0
Nominated Output 2
Behaviours of Wh-words in English Speakers’ L2 Chinese Wh-questions*
BOPING YUAN
University of Cambridge
Address for correspondence
Faculty of Oriental Studies, University of Cambridge, Sidgwick Avenue, Cambridge
CB3 9DA, the United Kingdom
Email: [email protected]
The running head: Behaviours of wh-words in L2 Chinese
*
The research reported in this paper is part of a project on mental representations of wh-words in nonnative grammars of Chinese, which is funded by the Economic and Social Research Council in
England (Grant reference number: RES-000-22-0180). I gratefully acknowledge the financial support
for the project from the ESRC. I would also like to thank the following people for their invaluable
assistance in my data collection: Shio-yun Kan, Yang Song, Jing Fang, Lianyi Song, Dian Huang,
Guohua Chen, Yang Zhao, Limin Jin, Xue Gu and Bin Yu. I am also in debt to students and teaching
staff from Oxford, London, Westminster, Leeds, Durham, Newcastle and Edinburgh Universities in the
U.K. for their participation in my empirical study. Without their help, this research project would have
been impossible. Earlier versions of this paper were presented at GALA 2005 Conference at
University of Siena, Beijing Foreign Studies University, Hunan University. I thank the audience for
their valuable and helpful comments and suggestions.
1
This paper presents an empirical study of how different wh-words behave in English
speakers’ L2 Chinese wh-questions. Results indicate that C0 in L2 Chinese whquestions is valued by merging the wh-particle “ne” into C0 and that there is no L1
transfer in this aspect of L2 grammars as no wh-movement is found in English
speakers’ L2 Chinese. Our results also indicate that English speakers accept Chinese
wh-questions with wh-arguments embedded in a complex NP (CNP) or a sentential
subject. However, there is evidence that wh-words do not develop in a uniform
fashion and that some wh-words behave variably at different Chinese proficiency
levels. The variability is believed to result from optional attachments of an operator
feature to certain wh-words and from unsuccessful removal of an L1 representation
from learners’ L2 Chinese lexicons. The results will be analysed on a basis of a
lexical morphological feature defect account, and we will propose that the lexical
morphology-syntax interface can be a locus of non-native-like behaviours in L2
acquisition.
1. Introduction
In second language (L2) research, Chinese is generally treated as a “pure” wh-in-situ
language, that is, wh-words remain in situ and there are no island effects. Many L2
studies have looked at L2 acquisition of wh-movement by speakers of a wh-in-situ
language (Hawkins and Chan, 1997, Hawkins, 2004, Klein, 1995, Martohardjono,
1993, Martohardjono and Gair, 1993, Schachter, 1990, White and Juffs, 1998), or vice
versa (Kim, 2003, Yuan, 2005). Few people have done research investigating whether
all wh-words behave similarly and develop uniformly in L2 acquisition of whquestions. Following recent findings in linguistic research that wh-nominals and whadverbs have different behaviours in Chinese and that wh-movement and island
effects do exist in Chinese wh-questions with wh-adverbs, this paper will report on an
empirical study examining how different wh-words behave in English speakers’ L2
Chinese wh-questions.
Unlike English wh-questions, in which the wh-word has to move to the initial
position of the sentence, as can be seen in the English translation of (1), the wh-word
2
in Chinese wh-questions remains in situ, as shenme in (1)1. In Chinese, there are two
question particles ma and ne, with ma being a yes-no-question particle (hereafter yesno particle) and ne a wh-question particle (hereafter wh-particle). The yes-no particle
is generally obligatory2, as in (2), but the wh-particle is optional in Chinese whquestions, as in (1).3 Another characteristic of Chinese wh-questions is that the in-situ
wh-word can be located inside islands, such as a complex NP (CNP) as in (3) and a
sentential subject as in (4). Obviously, wh-words cannot be extracted from these
islands in English.
(1) Ni xiang chi shenme (ne)?
you want eat what
Q
‘What would you like to eat?’
(2) Ni xihuan ta ma?
you like him Q
‘Do you like him?’
(3) Ta xihuan [shei xie de shu] (ne)?
he like who write DE book Q
*‘He like the book who wrote?’4
(4) [Shei qu Beijing] bijiao
heshi (ne)?
who go Beijing relatively suitable Q
*‘Who goes to Beijing is more appropriate?’
In this paper, we follow Cheng and Rooryck (2000, 2002), Simpson (2000) and Rizzi
(1997) in assuming that C0 in wh-questions is essentially ambiguous and unvalued
with respect to specification of force, and that unvalued C0 must be valued. The
valuation can be carried out by merging a wh-particle into C0 or by moving a whphrase to Spec CP. Chinese adopts the former and English the latter. In Chinese whquestions, the valuation of force is carried out by the Chinese wh-particle ne, which is
merged into C0. In this case, the wh-particle ne, whether phonetically realised or not,
values C0 as C0 with the [+Q, +wh] features, which licenses the wh-word in situ in
Chinese. We make a distinction between wh-nominals, wh-arguments and wh1
The abbreviations used in the glosses are: Q = a question particle; DE = a modifying marker that
occurs at the end of a pre-nominal modifier; PART = a particle.
2
The yes-no particle is usually not used in questions such as echo questions or rhetorical questions.
3
The yes-no particle and the wh-particle can only appear in matrix clauses and cannot be used in an
embedded question.
4
The English translations for the Chinese wh-questions in this paper are direct translations, and they
are not always grammatical in English.
3
adverbs. We assume by following Cheng and Rooryck (2002) that wh-adverbs in
Chinese have an operator feature, but wh-nominals do not and that the operator
feature has to undergo movement to Spec of CP for feature checking.
Results of our study indicate that C0 of the wh-questions in English speakers’ L2
Chinese grammars is valued by the merging of the Chinese wh-particle ne into C0,
and that there is no L1 transfer in this aspect of L2 grammars as no wh-movement is
found in English speakers’ L2 Chinese. Our results also indicate that English
speakers accept Chinese wh-questions with wh-arguments embedded in a CNP or in a
sentential subject. This suggests that there is no movement of wh-argument in L2
Chinese grammars as any movement of wh-words out of a CNP or a sentential subject
would violate Subjacency. However, there is evidence that in English speakers’ L2
Chinese grammars, wh-words do not develop in a uniform fashion and that different
wh-words behave differently at different Chinese proficiency levels. Some wh-words
are found to have variable behaviours in L2 Chinese grammars, and some of these
variable behaviours are persistent at advanced levels and are likely to become
candidates for fossilization. We will analyse the results on the basis of a lexical
morphological feature defect account.
In resent years, many L2 studies have focused on the relationship between functional
categories and their morphological realizations in L2 acquisition (cf. Hawkins and
Chan, 1997, Hawkins, 2001, Hawkins and Liszka 2003, Lardiere, 1998a,b, 2000,
Prévost and White, 2000, Smith and Tsimpli, 1995, Tsimpli, 2003, White, 2003,
among many others), and the relationship between lexical morphology and syntax has
not received much attention. In this paper, we will demonstrate that we cannot take it
for granted that features of lexical items can be acquired by L2 learners. The lexical
morphology-syntax interface can be a locus of non-native behaviours in L2
acquisition.
2. Wh-questions in English and Chinese
Huang (1982a) is among the first to propose that wh-words in Chinese undergoes whmovement at LF. He argues that there is a parallelism in scope and selection between
4
overt wh-movement in languages like English and covert wh-movement in languages
like Chinese.
It is well-known that overt wh-movement is subject to Subjacency constraint
(Chomsky, 1973), which does not allow extraction of a wh-word from islands such as
a CNP or a sentential subject, as can be seen from the English translation of (3-4).
However, the Chinese counterparts, as in (3-4), are perfectly grammatical, even
though the wh-words in these sentences are all located inside islands (in (3), it is
inside a CNP; in (4), it is inside a sentential subject).
However, when the wh-word in the island is a wh-adverb rather than wh-argument,
the sentence becomes ungrammatical, as shown in (7), which is in contrast with (5)
and (6), where the wh-words in the island are wh-arguments.
(5) Ni xihuan [shei xie de] shu (ne)?
you like who write DE book Q
*‘You like the book who wrote?’
(6) Ni xihuan [ xie shenme shu de] ren (ne)?
you like write what book DE person Q
*‘You like people who wrote what book?’
(7) *Ni xihuan [ta weishenme xie de] shu (ne)?
you like he why
write DE book Q
*‘You like the book he wrote why?’
The distinction above between wh-arguments and wh-adverbs only apparently
provides a solution to the problem. Chinese wh-words corresponding to “when” and
“where”, which are generally regarded as adjuncts, can appear in islands and do not
cause ungrammaticality to the sentences, as shown in (8) and (9), which form a
contrast to the sentence in (7), in which the wh-word weishenme “why” is located
inside a CNP island.
(8) Ni xihuan [ta (zai) shenmeshihou xie de] shu (ne)?
you like he (at) when
write DE book Q
*‘You like the book he wrote when?’
(9) Ni xihuan [ta zai nar xie de] shu (ne)?
you like he at where write DE book Q
*‘You like the book he wrote where?’
