Organic Grammar as a measurement of development Martha Young-Scholten University of Newcastle (England) Colleen Ijuin Georgia Perimeter College Anne Vainikka Johns Hopkins University Contact address Martha Young-Scholten School of English Literature, Language and Linguistics University of Newcastle Newcastle upon Tyne NE1 7RU England daytime phones +44-191-222-7751/334-3015 evening phone +44-191-386-5130 fax +44-191-222-8708 martha.young-scholten@newcastle.ac.uk/martha.young-scholten@durham.ac.uk Biosketches Martha Young-Scholten’s work ranges from the adult second language acquisition of morpho-syntax and phonology to unschooled adults who learn to read for the first time in their second language. She is senior lecturer and an MA director at the University of Newcastle in England. Fluent in Arabic, Persian and Japanese (as well as her native English), Colleen Ijuin teaches ESL at the Dunwoody campus of the two-year George Perimeter college system where she is leading efforts to find better placement and promotion tests. Anne Vainikka is a researcher at Johns Hopkins University. She works on theoretical syntax (primarily Finnish) and language acquisition, and has co-authored a series of articles with Martha Young-Scholten on the L2 development of morpho-syntax. Organic Grammar as a measurement of development Abstract (new – please read) This paper reports on a study applying a combined measure of morphological and syntactic complexity to assess the written proficiency of 47 community college ESL students at three program levels. Based on oral production data from both child and adult second language learners, the theory of Organic Grammar (OG) holds that morphosyntactic development is primarily driven by innate linguistic mechanisms. Under OG, route of development is common across learners from different native language backgrounds, and proceeds in stages defined by both syntactic and morphological criteria. We claim that these criteria can straightforwardly be applied as an assessment tool. Although similar syntactic criteria were proposed by Pienemann et al. in 1988, even subsequent refinements of their Rapid Profile (e.g. Pienemann 2001) have involved assessment of oral production entailing special training. Based on our written production data, we make three claims: (1) a tool involving both inflectional morphology and syntax provides finer grained delineation of stages of development; (2) these stages translate into ESL program levels; (3) application of a limited set of clear criteria to written samples dispenses with the need for any special training, indicating potential for adoption by classroom practitioners. INTRODUCTION In this paper we explore the usefulness of morpho-syntactic analysis as a tool to assess the written production of adult English language learners. While this is by no means the first such attempt, the method of analysis we adopt is distinctive in its application of both morphological and syntactic complexity to the assessment of second language production. We begin our paper by discussing current practice in the assessment of writing by English as a second language (ESL) learners and suggest that the prevailing view is that neither holistic evaluation nor discrete item tests are adequate. We then move on to a review of the research findings on the second language acquisition (L2A) of inflectional morphology and syntax that underpin the assessment tool, based on the theory of Organic Grammar, which our study investigated. In so doing we take up the challenge in TESOL Professional Papers #5 to base assessment ‘on current educational research, especially in the areas of first and second language acquisition’ (Principle 2 of 8; 1998:7). The heart of our paper is a case study of 44 adult ESL learners at a US community college from whom written samples were obtained during Semester 1. Applying Organic Grammar stage criteria to the data we show that assessment is accurate for 42 of these learners in terms of their program level in Semester 2. ASSESSMENT in ESL Qualitative assessment in ESL writing Two diametrically opposed types of assessment seem to exist in ESL assessment with little in the middle. At one end of the spectrum are norm-referenced, standardized discrete item tests (e.g. parts of the TOEFL) and at the other end the practice of qualitatively assessing written or spoken samples. The latter constitute standardized measures only when highly trained assessors are used as for the IELTS. Because we are interested in assessment in the type of ESL context from which our written samples come, we only consider here the literature on the qualitative assessment of written production by teachers rather than trained assessors. Brown (1998) is a volume presenting examples from programs around the world of the qualitative assessment of second language learners. The following terms appear in connection with the ranking of written production (all italics added): 1) Terms used in holistic assessment the students have ‘adequately described the situations…’ p. 39 ‘words are phrases combined effectively’ p. 43 ‘words and phrases exhibit good patterns of language’ p. 43 ‘focus your evaluations on the quality of the sentences they make’ p. 162 Morpho-syntax appears as the explicit focus of assessment, where accuracy is evaluated using the following relative terms (in italics) in connection with learner level (p. 180): 2) elementary = very limited grasp of grammatical and relational patterns intermediate = limited grasp of grammatical and relational patterns upper intermediate = moderate grasp of grammatical/relational patterns advanced = competent grasp of grammatical and relational patterns special purpose = confident and generally accurate use In rating the intelligibility of a sample of writing, the terms adopted are also usually relative rather than absolute (p. 181), where the learner demonstrates: (3) simple meanings basic meanings familiar topics can be handled, but not detailed explanations competent communication communicates well in general and specialised areas When errors are the focus, this appears to situate assessment at the discrete item testing end of the spectrum, yet the use of relative terms belies this. 4) under spoken grammar (p. 364) unacceptable: no control, except with simple stock phrases adequate: some control of basic constructions yet major or repeated errors good: generally good control, with some grammatical errors excellent: sporadic minor grammatical errors Our study is based on the conclusion that there is a need for a classroom assessment tool that goes beyond holistic evaluation yet does not penalize learners for committing errors in the process of complexifying their morphology and syntax. There other alternatives similar to what we will propose here: Pienemann, Johnson and Brindley’s (1988) Rapid Profile is one such tool, and a more recent proposal by Kozloski and Donahue (2005) is another. However, a syntax only measure such as Pienemann et al.’s is less useful, and Kozloski and Donahue’s combined measure is more complex than required. Current assessment practices in one ESL program The students from whom our writing samples were obtained represent a population of ESL learners typical of those attending community or junior colleges in the USA. Recent demographic data indicate the following status of ESL learners at the metropolitan area community college system from which our sample comes: 58% permanent residents; 18% student/F-1 visa holders; 15% naturalized citizens; 8% visitor/on non-immigrant visa holders; 2% refugees. Citation for these data. Representing a range of different language learning