Knowledge of syntax and verbal morphology in adolescent L 2 English 1 : A Feature Reassembly account

This study investigates knowledge of both syntax and verbal morphology by L2 English classroom learners with L1 Japanese in affirmative sentences with VP-adverbs (e.g., She usually eats breakfast at nine). It proposes that the findings are consistent with the Feature Reassembly Hypothesis (Lardiere, 2008, 2009) which the Full Transfer Full Access Hypothesis underlies. Results are obtained from an elicited production task: for written data with 90 junior high school students (12-15 years old) and 30 university students (19-20 years old), as well as for spoken data with 12 junior high school students (12-13 years old). It is observed that the pattern of use of verb forms is different from that found in other L2 English studies. The key differences are: (1) high omission of copula is in copula is+adverb contexts; (2) two kinds of commission error where tense/aspect forms replace each other. It is argued that such variability in verbal morphology could be accounted for by the differences in the processes and conditions by which relevant morphosyntactic and semantic features are assembled between L1 and L2, which is consistent with the key claim of the Feature Reassembly Hypothesis.


Introduction
There has been considerable debate over what the sources of morphological variability are in L2 acquisition.A number of generative L2 studies 5 have reported that L2 learners make omission and misuse/overuse errors in the production of inflectional morphology, regardless of differences in age, L1 (first language) background, and L2 (second language) proficiency.Recently, Lardiere (2008Lardiere ( , 2009) ) has proposed a new feature-based account for L2 learners' difficulty in morphological representation.This study aims to explain L1 Japanese learners' variability in the production of verbal morphology, applying Lardiere's new approach which focuses on L2 speakers' failure to re-assemble morphosyntactic and semantic features into L2 lexical items in a target-like way.Participants were Japanese adolescent classroom learners of L2 English in both early stages and later development.A picture-stimulus task was designed to elicit both spoken and written production data and to allow a comparison to be made between verb morphology and syntactic properties (verb placement and subject raising with Nominative Case marking) in the same obligatory contexts (affirmative sentences with-VP-adverbs).The findings suggest that the Feature Reassembly Hypothesis could explain Japanese learners' systematic variability in: (1) the omission of copula is in copula is+adverb contexts; (2) two kinds of commission error where tense/aspect forms replace each other.
The study is composed of six sections, including this Introduction.Section 2 outlines the Feature Reassembly Hypothesis account of the morphology produced by L2 learners and the differences that exist in verbal morphology between English and Japanese.Section 3 is concerned with the methodology used.In Section 4, the results are presented, separating syntax and morphology.Section 5 is a discussion of the findings.Finally in Section 6, a conclusion is drawn, including implications for future research on variability in verbal morphology by L2 learners.

