pragmatic or syntactic ziji? evidence from language...

Pragmatic or Syntactic Ziji? Evidence from Language Transfer

Darcy Sperlich

National Kaohsiung University of Applied Sciences, Taiwan

[email protected]

This article aims to understand the place of syntax and pragmatics in regard to Chinese ziji. While there has been continued theoretical debate on this topic, there is little experimental evidence to verify whether syntax or pragmatics plays the defining role in the regulation of ziji. This study achieves this by investigating the interpretation of the long-distance reflexive ziji by English and Korean learners of Chinese. English learners of Chinese are hypothesized to negatively transfer their syntactic anaphoric strategies into their L2 Chinese, while Korean learners of Chinese are hypothesized to positively transfer their pragmatic anaphoric strategies in their L2 Chinese, as Chinese reflexive pronouns are pragmatically regulated. The data gathered includes antecedent judgements and psycholinguistic observations, finding that Korean learners of Chinese are at an advantage over their English counterparts. This is discussed against current syntactic and pragmatic anaphoric theory, arguing that this evidence supports the theory that ziji is pragmatic in nature.

Keywords: Chinese, ziji, transfer, syntax, pragmatics, binding, anaphora

Journal of Cognitive Science 17-4: 607-652, 2016Date submitted: 10/05/16 Date reviewed: 12/17/16 Date confirmed for publication: 12/27/16©2016 Institute for Cognitive Science, Seoul National University

608 Darcy Sperlich

1. Introduction

Modern perspectives on the treatment of anaphora have appealed to multifaceted theoretical approaches due to its complexity across languages. Dominant past paradigms such as Chomsky’s (1981) Binding theory limited anaphoric relations to syntax. This, however, has been supplanted by flexible approaches advocated by Huang (2000) a pragmatist, and Reuland (2011) a syntactician, who both argue that syntax and pragmatics are important components of anaphora. In spite of this, a wealth of studies on reflexive pronouns continue to be conducted through a syntax-only approach which only serves to entrench past notions of syntactic dominance. The present study offers a new direction in the study of reflexive pronouns by examining the syntactic and pragmatic computations involved, the latter being greatly under-represented in the literature. This is achieved by focusing specifically on the Chinese reflexive pronoun ziji, which has been the center of debate between syntacticians and pragmatists.

The theoretical debate on the nature of ziji extends back to the 1980s, where there was a concentrated effort to understand the distribution of ziji through pure-syntactic means (such as Battistella 1989, Tang 1989, Battistella and Xu 1990, C.-T.J. Huang and Tang 1991, Cole and Hermon 1990, Sung and Cole 1991, Cole and Sung 1994, Xue et al. 1995, Cole and Wang 1996 and Cole and Sung 1997). However, these approaches met with difficulty in trying to capture ziji’s distribution syntactically due to ziji’s pragmatic elements. This lead syntacticians to change course and to adopt a dual ziji solution as expounded by C.-T.J. Huang and Liu (2001);1 ziji is an anaphor while locally bound and a logophor while long-distance (LD) bound. While there has been further work on ziji, such as Yu (2000), Liu (2003) and Cole et al. (2006), C.-T.J. Huang and Liu’s (2001) dual approach is the most accepted among syntacticians, with Huang et al. (2009) adopting their theory.2 Huang (1991, 1994, 2000, 2000a, 2004, 2006, 2014) on the other hand, building on the work of Levinson (1987, 1991, 2000) has strongly

1 This is first seen in Liu (1999). 2 The syntactic post-2000 references represent the most up-to-date literature on the syntactic analysis of ziji.