and academic histories, the ESL program attracts (1) students who will have completed secondary school in their native country; (2) students who will have completed secondary school in the USA, but who were not born in the USA and for whom English is not their native language; and (3) students whose secondary education was compromised due to the circumstances that led to their immigration. For admission to the community college system, non-native English speakers must meet a threshold, as determined by the college’s board of regents: either 330 on the SAT verbal component or 460 on the TOEFL. An applicant with an SAT verbal score below 480 (certainly below 460 – my typo, probably) must take a college-designed (Correct Colleen?) ESL placement test whose components are a reading task and a timed writing sample; there is no oral component. Progression through this ESL program involves at the two intermediate levels and an advanced level. Courses (three at the intermediate level and two at the advanced level) are context based. Promotion from one level to the next level is by the program-internal assessment of a written sample along with a reading test. need to say a bit more about the 2004-2005 departmental assessment measures – esp. the reading This ESL program illustrates both ends of the spectrum with respect to the assessment of written production. At the quantitative end, assessment of the written sample is in terms of attempts vs. errors, and at the qualitative end, the same relative terms shown above in (1) – (4) such as ‘control’ are used. Teachers on the program have observed a situation where based on their scores, students who demonstrate the ability in the classroom are not promoted to the next level based, and students lacking the ability are prematurely promoted. Considerable variation exists within the two intermediate levels, where teachers on the program express concerns not only about the assessment used for progression but also for initial placement. We propose that research findings from the last several decades from the study of second language (L2) acquisition can fill the gap between discrete-item testing and holistic evaluation.1 L2 DEVELOPMENT Since Dulay and Burt’s (1974) and Bailey, Madden and Krashen’s (1974) landmark study of the development of grammatical functors or morphemes by child and adult ESL learners, respectively, numerous studies of the acquisition of various second languages have pointed to the conclusion that all L2 learners follow a predictable route of development largely independent of the learner’s age at initial exposure, their native 1 One might ask why if these research findings are indeed pedagogically relevant, ESL professionals rarely hear about them. We believe the answer is sociological. Throughout the 1970s, researchers working on second language acquisition felt obliged to consider the pedagogical implications of their findings. But since the 1980s, the establishment of (i) separate journals such as Second Language Research and (ii) separate conferences such as GASLA (Generative Approaches to SLA) along with (iii) the increasing location of MA TEFL programs outside of departments of linguistics have led to the current situation, where SLA research findings are neither readily available nor accessible to outsiders. While much of the SLA research carried over the last 25 years has initially had no obvious pedagogical implications, when findings turn out to be robust nowadays they fail to get conveyed because the old paths of communication are no longer there. language, the type of exposure or educational background (Hawkins 2001). This and other studies carried out at the time involved the analysis of errors in terms of absence or suppliance of forms in so-called obligatory contexts as in Lily eat_ scrambled egg_ for breakfast where the two obligatory contexts are for third person singular –s and plural –s. This idea originated in Roger Brown’s (1973) longitudinal study of three children learning English as their first language who were found to develop 14 grammatical functors, ranging from the bound morphemes –ing and third person singular –s to the free morphemes copula be and articles, in a common order despite external differences such as environmental input.2 Based on a cross-sectional study by deVilliers and deVilliers (1973) confirming Brown’s longitudinal order, Dulay and Burt’s and Bailey et al.’s crosssectional studies found a common order in L2 English based on learners’ suppliance of grammatical functors in obligatory contexts. Along with 1970s studies of English learners’ development of negation and question formation (see Hawkins 2001 for a review) were studies of European studies syntactic development by adult guest-workers and immigrants acquiring target in addition to English languages. These studies – always of oral production data collected through interviews/conversation or elicitation tasks such as picture description and narrative (e.g. silent film) retelling - have driven debate in adult second language acquisition research. Table 1. Major adult second language acquisition studies Study L1 and L2 learners type of study Bailey et al. 1970s 2 L1 Spanish and 11 other L1s L2 English 73 adults crosssectional Issues/hypotheses introduced Interlanguage and Error Analysis replace the Contrastive Analysis Thirty years on the conclusion that no fundamental differences exist across children in their development of morphology (and syntax) still stands. Cazden et al. 1970s L1 Spanish L2 English 2 children 2 adolescents 2 adults 45 adults→ ZISA 1980s (Clahsen & Muysken; Pienemann) L1 Spanish, Portuguese and Italian L2 German ESF 1990s (Klein & Perdue; Schwartz & Sprouse) LEXLERN 1990s (Clahsen, Vainikka & Young-Scholten) VYSA 1990s/2000s (Vainikka & YoungScholten) six L1s five European L2s 40 adults 2 ½ years longitudinal L1 Korean, L1 Turkish L2 German 17 adults crosssectional L1 English L2 German 3 adolescents 1 year longitudinal 12 adults→ 10 months longitudinal crosssectional 2 years longitudinal Hypothesis Social factors (Schumann’s Pidginization Hypothesis) Do adults and children use same mechanisms? Universal Grammar Access vs. Fundamental Differences; Processability The Basic Variety; Full Transfer/Full Access Role of L1 reconsidered: Minimal Trees; significance of morphology: Structure Building Organic Grammar Linguistic development under Organic Grammar In the spirit of Bailey et al.’s ideas on a common path of development as determined by production of certain grammatical morphemes, and based on further ideas from the study of first language acquisition (e.g. Clahsen (1990); Radford (1995), Vainikka and YoungScholten (1994) posited that the second language learner’s initial state of development involves only the basic syntactic relations obtaining between the non-finite verb and its complement, that is the syntactic phrase VP. They claim that as the L2 learner develops, s/he ‘builds up’ syntactic structure much like children do, based on the interaction between the ambient, linguistic input and the learner’s internal, innate linguistic mechanisms (i.e. Universal Grammar; Chomsky 1981). Under Organic Grammar, inflectional morphology emerges in connection with syntax.3 Across the columns in 3 In Clahsen and Muysken (1986), Klein and Perdue (1997) and Pienemann (1998) the development of L2 German is presented without this well-attested early L1-based stage. Schwartz and Sprouse (1996) on the other hand, claim that the entirety of the learner’s L1 constitutes the initial state and view subsequent Table 2, development is shown to be a process of increasing morphological variety