eJournals Fremdsprachen Lehren und Lernen 28/1

Fremdsprachen Lehren und Lernen
flul
0932-6936
2941-0797
Narr Verlag Tübingen
Es handelt sich um einen Open-Access-Artikel, der unter den Bedingungen der Lizenz CC by 4.0 veröffentlicht wurde.http://creativecommons.org/licenses/by/4.0/121
1999
281 Gnutzmann Küster Schramm

An L2 subset parameter value in the grammar of L2 (G2)

121
1999
Wilfried Weigl
This paper investigates the acquisition of the position of the English verb by students of a (Bavarian) Gymnasium. This acquisition requires resetting the verb movement parameter from a ‘superset’ to a ‘subset’ value. Holding that the Subset Principle is inoperative in L2, and that G1 [= grammar of L1] is strongly effective (‘transfer’), this resetting was assumed to be difficult, with the difficulty reflected in the span of time required and in the percentage of learners that fail to achieve it. This assumption was verified by the results of a test: The verb movement property ‘*Verb in ARG’ had been acquired by 70 %, the property ‘*Verb in C’ by only 50 % of the students, with acquisition of the former growing throughout the course, acquisition of the latter not growing at all. It was concluded that the teaching of L2 needs improving. It can possibly be provided through a shaping of the input that makes the learner ‘focus on form’.
flul2810181
...__ _____ N_1_· c_h_t-_t_h_e_m_a_t_i_s_c_h_e_r_T_e_i_I _____ __.I Wilfried Weigl An L2 subset parameter value in the grammar of L2 (G2) The verb position in the English of German Gymnasium students Abstract. This paper investigates the acquisition of the position of the English verb by students of a (Bavarian) Gymnasium. This acquisition requires resetting the verb movement parameter from a 'superset' to a 'subset' value. Holding that the Subset Principle is inoperative in L2, and that G 1 [= grammar of Ll] is strongly effective ('transfer'), this resetting was assumed tobe difficult, with the difficulty reflected in the span of time required and in the percentage of learners that fail · to achieve it. This assumption was verified by the results of a test: The verb movement property '*Verb in AGR' had been acquired by 70%, the property '*Verb in C' by only 50 % ofthe students, with acquisition ofthe former growing throughout the course, acquisition of the latter not growing at all. lt was concluded that the teaching ofL2 needs improving. lt can possibly be provided through a shaping ofthe input that makes the learner 'focus on form'. 1. Introduction Over the years, it has become clear that the acquisition of a second language can best be explained through assuming the existence of (some variant of) a language module which is autonomous within man's overall cognitive apparatus, but dependent, with regard to effectiveness, upon L2 input. The language module' s (main) components are Universal Grammar (UG) and the Grammar of Li (Gl), the input's main feature is its quality (vs quantity). 1 Given that situation, it seems plausible to hold the view that, instead of continuing to deal with broad questions, "it is now the turn of a somewhat more detailed focus on the precise linguistic competence oflanguage learners" (White 1996: 115). lt is against that background that we deal with the acquisition, by German high school (Gymnasium) students, of the position of the English verb. Of previous studies into L2 verb acquisition, only one is relevant here: White (1992). White (1992) is an investigation of the acquisition of the position of the English verb by learners with LI French. The one result relevant here is that "francophone learners of English realise the impossibility of long movement [= movement across not- W. W.] [...] in English" (277). What could not be investigated with LI French is the acquisition of the For theoretical support for this view, see its (comprehensive) exposition in the chapters, in Ritchie and Bhatia (eds.) (1996), by Gregg, White, Flynn and Gass (language module) and by Long (input). Empirical support comes from a number of studies into the acquisition of individual syntactic phenomena, such as White (1991a, 1991b), Eubank (1992), Roeper ( 1992), White (1992), Hawkins (1993), and Uziel (1993). IFlLlJIL 28 ( 1999) 182 Wilfried Weigl impossibility of verb movement to C (in contrast to what obtains in (Ll) German). White' s result will be given further consideration below. The remainder of this paper proceeds as follows: In the next section, a short exposition of our view of the 'sentence' 2 and of (its) L2 acquisition (2.1), and a characterization of the input available (2.2) is given. In the light of these facts, we then establish the learner's task and our assessment of its solvability (2.3). In section 3 we present, and analyze, our test results. In section 4 we consider the L2 competence attained in a teaching perspective, with a few remarks towards a potential improvement of instructed acquisition of L2. 2. Presuppositions 2.1 Theoretical assumptions 2.1.1 The structure of the sentence Our view of the sentence is that of the Minimalist Program. 3 In the present context, however, for clarity of presentation, we (1) continue to use the concepts 'X-bar' and 'movement', and (2) simplify the sentence-tree, omitting the specifier-level of NegP and of TP, and omitting altogether those categories that are irrelevant here (AGRP, Vp). On that basis, the structure of the sentence, for English, is as follows: (]) CP I \ Spec C' I \ C AGRP I \ Spec AGR'• I \ AGR NegP I \ Neg TP I \ T VP I \ Spec V' 1 V Being the general structure of the English sentence, (1) is neutral with regard to sentence type. lt captures all of the relevant types: The tenn sentence is used here for convenience. In a modern view, the basic syntactic unit is a VP dominated by projections of functional categories like AGR, Tand C. Aware of this, we continue to use sentence (which then does not need inverted commas). 