609Pragmatic or Syntactic Ziji? Evidence from Language Transfer

advocated for a pragmatic solution to ziji via his revised neo-Gricean pragmatic theory of anaphora (RNGPTA) given the failures of syntactic theory. In fact, the RNGPTA successfully accounts for ziji’s distribution without making any recourse to syntax. The pragmatic and syntactic debate surrounding ziji will be discussed further on, however it is suffice to say for now that the pragmatic nature of ziji has presented clear difficulties for syntactic approaches.3

Building on this theoretical division, this study aims to discover which theoretical paradigm successfully accounts for ziji, which is achieved by conducting an experiment on the second language acquisition (SLA) of ziji. The motivation for utilizing SLA is that through language transfer, one can experimentally identify the anaphoric strategies present in the mother and target languages. Previous studies on the SLA of ziji have drawn extensively on the syntax-only approach to ziji, especially in terms of its LD binding (Yuan 1997, 1998; Dugarova 2007, 2008; Kong 2009, 2011). An illustrating example is Yuan (1998), who researched English learners of Chinese’s acquisition of ziji; Yuan assumed a syntactic Binding theory approach whereby English learners of Chinese transfer their syntactic rules surrounding English reflexive pronouns (e.g., himself) into Chinese while also assuming that syntax is responsible for ziji’s regulation (argued here to be incorrect theoretically). Thus, SLA transfer in Yuan’s study is syntactically motivated and likewise explained while ignoring pragmatics; such approaches only serve to reinforce the syntactic approach to ziji without giving due consideration to pragmatics.

This study examines the SLA of the Chinese LD reflexive ziji by English and Korean learners of Chinese from a pragmatic viewpoint. This is motivated due to the similarities Korean reflexive pronouns share with Chinese, and differences Chinese and Korean have with English reflexives. Korean has two reflexive pronouns, caki and casin, which share similarities to Chinese ziji (all of which are simplex and can be LD bound). Research trends in Korean, like Chinese, have focused on accounting for caki

3 Alongside syntactic and pragmatic approaches to ziji, there are also semantic theories such as Pan (1997, 2001) Oshima (2007, 2011), Wang and Pan (2014, 2015), among others. The focus of this article however is on the syntactic versus pragmatic solutions to ziji.

610 Darcy Sperlich

and casin’s distribution syntactically (e.g. Cole et al. 1990; Sohng 2004), whereas more recently it has been noted that pragmatics plays an important role (e.g. Kim 1993; I.Kim 2013;4 Lee 2008; Madigan 2015; O’Grady 2013). Therefore, it is to say that there is increasing evidence in the literature that caki/casin are controlled pragmatically, and it will be assumed so in this article (our data further supports this position). English on the other hand only has the complex reflexive pronoun pronoun+self (e.g. himself), bearing no similarities to Chinese ziji.5 Furthermore, its distribution has been successfully captured by syntactic theories (such as the Binding theory, Chomsky 1995) given that himself is usually locally bound. To be clear, rather than focusing on the form of the reflexives as the main point of differentiation, the main thesis advanced is that Chinese and Korean reflexive pronouns operate within a pragmatic system, whilst English reflexive pronouns are regulated under a syntactic system.

The inclusion of these three languages juxtaposes syntactic and pragmatic anaphoric strategies against each other, namely English versus Chinese and Korean. There are two hypotheses; 1) English learners of Chinse will have their English syntactic anaphoric strategies present in their second language (L2) Chinese. This would be problematic for the learners, as it is assumed here that pragmatic plays the decisive role in the regulation of ziji. 2) Korean learners of Chinese will transfer their pragmatic anaphoric regulation strategies into their L2 Chinese. A pragmatic language transfer situation would benefit the Korean learners of Chinese, compared to the syntactic transfer in English learners of Chinese. If this was found to be correct, then this would indicate experimentally that ziji is pragmatic in nature, in contrast to the syntactic thoughts about ziji offered in the literature. The article is structured as follows: Section 2 investigates the empirical facts surrounding reflexives in these three languages and their theoretical treatment; Section 3 introduces the methodology of the experiment; Section 4 analyzes the results; Section 5 presents the discussion

4 I thank the reviewer for bringing this article to my attention.5 Chinese also has a complex reflexive taziji, but it is not focused on here as it is typically not LD bound. Moreover, other effects associated with LD binding (such as the blocking effect) are not the focus of this article.


and finally Section 6 concludes the article.

2. Review of reflexive pronouns

This section discusses what long-distance (LD) binding is, how this is realized in the languages studied here, and finally appraises the theoretical and SLA approaches to LD reflexives.