and complexity and that parallels increasing syntactic complexity in the target language (TL). Table 2. Organic Grammar: Criteria for stages in L2 English Stage word order types of verbs verbal pronouns agreement; tense marking initially thematic (main) none subject, 1 resembles that verbs only object of the native pronouns language absent no longer thematic verbs; none pronouns 2 resembles the copula ’is’ appears forms begin native to emerge language declaratives thematic verbs, no agreement; more 3 resemble the modals; copula some tense, pronoun target forms beyond ‘is’ some aspect, forms, but language but not they can productive still be missing resembles the thematic verbs, productive pronouns 4 TL apart from modals, copula tense, aspect; obligatory, complex forms beyond ‘is’ some ‘there’ and syntax and auxiliaries in agreement, existential all forms and tenses esp. forms of ‘it’ emerge emerge ‘be’ always complex tense, forms are use of 5 resembles the aspect forms; usually ‘there’ and TL passives; range of correct, apart ‘it’ beyond thematic verb, from newly stock modal, auxiliary attempted phrases forms ones complex syntax None formulaic or intonationbased Qs formulaic whQs; yes/no Qs w/o inversion; conjoined clauses productive wh-Qs, but all Qs may lack inversion; simple subordination all Qs with inversion; complex subordination One can observe that beginning L2 learners both in naturalistic and classroom contexts produce verb-less or single word utterances (see Myles 2005). These utterances could be said to reveal an initial stage of development much like the child’s one-word stage, but development as a process of overcoming L1 influence. They offer no explanation for the well-attested emergence of inflectional morphology. about which little can be said regarding syntax or morphology. Stage 1 is characterized by the production of multiword utterances, along the lines of the young child’s two-word and ‘telegraphic’ stages. Under Organic Grammar, this stage is also referred to as the ‘minimal tree’, and it also involves a sub-stage when the learner’s native basic word order within the VP (e.g. object – verb vs. verb-object) does not match that of the target language tree. At this stage, L2 learners, especially instructed ones, also produce longer memorized (unanalyzed) chunks such as My name is X (see Myles 2005). Data from a Japanese boy acquiring English (Yamada-Yamamoto 1993) show the influence of Japanese at his earliest stage of syntactic development. In Japanese, the object precedes the verb, while in English it follows the verb. Hence this boy’s first minimal tree in English displays Japanese word order. After several months of additional English input, the boy reaches a second sub-stage where his minimal tree switches to English word order, with the verb preceding the object (see also Hazendar’s 1997 similar findings from a Turkish boy learning English). At both sub-stages, the boy produces nonfinite forms, either (presumably, but see below) bare forms like ‘eat’ or participles like ‘eating’. (6) Stage 1a: Japanese object-verb (OV) order bread eat bananas eating Stage 1b: English verb-object (VO) order eating banana wash your hand The rich verbal morphology of German allows us to observe much more clearly than in English the morphological development that parallels the L2 learner’s syntactic development from these simple minimal trees with their non-finite forms to more complex syntax and elaborated morphology. In fact, the richness of German morphology along with the variety of verb positions propelled research on the L2 acquisition of German into prominence, as shown in Table 1, where four out of the six major studies were either entirely on German or included German. Data from a smaller, cross-sectional study by Dimroth (2002) of immigrants in Germany includes eight Stage 1 learners. Story retelling data from the Russian adults provide a mirror image of the Japanese learners of English where in the least advanced Russian-German inter-language, the minimal tree is Russian, i.e. VO (speaker R31), and in slightly more advanced Russian-German inter-language, the tree becomes the German OV (speaker R10).4 Unlike English, where only third person singular is marked for agreement with the verb (and all other present tense main verbs resemble bare forms), German has a full paradigm, allowing the researcher to conclude that the main verbs in (7) are non-finite (in this case infinitive) forms, showing no agreement with the (intended) subject. (7) Stage 1a Gehen Wald. (learner R31) go –INF forest [Er geht in den Wald.] ‘He's going into the forest.' Stage 1b 4 German has mixed headedness in its syntax and differs from Japanese in that finite verbs appear in second position in matrix, declarative clauses, as in the target utterance in (7) Er geht in den Wald. These facts about German have made it an extremely interesting language to study the development of. Rote man Bier trinken. (learner R10) red man beer drink -INF [Der rote Mann trinkt Bier.] ‘(The) red man is drinking beer.' Under Organic Grammar, after the learner’s initial reliance on his/her native language, inflectional morphology and syntax begin to develop following a common order for all learners of a given language. The syntactic development of all language held to be driven by internal linguistic mechanisms, in response to the linguistic input the learner hears (or reads). Crucial for the application of these stages to the assessment of any L2 learners is the idea that development of inflectional morphology entails the development of the syntax associated with that morphology. The examples in (8) from a cross-sectional study of primarily low-literate Somali-speaking learners of English (Young-Scholten and Strom 2004) illustrate further, post-minimal tree development in English, where functional morphology begins to emerge with the development of syntactic complexity starts to develop. At Stage 4 additional functional morphology emerges along with simple subordination involving relative clauses and complementizer clauses without that. These clauses are adjuncts – their absence does not result in an ungrammatical sentence. And at Stage 5, there is a range of productive functional morphology and complex syntax, including purpose clauses (adjuncts with complex internal structure), as well as complementizer clauses with that and clauses following seem. The latter two clause types are arguments of a higher verb (He thinks that….); their absence results in an ungrammatical sentence. Passives are included here, too, as they involve a displaced element: the object becomes the subject. Stage 5 also includes wh-object questions such as What did you see? and inverted yes/no questions. Importantly, the production of a new form does not equate with its productivity, one gauge of which is the use of a form to consistently distinguish a different meaning (see e.g. Wagner-Gough 1978 on form use without meaning distinction in child L2 acquisition). Continued non-target use of forms at Stage 4 and Stage 5 may partly be the result of the absence of a particular grammatical feature in the learner’s first language. Hawkins (2001) claims such features (e.g. articles for Chinese learners of English and grammatical gender for English learners of German), are not acquirable by adult L2 learners.5 (8) The initial functional syntax stage (Stage 3) The woman is cry. auxiliary, without –ing Because too bad. subordinating conjunction, without a verb Elaborated functional syntax (Stage 4) Someone’s die because he have accident. present perfect, -ed missing productive simple subordination Car hit the kid that’s