3 For the exposition of the Minima/ ist Program, see Chomsky (1995); for an introduction, see Radford (] 997). IF]Lm, 28 ( 1999) An L2 subset parameter value in the grammar of L2 (G2) 183 Spec C Spec AGR Neg V (2) E John does not play (3) E Does John play? (4) E AGR Today John does not play (5) E John has not played (6) E Has John played? (7) E AGR Today John has not played Table I: Place of sentence constituents (E) (In sentences (4) and (7), the adverb is in a place created ad hoc, through adjunction to AGRP. do is base-generated in AGR, have in T.) There is one more sentence type, (8). In it, the verbis in AGR, but is has gone there only after Spelt-Out: (8) John plays. The sentence types in Table I have the following properties: (i) The verb is in V. (ii) The inflectional affix ('lnfl'), generated in AGR, surfaces in AGR in statements; in C it surfaces only in questions: English is 'non-Infl 2' (as also shown by *Today does John not play).4-The realization oflnfl (if there is no have) has recourse to do ('do-support'). (iii) The verb is not in C, not even in questions: English is (not only non-Infl 2, but) 'non-V2'. For German, the structure of the sentence is as follows: (9) CP / \ Spec C' / \ C AGRP / \ Spec AGR' / \ NegP AGR / \ Neg TP / \ VP T / \ Spec V' 1 V 4 According to the notion underlying the V2 concept, "having the finiteness operator [+F] in C" (Platzack/ Holmberg 1989: 52), there should be no obstacle to there being Infl 2, too. There indeed is none, as shown by sentences such as Heid dud der Fritz ned kumma (Heute tut Fritz nicht kommen). IFLllllL 28 ( 1999) 184 Wilfried Weigl lt captures the following sentence types: Spec (2) G Hans (3) G (4) G Heute (5) G Hans (6) G (7) G Heute C Spec AGR Neg spielt nicht spielt Hans? spielt Hans nicht hat nicht hat Hans hat Hans nicht Table II: Place of sentence constituents (G) V gespielt gespielt? gespielt (In sentences (4) G and (7) G, the adverb is in SpecC.) The sentence types in Table II have the following properties: (i) The (finite) verbis not in V. (lt is in V in subordinate clauses, which is probably drawn upon in the acquisition process.) (ii) Infl is in C, not only in questions, but also in statements (cf. also note 4): German is 'Infl 2'. (iii) The (finite) verbis in C, too: German is (not only Infl 2, but also) 'V2'. 2.1.2 The structure of L2 acquisition As stated above, the language module assumed here is (1) complex and (2) autonomous. The specifics of this view, and their bearing upon the mode of (L2/ ) G2 acquisition, are as follows: (1) Assuming that the main components of the language module are UG and 01 5, there must be structural interrelation between them. Concerning it, we assume that the operation of G 1 has priority over that of UG. For the acquisition process this means that whenever there is surface identity across an L2/ Ll sentence pair, the L2 sentence is interpreted in the light of Ll ('transfer'). 6 UG is only called upon if, and when, G 1 is not/ no longer usable. Transfer is not infrequent, but given that surface identity rarely corresponds to füll identity, it often results in an erroneous G2, with erroneousness being of two different kinds. The overall system of cases of transfer is as follows: (9) transfer result / \ * G2 G2 (1) / \ Has John played? erroneousn. erroneousn. concealed (2) evident (3) John plays Today plays John The main 'learning principle', the Subset Principle, posited in the research some time ago, we assume tobe inoperative (Cf. Gass 1996: 332). 6 For the different views of transfer, see Gass (1996: 329 f). fLllllL 28 ( 1999) An L2 subset parameter value in the grammar of L2 (G2) 185 Of the three cases of transfer, (1) is the one in which the resulting G2 is correct. In the 'Verb' (vs 'have') cases, the resulting G2 is erroneous. In case (3), G2-erroneousness is evident, because surface-perceptible, in case (2), it is not. For the learner this means that (3) is easier to repair than (2): The ungrammaticality of (3) can be gathered, in the course of time, from 'indirect negative evidence', from the fact that Today plays John does not ever occur as an L2 utterance. 7 The ungrammaticality of (2), however, cannot be gathered from immediately relevant evidence, because it is that evidence, direct positive evidence, that has failed to prevent its establishing. Repair of (2), however, can be effected via the repair of (3): Having realized that Today plays John is ungrammatical, the learner may have realized that this is because the English verb cannot be in C. This insight, in turn, could reveal the ungrammaticality of Johnplays. Consider now (2), the autonomy of the language module and its bearing upon L2 acquisition. Structurally, this autonomy means that there is no interface between the language module and cognitive processors. 8 What the language module organizes is language 'acquisition', resulting in 'competence', what the cognitive processors organize is language 'learning', which results in 'knowledge' . 