2.1 What are long-distance reflexives?

Reflexives can be classified into two types, polymorphemic (e.g. himself) and monomorphemic (e.g. ziji). The former (a complex reflexive) is usually locally bound and the latter (also called a simplex reflexive) has the ability to be LD bound. In morphosyntactic theories of binding (e.g. Reuland 2011) the phi-feature specification of reflexives is crucial to LD binding—complex reflexives are fully specified while simplex reflexives are underspecified, thus allowing the latter to be LD bound. Other than phi-features, logophoricity is another key factor in LD binding (Huang 2000). Logophoricity is the phenomena whereby the voice of the sentence/discourse internal speaker is represented using a logophoric pronoun (Clements 1975), found in some African languages where there is a specific logophoric pronoun to perform this function (e.g. Ewe; Pearson 2015). Huang (2000) theorized that LD binding in East Asian languages is achieved by reflexive pronouns being able to be used logophorically within a logophoric environment, as in (1) below. According to Huang (2000), there is a preference to bind long-distantly if there is a logophoric trigger in a sentence; for example in (1) the verb shuo ‘say’ introduces a logophoric predicate where the viewpoint of Lao Chen is being reported. Furthermore, our stereotypical beliefs of how a drunken person might behave reinforce the pragmatic bias on Lao Chen.

(1) Hezui de Lao Chen1 shuo Lao Wang2 burang ziji1/*2 kai che. Drunk POS Lao Chen say Lao Wang NEG-PERM self drive car ‘Drunken Lao Chen1 said that Lao Wang2 would not let self1/*2 drive the car.’

612 Darcy Sperlich

With the basic overview of LD reflexives completed, the focus shifts to the status of LD reflexives pronouns in Chinese, Korean and English.6

2.2 Chinese, Korean, and English Long-distance Reflexive Pronouns

The main simplex reflexive of Chinese is ziji. It is monomorphemic and deficient in some phi-features, characterized by being able to be bound outside a local clause, as in (2):

(2) Lao Chen1 shuo Lao Wang2 buxihuan ziji1/2. Lao Chen say Lao Wang NEG-like self ‘Lao Chen1 said that Lao Wang2 doesn’t like self1/2.’

As the sentence is referentially ambiguous, ziji can be bound by either antecedent. Moreover, ziji can be used logophorically as described above.7

Examining Korean, it has two simplex reflexives caki and casin which can be LD bound (Kim 1993), in a similar fashion to ziji. The key difference between caki and casin is that caki is most naturally bound by 3rd person NPs, while casin has no such restrictions (J.-H. Kim & Yoon 2009). Casin is a monomorphemic reflexive (as is caki), devoid of some phi-features as it is able to take the plural suffix –tul (Lee 2008). Huang (2002) noted the use of caki as a logophor, however he did not include casin in the analysis. In (3), which has an epistemic predicate, it would appear that casin is behaving like a logophor:

(3) Mary1ka Susan2i casin1/2ul wenmanghantako Mary-NOM Susan-NOM self-ACC blame-PRS-DELC-COMP sayngkakhanta. think-PRS-DELC ‘Mary1 thinks that Susan2 blames self1/2.’

(Lee 2008:99)

6 English has no LD reflexive pronoun as such. Furthermore, we do not investigate polymorphemic reflexives as they usually do not exhibit LD binding behavior.7 Cf. Huang (1994) and Yu (2000) for a full discussion on ziji.


English, in contrast, has no simplex reflexive (apart from self appearing in headlines), only having a polymorphemic reflexive, pronoun+self (e.g. himself), which is fully specified for phi-features. It is well known that syntax plays a strong local role in the regulation of himself unless placed in a logophoric environment, occupying a non-argument position (see Baker 1995 for an in-depth discussion).

In sum, Korean and Chinese simplex ref lexives are similar morphosyntactically and in binding patterns, while the English monomorphemic reflexive is very different.