lie down on the street. progressive, -ing missing subject relative clause Target-like functional syntax (Stage 5) The young boy was having fun with his bike. past progressive. When you reverse, you have to see anybody behind. complex subordination He doesn’t did that. variety of forms of do, confusion of tense/non-finite forms 5 In testing this hypothesis, it appears that it is not borne out. Adult German learners of Spanish, for example, have problems with gender marking in that language. Our own data (see below) suggest that feature absence simply increases a difficulty already present for all adult second language learners (also proposed by Zobl 1980). If this observation is correct, this may contribute to the necessary elaboration of Hawkins’ (2001) Modulated Structure Building, which allows for some continued first language influence. These examples reveal a further characteristic of development: it is not linear. As new forms and structures emerge, they often destabilize the learner’s current grammar, resulting new errors, and apparent backsliding. Every set of examples in (8) reveals destabilization where the learner omits an obligatory verb or complementizer or produces non-target non-finite forms. An assessment tool based on the analysis of errors illequipped to recognize the learner’s progress amidst such errors. Indeed since the decade in which error analysis was the height of fashion (see e.g. Richards 1974), researchers have recognized the non-linear characteristics of linguistic development; see for example Lightbown (1985) on the relevance of research to teaching and Pienemann et al. (1988) on non-linearity and assessment. MORPHO-SYNTAX-BASED ASSESSMENT As noted in our introduction, ours is not the first attempt to apply research findings from the acquisition of morphology and syntax to the assessment of language learners’ production. Older ideas include Crystal, Fletcher and Garmen’s (1976) language disability assessment tool, and more recent proposals such as Scarborough’s (1990) Index of Productive Syntax (IPS). The latter, for example, is based on emergence rather than mastery of both syntax and morphology, where emergence is signified by two productive (non-formulaic) uses of a form or construction. Applying the IPS to oral samples from two-to-four-year-old children, Scarborough was able to account for their later reading progress. Discussed briefly above, the Rapid Profile is a similar proposal, but for adult L2 assessment. Like Organic Grammar, Pienemann, Johnston and Brindley’s (1988) ideas originate in the study of the acquisition of L2 German by adults. Rapid Profile parallels Organic Grammar in several additional ways. Both RP and OG-based assessment assume for all L2 learners of a given language a common path of development, marked by progression through well-defined stages. Development is marked by emergence of complexity rather than disappearance of non-target structure. The further assumption that both approaches share is that stages are implicational, where arrival at stage Y implies passing through stage X. Several important differences between these two proposals favor Organic Grammar-based assessment. The first difference has already been mentioned: Organic Grammar assessment includes both morphology and syntax, allowing finer-grained assessment. Equally important is a final difference between Rapid Profile and Organic-Grammar-based assessment: practicality. Unlike the oral samples Rapid Profile is applied to (by trained assessors), written samples are easy to collect. Moreover, given the key contribution of writing to academic success, it is reasonable to expect that teachers will prefer a tool that assesses written proficiency over one that assesses oral proficiency. Organic Grammar criteria comprise a sort of ‘natural class’, where certain morphology is connected to certain syntactic projections.. If our suspicions are correct, knowledge of theoretical syntax is unnecessary for application of the criteria to samples of learner production. The combination of knowege ESL teachers possess about language and their experience with learner data allows them to readily make sense of OG criteria. 6 6 Both assumptions could, of course, be grounded in empirical data. However, the aim of this paper is to first determine the initial validity of OG-based assessment. Organic Grammar as an assessment tool The first application Organic Grammar criteria to the assessment of ESL learners’ production (oral in this case) was Young-Scholten and Strom’s (2004) cross-sectional study of 10 Somali and 7 Vietnamese adults with little or no primary schooling (see examples in (8). The study investigated cognitive and linguistic factors contributing to reading progress, where reading level was determined by scores on a matrix of reading sub-skills, and linguistic proficiency was based on the stages of Organic Grammar. And using OG as a measure of syntactic development but only single word reading as a measure of reading development, a significant overall correlation was found, and for the Somali group at p <.01; low numbers prevented reliable calculation for the Vietnamese. Table 3. Morphosyntax level and reading level Learner/sex/age V1 f 51 V6 f 70 V2 f 64 V5 m 34 V7 m 53 V4 f 43 V3 f 31 NL school ESL in USA 0 0 2 yrs 1; 4 yrs 5 yrs 3 yrs 3 yrs 1 yr 1 yr 2 yrs ½ yr ½ yr ½ yr 4 yrs S2 S10 S9 S8 S4 S6 S5 S7 S3 0 0 0 0 0 2 yrs 2 yrs 5 yrs 0 2 yrs 1 ½ yrs 1 yr 4 mns 3 yrs 1 yr 1 yr 1 ½ yrs 2 wks f f f f f f f f m 47 66 54 31 38 24 32 30 30 20 yrs 2 ½ yrs 8 yrs ¾ yr 3 yrs 13 yrs 12 yrs 5 yrs 3 yrs 4 yrs 9 yrs 9 yrs 2 yrs 2 yrs 9 yrs 2 yrs reading level 1 1 3 3 3 3 4 syntax stage 2 1 1 1 1 2 5 1 1 1 1 2 2 3 3 4 1 2 2 2 2 2 1 3 5 S1 m 26 4 yrs 0 1 yr 5 5 Although sample size is small, the data in Table 3 suggest that neither ESL instruction nor length of US residence is related to reading or syntax level. For example, of the eight unschooled adults, six were non-readers with no ability to read either single words in isolation or a paragraph, and all were at OG Stage 1 or 2, despite ESL classes ranging from 4 months to 1 ½ years and residence of ¾ of a year to 12 years. S3 somehow managed during his two years of US residence to reach level 4 in English reading (without the ability to read Roman-alphabet-based Somali) and OG stage 5 with only two weeks of ESL classes and two years’ US residence at the time of testing.7 THE STUDY At the community college context where the study took place, we sought to apply Organic Grammar stage criteria to a sample from the population of intermediate ESL students at that particular campus, where teachers had expressed concerns about current methods of assessment. Our aim was to determine the reliability of Organic Grammarbased criteria for level placement and progression, in order to consider it as an alternative possessing both intuitive appeal and practicality. The challenges such mixed background 7 This learner’s impressive achievements underscore the need for continued investigation by researchers of the linguistic development of immigrant adults who make rapid progress without the support of educational institutions. While longitudinal studies of this sort of population exist (see Table 1), we are aware of no longitudinal studies that track the linguistic development of completely unschooled adults. None of the longitudinal studies of immigrants tracked the literacy development of learners in their sample who might have had less than five years native language schooling. Thus while the linguistic achievements of S3 are not unexpected by those familiar with the studies in Table 1, his literacy achievements seem