This thesis is organized as follows. Chapter Two reviews several theoretical studies on language acquisition and empirical research on the acquisition of reflexives. Chapter Three introduces the research design and Chapter Four presents and discusses the results of the experiments. Finally, Chapter Five summarizes the major findings of the present study.
11
CHAPTER TWO LITERATURE REVIEW
In this chapter, linguistic properties of Italian and Chinese in theoretical viewpoints and empirical studies will be discussed. Section 2.1 shows the linguistic properties of Italian and Chinese reflexives from previous theoretical studies while Section 2.2 reviews the empirical studies related to the acquisition of reflexives. Finally, Section 2.3 is a summary of this chapter.
2.1 Linguistic Properties and Literature Review of Italian and Chinese Reflexives
Previous studies have pointed out that although Italian and Chinese are two genealogically unrelated languages, both of them have anaphoric reflexives and emphatic reflexives.
Moreover, researchers also indicated that the structure of anaphoric reflexives is more complex than the emphatic ones (Giorgi, 2006, 2007; Tsai, 2002; Yeh, 2000).
Section 2.1.1 introduces the anaphoric reflexives in both Italian and Chinese. Regarding anaphoric reflexives, three major properties long-distance binding, blocking effects, and subject orientation will be examined. In addition to anaphoric uses, Section 2.1.2 introduces the emphatic reflexives. Section 2.1.3 investigates the peculiar property of Italian reflexive verbs, which have been briefly mentioned in Chapter One. In the end, Section 2.1.4 is a summary of this section.
Before discussing the important properties between Italian and Chinese reflexives, let us have an overview of reflexive forms in the two languages.
12
As seen in (1a) to (1c), Italian reflexives can be divided into three groups, reflexive verbs, pronominal reflexives and possessive reflexives (Giorgi, 2006). The reflexive verb accusarsi
‘accuse’, pronominal se stesso ‘him-self’ and the possessive reflexive proprio all have reflexive
meanings; however, their distributions and tones are different. As can be seen in (1a), an Italian reflexive verb always is adjoined to the verb essere ‘be V’, which is a specific property in Italian. In (1b), se stesso can be considered an emphatic reflexive when a speaker tries to strengthen the meaning of self, in which se stesso is adjoined to the verb avere ‘has/have’.
13
Lastly, (1c) is the only anaphoric type in Italian reflexives that has long-distance (LD) binding and is a possessive reflexive.
With regard to Chinese reflexives, (2a) and (2c) are anaphoric uses and they both can be long-distance bound; furthermore, (2c) is a possessive reflexive. With respect to emphatic uses, (2b) is an example where the speaker uses a reflexive in a non-argument position to strengthen the meaning of self (Burzio, 1986, 1996, 2008; Giorgi, 2006).
2.1.1 Anaphoric Reflexives
Anaphoric reflexives are reflexives in argument positions and they can refer to antecedents in a context. Giorgi (2006:1012) claimed that ‘neither Italian possessive proprio nor Chinese
ziji can be used for deixis’; hence, both of them are anaphoric reflexives. However, Chinese taziji is excluded in the present study since we only focus on Chinese ziji and Italian proprio.
In Chapter One, we have already discussed the important properties of anaphoric reflexives, such as LD binding (see Seciton 2.1.1.1), blocking effects (see Section 2.1.1.2) and subject orientation (see Section 2.1.1.3). As mentioned previously, both Italian and Chinese anaphoric reflexives have long-distance binding; however, their binding conditions are attributed to different factors. Moreover, the LD binding is not free from restrictions in both languages, namely blocking effects. Lastly, these two languages behave differently with respect to subject orientation.
14
2.1.1.1 Long-distance Binding
Many researchers (Battistella & Xu, 1990; Giorgi, 2006, 2007; Rivero, 2001; Tang, 1989;
Yeh, 2000) have pointed out that Binding Theory proposed by Chomsky (1981) cannot ensure the proper binding of Italian proprio and Chinese ziji. In Italian, a first person pronoun cannot be an antecedent while the possessive proprio is always third person. Hence, when an antecedent-less anaphor appears, the sentence is ungrammatical (Giorgi, 2006:1033):
(3)
*Io amo la propria madre.I love the self mother ‘I love self’s mother.’