5
The sentences in (8) and (9) indicate that the wh-words shenmeshihou “when” and
nar “where” in Chinese pattern on a par with the wh-arguments shei “who” and
shenme “what” rather than the wh-adverb weishenme “why” in their extraction
possibility at LF. Huang (1982b) suggests that shenmeshihou “when” and nar
“where” are NPs in Chinese as nar5 “where” has to be preceded by a preposition zai
“at” as in (9) and shenmeshihou “when” can be optionally preceded by a preposition
zai “at”, as in (8). Huang further points out that shenmeshihou can be literally
translated as “what time” with shenme “what” modifying shihou “time”. On the basis
of this evidence, Huang assumes that shenmeshihou “when” and nar “where” are
dominated by NP in the position [PP P [NP__], where the preposition P may or may
not be phonetically realised, and that shenmeshihou “when” and nar “where” are
complements of prepositions, and are on a par with shenme “what” and shei “who” in
being NPs, thus wh-nominals. In this sense, shenmeshihou “when” (8) and nar
“where” (9) can be distinguished from weishenme “why” in (7) by the distinction
between NP and non-NP or between nominals and adverbs.
It has been observed in Lin (1992) and Tsai (1994a,b) that zenmeyang “how” patterns
with nominals in Chinese as well. As we can see in (10), the wh-word zenmeyang
“how” is located inside a CNP island and the sentence is grammatical. To answer a
question like (10), one can only use PPs like yong qiaokeli “with chocolate” or yong
naiyou “with cream”. This shows that at least with regard to the island constraint,
shenme “what”, shei “who”, shenmeshihou “when”, nar “where” and zenmeyang
“how” behave similarly but weishenme “why” behaves differently.
(10) Ni xihuan [ta zenmeyang zuo de] dangao (ne)?
you like he how
make DE cake Q
‘You like the cake he makes how?’
In his split-CP proposal, Rizzi (1997) argues that CP should be split up into separate
structural layers. Rizzi borrows Chomsky’s (1995) terminology and calls one of the
layers Force Projection,6 which can provide information about whether the sentence
is a declarative, an interrogative, an exclamative, a relative, an imperative, etc. The
ForceP functions as an interface of the clause with a higher clause in the case of
5
6
The Chinese translation of “where” can be nali as well.
The other layers that Rizz (1997) proposes in his split-CP hypothesis are TopP, FocP and FinP.
6
embedded ForceP or an interface with the superordinate structure of the discourse in
the case of matrix ForceP. According to Rizzi, force can be expressed by overt
morphological encoding on the head of ForceP, or by moving/providing a required
operator to ForceP. Chinese seems to have adopted the former and English the latter.
We can argue that in Chinese, the force of a sentence is generally specified by
particles, as shown in the sentences in (11-14)7. In (11), the yes/no question is
specified by the yes/no-particle ma; in (12), the wh-question is marked by the whparticle ne; in (13), the exclamative force of the sentence is specified by the use of
the particle a; and in (14), the imperative force is marked by the particle ba. The use
of particles in Chinese to specify the force of the sentence seems to support Cheng’s
(1991) Clause Typing Hypothesis, in which she suggests that each sentence has to be
“typed” as a declarative sentence or an interrogative sentence. As we are not dealing
with the other layers of CP, we will continue to use the traditional label CP for
ForceP for the sake of convenience.
(11) Ni hui shuo Hanyu ma?
you can speak Chinese Q
‘Do you speak English?’
(12) Ni xiang chi shenme (ne)?
you want eat what
Q
‘What do you want to eat?’
(13) Duome haode tianqi
a!
how good weather PART
‘What a nice weather!’
(14) Zanmen hui
jia
ba.
We go-back home PART
‘Let’s go home.”
In this paper, we follow Simpson (2000) in assuming that C0 is essentially ambiguous
and underspecified with respect to specification of force. We also follow Cheng and
Rooryck (2000, 2002) and Rizzi (1997) in believing that the under-specified or
unvalued C0 must be valued and specified. In Chinese wh-questions, the valuation of
force is carried out by the Chinese wh-particle ne, which is merged into the head of
CP. In this case, the wh-particle ne not only values C0 as C0 with a [+Q] feature but
7
Here we assume that CP in Chinese is head-final.
7
also specifies it as C0 with a [+wh] feature. Following Rooryck’s (1994) idea of
embedded Attribute-Value sets of phi features, we can have the featural
representations in (15).
(15) The wh-particle: ne[Q: +wh]
a. [CP [Ni xihuan shei] [C0 ]]
you like who
[F: ]
(before the force in C0 is valued)
b. [CP [Ni xihuan shei] [C0 ne ]]
you like who
[F: [Q: +wh] ]
(after the force in C0 is valued by ne)
In (15), the wh-particle ne has force value [+Q], which is further specified as having
a value of [+wh] (rather than [+yes/no]). In (15a), the head of CP is underspecified
and unvalued as to whether it is a declarative, an interrogative, or an exclamative. By
merging ne into C0, the force of C0 is valued as a head of CP with both [+Q] and
[+wh] features. The wh-particle ne in C0 has a dual function: it disambiguates the
force of the sentence and at the same time it also licenses the wh-word in situ. In this
sense, the wh-particle ne is the licensor of wh-in-situ in Chinese, whether ne is
phonetically realised or not. If we adopt Chomsky’s (2001a, b) concept of Agree, we
can assume that the wh-particle ne in C0 “probes” in its c-command domain for
compatible features. Once the features of ne and those of the in-situ wh-word (i.e. the
Goal) are matched (i.e. agreed), the in-situ wh-word can be checked and licensed at a
distance.
However, one question that arises is why, unlike sentences with in-situ wh-nominals,
sentences with an in-situ wh-adverb like (7), repeated below, are ungrammatical in
Chinese given that an in-situ wh-word can be licensed by the wh-particle ne in C0 and
the feature-checking can take place at a distance. The sentence in (7), where the whadverb weishenme “why” is inside a CNP, is in contrast to the one in (16), where the
same wh-adverb is outside the CNP. This shows that although a wh-adverb can stay
in situ in Chinese, it cannot appear inside an island, which suggests that a kind of
movement is involved in Chinese sentences with the wh-adverb.
8
(7) *Ni xihuan [ta weishenme xie de shu] (ne)?
you like he why
write DE book Q
‘You like the book he wrote why?’
(16) Ni weishenme xihuan [ta xie de shu] (ne)?
you why
like he write DE book Q
‘Why do you like the book he wrote?’
Following Tsai (1994a,b, 1999), Cheng and Rooryck (2002) make a clear distinction
between a wh-nominal and a wh-adverb and assume that the wh-adverb in Chinese
has an operator feature but the wh-nominal does not. However, they believe that
instead of category movement of the whole wh-adverb, only the operator feature of
the wh-adverb undergoes movement. The ill-formedness of (7) is then due to the
Subjacency violation by the movement of the operator feature.
It has been proposed by Tsai (1994a,b) that in Chinese a distinction should be made
between the instrumental “how” and the manner “how” on the one hand, and between
the purpose “why” and the reason “why” on the other as these wh-words behave
differently in Chinese syntax. According to Tsai, the instrumental “how” and the
purpose “why” are encoded with nominal properties and therefore should be treated
as DP or PP. As a result, they are allowed to remain in situ inside a CNP or inside a
sentential subject. This implicates that the instrumental “how” and the purpose “why”
in Chinese do not have an operator feature which requires raising. However, the
manner “how” and the reason “why” in Chinese are ruled out in a CNP or in a
sentential subject. This is because these Chinese wh-words are wh-adverbs; they have
an operator feature which requires raising but is unable to move out of a CNP or a
sentential subject due to the Subjacency constraint. Sentences in (17), (18) and (19)
can be used as evidence for the distinction between these wh-words.
(17) Ni bu xihuan [[ta zenmeyang zuo de] dangao] (ne)?
you not like she how
make DE cake
Q
a. ‘What is the means x such that you don’t like [cakes [which she makes by x]]?
(instrumental-“how”)
b. *‘What is the manner x such that you don’t like [cakes [which she makes in x]]?
(*manner-“how”)
(18) Ta yiban zenmeyang zuo dangao?
she usually how
make cake
a. “What is the means by which she makes cakes?”
(instrumental-“how”)
b. “What is the manner in which she makes cakes?”
(manner-“how”)
9
(19) a. Ni xihuan [[wei(le) shenme xuexi de] xuesheng] (ne)?
you like why
study DE students
Q
‘What is the purpose x such that you like [students [who study for x]]?
(purpose-“why”)
b. *Ni xihuan [[weishenme xuexi de] xuesheng] (ne)?
you like
why
study DE students Q
*‘What is the reason x such that you like [students [who study because of x]]?
(reason-“why”)
Answers to the question in (17) can be “With chocolate” or “With lots of cream”, but
answers like “Carelessly” or “Casually” are not possible. However all these answers
become possible when zenmeyang “how” occurs in a simple sentence like (18). In a
similar fashion, answers to the questions in (19) can be “Those students who study in
order to become linguists” or “Those students who study for the purpose of getting
into a PhD programme”. It is not possible to answer the question by saying “Those
students who study because they enjoy studying” or “Those students who study
because their parents want them to”.8 In fact, a better English translation for weile
shenme is “for what” rather than “why”. The Chinese word weile is a preposition as in
weile ni “for you” or weile guojian “for the country”. It should be mentioned that
some native Chinese speakers feel that the distinctions between these wh-words
sometimes can be rather vague in some Chinese sentences. Tsai (1994b) also admits
that some of his Chinese informants do not always find it easy to draw a clear line
between the instrumental “how” and the manner “how” and between the reason
“why” and the purpose “why”. Although we do not pursue these distinctions in our
empirical study, they may have implications in our L2 data analyses as L2 Chinese
learners are exposed to various types of Chinese wh-words in their input data.