2.3 Syntactic and Pragmatic Theoretical Perspectives

The theoretical viewpoint adopted in this article is that the distribution of anaphora can be best explained by a combination of syntax and pragmatics, as argued by Huang (1991, 1994, 2000, 2000a, 2004, 2006, 2014). This viewpoint has been driven by the difficulties understanding anaphora through a syntax-only perspective, particularly with reference to Chinese ziji. This is best demonstrated through the well-known Binding theory, which aims to capture the distribution of anaphora through three principles, the pertinent one here being Condition A in which anaphors (reflexive pronouns) must be bound locally. Exceptions arose (e.g. Culicover & Jackendoff 1995), especially regarding simplex anaphors (like ziji) which proved to be the greatest challenge given their binding across finite clauses, which is not predicted by the Binding theory. Using Chinese as a case example, we will see how generative syntactic theory has evolved but still faces difficulties with the pragmatic elements of ziji.

The syntactic solution (in the universal grammar (UG) generative paradigm) to ziji had been investigated by many scholars, such as Battistella (1989), Tang (1989), Battistella and Xu (1990), C.-T.J. Huang and Tang (1991), Cole and Hermon (1990), Sung and Cole (1991), Cole and Sung (1994), Xue et al. (1995) Cole and Wang (1996), Cole and Sung (1997), Yu (2000), Liu (2003) and Cole et al. (2006) with the main approaches being the Head (Cole and associates) versus Adjunction analysis (C.-T.J. Huang and associates). They both aimed to account for ziji’s LD binding through chain movement in conjunction with ziji’s lack of phi-features. However,

614 Darcy Sperlich

these theories have been refuted both on empirical and theory internal grounds (Huang 1994, 2000), and in fact Cole and Sung (1997) made specific reference to the pragmatic use of ziji that pure syntactic theories need to take into account.

In response to the difficulties a pure syntactic approach poses for ziji, syntacticians focused on Chinese have opted for a dual syntactic-pragmatic ziji solution (C.-T.J. Huang and Liu 2001). That is, ziji is an anaphor while locally bound, and a logophor while LD bound. Huang (2000:173) however rejects this analysis as it is essentially circular in nature, if it is not X then it must be Y and vice versa. C.-T.J. Huang and Liu (2001) differentiate the two by adopting the governing category as the dividing line, but such machinery (among others) is not available and cannot be reinterpreted in modern syntactic approaches, to be discussed.

The basic motivation of Huang’s (2000) RNGPTA is accounting for the distribution of anaphora through the competition of forms a position might take. For example, a single position in Chinese might take a null element, pronominal, noun phrase, complex reflexive or simplex reflexive shown in (4) below—all of which is acceptable syntactically, however pragmatically there is a difference in the message expressed through the relevant choice.

(4) Xiaoming shuo Ø/ta/ziji/taziji/Xiaoming zui xihuan gudian yinyue. Xiaoming say null/3SG/self/3SGself/Xiaoming most like classical music ‘Xiaoming says that Ø/he/self/heself/Xiaoming likes classical music most.’

(Huang 2000:257)

The exact mechanism behind the RNGPTA will not be covered here in detail, instead an overview will be given with explicit application to Chinese ziji.8 Considering (3), there are two competing antecedents for ziji, Lao Chen and Lao Wang. The key question is what allows ziji to be LD bound by Lao Chen? Huang theorizes that ziji can be used logophorically, noting that ziji is in competition with ta ‘he/she’. Both have the same ± coreferential feature, but ziji has an extra ± logophoric feature. There is also a logophoric trigger shou ‘say’, providing the necessary environment for ziji

8 Cf. Huang (2004).


to be LD bound, assuming the speaker intends for the message to include the logophoric element (otherwise ta would be used).9 Thus, there are no syntactic mechanisms needed to explain the binding of ziji.

In recent work concerning syntactic approaches, Binding theory has been reformulated into interactions between grammatical primitives rather than stipulated theoretical modules, as seen in Reuland (2011), Heinat (2008), Hicks (2009), Rooryck and Wyngaerd (2011) and Safir (2004, 2005). The common theme through these works is to rework the Binding principles into general grammatical interactions pertinent to anaphora in a language (e.g. Case and phi-features; see Safir 2013 for a review). Moreover, allowances for pragmatics are apparent in Reuland (2011), who theorizes that where syntax is not involved with anaphoric binding, discourse becomes the pervading factor. In a nutshell, if a reflexive pronoun cannot enter syntactic operations, then instead of trying to force the anaphor to fit the syntax it is more economical to appeal to pragmatics. While these newer theories have yet to be applied to ziji, the important message is that syntacticians are now open to pragmatic control over reflexives.