On the contrary, in Chinese the first person pronoun can be an antecedent for the reflexive ziji, as shown in (4).
(4) Wo ai ziji de mama.
I love self’s mother ‘I love my mother.’
With respect to their distributions, Italian proprio is a possessive anaphor (Giorgi (2006), as in (5):
(5)
Gianni
icrede che Paolo
jodi la propria
i/jmoglie.
Gianni believes that Paolo hates the self wife
‘Gianni believes that Paolo hates (SUBJ) self’s wife.’
Sentence (5) is ambiguous since both Gianni and Paolo are possible antecedents. As noted by Giorgi, Paolo is a co-argument of the DP containing an anaphor, which also satisfies the c-command requirement. Regarding Gianni, the extended projection to the MOODP1 of the verb
1 Giorgi and Pianesi (2001) and Giorgi (2009) have noted that there are two distinct complementizers singled
15
credere ‘believe’ is marked as unsaturated while the bearer-of-attitude
2 is locally available.Hence, Gianni can be selected as a possible antecedent for la propria. The unsaturated position of the MOODP is finally spelled out as a LDA and theta-identified with Gianni.
In Chinese, the long-distance binding with respect to the analogous structure also occurs, as in (6).
(6) Zhangsani xiangxin Lisij taoyan zijii/j de laopuo.
Zhangsan believe Lisi hate self DE wife
‘Zhangsan believes that Lisi hates his wife.’
In (5) and (6), we can see that the LD binding in the sentence triggers the ambiguity to the addressee. In addition to bi-clausal structures, both Italian and Chinese exhibit LD binding in tri-clausal structures, as in (7) and (8), respectively.
(7) Italian:
[
S1Mario
isupponeva [
S2che Gianni
jcredesse [
S3che Paolo
kodiasse la propria
i/j/kmoglie]]].
‘Mario supposed that Gianni believed (SUBJ) that Paolo hated (SUBJ) self’s wife.’
(8) Chinese:
[S1Zhangsani renwei [S2Lisij xiangxin [S3Wangwuk taoyan zijii/j/k de laopuo]].
‘Zhangsani supposed that Lisi believed that Wangwu hated his wife.’
out in embedded clauses. According to their split-C framework, the lower complementizer, called MOOD, occurs with an embedded subjunctive verb while the higher complementizer C is strictly associated with Double Access Readings (DAR), which is further explained in footnote 4.
2 The notion of the bearer-of-attitude is proposed by Giorgi and Pianesi (2001), who claim that some Italian verbs with communication functions, such as dire ‘’say, express the subject’s attitude. This is like an implication of the speech act.
16
Although both Italian and Chinese reflexives can select their antecedents across the binding domain, it should be noted that their domain of binding is constrained by blocking, which will be discussed in the following section.
2.1.1.2 Blocking Effects
When an anaphor is restricted to its antecedents under certain circumstances, the blocking effects occur. According to (Giorgi, 2006), the blocking conditions in Italian and Chinese reflexivization across domains are different. Italian is a language with mood distinction, and the domain of LDA can be defined by its indicative mood (IND) and subjective mood (SUBJ).
Italian blocking occurs with respect to indicative verbs and also the speaker’s coordinate, which is initiated by certain verbs, such as dire ‘say’. When the speaker’s coordination is initiated by the indicative verbs, the bearer-of-attitude is locally available and the saturated MOODP will not be extended, namely blocking. In (9), Italian LD binding is due to the subject’s coordinate and the bearer-of-attitude.
(9) [S1Giannii crede [S2che Paoloj supponga [S3che Mariok odi la propriai/j/k moglie.]]]
Gianni believes that Paolo supposes that Mario hates the self wife
‘Gianni believes that Paolo supposes (SUBJ) that Mario hates (SUBJ) self’s wife.’
As can be seen in (9), the subjective verbs ‘suppose’ and ‘hate’ cannot saturate MOODP and
Gianni is locally available as a possible antecedent. However, when the anaphor in the clause
17
containing an indicative verb, LDA will be blocked, as seen in (10).