Let’s next analyse wh-questions in English. Given that English does not have a whparticle like Chinese ne, it is not surprising that it resorts to a different means to
disambiguate unvalued C0 and specify it as C0 with the [+wh] feature. It has been
suggested (cf. Cheng and Rooryck 2000, 2002, Simpson 2000) that C0 in English whquestions is disambiguated by the wh-word which is moved from its base-generated
position to Spec CP, where the wh-word not only values the force of the C0 as a
We can use Reihart’s (1998) examples in (i) to show that in English it is also the availability of the
nominal feature of the wh-word that decides whether the wh-word can remain in situ inside an island.
(i)
a. *Who fainted when you behaved how?
b. Who fainted when you behaved what way?
8
10
question with the [+Q] feature but also specifies it as C0 with the [+wh] feature.
Cheng and Rooryck provide the following as featural representations of the valuation
process (2002 p. 25).
(20) a. [CP [C0 ] [IP John saw who[Q:wh]]
(before valuation of the force in C0)
[F: ]
b.[CP Whoi[Q: wh] [C0 ] [IP John saw ti]] (after overt movement of who and valuation)
[F: [Q: wh]]
In (20b), because of the wh-word raised to Spec CP, the head of CP is valued not
only as a question but also specifically as a wh-question. The raised wh-word
actually percolates the whole CP and makes it a CP with the [+Q] and [+wh] features.
The C0 activated this way becomes a wh-licensor and allows for all other wh-words
in its domain to be successfully licensed in a multiple wh-question. In other words,
once a single wh-word is moved to Spec CP, all the other in-situ wh-words are
licensed at a distance9, because Agree can check the [+Q] and [+wh] features long
distance. This makes movement of the other wh-words both unnecessary and
impossible.10 In this sense, the wh-word in Spec CP is the licensor of the wh-in-situ
in English.
3.
Empirical Study
3.1. Subjects
The empirical study includes 107 English speakers as subjects11. It also includes 20
native speakers of Chinese as controls. The English subjects were undergraduate
students, post-graduate students, lecturers and professors of Chinese from universities
9
Of course there are constraints (such as the superiority constraint) regarding which wh-word is to be
moved to value the ambiguous C0.
10
This can be seen in the two examples (from Watanabe 2001:207) below, where the subject wh-word
“who” is assumed to have moved to Spec CP, which makes it possible for the other wh-words to
remain in situ.
(i)
Who is reading a book that criticizes who?
(ii)
Who remembers where we bought what?
11
Actually, more English speakers participated in the empirical study. However, as some were later
found to be British-born Chinese, or Cantonese speakers, or only finish parts of the tasks, these
subjects’ data were discarded.
11
in the U.K. The native Chinese speakers were university students or office workers in
China. All the British universities involved teach Chinese from scratch and the
students involved in this study include both specialists and non-specialists of Chinese.
The weekly classroom teaching of Chinese for British undergraduate students ranges
from two to ten hours. All lecturers and professors involved were teaching either
Chinese or Chinese-related courses in their universities.
On the basis of their performance in a Chinese cloze test, the English speakers were
divided into five Chinese proficiency groups respectively: Beginner Group, PostBeginner Group, Intermediate Group, Post-intermediate Group and Advanced Group.
The native Chinese speakers are in the Native Chinese Group. Information about each
of the 6 groups is given in Table 1. An ANOVA result shows that there is a significant
difference between all groups in their performance in the cloze test, (F=855.619,
p<0.001) and the following-up scheffé tests indicate that except for the advanced
group, each of the learner groups is significantly different from the native Chinese
group and that all the learner groups are significantly different from each other.
Table 1: Information about each group
Average months Average months Mean scores in the
No. of
Groups
subjects
Average
of
in
cloze test (total=40)
Age
studying Chinese
China/Taiwan
(ranges in brackets)
English Beginner
20
22
4
1
4 (1-6)
English Post-beginner
20
23
10
3
11 (7-15)
English Intermediate
28
22
29
6
22 (16-25)
English Post-intermediate
25
27
83
18
30 (26-34)
English Advanced
14
36
207
44
36 (35-39)
Native Chinese
20
28
N/A
N/A
39 (38-40)
3.2. Research questions
The research questions we ask in our empirical study are:
a. How is C0 in L2 Chinese wh-questions disambiguated and valued? If the L1
transfer part of the Full Transfer/Full Access model (Schwartz and Sprouse
1994, 1996) is adopted, we can hypothesize that at the initial stage, C0 of wh-
12
questions in English speakers’ L2 Chinese would be valued by a wh-word
moved from its base-generated position to Spec CP;
b. Will the valuation of C0 of L2 Chinese wh-questions by means of the whparticle ne imply wh-in-situ in L2 Chinese? That is, will the insertion of ne
into C0 make wh-movement unnecessary and impossible in L2 Chinese whquestions?
c. Can L2 Chinese grammars make a distinction between wh-nominals and whadverbs so that wh-nominals can stay in situ inside islands such as CNPs or
sentential subjects while wh-adverbs cannot? If the distinction can be made,
we can assume that the wh-nominal and the wh-adverb have different sets of
morphological features attached to them in L2 Chinese grammars, with the set
of the latter having an operator, which is absent in the former. The operator of
the wh-adverb has to undergo movement and the movement is subject to
Subjacency.
3.3. Acceptability judgment test
Each of the subjects has to do an acceptability judgment test, which includes 18 types
of sentences related to the research questions stated above. Each type has four tokens
and in total, there are 72 test sentences relevant to our study here. There are also 48
sentences testing different aspects of L2 Chinese grammars, which will be reported
elsewhere. These 48 sentences also serve as distracters. All sentences are presented in
Chinese characters but all instructions are given in subjects’ L1 English. In order to
minimize any possible effect of vocabulary on the subjects’ judgment, efforts are
made to include only those basic words of daily life. In addition, English translation
and the Chinese pinyin (a Chinese phonetic system) are provided for some potentially
unfamiliar words. The 18 sentence types used in the test and their examples are given
in (21-26).
(21) Yes-no questions with or without ma
a. With ma (Control)
Ni shi Riben ren
ma?
you be Japan person Q
‘Are you Japanese?’
13
b. ?Without ma
Ni shi Riben ren ?
you be Japan person
‘Are you Japanese?’
(22) Wh-argument in situ
a. Wh-question with ne (control)
Ta xihuan shei ne?
she like who Q
‘Who does she like?’
b. *Moved wh-argument
*Shei ta xihuan ?
who she like
‘Who does she like?’
c. Wh-argument in situ without ne
Ta xihuan shei ?
she like who
‘Who does she like?’
(23) “When/where/how/why” in situ or moved
a. “when/where/how/why” in situ (Control)
Ta zhu zai nar?
he live in where
‘Where does he live?’
b. *Moved “when/where/how/why”
*Zai nar ta zhu?
in where he live
‘Where does he live?’
(24) Wh-word inside CNP
a. Control
Zhe shi Li Ming mai de zixingche.
This is Li Ming buy DE bike
‘This is a bike which Li Ming bought.’
b. Wh-argument
Zhe shi shei mai de zixingche?
This is who buy DE bike
14
*‘This is a bike which who bought?’
c. “When/where”
Zhe shi Li Ming zai nar mai de zixingche?
This is Li Ming in where buy DE bike
*‘This is a bike which Li Ming bought where?’
(25) “How”/“why” inside CNP
a. Control
Ni xihuan chi ta zuo de yu ma?
you like eat he cook DE fish y/n-Q
‘Do you like the fish he cooks?’
b. “How”12
Ni xihuan chi ta zenme(yang) zuo de yu?
you like eat he how
cook DE fish
*‘Do you like the fish he cooks how?’
c. *“Why”
*Ni xihuan chi ta weishenme zuo de yu?
you like eat he why
cook DE fish
*‘Do you like the fish he cooks why?’
(26) Wh-words inside sentential subjects
a. Control
Li Ming lai bijiao heshi.
Li Ming come quite suitable
?‘That Li Ming comes is quite suitable.’(i.e.‘It is quite suitable if Li Ming comes.’)
b. Wh-argument
Shei lai bijiao heshi?
who come quite suitable
*‘Who comes is quite suitable?’
c. “When/where”
Li Ming shenmeshihou lai bijiao heshi?
Li Ming when
come quite suitable
*‘Li Ming comes when is quite suitable?’
The English wh-word “how” can be translated into Chinese as either zenme or zenmeyang. In our
pilot study, our subjects accepted both. So in our main study, we present the Chinese wh-word as
zenme(yang) in our test sentences.
12
15
d. “How”
Li Ming zenme(yang) lai bijiao heshi?
Li Ming how come quite suitable
*‘Li Ming comes how is quite suitable?’
e. *“Why”
*Li Ming weishenme lai bijiao heshi?