miraculous. adult ESL learners present in the classroom become more manageable if the teacher has linguistically informative criteria to apply in the tracking of each individual’s progress. Methods and materials Despite the practice of collecting oral production data from adult learners to determine linguistic competence which the studies shown in Table 1 invariably adopted, we chose to collect written data. There were several reasons – all equally important - for this. First, although the route of development proposed under Organic Grammar is based on oral production, there is no a priori reason why OG criteria should not apply to written production. Certainly when writing the test subject may have more time to plan production than would be the case when speaking, and might therefore be expected to use constructions or forms not part of his/her linguistic competence. Here the question of whether a given construction or form is part of the learner’s linguistic competence becomes methodological. In an Organic Grammar-based analysis of a sample, use of multiple criteria and a productivity metric can override the effect of the learner’s production of constructions that may have been memorized. The second reason for our choice is that written expression is fundamental to academic success, and examining written assessment must be our starting point. The third and final reason we opted to collect written data relates to the ultimate application of Organic Grammar by teachers: a procedure that involves the collection and analysis of written samples is considerably less time-consuming than the collection and analysis of oral samples and thus more likely the one adopted by teachers. We were, however, aware of the possibility of achieving some level of spontaneity even in written production, therefore sought to deflect our test subjects’ attention from our real aim in asking them to write something. We therefore treated the written exercise the students were given as a memory task. To limit the range of lexical items and constructions the test subject might choose (thus facilitating data analysis), we used picture description. Our focal sample consisted of 44 intermediate-level ESL students from a variety of native language backgrounds along with 3 advanced ESL students and 4 nativespeaker adolescent control subjects expected to be comparable to non-ESL community college students. At the ----- of ---- quarter [Colleen to add], data were elicited from test subjects in groups of --- to ---- [Colleen to add group sizes]. Students were told they would first see a set of pictures in a PowerPoint slide show, and that they would have to write (in complete sentences) what they had just seen. They were told that they would see the slide show again and could then write down (in complete sentences) what they had forgotten after viewing the slide show for the first time. The slide show contained a series of photographs showing a river at flood stage, with a boat flipping upside down under a bridge and then righting itself, and the slide show concluded with a set of unrelated photographs. The two descriptions were combined for each student, resulting in written samples ranging in length from 48 to 195 words; sample size turned out to be unrelated to program level or OG stage. We also elicited biographic details from all test subjects, along with TOEFL, SAT and program in-house tests. The nature and general aims of the study were explained to the test subject, and each signed a consent form agreeing to participate. Data analysis In 5.2 the main thing has to do with initial non-subject measure and the there/it measure. --section 5, first line: missing 'to' (sought -- apply) --section 5.1, line 10 from end: -ing missing from 'after seeING the pictures' --section 5.2, 1st para: maybe write like this: (1) syntactic complexity in terms of (a) simple vs. complex coordination, (b) a non-subject (or purely grammatical subject) in initial position... and (c) constructions that involve displacement... Also, maybe (b) is not needed? The non-subject measure was useful in German, but for English (c) would already include anything that might fall under (b), right? Nor is it as easy to have a non-subject initially in English, unless we have a complex CP-structure, or passive. Also, I'm not sure about collapsing the non-subject with it/there subjects... may not be a useful measure -- but from the tables I see that the two were not collapsed. I'm a little confused. On the top of the next page, could we either list or define more precisely what belongs to (2)? (9) looks fine. After (9), 6 lines down: the first "produced" -produced where? Little unclear what the two parts of the ratio are. --section 5.3 -- very nice! (but again, did the there/it and the initial non-subjects columns tell you guys much? If we keep them, there should at least be some discussion on them somewhere) --section 5.4 -- nice! but we should include a sentence about the vocabulary measure, the last column (the article discussion is good) We considered both written samples (after both picture viewings) combined from each learner in terms of the Organic Grammar criteria as follows: (1) syntactic complexity in terms of simple vs. complex subordination, a non-subject (or purely grammatical subject) in initial position (relative to total declarative clauses in that sample) and constructions that involve displacement of elements from their canonical position (passives) and (2) morphological range of productive forms, including those that under Organic Grammar correlate with syntactic development. The examples in (9) illustrate our application of some of the criteria. The nature of the task meant that test subjects could choose between the narrative present and the past tense, and the actual unfolding of the event and the scenes depicted in the other pictures further lent themselves to use of progressive aspect. Both factors led to a degree of temporal reference shifting by many of the (particularly lower level) test subjects such that it was not always possible to determine for verbs such as hit (which turned out to be heavily used to describe the boat incident) whether the test subject’s intention was to indicate past tense (correct) present tense (incorrect). In (9a) it is clear from Subject 10’s use of looked that ‘hit’ is correct, while ‘fall’ is not. We concluded, however, from the entire sample that this learner’s morphology is limited. For example, his/her use of ‘fall’ for ‘fell’ represents a very early stage of past tense development (Berko 1958) prior to the point at which –ed is overgeneralized to irregular verbs such as ‘fall’, resulted in ‘felled’ or ‘falled’. This learner uses verb form invariantly, not marking tense or agreement. In other words, inflectional morphology is not productive. S5, on the other hand, produces both past tense (‘saw’) correctly and third person singular (‘goes’). While S18 produces a form (‘pass) incorrect with respect to either present (not marked for agreement) or past, his/her use of correct forms along with a relative clause indicate a learner more advanced than S10. Subject 20 is the most advanced: she exhibits a range of (mostly correct) morphology, and she produces a complex subordinate clause, with the complementizer ‘that’. (9) Stage 3- type utterances: Limited morphology; co-ordination: a) Boat hit the bridge and fall down and people looked that boat. (S10) Stage 4-type utterances: Some morphology; simple subordination b) First we saw the white ship goes to hit the bridge. (S5) c) I see a ship which was down to the river and pass the bridge and come out to other side