9 In L2 in the classroom, there usually is a certain amount of knowledge production, effected through grammar teaching. To the extent, however, that the language module is autonomous, grammar teaching is of little avail: The knowledge resulting from it is hardly transformable into competence. IO This takes us to the importance, for successful L2 acquisition, of (the quality of) the input. If grammar teaching (and error correction) are little appropriate because they appeal to cognition, input, to be appropriate, must be indirect/ inexplicit. In the L2 classroom, such input has long been used in the shape of 'comprehensible input', of speech "adapt(ed) [...] in ways that make models (of what is grammatical) comprehensible to the learner and thereby usable for acquisition" (Long 1996: 413). In L2 acquisition research, it has been investigated whether comprehensible input alone is sufficient. The result is that it is not: "(C)omprehensible input alone is insufficient, particularly with adults and if nativelike proficiency is the goal" (Long 1996: 423). Accordingly, it has been claimed that what is needed is 'input enhancement' (Sharwood Smith 1993). One way of achieving it is the use of means that make the learner (not 'focus on forms', but) 'focus on form' (Long 1991). We disregard 'direct negative evidence' here. For support for this view, see Felix (1985) and Zobl (1995). 9 In the present paper, we use acquire and learn as non-terms, hence interchangeably. 10 Moreover, L2 knowledge as such is of limited durability. (Cf. Lightbown/ Spada/ Wallace (1980), Pienemann (1984), White (1991a) and Weigl (1998).) f'LllllL 28 ( 1999) 186 Wilfried Weigl 2.2 Empirical reference The L2 acquisition referred to here is the one in a Bavarian high school (Gymnasium) course. In it, as elsewhere, the input is deterrnined, quantitatively, by the time available for the course and, qualitatively, by the teaching practice. Regarding time, the L2 course situation is as follows: form lessons/ week accum. number 5 6 216 6 611 432 7 4 576 8 4 720 9 3 828 Table III: Time allocated to the L2 course 10 3 936 11 4 1080 Against that background, consider input quantity. Crucially we assume here, on the basis of evidence from previous studies, that in the frame given L2 acquisition up to an acceptable degree of 'purity' of L2 competence 12 can extend from L2 beginning until after form 10. 13 For this possibility tobe realized, relevant input must be available throughout that period. For the input relevant here, data reflecting the setting of the English verb movement parameter, such availability can be assumed: Questions and negative statements are frequent in every-day language, and probably more frequent in L2 classroom use. And their frequency of occurrence can be assumed to be more or less constant throughout the course. There should thus be no shortness, in the Gymnasium L2 course, of the input data required. Regarding the quality of the input, we dismiss from consideration (cf. 2.1.2, above) the input from grammar teaching 14 and deal with the input that is indirectlinexplicit. For the learner, such input is the material from which he has to extract the information that he needs to build his G2. The question here is whether the simple making available of 'natural' data, sufficient for Ll, suffices to enable the learner to make that extraction. The answer, uncontroversial, is that it does not: The data must be selected and, if possible, presented in a way that makes the learner 'notice' the relevant information, makes him 'focus on form' (Long 1991). Measured against this requirement, the input in the Gymnasium L2 course is hardly sufficient: The relevant teaching material, 'grammar exercises', fails to narrowly focus on the difference between L2 and Ll, and to arrange the sentence types according to their learnability and their informativeness. The extraction of information from the L2 data has thus tobe made without much help from the side of teaching. II Recently changed to 5. 12 For an exposition of that concept, see below, 2.3. 13 In a different, more comprehensive perspective, what happens over the years is integration, "the development per se of one's second language grammar" (Gass/ Selinker 1994: 303). 14 The teaching of the relevant grammar is located as follows: do-support is taught in form 5, and 'revised' in forms 6 and 8. The have-form ('Present Perfect') is taught in form 6. IFlLllllL 28 (1999) An L2 subset parameter value in the grammar of L2 (G2) 187 2.3 The English sentence: acquirability Regarcling acquirability, we first consider the English sentence 'as such', as neutral with regard to L2, and then as object to be acquired by Gerrnan learners. The properties of the English sentence (cf. above, 2.1.1 ), and their interrelations, are as follows: PI: The verbis in V, hence does not even go to AGR. P2: Infl is in AGR in statements, in C only in questions. (Infl-realization, without have, has recourse to do.) P3: Beingin V, the verbis not in C, not even in questions. Consider now how Pl through P3 are surface-reflected, thus made acquirable through reference to surface facts. Pl, that the verbis ('has stayed') in V, can most unequivocally be taken from negative statements (sentence type (2)): (2) John does not play From (2) it is perceptible that Infl is affixed to do, immediately 'to the left of' not. Accordingly, the verb must be in V, which is, moreover, given positive expression through the fact that it is 'to the right of' not. (That the opposite solution, Infl in V instead of the verb in AGR, is precluded, too, is shown by the sentence * John not plays.) That have, in contrast to the verb, is not immovable can be taken from type (5): (5) John has not played In (5), the position of has, immediately 'to the