Li Ming why
come quite suitable
*‘Li Ming comes why is quite suitable?’
As we can see in (21-26) above, all test sentences can be categorised in 6 groups, and
in each group there is a control sentence and a corresponding experimental sentence
or more than one corresponding experimental sentence. The control sentence and the
experimental sentence are identical except for one difference. The difference can be
the use or non-use of yes-no question particle ma, as in (21), the use of wh-question
particle ne in relation to wh-in-situ or wh-movement, as (22), wh-in-situ vs. whmovement, as in (23), CNP with or without an embedded wh-word, as in (24) and
(25), a sentential subject with or without an embedded wh-word, as in (26). There is
also correspondence between experimental sentences, as in (24), (25) and (26), and
the only difference between these experimental sentences is the type of wh-words
used in the sentence (i.e. wh-arguments vs. shenmeshihou “when” / nar “where” vs.
zenme(yang) “how” vs. weishenme “why”). In this way, we can be certain that any
difference in the subject’ judgment between a sentence and its corresponding
sentences is due to the difference in the use of wh-words in the two sentences.
All test sentences are randomised, and subjects are asked to judge the acceptability of
each sentence by circling a number on a scale as given in (27). On the basis of the
scale in (27), we treat any score of “+1” or above that the subject assigns to a
particular sentence as a sign of accepting that sentence. Conversely any score of “-1”
or below as a sign of rejecting that sentence, and scores between “-1” and “+1” are
treated as signs that the subject is uncertain about the sentence.
(27)
___________________________________________________________
-2
-1
completely
probably
unacceptable
unacceptable
0
“I don’t
know”
+1
+2
probably
completely
acceptable
acceptable
16
3.4. Results
As we can see from Table 2, all groups accept Chinese yes-no questions with ma.
This suggests that like the native Chinese grammar, L2 grammars of these learner
groups allow the Chinese yes-no particle ma to value C0 of Chinese yes-no questions
and specifies it as C0 with the [+Q] feature. However, it is found that the native
Chinese group as well as some of the learner groups fail to reject Chinese yes-no
questions without ma, as shown in the third row in Table 2. Only the Post-beginner
Group and Intermediate Group reject Chinese yes-no questions without ma. This is
probably because in Chinese rhetoric or echo questions, ma is not required and the
Chinese speakers as well as many of the L2 learners take Chinese rhetoric questions
or echo questions into consideration when they make the judgment.
Table 2: Means scores of the judgment of yes-no questions with and without ma
Beginners Post-beginners Intermediate Post-intermediate Advanced Chinese
With ma
1.75
1.9
1.98
1.98
1.98
1.91
?Without ma
-0.71
-1.15
-1.33
-0.74
-0.95
-0.53
Most learner groups also accept Chinese wh-questions with ne. As shown in Table 3,
except for the Beginner Group, all learner groups, like the native Chinese Group,
accept Chinese wh-questions with ne, which can suggest that C0 of wh-questions in
these learners’ L2 Chinese grammars is specified as C0 with [+Q] and [+wh] by the
wh-particle ne which is merged into C0. Within the theoretical framework we have
adopted, the wh-particle ne merged into C0 of wh-questions should make any
movement of wh-word to C0 unnecessary and impossible. This seems to be indeed the
case in our study. As the data in the second and third rows of Table 3 show, those
groups who accept Chinese wh-questions with ne all reject Chinese wh-questions with
the wh-word moved from its base-generated position to the sentence initial position.
Table 3 also shows that all learner groups including the beginner group, like the
Native Chinese Group, accept Chinese wh-questions without phonetically realised
“ne”. It seems that valuation of C0 of Chinese wh-questions by phonetically
unrealised “ne” is preferred by L2 Chinese grammars at least at beginner levels.
17
Table 3: Mean scores in the judgment of wh-questions with/without ne and with whargument moved
Beginner Post-beginner Intermediate Post-intermedia Advanced Chinese
With ne
0.7
1.13
1.46
1.53
1.89
1.89
*Wh-movement
-1.15
-1.25
-1.48
-1.72
-1.8
-1.49
Without ne
1.69
1.88
1.86
1.84
1.98
1.96
Similar results are obtained in the groups’ judgments of simple Chinese wh-questions
with shenmeshihou “when”, nar “where”, zenmeyang “how” and weishenme “why”.
All learner groups accept these wh-words in situ in Chinese wh-questions, as shown
in the second row in Table 4, and they, except for the Beginner Group, reject or tend
to reject wh-questions with this type of wh-words moved to the sentence initial
position, as shown in the third row in Table 4.
Table 4: Mean scores in the judgment of simple wh-questions with
“when/where/how/why” in situ or moved
Beginners Post-beginners Intermediate Post-intermediate Advanced Chinese
In situ
*Moved
1.1
1.76
1.88
1.82
1.71
1.9
-0.24
-0.89
-1.02
-1.59
-1.64
-1.69
The data in Tables 3 and 4 provide us with evidence that from the post-beginner level
onward, C0 of the wh-question in L2 Chinese is valued by a phonetically realised or
unrealised wh-particle ne and that wh-movement seems to have become unnecessary
and impossible in English speakers’ L2 Chinese grammars as a result.
Could there be any feature movement in L2 Chinese wh-questions? Obviously, the
data above cannot provide an answer to this question as they are data from subjects’
judgment of simple wh-questions. In order to find an answer to this question, let us
look at the judgment data of wh-questions with wh-words embedded in a CNP. The
data in the second row of Table 5 are from subjects’ judgment of sentences with a
CNP but without any wh-word. This type of sentence is used as a control to
distinguish subjects who have mastered the basic structure of the sentence with an
embedded CNP from those who have not. As we can see, all groups, except for the
18
Beginner Group, have mastered the basic sentence structure with an embedded CNP
in Chinese. So we have to ignore the Beginner Group’s judgment and concentrate on
the other groups’ judgment of wh-questions which correspond to the control sentences.
Table 5: Mean scores in the judgment of wh-questions with a wh-word inside CNP
Beginner Post-beginner Intermediate Post-intermedia Advanced Chinese
Control
0.68
1.43
1.94
1.87
1.89
1.93
Wh-argument
0.55
1.43
1.46
1.58
1.56
1.71
“When/where”
-0.31
0.15
0.55
1.06
1.13
1.16
From the third row in Table 5, we can see that all learner groups, except for the
Beginner Group, accept Chinese wh-questions with a wh-argument inside a CNP. It
seems that so long as L2 Chinese grammars can handle the basic sentence structure,
there is no problem for wh-arguments to stay inside a CNP in L2 Chinese whquestions. This suggests that Subjacency is irrelevant here as no movement is
involved in this type of wh-questions in L2 Chinese.
However, the wh-words shenmeshihou “when” and nar “where” inside a CNP do not
seem to be so acceptable to L2 Chinese grammars at post-beginner and intermediate
levels. As we can see from the fourth row in Table 5, the Post-beginner and
Intermediate Groups, who are shown to have mastered the basic sentence structure
and who accept wh-arguments inside a CNP, do not accept Chinese wh-questions
with shenmeshihou “when” and nar “where” embedded in a CNP. We can take this as
evidence that wh-arguments and the wh-words such as shenmeshihou “when” and nar
“where” do not develop in a uniform fashion in L2 Chinese grammars. It seems that
only at post-intermediate and advanced levels does it become acceptable in L2
Chinese grammars to allow shenmeshihou “when” and nar “where” to stay in situ
inside a CNP. L2 grammars at beginner, post-beginner and intermediate levels seem
to be unsure of the status of shenmeshihou “when” and nar “where” and they seem to
be uncertain whether these wh-words should share the same behaviour as wharguments in Chinese wh-questions.
How do wh-words zenmeyang “how” and weishenme “why” behave in L2 Chinese
wh-questions? The data in Table 6 suggest that all groups, except for the Beginner
19
Group, can handle the basic sentence structure for testing wh-questions with
zenmeyang “how” and weishenme “why”, as these groups accept the control sentences
with a CNP, as shown in the second row in Table 6.
However, as shown in the third row in Table 6, none of the learner groups, including
the advanced group, accepts Chinese wh-questions with zenmeyang “how” embedded
inside a CNP, which forms a striking contrast with the judgment of the native Chinese
group. A careful examination of the data reveals that with regard to wh-questions, the
status of zenmeyang “how” in L2 Chinese grammars remains variable throughout the
acquisition process as the subjects in all groups optionally accept and reject whquestions with zenmeyang “how” embedded inside a CNP. This variable status seems
to be permanent in English speakers’ L2 Chinese grammars as the variability
continues even at a very advanced level.
In contrast to the variable behaviours of zenmeyang “how”, the behaviour of
weishenme “why” in English speakers’ L2 Chinese wh-questions seems to be rather
consistent. As we can see from the fourth row in Table 6, all groups, except for the
Beginner Group, reject Chinese wh-questions with weishenme “why” embedded
inside a CNP. This could be due to the fact that as the case in the native Chinese
grammar, weishenme “why” in L2 Chinese grammars has an operator which has to
move to Spec CP in wh-questions. Since weishenme “why” in the test sentences is
embedded in a CNP and since the movement of the operator of weishenme “why” out
of a CNP would violate Subjacency, L2 Chinese grammars reject wh-questions with
weishenme “why” embedded in a CNP.