In sum, there are two competing Chinese-specific approaches to binding, C.T-J. Huang and Liu’s (2001) dual syntactic/pragmatic solution (based upon outdated theory) and a pure pragmatic solution (Huang 2000).10 Thus, in light of the role pragmatics plays in the regulation of ziji, SLA researchers investigating ziji need to take note of advances in theory, which is discussed in the next section.

2.4 Second Language Acquisition of Ziji

The current study radically differs to previous SLA studies on anaphora as the position adopted is that anaphora can be understood through syntax

9 This process occurs under the neo-Gricean constraints of the Quantity, Informativeness and Manner principles.10 The situation for Korean is similar to Chinese, with cues taken from Chinese anaphoric research (e.g. Sohng 2004), where generative approaches are dominant (see Madigan 2015 for an up-to-date review). English however, fits comfortably within a pure syntactic approach.

616 Darcy Sperlich

and pragmatics (rather than just syntax), language dependent. As this study investigates Chinese ziji, in light of current theoretical developments it is important to reassess past SLA studies on ziji which assume a syntactic orientation of ziji, and reanalyze their conclusions from a pragmatic perspective.

Studies on the SLA of ziji that assume syntactic regulation of ziji include Yuan (1997, 1998), Dugarova (2007, 2008) and Kong (2009, 2011).11 These studies investigate English learners of Chinese’s acquisition of ziji through UG (e.g. Schwartz and Sprouse’s 1996 FT/FA model). The key testing component is investigating binding in biased versus neutral sentences (which are usually biclausal with one antecedent per clause as in (2) above). Importantly, the biased sentences are designed so that there is linguistic pressure on the participant to select the LD antecedent, while in neutral sentences either antecedent are equally possible. Native Chinese speakers follow bias in the biased sentences and allow ambiguous binding in neutral sentences (as shown in Huang 2000), however learners whose first language’s (L1) anaphoric system is syntactically oriented struggle with this.

Yuan (1997, 1998) investigates the acquisition of ziji by English and Japanese learners of Chinese, examining if Japanese acquire ziji more easily than English speakers, due to Japanese zibun (‘self’) and ziji both being Xo anaphors, contrasted to English which only has Xmax anaphors. Yuan found the Japanese transfer their L1 knowledge into their L2 Chinese successfully, however the English learners must acquire the LD properties of ziji while dealing with the L1 interference of a local binder. Yuan’s approach is accounted for in purely UG terms (making recourse to syntactic mechanisms in transfer) without considering pragmatic effects. An alternative explanation is that the Japanese transfer their L1 pragmatic strategies into their L2 Chinese as Japanese reflexive pronouns appears to be also pragmatically regulated like Chinese and Korean (see Iida 1996). The English speakers on the other hand need to acquire the pragmatic aspects surrounding ziji and not syntactic ones as assumed by Yuan.

Dugarova (2007, 2008) emulates Yuan’s work, assuming that ziji is

11 To date, there are no Korean learner of Chinese studies.


syntactically regulated and is likewise acquired. However unlike Yuan’s participants, Dugarova’s English learners of Chinese acquire ziji better, as Dugarova’s beginner group appears to LD bind more than the advanced group, regardless of neutral or biased conditions. However, the advanced group does choose the LD antecedent more (by itself without concurrently selecting the local antecedent) in biased contexts. Thus, the advanced group is more responsive to bias, showing that they have acquired ziji’s pragmatic conditions, unlike the beginners. The intermediate group does not show specific progression as their binding patterns cannot be clearly distinguished from the beginner group. All groups have a high degree of local binding.12 In sum, the English learners suffer from L1 syntactic transfer and find it difficult to acquire the pragmatics of ziji. However, these results may not be entirely reflective of the learner’s grammar due to the multiple choice methodology used, as it is known to tap learner preference rather than their full grammar.13

Finally, Kong (2009, 2011) finds that English learners of Chinese select for LD ziji at the beginning stages, which improves as proficiency increases. Kong hypothesizes that the low proficiency learners interpret ziji as a pronoun, and the higher proficiency learners interpret ziji as an anaphor. However, the issue with these results is that Kong mixed the neutral and biased sentences together in the analysis, which does not allow clear conclusions to be made. In other words, with the biased and neutral stimuli being analyzed as a single type of result, this cannot be considered a reliable analysis of the data. Moreover, Yuan (1998) considered the interpretation of ziji as a pronoun, but rejected this because ziji was locally bound, contradicting how a pronoun is used.