left of' not, is not its base position. This can be taken from a construction like Not to have played [...], in which have is 'to the right of' not. P2, that in statements lnfl is in AGR (vs in C), can best be taken from adverb-headed (negative) statements: (4) Today (,) John does not play From (4) it is perceptible that in statements the lnfl-expression, does, is not 'to the left of' the subject, John, and, accordingly, not 'to the immediate right of' the adverb (where it is in an Infl 2-sentence like Heid dud der Fritz ned kumma). have, too, is in AGR (vs C) here: (7) Today (,) John has not played Sentence (7) has the same surface facts as sentence (4). P3 finally, that the verb does not 'go' to C, can most unequivocally be taken from questions: (3)' Does John not play? (3) Does John play? From these sentences it is perceptible that the position 'to the left of' the subject, occupied by the Infl-expression, is not occupied by the verb. Sentence (3)' is even more informa- JFLU! L 28 (1999) 188 Wi[fried Weigl tive: lt shows not only that the verb is not in C ('to the left of the subject'), but also that it is in V ('to the right of' not), have, not being a verb, goes to C. This is perceptible from the counterparts of sentences (3)' and (3), with has replacing does: (6)' Has John not played? (6) Has John played? If the English sentence is acquirable 'as such', its acquirability by speakers ofGerrnan is (co-)determined by the properties of the Gerrnan sentence. These properties, however, are in contrast to those in English (cf. 2.1.1). Given that fact and given, in addition, our view that G 1 as co-determinant is particularly effective in the early stage of L2 acquisition, the acquisition of the English sentence should consist in increasingly 'getting away' from the Gerrnan sentence, reflected in an increase of the 'purity' of the G2 attained. In that perspective, the acquisition of properties Pl through P3 figures as follows: Pl, non-movement of the verb: Relevant here are the sentences (2) and (10): (2) John does not play (10) *John plays not In the earliest stage of acquisition, (2) might be rated, in a 'grammaticality judgement' task, as ungrammatical, (10) as grammatical. The incorrect rating of (10) can be explained as resulting from the transfer of (2)G: (2) G Hans spielt nicht The grammarunderlying the two incorrectjudgements is 'pure' Gl, applied to L2. Pure Gl, if occurring at all, can be assumed to be very .short-lived. lt will soon be superseded by Gl/ G2, on the basis of which both sentences (2) and (10) are rated.as grammatical. The correct judgement of (2) here shows that the learner has established an L2 competence, the incorrectjudgement of (10) shows that this competence is notyet pure, not yet disconnected from G 1. G1/ G2 should be the longest stage in the L2 acquisition process, with a gradual increase of the strength of G2 relative to that of G 1. Probably, most of the learners end up there, without achieving pure G2. For the second Pl-aspect, acquisition of the moving of have, consider sentences (5) and (5)G: (5) John has not played (5) G Hans hat nicht gespielt (5) is no more acquirable through transfer from (5)G than (2) through transfer from (2)G. However, whether transfer has occurred here cannot be established through judgement of grammaticality: Correct judgement of (5) may conceal that the view of the sentence structure (vs surface) is erroneous, because Gl. At the same time, a Gl-view of (5) should gradually give way to a correct (G2-)view, establishable through cross-referring to other sentence types. Properties P2 and P3 should be acquired in the same way as Pl. The sentences relevant for them are the following: IFLUJL 28 (1999) An L2 subset parameter value in the grammar of L2 (G2) P2: (4) Today (,) John does not play (11) *Today plays John not (7) Today (,) John has not played (12) *Today has John not played P3: (3)' Does John not play? (13) *Plays John not? P3 holds the only case in which L2 can be acquired through transfer: (6r Has John not played? (6) Has John played? 3. A case of L2 sentence acquisition 3.0 The test 189 The test was designed so as to capture our view of L2 sentence acquisition, regarding both its process character and the purity of the G2 attained. The process character of L2 acquisition was captured by selecting groups of learners each of which could be assumed to represent a certain stage of G2 attainment. The groups selected were one form 6 (n = 19), 8 (21), 10 (19) and 12 ('Grundkurs'), students, accordingly, in their L2 years 2, 4, 6 and 8, aged 10, 12-13, 15 and 17-18. Regarding G2 purity, its content, and the basics of implementation, were stated above (2.3). In some more detail, the reasoning there, for Pl, is as follows: In English as in German, lnfl must be realized in AGR. The English verb, however, does not leave V, except after Spell-Out, in statements without not. That the English verb in no case goes across not is shown by sentence type (2), contrasted by (10), the literal translation of (2) G: (2) John does not play (2) G Hans spielt nicht (10) *John plays not The purity of a learner's G2 can now be established on the basis of the patterning of the judgements of (2) and (10). The following patterns are conceivable: sentence (2) sentence (10) -cor -cor 2 +cor -cor 3 -cor + cor Table IV: Patterns of judgement of related L2 sentences These patterns express the following variants of competence: 4 + cor + cor Pattern 1: pure G 1; pattern 2: G 1 and G2 held simultaneously (as would-be G2); pattern 3: neither G 1 nor