Table 6: Mean scores in the judgment of wh-questions with “how” or “why” inside a
CNP
Beginners Post-beginners Intermediate Post-intermediate Advanced Chinese
Control
0.73
1.71
1.89
1.97
1.82
1.93
“How”
-0.15
-0.09
0.06
0.22
0.27
1.19
*“Why”
-0.41
-1.16
-1.47
-1.49
-1.59
-1.65
Can we get similar results from wh-questions with a wh-word embedded in a
sentential subject? This type of sentence is also useful for testing possible wh-
20
movement in L2 grammars as any movement out of the sentential subject would
violate Subjacency. The data in the second row in Table 7 are mean scores of the
groups’ judgment of control sentences, i.e. sentences with a sentential subject but
without a wh-word. These data suggest that sentences with a sentential subject are
acquired very late and that only those subjects at post-intermediate and advanced
levels can handle this type of sentences. This forces us to concentrate only on these
two groups’ judgment of wh-questions with a wh-word embedded in a sentential
subject.
As we can see from the third row in Table 7, the Advanced Group accept Chinese whquestions with a wh-argument embedded in a sentential subject. A similar result can
also be seen in the fourth row in Table 7, which indicates that the Advanced Group
accept Chinese wh-questions with shenmeshihou “when” or nar “where” embedded in
a sentential subject. The post-intermediate group also accept this type of sentences
although not as strongly as the Advanced Group.
Table 7: Mean scores in the judgment of wh-questions with a wh-word inside a
sentential subject
Beginners Post-beginners Intermediate Post-intermedia Advanced Chinese
Control
0.43
0.26
0.33
1.11
1.48
1.84
1.11
1.73
Wh-argument
0.24
-0.35
-0.45
0.3813
“When/where”
0.06
-0.15
0.22
1
1.54
1.81
“How”
-0.25
-0.06
-0.02
0.52
0.36
1.4
*“Why”
-0.09
-0.11
-0.46
-0.94
-1.04
-1.34
However, when the wh-word zenmeyang “how” is embedded in a sentential subject of
a wh-question, the judgment of the sentences by the learner groups becomes variable,
as shown in the fifth row in Table 7. A careful examination of the data reveals high
degrees of variability in the judgment of this type of sentences by the subjects as the
subjects in these groups optionally accept and reject this type of sentences. The results
suggest that the wh-word zenmeyang “how” in English speakers’ L2 grammars is far
13
It is not clear why post-intermediate learners accept the control sentences and wh-questions with
shenmeshihou “when” and nar “where” embedded inside a sentential subject but do not accept whquestions with wh-arguments embedded inside a sentential subject.
21
from being native-like even at very advanced levels and that it can serve as an
example of fossilization in English speakers’ L2 Chinese grammars.
The wh-word weishenme “why” seems to behave differently from the other wh-words
in English speakers’ L2 Chinese grammars. As we can see from the sixth row in Table
7, the Advanced Group and Post-intermediate Group reject or tend to reject whquestions with weishenme “why” embedded in a sentential subject. The native
Chinese group also reject this type of sentences.
The results from the judgment of wh-questions with a sentential subject seem to be
compatible with those from wh-questions with a CND. Wh-words in English
speakers’ L2 Chinese do not behave in a uniform fashion. The results of both types of
sentences indicate that wh-arguments stay in situ and do not undergo any type of whmovement in L2 Chinese. The behaviours of wh-words shenmeshihou “when” and
nar “where” are variable in English speakers’ L2 Chinese grammars until postintermediate and advanced levels when the two wh-words are allowed to stay inside a
CNP or a sentential subject. The behaviours of the wh-word zenmeyang “how” remain
variable throughout the L2 Chinese acquisition process and it seems that it can never
be native-like in English speakers’ L2 Chinese grammars. The wh-word weishenme
“why” behaves differently; as the learner’s Chinese language proficiency improves,
the wh-word weishenme “why” inside a CNP or inside a sentential subject becomes
increasingly unacceptable in L2 Chinese. The results suggest that by the time when
learners have reached post-beginner levels, their L2 grammars start to reject sentences
with the wh-word weishenme “why” embedded in a CNP or in a sentential subject.
Table 8 provides a summary of the groups’ judgment of different Chinese wh-words
inside a CNP and inside a sentential subject.14
Table 8: Summary of English speakers’ judgment of different Chinese wh-words
inside CNPs and inside sentential subjects
To a large extend, the data in Table 8 are based on the groups’ judgment of Chinese sentences with
wh-words embedded inside a CNP. This is because English speakers seem to be able to handle Chinese
sentences with a CNP earlier than sentences with a sentential subject.
14
22
shei “who”,
shenmeshihou
Zenme(yang)
weishenme
shenme “what”
“when”, nar “where”
“how”
“why”
N/A
N/A
N/A
N/A
post-beginners
√
?
?
X
intermediate
√
?
?
X
post-intermediate
√
√
?
X
advanced
√
√
?
X
native Chinese
√
√
√
X
beginners
Note: N/A=learners have not mastered the basic sentence structure; “√” = accept; “X”=reject; “?”=
indeterminate and variable.
4. Discussion
4.1. Valuation of L2 Chinese wh-questions
Our results indicate that C0 of the wh-question in English speakers’ L2 Chinese
grammars is valued by the merging of the Chinese wh-particle ne into C0 and that in
the yes-no question, C0 is valued by the Chinese yes-no particle ma. There is no L1
transfer in this aspect of L2 grammars as no wh-movement is found in English
speakers’ L2 Chinese although it is generally required in their L1 English. This
finding does not lend support to the transfer part of the Full Transfer and Full Access
model proposed by Schwarts and Sprouse (1994, 1996). Instead, it provides further
evidence for the argument that L1 transfer is not everywhere at the initial stage
although it is a common phenomenon in L2 acquisition (cf. Yuan, 2001, 2004). Our
results suggest that “merge” is given priority over “move” in English speakers L2
Chinese grammars and this is probably because “merge” is more economical than
“move”. An important theme in the Minimalist Program (Chomscky, 1995, 1998) is
that linguistic operations are subject to the principle of economy and that the
derivations should be as small as possible and be applied in a way that minimizes
search. In this sense, “merge” is more economical and less costly than “move”. If
Universal Grammar is available in L2 acquisition, it should not be surprising that the
principle of economy overrides the L1 transfer of wh-movement in English speakers’
L2 acquisition of Chinese wh-questions, even in the initial state. Almost from the
23
outset, L2 Chinese learners are exposed to yes-no questions and wh-questions in their
input data. The use of ma and ne in Chinese yes-no questions and wh-questions in the
input data can inform L2 Chinese grammars how C0 in Chinese yes-no questions and
wh-questions is valued. The principle of economy will dictate that any movement of a
wh-word to Spec CP will become unnecessary and impossible once the head C0 of L2
Chinese wh-questions is valued by ne. This seems to be what is suggested by the
results presented in Tables 3 and 4. When C0 of the wh-question is appropriately
specified, it is able to license the wh-word in situ in the sentence and features of the
in-situ wh-words can be checked at a distance by Agree.
Our data also indicate that English speakers accept Chinese wh-questions with wharguments embedded in a CNP or in a sentential subject so long as they have
mastered the basic structures of these two types of sentences. This suggests that there
is no movement of wh-argument in English speakers’ L2 Chinese grammars as any
movement of these wh-words out of a CNP or a sentential subject would violate
Subjacency. The findings here support our argument above that once C0 of the
Chinese wh-question is specified and valued as C0 with [+Q] and [+wh] by the whparticle ne, whether phonetically realised or not, the C0 thus specified and valued will
become a wh-licenser for the wh-word in situ and the principle of economy in L2
Chinese grammars will make any wh-movement unnecessary and impossible in L2
Chinese.
However, there is evidence that in English speakers’ L2 Chinese grammars, wh-words
do not develop in a uniform fashion and that different wh-words behave differently at
different Chinese proficiency levels, as shown in Table 8. Except for the Beginner
Group who have not mastered the basic structure of the test sentences, all learner
groups allow the Chinese wh-arguments shei “who” and shenme “what” to be inside a
CNP or a sentential subject. The same applies to the Post-intermediate and Advanced
Groups’ judgment of shenmeshihou “when” and nar “where”. The status of
zenmeyang “how” inside a CNP or a sentential subject seems to be variable and
indeterminate in all learner groups’ L2 Chinese grammars, and the variability and
indeterminacy are also found in the Post-beginner and Intermediate Groups’ judgment
of shenmeshihou “when” and nar “where”. Except for the Beginner Group, all the
24
learner groups, like the Native Chinese Group, seem to correctly reject sentences with
weishenme “why” inside a CNP or inside a sentential subject.
Why are there differences in the behaviours of wh-words in English speakers’ L2
Chinese? Why do Chinese wh-words behave differently at different Chinese
proficiency levels? And what are the underlying mental representations of the whwords in English speakers’ L2 Chinese wh-questions? To facilitate our analyses, let
us first make distinctions between three types of wh-words in the native Chinese
grammar in (28) and then try to answer the above questions with regard to learners’
Chinese proficiency levels.