2.5 Summary

The SLA theme is that negative syntactic language transfer inhibits the

12 As Dugarova adopts similar explanations as Yuan, our comments on Yuan apply here also.13 Methodological issues in SLA anaphoric testing can be read in Demirci (1997), White et al. (1997) and Sperlich (2015).

618 Darcy Sperlich

acquisition of ziji by English learners of Chinese. Beginners overall prefer local binding due to local syntactic binding transfer from their L1, which gradually diminishes as proficiency improves (as they slowly acquire Chinese pragmatic binding strategies), allowing more LD binding. However, a prevailing issue is that the methodology of these studies did not elicit fully valid data, resulting in a less than full picture of their binding ability. Finally, these studies only appeal to a syntactic ziji, which is a clear weakness in light of the theoretical review. The theoretical review undertaken by this study motivates an investigation into ziji through a pragmatic approach rather than a syntactic one, allowing reassessment of this misunderstood problem.

3. Methodology

An experimental approach is adopted to understand language transfer patterns in the learners’ interlanguage. Firstly, as noted previously, SLA anaphoric studies have suffered from validity problems as their methodologies tap learner preference rather than their full interlanguage grammar. Sperlich (2015) addressed these problems in detail developing a new methodology to elicit anaphoric judgements which are highly representative of a learner’s interlanguage.14 This methodology is adopted here and its use is discussed below.

Beginning with the test items’ structure, the sentences are biclausal (to test one LD binder) and triclausal types (to test two different LD binders), which are either neutral or biased towards certain antecedents. Biased sentences encourage binding by the LD antecedent, while in the neutral sentences there are no such restrictions. Key is the neutral versus biased dichotomy to judge whether binding is syntactic or pragmatic, which will be illustrated when discussing the actual items. The focus is on intersentential relations, which is important to note as syntax is usually considered to regulate these relations, not pragmatics (which has been more focused on intrasentential relations, with the RNGPTA being an exception). The reflexive pronoun tested occurs after all antecedents, termed forward reflexivization. The

14 For reasons of space this cannot be fully discussed here.


embedded clauses are either finite or nonfinite (but only finite in triclausal sentences). This dichotomy, however, applies only to English; even though the label is used to describe Chinese and Korean sentences, this does not imply that the distinction is made in these languages.15

In total there are 11 different sentence types tested with 2 representative tokens each, for a total of 22 target items assessed. Equal amounts of distracter items are included (22), therefore in total 44 different sentences were judged. The biclausal sentences are named as: N-F/NF (No Pragmatic and Semantic Bias-Finite/nonfinite), P-F/NF (Pragmatic Bias-Matrix subject-Finite/nonfinite) and PS-F/NF (Pragmatic and Semantic Bias-Matrix subject-finite/nonfinite). Triclausal sentences are: T-N (No Pragmatic and Semantic Bias-Finite), P-LD1 (Pragmatic Bias-Matrix subject-Finite), P-LD2 (Pragmatic and Semantic Bias-1st Embedded subject-Finite), PS-LD1 (Pragmatic and Semantic Bias-Matrix subject-Finite) and PS-LD2 (Pragmatic and Semantic Bias-1st Embedded subject-Finite). Below are examples of two sentence types, Appendix A containing all the test sentences.

(P-F/NF) Biclausal, Pragmatic Bias to the Matrix Subject, Finite/nonfinite. In this sentence type, there are pragmatic factors which help push the antecedence to the matrix subject. In example (5) below it can be reasonably assumed that the apple is given to Laoshi (teacher), as a stereotypical ‘good’ student would do.