G2; pattern 4: pure G2. Of these patterns, 3 is unlikely to really occur; 1, JFJLd 28 ( 1999) 190 Wilfried Weigl if it occurs, is of short duration, hence of marginal interest. We will thus concentrate on pattems 2 and 4. The (development of) G2 attainment can then be established by viewing the test results in terms of these patterns. Regarding sentences (10) and (2), identical with test sentences (TSs) 9 and 30, this view works as follows: We assume that those who judge TS 30 correctly know that 'L2 = L2', while those that judge TS 9 corr_ectly know that 'L2 * Ll '. Accordingly, those that judge TS 9 incorrectly assume that 'L2 =LI'; they thus have a G 1 component in their L2 grammar. The percentage of learners that do not have such a component then results from the subtraction of 'L2 = Ll' from 'L2 = L2': form 6 8 10 12 L2 * L1 (TS 9) 10 28 42 70 L2=Ll 90 72 58 30 L2 = L2 (TS 30) 95 95 100 100 G2 pure 5 23 42 70 Table V: Pure G 1, inferred In that perspective, the capturing of properties PI through P3 by TS pairs/ TSs is as follows 15: property TS pair/ TS property TS pair/ PS *Verb across not 9-30, 42-30 16 Verb to AGR 18,40 *lnfl across not 39-30 have across not 10, 43 Table Via: Properties PI in TS pairs/ TSs property TS pair/ TS property TS pair/ PS *lnf across subj. *lnfl accros subj. 47-23 *Verb across not+ subj. 22-23 *lnfl across subj. (have) 7-8 *Infl across subj. have across not *Verb across subj. 26-23 *lnfl across subj. 5-38 Table Vlb: Properties P2 in TS pairs/ TSs 15 For the total of the test sentences used, see 5. Appendix (on page 199). 16 In six cases, TSs with final today were used in addition to otherwise identical TSs, to make up for the occurrence of TSs with initial today. JFLIIIL 28 (1999) An L2 subset parameter value in the grammar of L2 (G2) property *Verb across not+ subj. *Verb across subj. have across subj. TS pairffS 28-4,25-31 35-37,36-20 24 Table Vlc: Properties P3 in TS pairs/ TSs 191 Sentences 9 (-30) through 24 were presented to the test subjects, for 'judgement of grammaticality'. 17 3.1 The test results The test results are as follows: *Verb across not 9-30,42-30 2 * 1 10 28 42 70 16 38 58 59 2 = 1 90 72 58 30 84 62 42 41 2=2 95 95 100 100 95 95 100 100 G2 5 23 42 70 11 33 58 59 *lnfl across not 39-30 2 * 1 95 100 100 100 2=1 5 0 0 0 2=2 95 95 100 100 G2 90 95 100 100 *Verb to AGR 18,40 2 * l 2=1 84 95 100 100 2=2 84 95 100 100 G2 84 95 100 100 *have across not 10,43 2 * 1 - 2=1 - 95 95 100 - 95 95 100 2=2 - 95 95 ! 00 - 95 95 ! 00 G2 - 95 95 ! 00 - 95 95 100 Table Vlla: Acquisition of properties PI 17 For the acceptability of 'judgement of grammaticality', see Sorace ( l996). f'ILlllllL 28 ( 1999) 192 Wilfried Weigl *Infl. across subject + *Verb across not+ subj. 22-23 2 * 1 68 81 100 100 2=1 32 19 0 0 2=2 95 90 95 92 G2 66 71 95 92 *lnfl. across subject + *Verb across subj. 26-23 2 * 1 42 62 79 67 2=1 58 38 21 33 2=2 95 90 95 92 G2 37 52 74 59 *lnfl. across subject 47-23 2 * 1 26 52 74 89 2 = 1 74 48 26 11 2=2 95 90 95 92 G2 21 42 69 81 *Infl. (with have) across subject 7-8 2 * 1 - 43 89 89 2 = 1 - 57 11 11 2=2 - 67 84 89 G2 - 100 73 78 *have across not+ *Infl (with have) across subject 5-38 2 * 1 - 43 79 85 2 = 1 - 57 21 15 2=2 - 95 84 92 G2 - 38 63 77 Table Vllb: Acquisition of properties P2 *Verb across not+ subj. 28-4,25-31 2 * 1 53 81 89 92 42 90 100 96 2 = 1 47 19 11 8 58 10 0 4 2=2 68 81 58 48 74 81 100 96 G2 21 62 42 40 16 71 100 92 fLlllL 28 (] 999) An L2 subset parameter value in the grammar of L2 (G2) 193 *Verb across subject 35-37, 36-20 2 * 1 68 57 58 52 42 48 37 48 2=1 32 43 42 48 58 52 63 52 2=2 84 95 100 100 100 86 100 100 G2 52 52 58 52 42 34 37 48 *have across subject 24 2 * 1 2=1 - 86 95 89 2=2 - 86 95 89 G2 - 86 95 89 Table Vllc: Acquisition of properties P3 3.2 Features of L2 sentence acquisition 3.2.1 Structure-related features 3.2.1.1 The degrees of L2 sentence property acquisition Regarding the (ultimate) degrees of acquisition of properties PI through P3, the previous figures warrant the following statements 18 : PI: G2 contains an 98% knowledge ofthe fact that the verb goes to AGR (TSs 18, 40), although it is probably not known that the verb moves only after Spell-Out. 19 Complementarily, there is a 100% knowledge ofthe fact that Infl does not go to V (TS 39). More important; however, the essential fact, that the English verb basically (before Spell-Out, and throughout in sentences with not) stays in V, is known by at most 70 % of the acquirers (TS 9). 20 At the same time, it is known by 100% that have does not stay in its birth place, but go to AGR, in any case (TSs 10, 43). P2: That in statements Infl does not surface in C but in AGR is G2-known by 81 % if the Inflbearer is do (Ts 47) and by 77 % ifthe Infl-bearer is have (TSs 7, 5). P3: That the verb does not go to C, not even in questions, is G2-known by only 50% of the acquirers (TSs 35, 36) and, surprisingly, by an even lesser percentage of 40 for a sentence with, additionally, not (TS 28). At the same time, it is known by 89 % of the acquirers that, in contrast to the verb, have goes to C (TS 24). In sum, then, the degrees of ultimate attainment of L2 sentence properties are as follows: 18 In the present section, we use the figures from form 12. 19 Whether or not this is known could not be captured by our test (and probably cannot be captured by any test). 