Theoretical background
2.1.Morphology production in the Feature Reassembly Hypothesis (Lardiere, 2008(Lardiere, , 2009) ) Lardiere argues that persistent L2 variable phenomena (e.g., omission, misuse, and overuse of inflectional morphemes) cannot be accounted for by binary parameter settings which represent "all-or-nothing phenomena" (2008:108).The Feature Reassembly Hypothesis (henceforth, the FRH), which is built on the claims of the Full Transfer Full Access (FTFA), proposes that feature re-assembly might be a source of persistent difficulty for L2 speakers, although any feature contrasts can be detected and ultimately acquired.Framed within the Minimalist Program (Chomsky 1995(Chomsky , 1998(Chomsky , 2001(Chomsky , 2005)), the FRH attributes L2 morphological variability to different, language-specific manners and conditions in which features are selected and assembled 6 , not to failure to select parameterized features causing permanent representational deficit in the L2 7 .It is argued that L2 speakers have to select and reassemble "the right combination of features" (2009:215) from an L1 "fully developed system of assembled lexical items and functional categories" (2009: 185) "into the right lexical items" under "the appropriate conditioning environments for their expression" (2009:215).   1 summarises the differences between the three languages.
In English, plural marking is obligatory: [+plural] is combined with [±human]/[±definite] and agrees with quantifiers denoting plurality (e.g., six students, several students, and both students, 2008:122).By contrast, in Chinese, plural marking is optional: [+plural] is represented only by suffix-men, which allows only two features [+human]/[+definite] and also prohibits co-occurrence with quantifiers (e.g., *san-ge xuesheng-men 'three-CL student-PL ', 2009:196) 11 .Given the differences between English and Chinese, Lardiere assumes that L1 English speakers of L2 Chinese are "initially likely to overgeneralize the applicability of plural marking in Chinese " -men (2009:198), while L1 Chinese speakers of L2 English show "developmental undersuppliance of plural marking" due to "non-obligatoriness" (2009:196).It is argued that L1 English speakers would need to reassemble [+plural] with [+definite] 12 into an L2 Chinese plural suffix -men, while L1 Chinese speakers of L2 English would need to learn obligatoriness and extend co-occurrence with [-human]/[-definite](2008:123), and quantifiers.In Korean, the plural suffix-tul has more complicated distribution than in English and Chinese.Lardiere points out that the reassembling of a Korean plural item -tul requires L1 English speakers to determine "possible conditioning environments" (2009:209) and to understand multiple interpretations of lexical semantic features.

Verbal morphology in English and Japanese
English and Japanese have tense/grammatical aspect markers in common, as shown in Table 2.The only difference lies in the surface form of the imperfective aspect markers: in English, progressive is marked by the discontinuous morpheme be + -ing, while in Japanese there is a continuous verb-final affix tei-ru/-tei-ta.Because of this difference, Japanese learners of L2 English may have difficulty in reassembling features [-past] [+progressive] from a continuous morpheme in the L1 to a discontinuous one in the L2.In spite of the similarity in overt forms, there are crucial differences in the interpretation of tense/aspect between English and Japanese.Japanese exhibits multiple aspectual interpretations of each verbal inflection.With the non-past imperfective marker, a clear semantic asymmetry is observed between English 'is+Ving' and Japanese 'V+teiru' (Slabakova, 2008:162).Seen from a different perspective, this suggests multiple representations of each semantic feature in Japanese: adverbs/adverbial phrases enable multiple verbal morphemes to represent one aspectual property in Japanese.An adverb maiasa 'every morning' realises a semantic feature [+habitual], with not only non-past marker-ru but also imperfective marker-teiru, as given in example 5 (a/b) (see 2 b, 3 a).
Likewise, the combination of an adverb moo 'already' and overt numeral+classifier 3 satu 'three books' represents a semantic feature [+telic] with imperfective -teiru as well as past/perfective -ta/da, as given in example 6 (a/b).( 6) [+Telic] =She has already read three books.a. Past/perfective-da Kanojyo-wa moo hon-o 3 satu yom-da She-TOP already book-ACC 3-CL read-PERFECTIVE 'She has already read three books.'b.Imperfective-deiru Kanojyo-wa moo hon-o 3 satu yom-deiru She-TOP already book-ACC 3-CL read-PROGRESSIVE 'She has already read three books (and is still reading).'*She is already reading three books.
In addition, differences in derivation are found between English state verbs and the Japanese counterparts, as shown in Table 4.