river. (S18) Stage 5-type utterances: Morphological range; complex subordination d) I am so happy that I found my grandfather and grandmother in their store. e) I was in the river contoured with beautiful green trees. (S20) We looked at two additional factors – vocabulary and article use - in the sample on assumption that these two factors might turn out to be indicators of level. Our idea was that a relatively larger range of vocabulary might compensate for a lower level of morpho-syntactic development or it might actually mask a lower level of development. As a measure of vocabulary range we looked at the number of different nouns the learner produced relative to the total number of nouns they produced in the entire sample. A learner for whom the ratio exceeded 2:1 was considered to have a good vocabulary range. As a morphological feature not present in many of the learners’ first languages, the article has been considered to be particularly difficult to acquire by L2 learners. Hawkins (2001) proposes that the absence of a feature in the learner’s native language results in its resistance to acquisition in a second language. To measure the acquisition of articles, we adopted methodology dating back to Brown (1973) and tabulated article suppliance in the obligatory contexts created by each student as they wrote their picture descriptions. As noted in Section 2.2., the impetus for this study was dissatisfaction at Georgia Perimeter College with the current holistic-criteria-based assessment used for both initial placement and for level progression which had led to the misplacement of students. In the process of placing the learners at stages, the second author only revealed which learners she knew to be misplaced after learners had been placed at three Organic Grammar stages or levels. Let us now turn to the results. 5.3. Results Our two sets of control group subjects, the advanced students and the native speaking adolescents, fully met the criteria for Organic Grammar Stage/Level 5 confirming the reliability of Organic Grammar-based assessment. While the non-native speakers made some lexical errors as in (10a), such errors were rare, and naturally not made by the native speakers (10b) who used a higher proportion of complex constructions and more precise and abstract vocabulary than the non-native speakers. (10) a) A ship passing by goes under the bridge and somehow gets banged with the bridge (S63). b) The pictures at first were related to the same event and then were of random origin (S65) We take Intermediate level (I) at GPC to equate with OG Stage 4, and High Intermediate (HI) with (unstable) Stage 5. The difference between High Intermediate and Advanced GPC ESL students is that the former have just entered Stage 5, while the latter will have firmly established the complex syntax and morphological range and can begin to take the sort of control of their writing craft that native speakers typically gain in college preparatory classes in secondary school. We will see that that of the 44 intermediate learners, xxx are not at the level at which they should be if Organic Grammar Stages equate with GPC levels. (See the Appendix for all students’ sample sizes, TOEFL, SAT and GPC test scores.) Importantly, our analysis reveals that most students are at the right level. But what Organic Grammar allows us to do is ground the intuitions that holisticcriteria give rise to. We began our analysis by placing students at levels solely on the basis of their complex syntax on the basis of their production of no subordinatation (Table 4), simple subordination (Table 5) and complex subordination (Table 6). By these criteria, four of the 44 students in our sample should not even be at the Intermediate Level. For two of these students, the Intermediate Learner S7 and the High Intermediate Learner S26, there were suspicions based on observations of their performance in class that neither belonged in the GPC ESL program. [Colleen to add more details.] If we look more closely at their syntax and also consider morphology, a case can be made that learners S10 and S31 are (low) Stage 4 learners. Both produce ‘it’ and/or ‘there’ subjects, both produce at least some initial non-subjects, and they display some range of morphology. We shall have more to say below about articles and vocabulary. Table 4. GPC students with no subordination Subjects S7/ GPC there level /it I none I it Initial subordination non- . morphology article vocab errors subjects simple complex range errors 0 0 0 some errors 1/12 10/20 1/9 0 0 some many 8/16 15/19 Amharic S10/ Korean S26/ errors HI none 2/11 0 0 good few 1/17 13/23 it; 5/10 0 0 some errors 1/23 17/26 Russian S31/ I Spanish there Turning to the table showing learners placed at the Intermediate Level based on their production of simple subordination, we find that morphology plays an even more important role. Learners are arranged here in order of morphological range, from none at the top (S110 to learners who produce passives at the bottom of the table. All those for whom there was a morphological range that tended to also include passive or present perfect were either at the High Intermediate level (as shown on the table) at the time of testing or were subsequently promoted to this level the quarter following data collection. In this group there are also two learners that an OG-based criteria would not place at the High Intermediate level, yet who were also at that GPC level: S4 and S28. S28 it turns out should not be at that level [Colleen elaborate], and the data support this. S4 is less clear; based on OG criteria she is at Stage 4. However, it may be that her vocabulary range and her lack of article errors indicate a high level of overall proficiency. Thus, in addition to correctly placing eight Intermediate Level and four High Intermediate Level students, Organic Grammar criteria successfully identified the six learners who were promoted to High Intermediate, and one who should not have been. Table 5 GPC students with simple subordination Subjects S11/ Vietnamese GPC there initial level /it non- I It subordination. morphology article vocab errors subjects simple complex 5/8 5 0 range No errors 0/12 12/18 S2/ I there 5/7 6 0 some few 2/13 11/17 I it/ 7/10 4 0 some few 2/20 20/28 there 5/11 5 0 some none 0/15 16/22 it/ 10/14 5 0 some few 0/26 18/35 8/12 5 0 some errors 3/16 23/34 Fula S3/ Russian there S4/ HI French S6/Farsi I there S15/ I it/ Polish+ there S28/ HI there 3/12 1 0 some errors 6/13 14/24 I 0 yes 1/20 13 0 some few 13/19 14/36 Korean S12/ Turkish pass S38/ pass. I there 2/11 2 0 good errors 7/34 17/36 I there 5/13 3 0 good none 7/24 18/32 I it 12/17 5 0 yes few 0/36 23/44 HI it, 10/21 2 0 yes few 1/28 22/42 none 0/9 8/10 few 1/25 25/33 some 1/16 14/21 Korean S44/ Korean S1/ Fula S13/ Wolof there pres perf S16/ HI there 4/8 3 0 some Urdu pass. S19/ HI there 2/13 1 0 yes Korean pass. S22/ Gujarati HI it,t 5/10 5 0 yes here S24/ I it pass 2/6 5 0 yes some 1/12 16/20 some 14/24 17/43 some 0/24 17/31 some 3/21 16/24 few 2/21 21/26 Russian pass S25/ I 0 1/16 8 0 yes Farsi pass S35/ I it 1/13 10 0 yes Vietnamese pass S37/ I it/ 2/9 3 0 yes Vietnamese there S41/ I there pass 5/6 6 0 some Spanish pass Finally, Table 6, which shows learners categorized on the basis of their production of some complex subordination in addition to simple subordination, consists almost completely of learners who subsequent to testing were promoted to High Intermediate (indicated by shading). S42 was not promoted to High Intermediate, but this was due to his failure to turn in assignments rather than his English. Students are again arranged in order of