20 If we include the ... today-sentence, TS 42, attainment is a mere 64%. IFILUllll., 28 ( 1999) 194 Wilfried Weigl *Verb in AGR *Infl in C (in statements) Verb in C (in questions) 70 81 50 100 77 89 have inAGR *Infl in C (in statements) have in C (in questions) Table VIII: Degrees of acquisition of sentence properties 3.2.1.2 ls there L2 sentence (vs property) acquisition? While the G2-knowledge of the sentence properties was established as such, by means of the test sentences, knowledge of the sentence as a unit cannot be established that way. lt can, however, be established indirectly, through inference from the property knowledge facts. 21 The reasoning here is as follows: Knowledge of the sentence comprises knowledge of the sentence positions AGR and C, and of the constraints on the respective occurrence of the verb in those positions. The possible constellations here are as follows: constellation Verb in AGR Verb in C 1 + + 2 + 3 *4 + Table IX: Constraints on the occurrence of the verb In words: If the verb can go to AGR, it may or may not also go to C (German and French, respectively). If the verb cannot go to AGR, it cannot go to C either (English). The latter fact, however, is ignored by a considerable number of our learners: That the verb in English does not go to C is known by a considerably lower number (50%) than that the verb does not go to AGR (70%). lt can thus be concluded that the English sentence (without have) is acquired by only half of the learners. 3.2.2 Process-related features 3.2.2.1 The structure of the growth of G2 In a G2-growth view, there are two categories of L2 sentence property acquisition: (1) that growth is not called for (nor really possible), because there is a high degree of attainment at an early stage of the course (cf. TSs 10, 43), and (2) that growth is called for, because there is a low degree of early attainment. In the present section, we deal with the key cases of category 2. 21 We concentrate here on 'Verb' and 'have', leaving aside 'Infl'. lFLUIL 28 ( 1999) An L2 subset parameter value in the grammar of L2 (G2) 195 The key cases coming under category 2, and the G2-growth of their respective properties are as follows: TS 9 5 23 42 70 TS42 11 33 58 59 Table Xa: G2-knowledge of *Verb in AGR TS47 21 42 69 81 TS 26 37 52 74 59 TS 22 66 71 95 92 Table Xb: G2-knowledge of *lnfl in C TS 35 52 52 58 52 TS 36 42 34 37 48 Table Xe: G2-knowledge of *Verb in C The rates of growth derivable from these figures are as follows 22 : 6 to 8 8 to 10 IOto 12 total *Verb in AGR 6 4,5 2,5 13 *Infl. in C 2 3 0 4,7 *Verb in C 0 0 0 0 Table Xd: G2-growth of PI through P3 As shown by these figures, G2-growth has a clear-cut structure, both 'vertically', in terms of properties, and 'horizontally', in terms of acquisition 'phase'. 'Vertical structure': The total of the growth effected, throughout the course, is (significantly) higher for '*Verb in AGR' than for '*Infl in C', and (significantly) higher for the latter than for '*Verb in C'. 'Horizontal structure': The strongest growth occurs between forms 6 and 8, the second strongest between 8 and 10, and the weakest between 10 and 12. '*Verb in C' is the exception here, with no growth at all. 22 Statistically, the growth rates are established as follows, for TS 9 for example: The basic fact is the significance of the G2-differences between every two subsequent forms. For TS 9, these differences are significant on the probability-levels .001 (form 6 to form 8), .02 (8 to 10) and .01 (10 to 12). P-levels are then converted into growth rates by expressing them through figures: .001 through 6, .01 through 5 etc. The growth rates for TS 9 then are 6 (6 to 8), 4 and 5. IFIL1llllL 28 ( 1999) 196 Wilfried Weigl 3.2.2.2 L2 ultimate attainment For the key cases, the ultimate attainment of L2 sentence knowledge is as follows (figures from Table VIII): * Verb in AGR * Infl in C with verb *Verb in C 70 81 50 100 77 89 have in AGR *Infl in C with have have in C Table XI: Degrees of ultimate attainment According to these figures, the structure of L2 ultimate attainment is as follows: (1) Across the properties with 'Verb' (vs 'have'), there is a difference which, except for '*Infl in C', corresponds to the difference in growth: '*Verb in AGR' is attained to a (significantly) higher degree than '*Verb in C'. (2) Except for '*Infl in C', again, there is a difference between Infl-realization with 'verb' and realization with 'have ': A property is attained to a (significantly) higher degree if it has 'have' than if it has 'Verb'. 3.3 A major L2 acquisition determinant: Gl Above (2.1.2) it has been assumed that transfer is practised in the case of L2/ Ll surface identity; in that case G 1 is assumed to be the major determinant of acquisition. That this assumption is correct is bome out by the figures for the acquisition of the relevant sentences. They are as follows: PI (TSs I 0, 43) / Zeile '2 = l' P3 (24) 95 86 Table Xlla: Transfer case ( 1) PI (9, 42) / Zeile '2 = l' 87 67 P2 (5,7) - 57 (22,26) 45 28 P3 (28, 25) 52 14 (35,36) 45 47 Table Xllb: Transfer case (3) 95 95 50 16 10 5 52 100 89 35 13 16 6 50 Transfer case ( 1) is straightforward: In it, L2/ Ll surface identity corresponds to full identity. Hence, a sentence like Has John played? is acquirable through transfer, immediately, and such acquisition really occurs, as shown by the earliness of acquisition, and by the high degree of acquisition attained in the first phase. IFILllllL 28 (1999) An L2 subset parameter value in the grammar of L2 (G2) 197 Case (3) is the case where transfer first results in an erroneous G 1, whose erroneousness, however, because evident, is repairable. Repair is indeed what happens in the acquisition of the case (3) sentences, with the exception of sentences 