(28) a. Wh-words with [+ argument] and [+ nominal] features
(e.g. shei “who” and shenme “what”),
b. Wh-words with [- argument] and [+ nominal] features
(e.g. shenmeshihou “when”, nar “where” and zenmeyang “how”)
c. Wh-words with [- argument] and [- nominal] features, i.e. wh-adverbs
(e.g. weishenme “why”).
4.2. Behaviours of Chinese wh-words at beginner and intermediate levels
It is likely that at beginner and intermediate levels, English speakers’ L2 Chinese
grammars are indeterminate as to whether the operator should be attached to whwords with the [-argument] feature or to wh-words with the [-nominal] feature. This
indeterminacy leads to optional attachment of the operator to wh-words in their L2
Chinese lexicons. When wh-words with the [-argument] feature are attached with the
operator, wh-words such as shenmeshihou “when” and nar “where” are not allowed
to stay inside a CNP or inside a sentential subject because the operator attached to
them requires raising to Spec CP for feature checking but the Subjacency constraint
makes that impossible. However, when the operator is attached to wh-words with the
[-nominal] feature, the wh-words shenmeshihou “when” and nar “where” are allowed
to stay inside a CNP or inside a sentential subject in L2 Chinese grammars as both the
learners’ L1 English and the L2 Chinese input data can tell their L2 Chinese lexicons
that these wh-words have the [+nominal] feature and therefore have no operator. As
25
shown in (29) and (30), these two wh-words in both languages can appear as object of
a preposition, which shows clearly that the wh-words shenmeshihou “when” and nar
“where” in these two languages have the [+nominal] feature as only words with the
[+nominal] feature can serve as object of a preposition.
(29) a. From where can we get some money?
b. Since when have you been living here?
(30) a. Women cong nar neng gaodao yixie qian?
We
from where can get some money?
‘From where can we get some money?’
b. Ni cong shenmeshihou kaishi zhu zai zheli?
You since when
start live in here
‘Since when have you been living here?’
If our analysis here is on the right track, we can assume that the Post-beginner and
Intermediate Groups’ variable judgment of Chinese wh-questions with shenmeshihou
“when” and nar “where” embedded inside a CNP or inside a sentential subject is due
to the optional attachment of the operator to the wh-words with the [-argument]
feature and to those with the [-nominal] feature in beginner and intermediate learners’
L2 Chinese lexicons. In other words, the wavering between the [-argument] feature
and the [-nominal] feature for the operator attachment is believed to be the source of
the variability at the surface level in these groups’ L2 Chinese wh-questions.
Let us put the behaviour of zenmeyang “how” aside for the time being and look at the
behaviours of the wh-words weishenme “why”, shei “who” and shenme “what” in
English beginner and intermediate learners’ L2 Chinese wh-questions. The wh-word
“why” has both the [-nominal] feature and the [-argument] feature in learners’ L1
English and there is no positive evidence in their L2 Chinese input data which
suggests to L2 Chinese grammars that this is not the case in the target language. As a
result, Chinese wh-questions with weishenme “why” embedded inside a CNP or
inside a sentential subject is rejected by L2 Chinese grammars whether the operator is
attached to wh-words with the [-nominal] feature or to wh-words with the [-argument]
feature in their L2 Chinese lexicons as the wh-adverb weishenme “why” has both of
the two features.
26
The wavering between these two features for the operator attachment will not lead to
variable behaviours of shei “who” and shenme “what” at the surface level either. This
is because the two wh-words have both the [+argument] and [+nominal] features.
Consequently, learners at these early stages accept wh-questions with shei “who” or
shenme “what” embedded inside a CNP or inside a sentential subject. This amounts to
saying that the stable behaviours of these wh-words shei “who”, shenme “what” and
weishenme “why” in these groups’ judgment are only superficial because the
underlying mechanism concerning the behaviours of these wh-words is still
indeterminate in these English speakers’ L2 Chinese grammars.
The English wh-word “how” has both the [-argument] and [-nominal] features. If
English beginner and intermediate learners of Chinese take the wh-word zenmeyang
“how” as the Chinese counterpart of the English wh-word “how”, we would expect
the wh-word zenmeyang “how” to have both the [-argument] and [-nominal] features
in the learners’ L2 Chinese lexicons. This would allow the Chinese wh-word
zenmeyang “how” to behave in the same way as the wh-word weishenme “why” at
these early stages. However, learners are exposed to positive evidence in their L2
Chinese input data that the Chinese wh-word zenmeyang “how” can be embedded
inside a CNP or inside a sentential subject. The two sources of information, i.e. the
positive evidence and the L1 transfer, are likely to make the operator optionally
attached to zenmeyang “how” in these learners’ L2 Chinese grammars, which can
account for their wavering between accepting and rejecting Chinese wh-questions
with zenmeyang “how” embedded inside a CNP or inside a sentential subject. That is,
the variation in beginner and intermediate learners’ judgment with regard to
zenmeyang “how” may result from the optional involvement of the operator as the
operator is only optionally attached to zenmeyang “how” at early stages of English
speakers’ L2 acquisition of Chinese wh-questions. Table 9 provides a summary of
consequences of the optional operator attachment in beginner and intermediate
learners’ judgment of Chinese wh-questions with the wh-word embedded inside a
CNP or inside a sentential subject. It seems reasonable to assume that the variable
behaviours of Chinese wh-words at these early stages are a result of what Sorace
( 2003) calls “developmental optionality” in early L2 Chinese grammars as it reflects
27
a partial overlapping between developmental stages in L2 acquisition of Chinese whquestions.
Table 9: Summary of the consequences of wavering between the [-argument] feature
and the [-nominal] feature for the operator attachment in English beginner and
intermediate learners’ L2 Chinese wh-questions with the wh-word embedded inside a
CNP or inside a sentential subject.
shei “who”,
shenmeshihou Zenme(yang)
weishenme
shenme “what”
“when”, nar
“how”
“why”
“where”
[-argument]
X
√
?
√
[-nominal]
X
X
?
√
Consequences
Accept
Variable
Variable
Reject
Note: “√” stands for the attachment of the operator to the features listed in the first column,
“X” for no attachment and “?” for an optional attachment.
4.3. Behaviours of Chinese wh-words at post-intermediate and advanced levels
So far we have only discussed the behaviours of different wh-words in the Postbeginner and Intermediate Groups’ L2 Chinese. In the discussion, we have assumed
that these groups’ L2 Chinese grammars generally use the [-argument] and [-nominal]
features optionally in determining whether a wh-word has an operator which requires
raising. The optional selection between different features in the groups’ L2 grammars
results in variable behaviours of some wh-words in the groups’ L2 Chinese whquestions. However, as we can see in Table 8, the variability in the Post-beginner and
Intermediate Groups’ judgment concerning shenmeshihou “when” and nar “where”
disappears in the Post-intermediate and Advanced Groups. This suggests that from the
post-intermediate level onward, feature attachment to L2 Chinese wh-words has been
reassembled and that wh-words with the [-argument] and [+nominal] features no
longer have an operator. This implicates that in these learners’ L2 Chinese grammars,
the [+/-argument] feature has lost the function of distinguishing wh-words with an
operator from those without and that the only feature that the L2 Chinese grammars
use for making such a distinction is the [+/-nominal] feature. This reassembly of
features in Chinese wh-words can be triggered by learners’ observation that all whwords that appear in a CNP and a sentential subject in Chinese have one thing in
28
common, that is, they all have the [+nominal] feature. This type of positive evidence
can force L2 Chinese lexicons to reassemble the features attached to Chinese whwords and remove the operator from wh-words with the [-argument] feature and only
attach the operator to wh-words with the [-nominal] feature.
Although it is not clear from our data when exactly the reassembly takes place at
intermediate stages of L2 acquisition of Chinese wh-questions, the relevant positive
evidence is certainly available in L2 Chinese learners’ input data from the outset. It is
plausible that the generalization of the [+nominal] feature to all wh-words inside the
CNP or sentential subject is not induced by learners’ L2 Chinese grammars until a
relatively later stage. This probably can count as an example for the argument that
adult L2 learners are not sensitive to all sorts of evidence in the L2 input data from the
outset and that not all cues in the input are “good” cues to L2 acquisition mechanisms
at a particular stage (cf. Carroll, 1999a). In this sense, it seems necessary to follow
Corder (1967) in making a distinction between input and intake. Corder’s distinction
is further developed in Schmidt’s (1990, 1993) Noticing Hypothesis, according to
which, noticing L2 language forms is a prerequisite for acquisition and these forms
will not be acquired unless they are noticed. Recently, a more fine-tuned proposal is
made by Carroll (1999b, 2001) regarding input and intake, which argues that not only
do L2 learners perceive what they expect to perceive – based on their conceptual
models of the environment - but they also perceive what their linguistic systems
enable them to perceive – based on the categories and structures of the grammar. That
is, intake can be determined by learners’ L2 grammars. Following this line of analysis,
we can argue that due to the complexity of the sentence in the input data involving the
CNP and the sentential subject, earlier L2 Chinese grammars are unable to perceive
the generalization of the [+nominal] feature to all Chinese wh-words inside the CNP
or sentential subject, and as a result, this piece of positive evidence cannot be
perceived and taken in by L2 Chinese grammars until a later stage. This can probably
provide an explanation why the reassembly of features does not take place until
around the post-intermediate level.