(10) [S1
Gianni
isa [
S2che Paolo
jha detto [
S3che Mario
kama la propria
*i/*j/kmadre.
Gianni knows that Paolo said that Mario loves the self mother
‘Gianni knows that Paolo said that Mario loves (IND) self’s mother.’
The possibility for the possessive la propria in (10) to have the matrix subject as its antecedent is blocked since the verb ama ‘love’ is indicative. Moreover, the verb detto ‘said’
expressed the bearer-of-attitude, which is different from the verb ama. This indicative verb ama is attributed to the double access reading (DAR) effect
3in Italian contexts (Giorgi &
Pianesi, 2001). In DAR contexts, the elements in T and C must be the same. However, the indicative verb cannot satisfy the bearer-of-attitude since the attitudes are different; hence, the only possible antecedent in (10) is Mario. Therefore, we can conclude that
the possessiveproprio can have an antecedent over the subjective clauses while the blocking effect occurs in
the indicative mood (Burzio, 1996; Giorgi, 2006).As for Chinese, its blocking effect occurs according to the person asymmetry
(Giorgi, 2006; Huang & Li, 1996; Huang & Liu, 2001; Yeh, 2000). As pointed out by Huang and Liu(2001), a third person LD antecedent may be blocked by a first/second person pronoun, as shown below:
(11) Zhangsan
irenwei wo/ni
jzhidao Lisi
ktaoyan ziji
*i/*j/k. Zhangsan think I/you know Lisi hate self
‘Zhangsan thinks that I/you know that Lisi hates himself.’
3 The Double Access Readings (DAR) only occurs in Romance languages that have abundant verbal morphology (Giorgi & Pianesi, 2001). Giorgi (2006) argues that in Italian, the binding of LDA coincides with DAR contexts whereas Chinese does not have any of DAR since it lacks the verbal morphology.
18
Giorgi (2006) named the Italian blocking effect as the ‘verbal blocking effect’ since
Italian’s LDA is limited by the presence of indicative verbal forms; however, the Chinese blocking effect is ‘nominal’ because the blocking effect appears in the clause in which the matrix and embedded participants do not agree in person.42.1.1.3 Subject Orientation
Another important property of anaphoric reflexives is subject orientation, which means that the only possible antecedent of the anaphor is the subject rather than other arguments, as illustrated in (12) and (13).
(12) Italian:
Gianni
iha informato Maria
jche la propria
i/* jcasa era in fiamme.
Gianni informed Maria that the self home was in flame
‘Gianni informed Maria that self’s house was on flame.’
(Giorgi, 2006:1013)
(13) Chinese:Zhangsani gaosu Lisij zijii/*j-de jia shihuo le.
Zhangsan inform Lisi self’s home on-flame PERS
‘Zhangsan informed that Lisi his
house was on flame.’
4 Giorgi (2006:1018) also noted that Italian is a ‘mild’ nominal blocking since the person asymmetry causes the blocking effect. However, the nominal blocking is not as strong as the subjunctive/indicative verbal blocking.
19
However, the bound anaphors in Italian are not obligatorily subject-oriented. For example:
(14) Ho convinto Mariai del proprioi valore. (Italian) convinced.1st Maria of self value
‘I convinced Maria of self’s value.’
(15) I informed Bill that he himself was selected by the committee. (English)
(Giorgi, 2006:1013)
As shown above, Italian and English pronouns can refer to objects. In both sentences, the reflexive proprio/himself is in object position and bound by the antecedent Maria/Bill. Italian possessive proprio is not necessarily bound with the subject.
On the contrary, Chinese pronouns are always subject-oriented, as can be seen in (16).
(16) Zhangsani song Lisij yi-chang zijii/*j-de chaopian.
Zhangsan give Lisi one-CL self’s picture
‘Zhangsan gave Lisi a picture of himself (Zhangsan).’
(Tang, 1989)
Hence, we can conclude that Chinese pronouns are obligatorily subject-oriented whereas Italian pronouns are not obligatory.