35/ 36: With the non-35/ 36 sentences, there is relatively strong transfer in the first phase of acquisition, resulting in a relatively high degree of erroneousness. This erroneousness, however, undergoes continuous repair, hence decreases throughout the course. In its final phase (form 12), the degree of erroneousness is lowest, significantly lower than in form 6 (form 8 for have). Not so, however, with sentences 35/ 36. In their case,.there is no repair of erroneousness, and the final degree of erroneousness is even higher than the initial one. The reason for the difference may be that the non-35/ 36 sentences each contain sentenceinitial today or (land) not, each of which is an element external to any (potential) position of verb and of subject, hence usable, in the search for these positions, as a point of reference. As such, today and not seem to be indispensable. lt is because of that indispensability that, while Plays John not? is repairable, Plays John? is not. 3.4 Summary and synthesis In the case of the L2 sentence acquisition investigated, the facts are as follows: (1) That the English verb does not go to AGR is (ultimately) acquired by 70% of the learners (against 100% for have), that the verb does not go to C by only 50% (against 89% for have). From these figures it follows that knowledge of the L2 sentence, comprising the simultaneous knowledge of both AGRand C-facts, is acquired by only half of the learners. 23 (2) The growth ofG2 has the following features: The growth ofthe knowledge of '*Verb in AGR' is higher than that of '*Verb in C' ('vertical structure'), and, 'horizontally', O2growth is strongest between forms 6 and 8, second strongest between 8 and 10, and weakest after 10. (3) L2 sentence acquisition is largely determined by the effectiveness of GI, through transfer. The result of transfer tends to be negative: an erroneous G2 which, if it is repairable, it takes time to repair. In the perspective of our presuppositions, these facts figure as follows: (1) Our acquirability hypothesis was too undifferentiated. lt stated that the effectiveness of G 1, strong at an early stage, would in all cases be reduced in the course of L2 acquisition. Reduction did occur in the acquisition of_'*Verb in A_GR', but did not occur in the acquisition of '*Verb in C'. This latter fact can be taken to show ( 1) that transfer plays a decisive role ('Verb in C' being the essential property of the German sentence), and (2) that the L2 input made available to the Gymnasium students was insufficient. 23 lt is here that the compariso,i with White 1992 is called for. White dealt with 'francophone Ieamers of ESL', hence '*Verb in C' was outside her scope. The '*Verb in AGR' -result was that "movement of a main verb past a negative [...] is never assumed to be possible by francophone leamers of ESL" (283). Here, however, in contrast to our result, transfer is likely to have been (positively) effective: *John plays not may be acquired on the basis of Jean ne joue pas, if not is seen as equivalent of ne. IFILIIBL 28 ( 1999) 198 Wilfried Weigl (2) All this is in accord with our general view of L2 acquisition: There is a basic tendency towards transfer, towards the adopting, by the learner, of L1 parameter setting values as part of the grammar of L2. Two things matter then: (1) The relationship between the L2/ Ll values, and (2) the inoperativeness of the Subset Principle. In the present case, the L2 parameter setting value is 'subset', the LI value 'superset'. Thus, the German learner has to change from a superset to a subset value. This would be easy if the Subset_ Principle were available. As it is not, the change required, from superset to subset, is the most difficult one of the changes imaginable. Accordingly, it could only be achieved if its making were supported by measures against the effectiveness of G 1. 4. L2 sentence acquisition in a teaching perspective In view of the time allotted to L2, the L2 sentence acquisition achieved must be rated as unsatisfactory. Whether it can be improved is an empirical question, which cannot be answered here. What can be done, however, is to name the direction in which measures towards an improvement should go, with such measures referring (1) to the arrangement of the matter taught, and (2) to the mode of its teaching. The arrangement of the matter taught, of 'the English sentence', should be made in terms of sentence types. 24 What matters here is the difference between sentence types with regard to learnability: The type whose L2 (vs Ll) structure is reflected in 'surface facts' (Today John plays) is learnable 'directly', without recourse to another type, whereas the type without such reflection (John plays) is not. Given that constellation, Today John plays should be given priority as object of teaching. lts knowledge should be generalizable upon John plays and thus promote the acquisition of that latter type. (Complementarily, it should contribute to the acquisition of *Plays John? , which has of course tobe induced, mainly, through the presentation of Does John play ? ). 