be attributed to postulates according to which the perception of a bilingual as two monolinguals in one (the fractional view vs. holistic view) is misguided (Grosjean 2008).Such claims have evolved mainly based on the observation that a bilingual's languages are in constant interaction, L1 affecting L2 and conversely.As a result, it is suggested that bilinguals constitute a unique type of a speaker-hear with a complete, integrated language system.In line with this argument, typical aspects of bilingual behavior, such as code-switching, are not considered deviations from the monolingual norm, but a direct consequence of a bilingual's linguistic repertoire.Cook (1991Cook ( , 1997Cook ( , 1999Cook ( , 2008) ) proposes the notion of multicompetence, defined as "the compound state of a mind with two languages" (1999: 103), which suggests that L2 speakers ought to be treated as independent, multicompetent L2 users.This is due to the fact that bi-and multilingual minds are subject to the interaction of L1 and L2(Ln), which manifests itself in the multi-directional transfer between two or more systems.The effects of L1 transfer to L2(Ln) appear to have received far more attention from SLA scholars, however, research (e. g.Latomaa 1998;Pavlenko/Jarvis 2002;Stoessel 2000;Zampini/Green 2001) has shown that L2 might also influence L1 in various domains of language use, a process that becomes increasingly salient in cases where gradual L1 attrition can be observed.Cook (2002) argues that the languages in the mind of an L2 user are neither completely separated nor completely integrated, and hence, proposes an integration continuum that encompasses various degrees and types of interconnection.Such an interactional perspective, which seems to be of central importance in the study of bilingualism, is adopted by only a few studies on nativelikeness (cf.Belletti et al. 2007;Hopp 2009).
Another concept that emphasizes the differences between bilinguals and monoglots is the dual or multilanguage model developed by Kecskes (2010).Although it shares some similarities with Cook's multicompetence, its main focus is on the conceptual instead of the linguistic system of a user.In other words, bi-and multilinguals possess a consolidated conceptual system, which results in significant differences in linguistic behavior in comparison to monolinguals.Therefore, the acquisition of a new language presupposes not only quantitative but also qualitative changes.Kecskes affirms that "multilingualism primarily appears to be a conceptual rather than a linguistic issue.It is not the languages that are integrated but the conceptual systems.Bi-and multilinguals have one conceptual system that operates two or more language channels" (2010: 106).Adding a new language to the conceptual base entails not only the acquisition of a new linguistic code, but, above all, changes in the existing system of socio-cultural knowledge encoded in the L1 and formed during L1 socialization.
The approach proposed by Kecskes can be viewed as an interesting attempt to account for lower levels of L2 ultimate attainment in late bilinguals, as well as the nature of differences between advanced L2 users and native speakers.L2 users cannot go through the same stages of L2 acquisition, as is the case for L1 development combined with socialization.The lack of prior sociocultural experience tied to L2 has direct implications for target language use, as can be observed in the domain of pragmatics.Unfortunately, the pragmatic skills of NNSs, in particular those passing as native speakers, have not been extensively investigated in a context that contrasts them with advanced L2 learners.The issue of (double) socialization and secondary acculturation is closely related to the view that exogenous factors might be responsible for the exceptional success of NNSs in L2 acquisition.For this reason, it is fair to say that this question should receive more attention from UA researchers, particularly those who assume that nativelike mastery of an L2 is not feasible, even for early bilinguals.The complex interrelationship between factors of a social/psychological nature and age of onset is still fraught with question marks.
Not all researchers agree that the inability of L2 users to attain native levels of proficiency can be accounted for by biological limitations.As mentioned in the introduction, contrary to claims that early L2 beginners are highly likely or bound to achieve near-nativeness/nativelikeness, minor differences between young bilinguals and native speakers have been found, both at the morphosyntactic (see 1) and phonological level (e. g.Flege 1999;Flege et al. 1997;Piske et al. 2001).These findings led some UA researchers (Cook 1995;Flege 1999;Grosjean 1982;Ioup 2005) to assume that the role of maturational constraints might not be as important as simply the fact of being bilingual and the above-mentioned differences may arise from what Birdsong (2005: 324) describes as "artifacts of bilingualism".
Based on this line of reasoning, as well as issues arousing controversy over the notion of the native speaker, Singleton/Muñoz (2011) argue that the application of the monoglot nativespeaker yardstick in the study of L2 ultimate attainment may be misguided (see also Singleton 2012).Instead, it is suggested that a comparison between early L2 starters and post-pubescent L2 learners would be far more adequate.This proposal deserves full consideration as it would enable SLA researchers to compare bilinguals with other bilinguals, thereby avoiding the monolingual bias and acknowledging the integrity of L2 users.
As pointed out by Cook, "the monolingual perspective in essence restricts the field of SLA research to enumerating the similarities and dissimilarities between L2 users and native speakers" (2016: 3).It is therefore argued here that the perception of NNSs as bilinguals in their own right might facilitate developments in future research, especially in terms of methodology, if such aspects as language mode, language dominance and L1 attrition or the interaction of L1-L2(Ln) are taken into account to a larger extent in study designs.

Conclusion
The main objective of this article was to highlight terminological inconsistencies in research on ultimate attainment, an issue which is of crucial importance for further research investigating the upper limits of L2 acquisition and the role of maturational constraints in this process.As shown in the present article, a lack of uniform terminology carries major implications for methodology employed in nativelikeness studies, as well as the interpretation of their results.Consequently, such a status quo hinders the collection of valid empirical data that could ensure a higher degree of generalizability of research findings; this is all the more so as the population of near-native/nativelike speakers is rather limited, in particular if the criterion of indistinguishability is applied.
Another issue which might need reconsideration is the validity of the criterion of native speaker and its perception in research on L2 ultimate attainment, especially in the light of the results of studies claiming that nativelikeness is unachievable, as well as terminological and methodological issues which might undermine such findings.Likewise, it seems that the phenomenon of indistinguishability plays a significant role in defining near-nativeness, and hence, necessitates further investigation and possibly a novel approach.
Finally, it is also argued that adopting a broader perspective which to a greater extent integrates theoretical assumptions originating from the field of the study of bilingualism, such as the interaction between L2 and L1, as well as intercultural communication might be advantageous to building the theory of UA, and as a result could contribute to the understanding of some of the most complex issues addressed by SLA research.