their morphological complexity. However, unlike for those learners who only produce simple subordination, this does not result in an additional division among them. Table 6. GPC students with some complex subordination Subjects GPC there initial level /it non- subordination. simple complex morphology range errors article errors vocab subjects S5/ I it 3/17 3 1 Chinese S18/ HI I 0 0 1 1 it/ 7/12 5 2 there I Vietnamese S9/ it/ I 1/24 25/38 it 2/7 *0 1 some; few 1/10 13/19 2/17 16/30 2/23 19/42 1/15 13/21 3/19 22/41 2/11 20/26 errors I there 6/14 6 1 Chinese some; many errors I 0 0 3 1 some, with errors HI there 4/12 4 1 good; some errors I it 2/15 3 2 good; some errors I it/ 2/11 3 2 there I Farsi S42/ 19/33 there Tigrinian S39/ 2/27 some; errors Indonesian S36/ limited; some 2 Chinese S33/ 18/21 3 Vietnamese S23/ 6/11 7/12 e S34/ limited, but no errors Cantonese S21/Chines 17/41 errors Spanish S8/ 3/35 errors Korean+ S29/ little, with it/ good; some errors 7/14 4 2 good; errors 2/24 21/33 2/5 8 1 good, but errors 1/26 16/31 there I there S43/ I 0 3/9 4 1 good, but errors 11/25 22/36 I it/ 4/7 6 1 good; no errors 0/19 12/22 5/11 7 3 yes; some 0/22 19/24 2/18 15/26 6/20 15/25 Farsi S40/ Spanish S27/ there I Vietnamese S30/ it/ there I there errors 6/12 6 3 Spanish S32/ yes; some errors I there 11/13 3 2 Korean passives; with errors 5. 4, Discussion The syntactic and morphological Organic Grammar criteria have successfully identified the GPC level at which 43 of the 44 students in our study belong. This is no small feat when one considers the relatively limited piece of writing each student produced as compared with what students produce for writing tests or what learners produce for oral samples (e.g. 100 utterances). Yet based on the notions of productivity and emergence along with the concept of implication, we have shown it is possible to accurately assess learners using a simple to administer elicitation procedure which yields data that are straightforward to analyze. Importantly, neither morphological nor syntactic criteria alone would have sufficed to accurately assess all learners in the study. The idea underpinning Organic Grammar is that inflectional morphology and syntax are intimately related to each other during development. In looking at the results presented in Tables 4,5 and 6, we can also note that there is little evidence of support for Hawkins’ prediction that articles will be impossible to acquire by those learners who native language backgrounds lack them. Those learners whose native languages have articles invariably make the fewest errors while those languages do not make the most (e.g. Spanish speakers vs Turkish speakers), yet it is not the case that every speaker of article-less languages makes more errors than speakers whose languages have articles. Compare for example Korean speaker S19 and Spanish speaker S41, who made 1 and 2 article errors, respectively. The great majority of article errors all learners committed were omission errors rather than confusion of indefinite, definite and zero article. CONCLUSION We have taken an initial step in re-introducing Pienemann et al.’s idea of implicationalhierarchy- or stage-based analysis to ESL assessment. By adopting a more recent analysis of linguistic development that incorporates morphology and syntax, namely Organic Grammar, we have been able to fill the gap between standardized discrete-item testing and holistic-criteria-based analysis of elicited written samples. Doing so does not require sophisticated equipment or programs to sort through scores of constructions produced by the learner. There is potential for refinement of our testing instrument through elicitation of more contexts for complex syntax, such as Wh-questions, and for further exploration of the accuracy of OG-based assessment through collection of a greater range and quantity of cross-sectional data. We envision that with such fine- tuning, OG-based assessment will be a highly usable tool for the assessment of both oral and written language for program placement, progression and exit. References Bailey, N, C. Madden and S. Krashen. (1974) Is there a ‘natural sequence’ in adult second language learning? Language Learning 24:235-243. Berko, J. (1958) The child’s learning of English morphology. Word 14:150-177. Brown, J. D. (ed.) (1998) New Ways of Classroom Assessment. Alexandria, VA: TESOL. Brown, R. (1973) A First Language. The Early Stages. Cambridge, MA: Harvard University Press. Chomksy, N. (1981) Principles and parameters in syntactic theory. In N. Hornstein and D. Lightfoot (eds.) Explanation in Linguistics: the Logical Problem of Language Acquisition. London:Longman. pp. 32-75. Clahsen, H. 1990. The comparative study of first and second language development. Studies in Second Language Acquisition 12:135-153. Clahsen, H. and P. Muysken. (1986) The availability of Universal Grammar to adult and child learners – a study of the acquisition of German word order. Second Language Research 2:93-199. Clahsen, H., J. Meisel and M. Pienemann. (1983) Deutsch als Zweitsprache. Narr: Tuebingen. Crystal, D., P. Fletcher and M. Garman. (1976) The Grammatical Analysis of Language Disability: A Procedure for Assessment and Remediation. London:Edward Arnold. deVilliers, J. and P. deVilliers. (1973) A cross-sectional study of the acquisition of grammatical morphemes in child speech. Journal of Psycholinguistic Research 2: 267-278. Dulay, H. and M. Burt. (1974). Natural sequences in child second language acquisition. Language Learning 24:37-53. Franck, J., J. S. Bowers, U. H. Frauenfelder and G. Vigloiocco (2003). Orthographic influences on agreement: A case for modality-specific form effects on grammatical encoding. Language and Cognitive Processes 18:61-79. Genesee, F., M. Gottlieb, A. Katz and M. Malone. (1998) Managing the Assessment Process. TESOL Professional Papers #5. Hawkins, R. (2001). Second Language Syntax. A Generative Introduction. Malden, MA: Blackwell. Haznedar, B. (1997). L2 acquisition by a Turkish-speaking child. Unpublished PhD thesis. University of Durham. Klein, W. and C. Perdue. (1992) Utterance Structure. Developing Grammars Again. Amsterdam: Benjamins. Klein, W. and C. Perdue. (1997). The Basic Variety (or: Couldn’t natural languages be much simpler?) Second Language Research 13: 301-347. Kozloski, W. and S. Donahue. (2005) The sophisticated sentence. language)( February 32-35. Lardiere, D. (2000). Mapping features to forms in second language acquisition. In J. Archibald (ed). Second Language Acquisition and Linguistic Theory, Malden, MA: Blackwell. 