2.1.1.4 Summary
First of all, both Italian reflexive proprio and Chinese reflexive ziji demonstrate long-distance binding. Italian anaphor proprio is a possessive form while Chinese reflexive can be either morphological ziji or possessive ziji-de. However, the Italian and Chinese LDAs are blocked by different conditions: in Italian the blocking effect is verbal, but in Chinese the
20
blocking is nominal (Burzio, 1996; Giorgi, 2006). However, as also noted by Giorgi (2006), Italian is mild nominal since the person asymmetry can also cause blocking. Regarding subject orientation, Chinese is strictly subject oriented because the pronouns are always bound to the subject; however, Italian pronouns can refer to both participants and objects, which is similar to English. Their differences may be attributed to the typological difference between Italian and Chinese. Italian has abundant verbal morphology and the DAR effect while Chinese lacks these properties, as shown in Table 2-1.
Table 2-1 Comparisons between Italian and Chinese Anaphoric Reflexives
Italian proprio Chinese ziji-de
Emphatics reflexives are different from anaphoric reflexives since the former ones only occur in non-argument position while the latter ones must appear in argument position. In Italian, the morphologically complex anaphor se stesso ‘lit. self-same’ is a strictly local anaphor and the interpretation focuses on ‘self’ instead of the action of the statement, as in (17) and (18) (Burzio, 1996, 2008).
21
(17) Gianni si lava.
Gianni himself washes
‘Gianni washes himself.’ (Burzio, 2008:16) (18) Gianni lava se stesso.
Gianni washes him-self
‘Gianni washes himself.’
In (17) and (18), the reflexives have the same interpretation. However, the complex anaphor se stesso in (18) strengthens the ‘self’ meaning while the sentence in (17) is simply to describe a fact.
Regarding the emphatic use of Chinese, the reflexive ziji usually occurs in the preverbal
position (Tsai, 2002; Yeh, 2000).
(19) a. Dodoi zijii hui hui-jia. (Appositive) Dodo self will go home
‘Dodo himself will go home.’
b. Dodoi hui zijii zuo zuo-ye. (VP-adverbial) Dodo will self do homework
‘Dodo will do homework (by) himself.’
The Chinese emphatic ziji only occurs in non-argument position and it must be adjoined to participants; furthermore, the distributions of emphatic use can be either appositive or VP-adverbial.
However, emphatic reflexives are not as complex as anaphoric reflexives since they do not have the property of LD binding. Moreover, they do not exhibit any blocking effect or subject orientation since both of them are properties of anaphoric uses, as demonstrated in Table 2-2.
22
Table 2-2 Comparisons between Italian and Chinese Emphatic Reflexives
Italian se stesso Chinese ziji
Long-distance Binding No No
Blocking effects No No
Subject Orientation No No
Position Postverbal Postverbal/Preverbal
2.1.3 Linguistic Properties of Italian Reflexives
An idiosyncratic property of Romance languages (i.e., Italian and Spanish) is their specific reflexive verbs. Italian reflexives can be divided into two types according to their arguments:
one is nominal while the other is verbal (Miličević, 2007: 93). Moreover, the difference between the first two types of verbs is in their distinct choices of aspectual auxiliaries that the former chooses the essere ‘be’ while the latter coincides with the avere ‘have’, as shown in (20c) and (20d) (Burzio, 1986).
(20) a. Luigi si è accusato.
Luigi himself is (ESSERE.3rd) accuse.PERS ‘Luigi has accused himself.’
b. Luigi ha accusato se stesso.
Luigi has (AVERE.3rd) accuse.PERS him-self ‘Luigi has accused himself.’
(Burzio, 1986; Miličević, 2007)
Both Italian reflexive and unaccusative verbs select aspectual auxiliary essere ‘be’ while Italian nominal reflexives and unergative verbs select the auxiliary avere ‘has’; hence, L3 learners may have difficulties selecting the right auxiliaries for these types of verbs.
23
We have mentioned that Italian reflexive verbs are composed of verb roots and the clitic pronouns, such as si. Burzio (1986) discusses the reflexive si by providing the examples below:
(21) a. Maria si guarda.
Maria herself watches
‘Maria watches herself.’
b. Il vetro si rompe.
the glass itself breakes
‘The glass breaks.’
c. Giovanni si sbaglia.