The reassembly of features attached to L2 Chinese wh-words, in general, enables the
L2 lexicon to make a native-like distinction between wh-words with an operator and
those without. That is, wh-words with the [+nominal] feature do not have an operator
29
while wh-words with the [-nominal] feature have an operator which is required to
raise to Spec CP for feature checking. This can account for Post-intermediate and
Advanced Groups’ judgment of Chinese wh-questions with shenme “what”, shei
“who”, shenmeshihou “when”, nar “where” and weishenme “why” embedded inside a
CNP and inside a sentential subject. As the wh-words shenme “what”, shei “who”,
shenmeshihou “when”, nar “where” all have the [+nominal] feature, they do not have
an operator. Therefore, they are allowed to remain in situ inside a CNP or inside a
sentential subject. In contrast, the wh-word weishenme “why” has the [-nominal]
feature, and as a result, it has an operator which is required to raise to Spec CP for
feature checking. If the wh-word weishenme “why” stays inside a CNP or inside a
sentential subject, the operator will not be able to move out of the CNP or the
sentential subject because of the Subjacency constraint. This is believed to be the
reason why Post-intermediate and Advanced Groups correctly reject Chinese whquestions with weishenme “why” embedded inside a CNP or a sentential subject and
accept those with shenme “what”, shei “who”, shenmeshihou “when” and “nar”
“where”, as shown in Table 8.
However, it is clear from Table 8 that learner groups’ judgment of Chinese whquestions with zenmeyang “how” is variable even at post-intermediate and advanced
levels, which can be taken as evidence of fossilization in this aspect of L2 Chinese
grammars. There seem to be two possible explanations for the persistent variable
behaviours of zenmeyang “how” in English speakers’ L2 Chinese wh-questions. One
is that there are two different representations of the wh-word zenmeyang “how” in
English speakers’ L2 Chinese lexicons, with one having the [+nominal] feature and
the other the [-nominal] feature. In the English lexicon, the wh-word how is grouped
in the same category as the wh-word why as both have the [-nominal] feature and both
have similar behaviours. However, the wh-word zenmeyang “how” in the native
Chinese lexicon is categorized differently from the wh-word weishenme “why”, with
the former having the [+nominal] feature and the latter the [-nominal] feature. There
seem to be reasons to assume that the word zenmeyang “how” may have two different
representations co-existing in English speakers’ L2 Chinese lexicons; one is a Chinese
counterpart of the English wh-word “how” with the [-nominal] feature and the other is
the proper Chinese wh-word zenmeyang “how” which has the [+nominal] feature. The
acquisition of the latter is made possible by the positive evidence that the Chinese wh-
30
word zenmeyang “how” can stay in situ inside a CNP or inside a sentential subject. As
there is no positive evidence in the input data which can help remove the Chinese
counterpart of the English wh-word how from English speakers’ L2 Chinese lexicons,
the two representations of zenmeyang “how” permanently co-exist in English
speakers’ L2 Chinese lexicons, leading to persistent optionality. This seems to support
Sorace’s (2003) analysis that “Advanced optionality is the consequence of
unsuccessful restructuring of the L2 grammar (and the related failure to expunge the
L1 setting)”. This kind of optionality may never disappear as it is not a developmental
phenomenon any more.
Another explanation, although it is a tentative one, is that the optionality may be due
to English learners’ difficulty in making a distinction in Chinese between the
instrumental “how” and the manner “how” at the morphosyntax-semantics interface.
Recall that in Chinese the instrumental “how” is encoded with a nominal property and
therefore has the [+nominal] feature while the manner “how” has an operator which
requires raising (cf. Lin, 1992 Tsai, 1994a). It is believed that English speakers are
able to make the distinction semantically as this distinction does exist in their L1
English, as shown in (31) (from Tsai, 2004), although the semantic difference does
not lead to syntactic difference in English, as shown in (32).
(31) A: How did John handle this matter?
B: a. Quite skilfully, I think.
b. By pulling quite a few strings.
(manner)
(instrumental)
(32) *Howi do you like [the cake [she makes ti]]
a. *‘What is the means x such that you like [cakes [which she makes by x]]?
(*instrumental-how)
b. *‘What is the manner x such that you like [cakes [which she makes in x]]?
(*manner-how)
In English, the instrumental “how” shares the same syntactic behaviour as the manner
“how”, as shown in (31) and (32). However, in Chinese the instrumental “how” has a
morphosyntactic implication different from the manner “how”; the former has a
nominal property and can remain inside a CNP or a sentential subject while the latter
31
has an operator which requires raising and makes it impossible for the wh-word to
stay inside a CNP or inside a sentential subject. This means that it is the
morphosyntax-semantics interface conditions that determine the behaviours of the
instrumental “how” and the manner “how” in the native Chinese grammar. It is
possible that the variability found in Post-intermediate and Advanced Groups’
judgment of zenmeyang “how” is due to insufficient knowledge of these interface
conditions in their L2 Chinese grammars. Although English learners are able to make
an interpretive distinction between the instrumental “how” and the manner “how” as it
is available in their L1 English (and it is probably also available universally), the two
semantic meanings of “how” in English do not have any different morphosyntactic
implications. In order for English learners to acquire the distinctive behaviours of the
instrumental “how and the manner “how” in Chinese, not only does the semantics
condition but also the morphosyntax conditions have to be met in their L2 Chinese
grammars. To achieve this, L2 Chinese grammars seem to have to soly rely on the
input data.15 However, these distinctive morphosyntax conditions are
underdetermined by the L2 Chinese input data. As we can see in (17), repeated below,
questions like (17) and any instrumental answer to it in the input data CANNOT tell
L2 Chinese grammars that the manner “how” and the instrumental “how” in Chinese
are subject to different morphosyntax conditions. In this situation, the interface
between the semantics condition and the morphosyntax conditions breaks down and
as a result, the operator will not be properly specified and the behaviours of
zenmeyang “how” will remain variable even in the final state of English-speaking
learners’ L2 Chinese grammars.16
(17) Ni bu xihuan [[ta zenmeyang zuo de] dangao] (ne)?
you not like she how
make DE cake
Q
a. ‘What is the means x such that you don’t like [cakes [which she makes by x]]?
(instrumental-how)
b. *What is the manner x such that you don’t like [cakes [which she makes in x]]?
(*manner-how)
15
It would be extremely rare if knowledge of these interface conditions could be imparted to learners in
classroom instruction.
16
The aspect of L2 grammars governing the behaviours of zenmeyan “how” is made more vulnerable
to optionality by the fact that the distinctions between the instrumental “how” and the manner “how”
sometimes can be rather vague in some Chinese sentences. As we mentioned earlier, even some native
Chinese speakers do not always find it easy to draw a clear line between the instrumental “how” and
the manner “how” in some Chinese questions.
32
A question may arise: Why is there no such a problem with weishenme “why” in
English post-intermediate and advanced learners’ L2 Chinese grammars, given that
there is also a distinction between the purpose “why” and the reason “why” in
Chinese? The answer to this question has to be lexical in nature. The purpose “why”
in Chinese is in fact spelt out as two words wei(le) “for” and shenme “what”. The
former is a preposition as in wei(le) ni “for you” and wei(le) xuexiao “for the school”.
If the purpose “why” is distinctively stored in the L2 Chinese lexicon as wei(le)
shenme and the reason “why” as weishenme, the optionality that occurs to zenmeyang
“how” will not happen to the purpose “why” and the reason “why” in L2 Chinese as
the latter two are treated as two separate and unrelated lexical items in L2 Chinese
lexicons. This can probably provide an account for post-intermediate and advanced
learners’ rejection of Chinese wh-questions with weishenme “why” embedded inside a
CNP or a sentential subject, as weishenme “why” has the [-nominal] feature and is,
therefore, attached with an operator which has to raise to Spec CP and does not allow
the wh-word to stay inside any island due to Subjacency.
It should be pointed out that the study reported in this paper is not designed to
investigate the distinction between the instrumental “how” and the manner “how” in
L2 Chinese. So, our analyses above about the distinction in relation to the behaviours
of zenmeyang “how” in L2 Chinese are tentative in nature and have to be treated with
caution. However, we believe that in L2 grammars, a breakdown at the
morphosyntax-semantics interface, as we discussed above, can be a cause of
optionality. Of course, studies specifically designed to address this issue are needed to
help us understand this under-studied area in L2 acquisition research.
5. Conclusion
Recent developments in linguistic research have revealed that raising an operator in
Chinese wh-questions can be blocked not only by islands such as the CNP and a
sentential subject but also by interveners such as zhi “only”, bu “not”, aspect markers,
etc. (cf. Soh, 2005, Tsai, 2004, Yang 2005). While further research is needed, by
using these interveners as well as the islands, to help deepen our understanding of
33
behaviours of Chinese wh-words with the [-nominal] feature and those with the
[+nominal] feature, we hope that the study presented in this paper has shown the
necessity to look at the lexical morphology-syntax interface as a possible locus for
variable behaviours in L2 acquisition.