Achievement
+-teiru =resultative state know believe have love In English, some state (e.g., know, believe, have, love13 ) verbs, as well as achievement (e.g., recognise, notice, find) verbs, never incorporate the imperfective marker because is+Ving is restricted to progressive interpretation.On the contrary, the Japanese counterparts have to take -teiru to denote state (e.g., sit-teiru) because the non-finite forms (e.g., si-ru) are interpreted as a dynamic, punctual action, like achievement verbs.This suggests that the Japanese corresponding verbs can take -teiru to denote a resultative state: a duration of state [+durative] [+stative], as a result of a completed action [+punctual] [+dynamic]14 .Furthermore, a Japanese imperfective marker-teiru can denote a resultative state, by attaching to verbs in other lexical aspect class than state (example 7): (1) accomplishment verbs with adverbs/adverbial phrases imply a resultative state after a completed action (example 8, repeated 2c); (2) achievement verbs denote a process directing to a natural endpoint (example 9 a/b).In particular, the absence of overt mechanism for telicity marking allows adverbs/adverbial phrases and overt numerals to play a crucial role in Japanese.In English, telicity is encoded by a system for cardinality on nominal features [+/-definite][+/-number]: articles a/the and obligatory plural-s represent a semantic feature [+/-telic] in verbal morphology -d/t, as given in example 10 (a/b).

Research questions and predictions
This study aims to test the generalizability of Lardiere's recent approach to different grammatical properties and to a new group of L2 English learners.The research question addressed in this study is whether the FRH can account for variability in verbal morphology production by Japanese adolescent classroom learners.The prediction is that L1 Japanese learners will provide positive evidence for four key claims underlying the FRH in affirmative with-VP-adverb contexts: (1) transfer will equally affect functional and lexical categories; (2) development away from the initial state grammar will be consistent with Universal Grammar(UG)-constrained restructuring; (3) there will be no correlations between the acquisition of morphology and the acquisition of syntactic properties; (4) where Japanese and English differ in the realisation of a syntactic or morphological property, Japanese learners of English will transfer the Japanese property into their L2 grammars.

Participants
All of the 132 participants in this study were adolescent Japanese classroom learners of English from national and private educational institutions in urban areas of Japan.The participants were divided into two groups.First, a total of 102 junior high school students from the 1 st to 3 rd grades 18 were recruited to investigate the initial state 19 and early development in L2 English acquisition: 90 students from all the grades for the written task, and 12 students of the 1 st grade for the spoken task.Second, a group of 30 university students in the 2 nd year was selected to examine later L2 development.There are reasons why junior high school students and the 2 nd year university students were tested.First, the Japanese students started receiving intensive, formal English teaching in the 1st grade of junior high schools when the experiments were carried out 20 .Second, the compulsory TOEIC course was scheduled for the 2 nd year in the university which kindly participated in this study.A proficiency test was replaced by length of English exposure, grade/age 21 .In addition, for junior high school students, a linguistic background questionnaire was conducted to exclude learners who had received intensive, regular, and long-term English education in either Japan or English-speaking countries before and after entering junior high schools; the junior high schools were selected on the basis of a similar deviation value 22 .For university students, the TOEIC (Test of English for International Communication) 23 score was employed to recruit students whose latest score at the time of the experiment was between 650 and 680; the 2 nd year university students24 had the same non-English related major as well as the same number of English class hours25 in the 1 st and 2 nd year.As summarised in Table 5, the participants were subdivided into four groups, based on both the length of exposure to English and grade (for junior high schools) /age (for university students).Each group shares the same number of participants for the written task (30 students).The testing for all of the four groups was carried out in late January 2011, to make the difference in length of English exposure equivalent.
the same initial state: Schwartz andSprouse (1994, 1996:1 year); Vainikka and Young-Scholten (1994:1.5-24years, 1996a: 10-25 months); Eubank (1997: 4 months-7 years); Epstein et al. (1996: 7 years). 20Since April 2011, English teaching has started in the 5th grade of primary schools, two years earlier than that in the previous system. 21Because of their tightly organised curriculum, the junior high schools who offered kind cooperation asked the task: (1) to take less than 40 minutes (including distribution of materials and instructions); (2) to be written production data collected by their teachers.As a result, the experiment constituted a task (63 questions in 30minutes) and a questionnaire (5 minutes in 12 questions) and it was impossible to recruit the equivalent number of participants for the spoken task. 22In Japan, it is regarded as an indicator of schools' academic ability.