25 Consider, finally, the mode of teaching recommendable, specifically the shaping of the L2 input. This shaping is based upon the view that L2 can be learned "implicitly (via subconscious abstraction of patterns from input data)", but not "subconsciously (i. e., incidentally, without intention; while doing something else)" (Long 1996: 427). In this view, the learner, to succeed in L2 acquisition, must be made tofocus onform, "to attend to language as object during a generally meaning-oriented activity" (Long 1996: 429). The means to that end are referred to as 'input enhancement' (Sharwood Smith 1993). Enhancement measures that have been proposed are "prior instructions to attend to both form and meaning [...], showing rules applied to examples [...], multiple-choice margin glosses [ .. ], and other forms" (Long 1996: 428). In our view, such measures are flawed: Being L2-explicit, they appeal to cognitive processors rather than to the language module. Appeal to the language module, however, seems achievable, through abstention from 24 The implementation of 'arrange' and of 'give priority' (below) need not be specified here (as we are not dealing with syllabus construction). 25 The arrangement principle instantiated here could be referred to as 'Priority to phenomena that are both leamable and informative'. lFL1lllL 28 (1999) An L2 subset parameter value in the grammar of L2 (G2) 199 explicitness, and the use of sentences from Ll, in an L2/ Ll arrangement that makes the learner aware of the contrast between L2 and L 1. Becoming aware of that contrast, on the object- (vs meta-) language level, would probably allow the learner to improve his L2 acquisition. 5. Appendix: The test sentences TS 4 Does John not play? TS 5 Today has John not played. TS 7 Today has John played. TS 8 Today John has played. TS 9 John plays not. TS 10 John has not played. TS 18 John plays. TS 20 Does John play today? TS 22 Today plays John not. TS 23 Today John does not play. TS 24 Has John played? TS 25 Plays John today not? TS 26 Today plays John. TS 28 Plays John not? TS 30 John does not play. TS 31 Doesn't John play today? TS 35 Plays John? TS 36 Plays John today? TS 37 Does John play? TS 38 Today John has not played. TS 39 John not plays. TS 40 John plays today. TS42 John plays not today. TS 43 John has played~ TS47 Today does John play. References CHOMSKY, Noam (1995): The Minima/ ist Program. Cambridge, Mass.: The MIT Press. EUBANK, Lynn (1992): "Verb movement, agreement, and tense in L2 acquisition". In: MEISEL, Jürgen M. (ed.): Acquisition ofVerb Placement. Dordrecht: Kluwer, 225-244. FELIX, Sascha W. (1985): "More evidence on competing cognitive systems". In: Second Language Research 1, 47-72. GASS, Susan (1996): "Second Language Acquisition and Linguistic Theory: The Role of Language Transfer". In: RITCHIE/ BHATIA (eds.), 317-345. GASS, Susan M. / SELINKER, Larry (1994): Second Language Acquisition: An lntroductory Course. Hillsdale, N. J.: Lawrence Erlbaum. HA WKINS, Roger/ TOWELL, Richard/ BAZERGUI, Nives ( 1993): "Universal Grammar and the acquisition of French verb movement by native speakers of English". In: Second Language Research 9, 189-233. LIGHTBOWN, Patsy M. / SPADA, Nina/ WALLACE, Robert ( 1980): "Some effects of instruction on child and adolescent ESL learners". In: SCARCELLA, Robin C. / KRASHEN, Stephen D. (eds.): Research in Second Language Acquisition. Selected Papers of the Los Angeles Second Language Acquisition Forum. Rowley, Mass.: Newbury House, 162-172. LONG, Michael H. (1991 ): "Focus on Form: A Design Feature in Language Teaching Methodology". In: DE BOT, Kees/ GINS BERG, Ralph B. / KRAMSCH, Claire (eds.): Foreign Language Research in Cross- Cultural Perspective. Amsterdam/ Philadelphia: John Benjamins, 39-52. LONG, Michael H. (1996): "The Role ofthe Linguistic Environment in Second Language Acquisition". In: RITCHIEIBHATIA (eds.), 413-468. IFLIIL 28 ( I999) 200 Wilfried Weigl PIENEMANN, Manfred (1984): "Psycholinguistic constraints on the teachability oflanguages". In: Studies in Second Language Acquisition 6, 186-214. PLATZACK, Christer/ HOLMBERG, Anders (1989): "The Role of AGR and Finiteness". In: Working Papers in Scandinavian Syntax 43, 51-76. RADFORD, Andrew (1997): Syntax: a minimalist introduction. Cambridge/ Melbourne: Cambridge University Press. RITCHIE, William C. / BHATIA, Tej K. (eds.) (1996): Handbook of Second Language Acquisition. San Diego/ New York/ Boston/ London/ Sydney/ Tokyo/ Toronto: Academic Press. ROEPER, Thomas (1992): "From the Initial State to V2: Acquisition Principles in Action". In: MEISEL, Jürgen M. (ed.): The Acquisition ofVerb Placement. Dordrecht: Kluwer, 333-370. SHARWOOD SMITH, Michael (1993): "Input enhancement in instructed SLA: Theoretical bases". In: Studies in Second Language Acquisition 15, 165-179. SORACE, Antonella (1996): "The Use of Acceptability Judgements in Second Language Acquisition Research". In: RJTCHIE/ BHATIA (eds.), 375-409. UzIEL, Sigal (1993): "Resetting Universal Grammar parameters: evidence from Second Language Acquisition of Subjacency and the Empty Category Principle". In: Second Language Research 9, 49-83. WEIGL, Wilfried (1998): "G2 constrained by G l: The expression of dativity acquired by German learners of English" [Unpublished Manuscript]. WHITE, Lydia ( l 991 a): "Adverb placement in second language acquisition: some effects of positive and negative evidence in the classroom". In: Second Language Research 7, l 33-l 6 l. WHITE, Lydia (1991b): 'The Verb-Movement Parameter in Second Language Acquisition". In: Language Acquisition 1, 337-360. WHITE, Lydia ( I992): "Long and Short Verb Movement in Second Language Acquisition". In: Canadian Journal of Linguistics/ Revue canadienne de Linguistique 37, 273-286. WHITE, Lydia ( 1996): "Universal Grammar and Second Language Acquisition: Current Trendsand New Directions". In: RITCHIEIBHATIA (eds.), 85-120. ZOBL, Helmut (1995): "Converging Evidence for the 'Acquisition-Learning' Distinction". In: Applied Linguistics 16, 35-56. IFLwL 28 ( 1999)