In this study, we have analyzed our results on the basis of a lexical morphological
feature defect account. If we adopt the view of the Minimalist Program that the
computational system takes words from the lexicon along with their morphological
information and forms LF and PF representations via a derivational procedure, we can
assume that a wh-word selected from L2 Chinese lexicons with a defective
morphological feature can have syntactic implications. In our study, this defect is
manifested in beginner and intermediate learners’ optional attachment of the operator
to wh-words with the [-argument] feature and wh-words with the [-nominal] feature,
which leads to variable behaviours of wh-words such as shenmeshihou “when” and
nar “where” in their L2 Chinese wh-questions. At post-intermediate and advanced
levels, the defect can be seen in the co-existence of two representations of zenmeyang
“how” in English speakers’ L2 Chinese lexicons; one is a Chinese counterpart of the
English wh-word “how” with the [-nominal] feature and the other the Chinese whword zenmeyang “how” with the appropriate [+nominal] feature, the result of which is
variable behaviours of zenmeyang “how” in post-intermediate and advanced learners’
L2 Chinese wh-questions. We hope that our analyses of the results in this study will
draw people’s attention to the lexical morphology-syntax interface as a possible locus
of non-native-like behaviours in L2 acquisition. It is important to look at the
relationship between functional categories and their morphological representations in
L2 grammars, and it is equally important not to take it for granted that features of
lexical items can be acquired in L2 acquisition.
References:
Aoun, J., and Li, A. Y-H. 1993a. Syntax of Scope. Cambridge: MIT Press.
Aoun, J., and Li, A. Y-H. 1993b. Wh-elements in-situ: syntax or LF? Linguistic
Inquiry 24:199-238.
Carroll, Susanne. 1999a. Adults' sensitivity to different sorts of input. Language
Learning 49:37-92.
34
Carroll, Susanne. 1999b. Putting "input" in its proper place. Second Language
Research 15:337-388.
Carroll, Susanne. 2001. Input and Evidence: the Raw Material of Second Language
Acquisition. Amsterdam: John Benjamins.
Cheng, L. L.-S. 1991. On the Typology of Wh-questions, MIT: Ph.D dissertation.
Cheng, L. L.-S. and Rooryck, J. 2000. Licensing wh-in-situ. Syntax 3, 1-19.
Cheng, L. L.-S. and Rooryck, J. 2002. Types of Wh-in-situ. Ms. Leiden University.
Chomsky, N. 1973. Condition on transformations. In A Festschrift for Morris Halle,
eds. S. Anderson and P. Kiparsky. New York: Holt, Rinehart and Winston.
Chomsky, N. 1995. The Minimalist Program. Cambridge, Mass: MIT Press.
Chomsky, N. 1998. Minimalist inquiries: the framework. MIT Occasional Papers in
Linguistics 15:1-56.
Chomsky, N. 2001a. Derivation by phrase. In Ken Hale, A Life in Language, ed. M.
Kenstowics. Cambridge, Mass: MIT Press.
Chomsky, N. 2001b. Beyond Explanatory Adequacy. Ms. MIT.
Corder, S. P. 1967. The significance of learners' errors. International Review of
Applied Linguistics 5:161-170.
Hawkins, R., and Chan, C. Y.-H. 1997. The partial availability of Universal Grammar
in second language acquisition: The 'failed functional features hypothesis'.
Second Language Research 13:187-226.
Hawkins, R. 2001. Second language syntax: A generative introduction. Oxford:
Blackwell.
Hawkins, R., and Liszka, S. 2003. Locating the source of defective past tense marking
in advanced L2 English speakers. In The lexicon-syntax interface in second
language acquisition, eds. R. van Hout, A. Hulk, F. Kuiken and R. Towell,
21-44. Amsterdam: John Benjamins.
Hawkins, R. 2004. Revisiting Wh-Movement: the Availability of an Uninterpretable
[wh] Feature in Interlanguage Grammars. Ms. University of Essex.
Huang, C.-T. J. 1982a. Move WH in a language without WH movement. The
Linguistic Review 1:369-416.
Huang, C.-T. J. 1982b. Logical Relations in Chinese and the Theory of Grammar,
MIT: Ph.D dissertation.
Kim, J. 2003. L2 Initial Syntax: Wh-movement and the Most Economical Syntactic
Derivation. In BUCLD 27 Proceedings, ed. B. Beachley, et al. Somerville,
MA: Cascadilla Press.
Klein, E. 1995. Evidence for a 'Wild' L2 Grammar: When PPs Rear their Empty
Heads. Applied Linguistics 16:87-117.
Lardiere, D. 1998a. Case and tense in the 'fossilized' steady state. Second Language
Research 14:1-26.
Lardiere, D. 1998b. Dissociating syntax from morphology in a divergent end-state
gramma. Second Language Research 14:359-375.
Lardiere, D. 2000. Mapping features to forms in second language acquisition. In
Second language acquisition and linguistic theory, ed. J. Archibald, 102-129.
Malden, MA: Blackwell.
Lin, Jo-Wang. 1992. The syntax of zenmeyang 'how' and weishenme 'why' in
Mandarin Chinese. Journal of East Asian Languages 1:293-331.
Martohardjono, G. 1993. Wh-movement in the Acquisition of a Second Language: A
Crosslinguistic Study of Three Languages with and without Movement,
Cornell University: Doctoral dissertation.
35
Martohardjono, G., and Gair, J. 1993. Apparent UG Inaccessibility in Second
Language Acquisition: Misapplied Principles or Principles Misapplication. In
Confluence: Linguistics, L2 Acquisition and Speech Pathology, ed. F Eckman.
Philadelphia: John Benjamins.
Prévost, P., and White, L. 2000. Missing surface inflection or impairment in second
language acquisition? Evidence from tense and agreement. Second Language
Research 16:103-133.
Reinhart, Tanya. 1998. Wh-in-situ in the Framework of the Minimalist Program.
Natural Language Semantics 6:29-56.
Rizzi, Luigi. 1997. The fine structure of the left periphery. In Elements of Grammar:
Handbook of Generative Syntax, ed. Liliane Haegeman, 281-337. The
Netherlands: Kluwer Academic Publishers.
Rooryck, Johan. 1994. On two types of underspecification: towards a feature theory
shared by syntax and phonology. Probus 6:207-233.
Schachter, J. 1990. On the Issue of Completeness in Second Language Acquisition.
Second Language Research 6:93-124.
Schmidt, R. 1990. The role of consciousness in second language learning. Applied
Linguistics 11:129-158.
Schmidt, R. 1993. Awareness and second language acquisition. Annual Review of
Applied Linguistics 13:206-226.
Schwartz, B. and Sprouse, R. 1994. Word order and nominative case in non-native
language acquisition: a longitudinal study of (L1 Turkish) German
Interlanguage. In Hoekstra, T. and Schwartz, B., editors, Language acquisition
studies in generative grammar: papers in honour of Kenneth Wexler from the
1991 GLOW workshop, 317-368. Amsterdam: John Benjamins.
Schwartz, B. and Sprouse, R. 1996. L2 cognitive states and the full transfer/full access
model. Second Language Research 12, 40-72.
Simpson, A. 2000. Wh-movement and the Theory of Feature-checking. Amsterdam:
John Benjamins.
Smith, N., and Tsimpli, I.-M. 1995. The mind of a savant. Oxford: Blackwell.
Soh, H.-L. 2005. Wh-in-situ in Mandarin Chinese. Linguistic Inquiry, 143-155.
Sorace, Antonella. 2003. Near-nativeness. In The Handbook of Second Language
Acquisition, eds. Catherine J. Doughty and Michael H. Long. Oxford:
Blackwell.
Tsai, W.-T. D. 1994a. On nominal islands and LF extractions in Chinese. Natural
Language and Linguistic Theory 12:121-175.
Tsai, W.-T. D. 1994b. On Economizing the Theory of A-bar Dependencies, MIT:
Ph.D.
Tsai, W.-T. D. 1999. On lexical courtesy. Journal of East Asian Linguistics 8:39-73.
Tsai, Wei-Tien Dylan. 2004. Left Periphery and How-Why Alternations. Ms.
National Tsing Hua University, Taiwan.
Tsimpli, I.-M. 2003. Features in language development. Paper presented at EUROSLA
2003, Edinburgh.
Watanabe, A. 2001. Wh-in-situ languages. In Baltin, M. and Collins, C. editors, The
Handbook of Contemporary Syntactic Theory, 203-225. Oxford: Blackwell
Publishers.
White, L., and Juffs, A. 1998. Constraints on Wh-movement in Two Different
Contexts of Nonnative Language Acquisition: Competence and Processing. In
The Generative Study of Second Language Acquisition, ed. S. et al. Flynn.
Mahwah, NJ: Erlbaum.
36
White, Lydia. 2003. Fossilization in steady state L2 grammars: Persistent problems
with inflectional morphology. Bilingualism: Language and Cognition 6:129141.
Yang, C.-Y. Barry. 2005. Aspect as an intervener. Paper presented at 13th Conference
of International Association of Chinese Linguistics, University of Leiden, the
Netherlands.
Yuan, B. 2001. The status of thematic verbs in the second language acquisition of
Chinese: against inevitability of thematic-verb raising in second language
acquisition. Second Language Research 17:248-272.
Yuan, B. 2004. Negation in French-Chinese, German-Chinese and English-Chinese
interlanguages. Transactions of the Philological Society 102:169-197.
Yuan, B. 2005. Superficial Native-likeness and Fossilization in Japanese Speakers' L2
Chinese Wh-questions: A Lexical Morphological Feature Account. Ms.
University of Cambridge.