Materials
This study employed a picture-stimulus task, which was designed: (1) to elicit both written and spoken production data from the least proficient learner; (2) to allow a comparison to be made between syntactic and morphological properties in the same question item; (3) to elicit 3 rd person singular personal pronouns as subjects 27 .Each of the 63 test items consists of: (1) a Japanese question sentence, after which each Japanese word in brackets ('Answer' or 'Question') instructed the participants which type to write or speak: either making an 'answer' or forming a 'question'; (2) a picture28 with one to three English words to prompt participants to produce writing and speaking in English, as given in the example item below, in which English gloss is added for the purpose of illustration.

nine, usually, breakfast
This study focuses on 7 properties of affirmative sentences with 4 frequency VP-adverbs (always, often, sometimes, usually).As shown in

Procedure
This study collected elicited production data from learners who had been exposed to English via formal instruction in a classroom.To make the elicited production as spontaneous as possible, three main procedures were followed: (1) participants were informed of the time they had taken at 5-minute intervals, during which they should have answered 10 questions, to ensure that they provided answers to all of the 63 questions in 30 minutes, without returning to the previous questions; (2) no detailed oral instructions were offered before and during the task and also no revisions were allowed 29 , to prevent participants from drawing on their metalinguistic knowledge.Written instructions before two sample questions made sure that they were required to write or speak whatever they first thought of, without worrying about the correctness 30 ; (3) Japanese translations of 10 English prompt words (12.3 % of the total) were offered and spelling errors/Japanese Katakana 31 were allowed, to prevent them from being distracted by English words' meanings and spellings.As explained in section 3.1, this study had no proficiency test, and there were only a small number of participants in the spoken production task.To make the written production data more reliable, the number and age of participants were increased (see Table 1) so as to fully observe gradual development in each of the three early stages and to compare it with later development in L2 acquisition.The written production data were collected by each teacher of all four groups in the English class.In the spoken task, the author was allowed to collect the data after school.Each of the 12 participants recorded his/her own oral answers with a portable recorder.The students sat in alternate seats, in order that they should not hear other students' utterances, or disturb each other's recordings.The recorded spoken production data were transcribed and analysed in accordance with the same scoring criteria as the written data.

Scoring criteria
In English affirmative clauses with main verbs, adverbs precede main verbs (13 a), while in those with copula be, they follow copula be (13 b).As illustrated in Table 7, verb placement in affirmative sentences with VP-adverbs was given either 1 (for the expected orders above, regardless of missing/faulty verbal inflections), or 0 (for any other orders).The combination 'be+adverb+bareV' was not scored under the heading of 'syntax', but under the heading of 'morphology'.

Verb placement with adverbs
Written and spoken data showed high accuracy rates in the placement of both main verbs and copular be in relation to adverbs, as shown in Tables 10/11.The 7 th grade participants in the spoken task produced less accurately in all categories (particularly in past irregular contexts) than the same grade participants did in the written task: however, the accuracy rates in the spoken data fall beyond the 40%-60% range (which is interpreted as variability, Leung, 2006:179).

Subject raising and case marking
Written data showed 100% target-like suppliance rates of overt subjects with Nominative Case in obligatory affirmative-with-VP-adverb contexts.The successful production of overt subjects is replicated in other existing studies (Lardiere, 1998 a, b;Haznedar, 2001;Inonin and Wexler, 2002;White, 2003;Goad, White and Steele, 2003), regardless of the fact that the L2 learners' L133 permits null subjects.