102-129. Lightbown, Patsy M. (1985a) Great expectations. Second language acquisition research and classroom teaching. Applied Linguistics 6:173-189. Myles, F. (2005). The emergence of morphosyntax in French L2. In J-M. Dewaele (ed.) Focus on French as a foreign language: Multidisciplinary Approaches. Clevedon: Multilingual Matters. pp. Pienemann, M. (1987) Psychological constraints on the teachability of languages. In C. Pfaff (ed.) First and Second Language Acquisition Processes. Rowley, MA: Newbury House. p. 143-168. Pienemann, M.. (1998). Language Processing and Second Language Development. Processability Theory. Amsterdam:Benjamins. Pienemann, M., M. Johnston and G. Brindley. (1988). Constructing an acquisition-based procedure for second language assessment. Studies in Second Language Acquisition 10:217-243. Pienemann, M. (2001). Rapid profiling. Keynote address. EUROCALL 2001 Conference Report. Prévost, P. and L. White. (2000). Missing surface inflection or impairment in second language acquisition. Evidence from tense and agreement. Second Language Research 16: 103-134. Radford, A. (1995): Children - Architects or brickies? In D. MacLaughlin, and S. McEwen (eds.) Proceedings of the 19th Annual Boston University Conference on Language Development, Somerville, MA: Cascadilla Press, 1-19. Richards, J. C. (1974). Error Analysis. Perspectives on Second Language Acquisition. London: Longman. Scarborough, H. S. (1990) Index of Productive Syntax. Applied Psycholinguistics 11:1-22. Schumann, J. (1978) The Pidginization Hypothesis: A Model for Second Language Acquisition. Rowley, MA: Newbury House. Schwartz, B. D. and R. Sprouse. (1996). L2 cognitive states and the Full Transfer/Full Access model. Second Language Research 12:40-72. Vainikka, A. and M. Young-Scholten (1994) Direct access to X'-Theory: Evidence from Korean and Turkish adults learning German. In T. Hoekstra and B. D. Schwartz (eds.) Language Acquisition Studies in Generative Grammar: Amsterdam: Benjamins. pp. 265-316. Delete?? Vainikka, Anne and Martha Young-Scholten. (1996). Gradual development of L2 phrase structure. Second Language Research 12:7-39. Vainikka, A. and M. Young-Scholten (1998). Tree growth and morphosyntactic triggers in adult SLA. In M-L. Beck (ed.) The L2 Acquisition of Morphology. Amsterdam: Benjamins. Vainikka, A. and M. Young-Scholten (2005). The roots of syntax and how they grow: Organic Grammar, the Basic Variety and Processability Theory. In S. Unsworth, T. Parodi, A. Sorace and M. Young-Scholten (eds.) Paths of Development. Amsterdam: Benjamins. Wagner-Gough, J. (1978) Comparative studies in second language learning. In E. Hatch (ed.) Second Language Acquisition. A Book of Readings. Rowley, MA: Newbury House. Yamada-Yamamoto, A. (1993). The acquisition of English syntax by a Japanese- speaking child: With special emphasis on the VO-sequence acquisition. In J. Clibbens and B. Pendleton (eds.) Proceedings of the Child Language Seminar. University of Plymouth. pp. 109-120. Young-Scholten, M. and N. Strom (2004). Factors determining adult nonliterates’ reading progress. Paper delivered at the annual TESOL convention. Long Beach, March 31. Zobl, H. (1980) The formal and developmental selectivity of L1 influence on L2 acquisition. Language Learning 30:43-57. Appendix GPC student data in subject-number order Subject number+ NL TOEFL/ GPC total there/it initial non- simple complex morphology: article vocab SAT reading/ words subjects subjects subordination subordination range; accuracy errors 12/17 5 0 yes; few errors 0/36 23/44 writing S1/Fula+ T473 189/30 182 it S2/Fula+ T377 117/30 85 there 5/7 6 0 some; few errors 2/13 11/17 S3/Russian S340 306/20 103 it/there 7/10 4 0 some;few errors 2/20 20/28 S4/French T513 270/30 92 there 5/11 5 0 some; no errors 0/15 16/22 S5/Chinese T490 252/20 171 it 3/17 3 1 little, with errors 3/35 17/41 S6/Farsi T463 216/30 156 it/there 10/14 5 0 some; few errors 0/26 18/35 S7/Amharic T490 261/10 68 0 0 0 0 some;errors 1/12 10/20 S8/Vietnamese S280 189/30 156 it/there 7/12 3 2 some; errors 1/24 25/38 S9/Cantonese+ T467 207/30 86 it 2/7 0 1 some; few errors 1/10 13/19 S10/Korean T467 306/20 53 it 1/9 0 0 some; many errors 8/16 15/19 S11/Vietnamese T477 207/20 79 it 5/8 5 0 no; errors 0/12 12/18 S12/Turkish T413 189/30 193 0 1/20 13 0 some; passive 13/19 14/36 S13/Wolof+ T487 189/40 167 it, there 10/21 2 0 pres perf; few errors 1/28 22/42 S14/Arabic+ T460 225/30 105 0 4/8 5 3 yes; few errors 1/20 11/21 S15/Polish+ T470 279/30 146 it/there 8/12 5 0 some, with errors 3/16 23/34 S16/Urdu S390 324/30 51 there 4/8 3 0 pass. limited;errorless 0/9 8/10 S17/Spanish T470 234/30 189 there 7/17 3 2 passive, but errors 0/23 19/33 S18/Korean+ T463 234/30 66 0 0 1 1 limited, but no errors 6/11 18/21 S19/Korean S440 333/30 129 there 2/13 1 0 passive; few errors 1/25 25/33 S20/Arabic+ T470 243/30 48 it,there 4/16 4 1 passive;some errors 0/18 20/23 S21/Chinese T483 234/30 131 there 6/14 6 1 some; many errors 2/17 16/30 S22/Gujarati+ S390 234/30 102 it,there 5/10 5 0 passive; some errors 1/16 14/21 S23/Chinese S300 207/30 87 there 4/12 4 1 good; some errors 1/15 13/21 S24/Russian+ - 270/20 89 it 2/6 5 0 passives; some errors 1/12 16/20 S25/Farsi T470 135/30 153 0 1/16 8 0 yes; passive; errors 14/24 17/43 S26/Russian+ S250 207/40 82 0 2/11 0 0 good; few errors 1/17 13/23 S27/Vietnamese+ S310 234/20 120 it/there 5/11 7 3 yes; some errors 0/22 19/24 GPC students‘data cont’d subject number+ NL TOEFL/ GPC total there/it initial non- simple complex morphology: range; article vocab SAT reading/ words subject subjects subordination subordin accuracy errors writing S28/Korean T487 315/30 85 there 3/12 1 0 some; errors 6/13 14/24 S29/Spanish T487 261/20 151 it/there 7/12 5 2 limited; some errors 2/27 19/33 S30/Spanish S290 198/30 132 there 6/12 6 3 yes; some errors 2/18 15/26 S31/Spanish T537 288/20 111 it/there 5/10 0 0 some, with errors 1/23 17/26 S32/Korean T470 225/20 112 there 11/13 3 2 passives; with errors 6/20 15/25 S33/Indonesian T483 243/20 195 it 2/15 3 2 good; some errors 3/19 22/41 S34/Vietnamese T480 234/20 132 0 0 3 1 some, with errors 2/23 19/42 S35/Vietnamese T513 306/20 133 it 1/13 10 0 passive, some errors 0/24 17/31 S36/Tigrinian S220 171/30 146 it/there 2/11 3 2 good; some errors 2/11 20/26 S37/Vietnamese S320 162/30 109 it/there 2/9 3 0 passive; errors 3/21 16/24 S38/Korean T533 351/20 121 there 2/11 2 0 good; errors 7/34 17/36 S39/Farsi+ T520 207/30 153 it/there 7/14 4 2 good; errors 2/24 21/33 S40/Spanish T480 189/20 90 it/there 4/7 6 1 good; no errors 0/19 12/22 S41/Spanish T563 306/20 115 there 5/6 6 0 limited (passive) few errors 2/21 21/26 S42/Chinese+ T467 243/20 142 there 2/5 8 1 good, but errors 1/26 16/31 S43/Farsi T433 207/30 146 0 3/9 4 1 good, but errors 11/25 22/36 S44/Korean+ T467 279/20 115 there 5/13 3 0 good; no errors 7/24 18/32 Advanced learners: two (S61; S63) of the three in our sample exhibited use of complex subordinate clauses or other complex syntax (use of ‘seem’ and of passives). S64 did not exhibit such syntax. The three native speaking adolescent controls (S65, 66, 67, 68) produced passives, ‘appear to’ clauses and ‘that’ complementizer clauses, i.e. complex syntax. NOTES 1. Native language = a ‘+’ indicates at least another 2. T= TOEFL/S=SAT, where TOEFL score is given when student has taken TOEFL and SAT; otherwise SAT is given. A dash indicates neither test has been taken. 3. GPC reading/writing test scores 4. Total words in sample, as an indication of student’s written fluency 5. it/there = instances of subject ‘it’ or ‘there’ indicate emergence of more complex syntax 6. Non-subjects in initial position relative to total declarative clauses. This count includes existential and weather ‘it’ and ‘there’. Syntax involving such subjects is more complex than syntax with full noun or referential pronominal subjects. 7. Simple sub-ordinate clauses refers to subject relative clauses 8. Complex sub-ordinate clauses refers to ‘that’ complementizer clauses, use of ‘seem’, purpose clauses, etc. which involve the highest syntactic projection (the most complex syntax) 9. Morphology = no/little/some/yes refers to the range of tense and aspect marking which for this sample included simple present, present continuous, simple past (regular and irregular verbs), past continuous, present perfect and passive. many, few, some and no refers to level of accuracy. 10. Article errors = based on obligatory contexts for articles (zero article, as in the plural, was not included) typically involved omission; omission may persist to otherwise high levels of proficiency, depending on their presence in the learner’s L1 11. Vocabulary = different nouns produced as a function of total nouns produced