Giovanni himself mistakes
‘Giovanni is mistaken.’
The three sentences shown above differ in certain contexts. Burzio assumes that the clitic
si is a reflexive object clitic, which is base-generated and form an empty category in object
position, as in (22).(22) Maria si guarda [e].
Maria herself watches herself (base-generated) ‘Maria watches herself.’
The object position is assigned a theta role by the verb and the clitic is a spell-out by case-marking. Moreover, the case and theta role of reflexive structure are born by se stesso/a
‘him/herself (emphatic)’. Hence, Burzio considers si in (21a) a clitic and si in (22) as a reflexive.
However, the clitic si in (21b) has no reflexive meaning since it alternates with a subject, as shown in (23).
(23) a. Il vetro si rompe.
the glass itself breakes
‘The glass breaks.’
24
b. Giovanni rompe il vetro.
Giovanni breaks the glass
‘Giovanni breaks the glass.’
According to Burzio, the clitic si is used to generate with the ergative verbs like rompersi
‘break’. Hence, the clitic si in (21b) and (23a) is categorized as “ergative si”. As for the clitic
si in (21c), Burzio considers it “inherent (-reflexive) si” since it alternates with neither a direct
object nor a subject, as illustrate in (24).(24) a. Giovanni si sbaglia.
Giovanni himself mistakes ‘Giovanni is mistaken.’
b. Giovanni sbaglia Pietro.
Giovanni mistakes Pietro ‘Giovanni mistakes Pietro.’
The behavior of ‘inherent si’ is assumed to be either intransitive or unaccusative.
Furthermore, as pointed out by Burzio (1986), the counterpart of inherent-reflexive sbagliarsi
‘mistake’ can only take prepositional objects, as shown in (25).
(25) a. Giovanni sbaglia tutto.
Giovanni mistakes everything
‘Giovanni does everything wrong.’
b. Giovanni si sbaglia su tutto.
Giovanni himself mistakes on everything
‘Giovanni is mistaken on everything.’
c. *Giovanni si sbaglia tutto.
Giovanni himself mistakes everything
‘Giovanni is mistaken everything.’
25
As the examples shown above, reflexive verbs are composed of the verb roots and the clitic si, such as defender+si ‘difendere’. Since the Italian clitic si conjugates with person and gender, it is expected that the complexity of the acquisition of Italian reflexive verbs increase.
Since the language specific property of auxiliary selection and clitics that are absent in learners’ native language or their L2 (i.e. English), Italian verbal reflexives may hinder L3 learner’s linguistic ability, as in Table 2-3.
Table 2-3 Comparisons between Linguistic Properties of Italian and Chinese Reflexives Italian reflexive verbs Chinese reflexive verbs
Clitics V Ø
Verbal conjugations V Ø
Auxiliary selection V Ø
Hence, it is worth investigating whether L3 learners can conquer the complexity of Italian verbal reflexives and master the conjugations of both verbs and clitics.
2.1.4 Summary
In this chapter, we have seen two different types of reflexives: anaphoric and emphatic.
With regard to anaphoric reflexives, both Italian and Chinese anaphors demonstrate LD binding.
However, the LDAs between the two languages are attributed to their different properties of reflexives, such as Italian possessive proprio and Chinese morphological ziji and possessive
ziji-de. As mentioned, there are other two other important properties of anaphoric reflexives,
namely blocking and subject orientation. Regarding blocking, Italian is verbal whereas Chinese26
is nominal; the blocking of Italian reflexives is attributed to the indicative verbs and attitude-of-bearer while Chinese blocking is determined by disagreement of person (1st and 2nd) and number (singular and plural). Moreover, Chinese reflexive pronouns are obligatorily restricted to subject orientation whereas Italian reflexives are not subject to this constraint, as shown in Table 2-4:
Table 2-4 Types of Reflexives in Italian and Chinese
Italian Chinese
Note: vx xx (zero contrast: no such form in L1; syntactic issue) vv vx (zero contrast: no interpretation in L1; semantic issue)
27
Besides, we also have discussed the emphatic use of reflexives in both languages. Chinese emphatic reflexives are found to occur in non-argument position whereas an Italian compound reflexive like se stesso can only occur in object position. As for the specificity of Italian reflexive verbs, it is predicted that this property may hinder L3 learners’ acquisition most since the learners’ mother tongue does not exhibit the property.