Participant Groups
Written Spoken

4.1.3.'be+bare V' construction
The constructions 'be+adverb+non-finite verb' were observed in affirmative with-VP-adverb sentences (examples 14 a/b) in both written and spoken data.As shown in Table 13, the number of cases was small and showed a tendency to decrease as the length of L2 English exposure increases.( 14  As shown in Figure 1, L1 Japanese adolescent learners showed that omission errors are higher than commission ones, which is similar to the results of L2 child speakers with various L1 backgrounds 38 in Paradis' study (2005)  39 .
Figure 1: Distribution of error types in affirmative with VP-adverb sentences (%)

Two kinds of bidirectional misuse
This study found two kinds of bidirectional misuse involving errors of commission (i.e., using a form in a non-target context).The first bidirectional misuse was observed between copula/auxiliary be and main verbs with VP-adverbs (Table 18): bare/non-past tense V was used in progressive contexts where is+Ving is the target form (15 a/b), while in past/non-past contexts, progressive is/was+Ving were used (16 a/b), although the number of instances was small.[U 2 nd P10, Written] (No.8She is eating breakfast now.) 43 The results in this study found the high accuracy rates (76.8% in the written data; 88.6% in the spoken data), as in other existing studies: (1) Haznedar ( 2001) "nearly at-ceiling with correct be-Neg placement" (Ionin, 2012); (2) Ionin and Wexler (2002) "100% correct be-Neg placement (33 tokens)" (Ionin, 2013); (3) Lardiere, 2007 (see footnote 43).( 16) [Vs→is+Ving] a.She _ playing the piano every day.
[JH 9 th P29, Written] (No. 40 She plays the piano every day.) The second bidirectional misuse was found between the non-past tense marker and the past tense marker (-s↔-d)  [JH 9 th P1, Written] (No.2She often played the piano last year.)

1. Full Transfer in the L2 initial state
The results showed that syntactic properties of the L1 were transferred into the L2 in the initial stages of L2 acquisition.First, both main verbs and copula be were correctly placed with adverbs in the written and spoken task.This suggests that the feature which determines V-to-T movement is specified [-strong] in the Tense category 46 .Second, overt subjects were almost perfectly produced in clause initial position and nominative case marking was 100 % target-like in both types of production data.This might provide potential evidence for the presence of a Tense category with related features (case, agreement, finiteness, and EPP): nominative case is assigned via "an agreement relationship between a finite T probe 47 and a nominal goal" (Radford, 2009:283); case-marked subjects surface as a result of subject raising which is triggered by an EPP feature in Tense.

UG access in L2 development
The findings suggest that L2 initial grammars develop by interaction with UG.Firstly, L2 Japanese adolescent classroom learners showed a similar trend of error distribution (omission >commission) to that of the L2 child speakers with various L1 backgrounds in the Paradis' study (2005).In addition, the commission errors exhibited no randomness: (1) no cases of 45 Affirmative-with-VP-adverbs: Past: 2,23,30,36,31,62 (no temporal adverbials).Auxiliary (present/affirmative): No. 8, 33, 50 (now).46 Consensus is yet to be reached among linguists about whether Japanese has verb raising or not.47 A probe is a head trying to find a nominal goal which can delete any uninterpretable features on the probe within its complement (Radford, 2009: 475).failure in number in the use of copula be (e.g., He are, She am, They is); (2) two kinds of bidirectional misuse (Vs ↔is+Ving; Vs↔Vd).Such similarity and regularity in results might be a possible reflection of UG constraints.Secondly, the 'be +bareV' construction, which is found in neither L1 Japanese nor L2 English, was produced in affirmative-with-VP-adverb contexts ('be+adverb+bareV'), as in the other obligatory contexts in this study 48 .The overgeneration of be forms with non-finite verbs was also observed in other L2 English studies of early learners with different L1 backgrounds and ages (see Table 14).This suggests that universal linguistic principles guide L2 early leaners in identifying be forms as free all-purpose finiteness markers.This leads an assumption that the 'be+bareV' order suggests a process guided by UG in early L2 development. Th findings, indicating UG involvement in L2 developmental processes, as well as the presence of functional category Tense with specified features in the L2 initial state (see section 5.1), are consistent with the Full Transfer Full Access Hypothesis.

3. A relation between syntax and morphology
A noticeable asymmetry in accuracy rates was found between verb placement and verbal morphology production.Verb placement over adverbs exhibited much more accuracy than verbal morphology suppliance (see Tables 10/11, 15/16) in the same test item.
Success in verb placement, as well as subject raising with Nominative Case marking (see section 4.1.2.), could provide potential evidence for the absence of a deficit in syntactic knowledge.This suggests a dissociation between syntactic knowledge and morphological production: persistent problems in morphology production are not attributable to impaired grammar.This is an underlying assumption of the Feature Reassembly Hypothesis.

L1 effects
Verbal morphology production showed variability in affirmative with VP-adverb contexts: (1) high omission of copula is: (2) commission errors involving the substitution of one morphological form for another.The importance of both as evidence for L1 transfer is considered below.

High omission of copula is in 'copula is+adverbs'
Copula 'is' demonstrated high omission only in with-VP-adverb contexts (see Table 17).The phenomenon is suggestive of a failure in the insertion of the Vocabulary entry /is/ into a syntactic node, not of a deficit in syntactic knowledge.This might be attributable to an extraneous syntactic node for adverbs, which both creates "'syntactic' distance" (Wakabayashi and Yamazaki, 2009 49 ) and increases computational loads in activating the entry for /is/ with features (Hawkins and Casillas, 2008).The high omission of copula is with VP-adverbs provides possible evidence for a failure in reassembling features into L2 lexical items.In other words, an extraneous syntactic node for adverbs contributes to increasing computational loads, which causes L2 learners to fail to reassemble multiple features [+finite, -past, 3 rd person, +singular] into a single L2 lexical item is.This is consistent with response times in spoken data: Japanese learners at the earliest stages of learning took more time before producing 'is+adverb' sentences, whilst they produced 'is+not' sentences instantly, as well as 'is+Ø' sentences.Other findings provide further support for the argument that the omission in 'is+adverb' is not a direct reflection of impaired syntactic knowledge: (1) target-like is placement (see Tables 3/4); (2) rare cases (6.7%: 14/208) of misplaced is cases (e.g., He often is tired).

Two kinds of bidirectional misuse
The two kinds of commission errors could be a possible reflection of L1 effects: aspectual ambiguity in Japanese 50 .In English, verbal morphology interacts with aspectual properties inherent to verbs; nominal morphology serves to encode telicity.By contrast, in Japanese, rather than verbal or nominal morphology, adverbials play a central role in determining aspectual interpretations.The more prominent role of adverbials in Japanese allows for: (1) multiple interpretations of each verbal inflection (see examples 2-4), particularly -teiru (examples 7-9); (2) multiple representations of each semantic feature (see examples 5/6).In other words, English and Japanese have differences not only in aspectual interpretations of verbal morphology, but also in representations of related aspectual features.This suggests that L1 effects could be a trigger of feature-reassembly failures: Japanese adolescent learners of English failed to reassemble aspectual features, such as [+/-habitual] [+/-telic] into different L2 lexical items, which resulted in two kinds of bidirectional misuse.This is consistent with a claim of the Feature Reassembly Hypothesis.

Conclusion
This study investigated knowledge of both syntax and verbal morphology in affirmative-with-VP-adverb contexts.Elicited written and spoken data by adolescent L1 classroom learners provide possible evidence for the Full Transfer Full Access and the Feature Reassembly Hypothesis.Success in syntactic operations and variability in 49  Wakabayashi and Yamazaki (2009) found that not 'linear' but 'syntactic' distance, caused by intervening constituents, disrupts affix hopping, which results in greater difficulty in producing an affixal form of 3ps-s for L1 Japanese learners. 50The response 'I am loving it' to 'How do you like this linguistics class?' is "perfectly acceptable, although the state verb love is not allowed to be used in the form of progressive tense.This suggests aspectual ambiguity in English (Tsujimura, 2007:385).morphological production suggest that a functional category Tense with specified features and UG control are operative in adolescent L2 grammars.In addition, selective morphological variability suggests failures to reassemble semantic and morphosyntactic features onto morphological realisation.English and Japanese both have the same overt tense/aspect morphology on the surface (Sugaya and Shirai, 2007:4), but they represent lexical aspect differently: in English, lexical verbs entail their aspectual features, while in Japanese, the imperfective marker -teiru and temporal adverbials alternate interpretations in aspectual features.The two kinds of bidirectional misuse suggest that the different representations of L1 semantic features might be another source of reassembly problems, which could cause morphological variability.This study presented a different perspective on the source of morphological variability in L2 acquisition: appropriate aspectual interpretation underlies successful acquisition of verbal morphology.
To examine more closely how the syntax-semantics interface reflects morphological representation, further studies are needed.
usually eats breakfast at nine.(No.18) b.He is often tired at night.(No.25)