2.2 Previous Empirical Studies of Reflexives
This section reviews some previous studies on the acquisition of reflexives. However, due to the lack of empirical studies concerning the acquisition of Italian reflexives, the empirical studies reviewed in this section will be the acquisition of reflexives in different languages. For example, Thomas (1991) examined the interpretation of English reflexive pronouns and Japanese reflexive zibun. Tremblay (2006) investigated the second language acquisition of Spanish reflexive passives and reflexive impersonals by French- and English-speaking learners at an advanced level. Miličević (2007) reported a study of the second language acquisition of Italian reflexive and reciprocal forms by Serbian and English learners. Finally, Montrul et al.
(2010) addressed the issue of the third language acquisition (L3) on object expression in Brazilian Portuguese (BP).
2.2.1 Thomas (1991)
Thomas (1991) addressed the issue of whether L2 learners have access to Universal Grammar (UG) in constructing the grammar of the target language. The main purpose of the study was to investigate the interpretation of anaphors by L2 learners. In the study, she
28
explained the English reflexive pronouns by native Japanese and Spanish speakers and the Japanese reflexive zibun by native English and Chinese speakers.
Two experimental studies were conducted in her study. The first task was an elicited imitation task, which was used to elicit L2 learners to produce pronouns and anaphors in varied syntactic structures. The second task was a multiple choice task that investigated L2 learners’
interpretation of pronouns and anaphors according to their target language.
The participants recruited were 132 English adult learners and 41 Japanese adult learners.
English learners were composed of both Japanese and Spanish native speakers, while the learners of Japanese were native English and Chinese speakers. Except Chinese participants, all participants were divided into three levels, low, mid, high, according to their L2 proficiency tests. Moreover, all participants were studying their respective L2s in L2-speaking environments. In the experiment, the participants were asked to read sentences and responded to the following question by identifying the referent of pronouns or anaphors. Hence, when a subject read a question like ‘Mary heard that Sue told the doctor about herself’, then the participant had to identify the referent of ‘herself’. Thomas categorized the participants’
interpretations into six types, as in Table 2-5.
29
Table 2- 5 Types of English Stimuli and Japanese Stimuli
English Japanese consistently identified the pronouns and anaphors with local antecedents. While in Type E2, L2 learners overwhelmingly chose the short-distance binding antecedents but only few chose the long-distance binding antecedents. With regard to the single-clause Type E3, only few learners of English chose the non-subject antecedents. Hence, in the interpretation of English reflexives study, the L2 learners did not violate the proper antecedent parameter.
With respect to the interpretation to the Japanese reflexive zijbun, low- and mid-level English speaking learners tended to bind the reflexive locally in Type J1 whereas 31% of the high-level learners allowed either local or long-distance binding interpretations. On the contrary, 50% of the Chinese-speaking learners consistently permitted the long-distance binding interpretation for Type J1. Regarding Type J2, majority of the participants bound the Japanese reflexive zibun locally. In this type, there was only little variation among the native
30
speakers and experimental groups. Lastly, all native controls and 88% of the Chinese-speaking learners assigned reflexives to subject antecedents for Type J3. However, only 33% to 54% of the English-speaking learners bound zibun to participants.
Thomas concluded that her results supported the hypothesis that UG is available to L2 learners. In her study, majority of the Japanese learners of English, even at low proficiency levels, tended to bind English reflexives locally. Besides, the English speakers in the Japanese study also captured the governing parameter setting in L2 that was quite different form their Ll, showing that the L2 learners indeed had the ability to reset parameter values unlike their L1 values. In short, the study of Thomas showed that L2 learners had direct access of UG in their second language acquisition of reflexives.
2.2.2 Tremblay (2006)
Tremblay (2006) investigated the second language (L2) acquisition of Spanish reflexive passives and reflexive impersonals by advanced French and English-speaking learners.
Tremblay (2006) investigated the second language (L2) acquisition of Spanish reflexive passives and reflexive impersonals by advanced French and English-speaking learners.