Table 1 :
Plural marking in English, Chinese, and Korean Lardiere illustrates how differently relevant features are realised and assembled by L2 learners.Based on analyses byLi (1999:88)  9and various researchers 10 , Lardiere considers the different conditions under which plural marking occurs in English, Chinese, and Korean, all of which are assumed to select the [+plural] feature.Table

Table 2 :
Similarity of tense-aspect marking in English and Japanese

Table 3 :
Table 3 summarises the differences in interpretation between English is+Ving and Japanese V+teiru.Differences in interpretation of non-past imperfective marker

Table 4 :
Differences in derivation between English state verbs and the Japanese counterparts Instead of object-marking (English)/verb-marking (Bulgarian)16, Japanese employs a combination of adverbs (example 12 a, see examples 6 and 8)/ adverbial phrases 17 (12 b) and overt numerals, which specifies cardinality to represent telicity.

Table 6 :
Table 6, 4 morphological properties (3ps-s, regular past-d, irregular past forms, and copula is) and 3 syntactic properties (verb placement over adverbs, overt subject suppliance, and Nominative Case marking in a Tense Phrase) were examined in the same 11 test items.Grammatical Properties investigated in the production of verbal morphology

Table 7 :
Marking Criteria for verb placement with adverbs (S=subjects, V=main verbs, A=adverbs, C=complements)Table8shows marking criteria for overt subjects and Nominative Case.The suppliance and case marking of subjects were scored for their own grammaticality respectively, regardless of the ungrammaticality of other properties in the same sentence.

Table 8 :
Marking Criteria for overt subject suppliance and Nominative Case marking Based on the criteria (Table9), verbal morphology production in affirmative sentences with VP-adverbs (3ps-s, regular past-d, irregular past forms, and copula is) was scored.Only sentences with correct word orders could be scored either 1 or 0, which is different from the scoring criteria for verb placement.

Table 9 :
Marking Criteria for verbal morphology with VP-adverbs (A=adverbs)

Table 10 :
Accuracy rate of verb placement with adverbs in obligatory contexts (%)(Written Data)

Table 11 :
Accuracy rate of verb placement with adverbs in obligatory contexts (%) 32 (Spoken Data)

Table 12 :
Suppliance of Overt subjects with Nominative Case in affirmative sentences with-VP-adverbs (%) (Case=Nominative Case) ) a.She is usually eat breakfast at nine.[JH 7 th P11, Spoken]35(No.2Sheusuallyeats breakfast at nine.)36b.She is sometimes drink milk last year.

Verbal morphology in affirmative with VP-adverb sentences
All groups produced verb morphology much less accurately than verb placement.In particular, the 7 th grade participants in the spoken data showed the lowest suppliance rate in present 3ps-s (3.6% in Table16).

Table 15 :
Suppliance of verbal morphology in affirmative-with-VP-adverb contexts (Written Data)

Table 17 :
Copula is omission in the three obligatory contexts (%: omission/total cases) (Written and spoken data)

Table 18 :
Bidirectional misuse in affirmative copula /auxiliary be and main verbs with VP-adverb contexts (%)