(5) Laoshi xiangxin xuesheng mei yi tang ke dou hui gei ziji yi ke pingguo. teacher believe student every one CLS class all will give self one CLS apple ‘The teacher believes that the student will give self an apple in every class.’

15 Whether Chinese exhibits a finiteness distinction is debated in the literature, with various linguists such as C.-T.J. Huang (1982) and Tang (2000) striving to find the finite/nonfinite distinction in Chinese. Nevertheless, Hu et al. (2001) provided a strong argument concluding that one cannot truly distinguish between the two clause types. Similarly regarding Korean, Lee (2009) found similar problems, arguing for a role of Mood and Modality in its licensing. Hence, in light of future research the position taken here for Korean is as the position taken on Chinese.

620 Darcy Sperlich

Thus, in these types of sentences it is expected that Laoshi will be selected over xuesheng (student) as pragmatics is a strong force in Chinese anaphoric regulation. This should be observed with the Korean learners of Chinese given similar pragmatic processes in their L1; however for English learners of Chinese, due to local binding syntactic transfer from their L1, they should select xuesheng.

(T-N) Triclausal, No Semantic or Pragmatic Bias, Finite. This sentence is ‘neutral’, i.e. without any pragmatic or semantic pressure on the binding of ziji in (6):

(6) Lao Wang tingshuo Lao Chen renwei Lao Li shanghaile ziji. Lao Wang hear Lao Chen think Lao Li hurt-ASP self ‘Lao Wang heard that Lau Chen thinks that Lao Li had hurt himself.’

In the neutral sentences, without pragmatic highlighting of a preferred antecedent it is predicted that no one antecedent will clearly dominate the binding of ziji. In Korean learners of Chinese this is predicted to hold (due to L1 positive transfer), but again with English learners of Chinese negative L1 transfer will cause them to bind locally, in-line with English syntactic anaphoric strategies.

Moving onto the judgement data elicitation methodology,16 using (6) above as the guiding example, the participant reads the stimuli and then he/she is asked to read the accompanying question below (7) (without comparing it to another question next to it), i.e. that ziji is bound by Lao Wang or not:

(7) Ziji = Lao Wang Y/N

Subsequently, the participant assesses the same stimulus later on inquiring about the other antecedent(s) in the sentence. Thus, the combination of stimuli and questioning methodology form the interpretative judgement test. Additionally, this study makes use of a second test, the truth-

16 The test is fully computerized.


value judgement test (again a combination of sentence and questioning methodology), but the questions are presented differently in (8):

(8) Lao Li shanghaile Lao Wang Y/N

In addition to antecedent judgement data, three psycholinguistic measures are used to amplify these judgements. These are reaction timing, confidence scales, and knowledge source attribution. These measures used alongside traditional antecedent judgement data provide a processing perspective on what lies behind the judgement. That is, these measures can show if an increased processing load occurs when judging something that might prove difficult, and vice versa. For example, if an English learner of Chinese is confronted with a sentence biased to an LD antecedent, and they are under syntactic L1 influence, this should cause processing difficulties as their L1 anaphoric strategies are incompatible with the target language. Thus, not only do we have judgement data to understand the status of their interlanguage, but also a psycholinguistic processing perspective as well.

The first measurement is reaction timing, recorded from when the stimulus first appears, to when the participant makes a Yes/No judgement. How timing correlates to judgements could show that quick timing represents a smooth decision made, while longer timing could show more processing involved. After the participant has made a Yes/No judgement, he/she selects his/her confidence from a scale below in (9):17

(9) 4 – Certain, you have no doubts that you are right3 – Quite confident2 – More or less guessing1 – Guess (i.e you could have flipped a coin, 50/50 chance of

getting it right)

If the confidence level selected is high, then this can show that the judgement is made without difficulty. The final technique is concerned with

17 This scale is based on Dienes (2008). All scales are provided in the participant’s mother tongue.

622 Darcy Sperlich

eliciting the participant’s belief on what he/she bases his/her judgement knowledge on. The scheme used to capture judgement knowledge is termed ‘source attribution categories’ (referred to here as ‘knowledge source’), introduced by Dienes and Scott (2005) in (10):

(10) G = Guess – you guessed the answer, just like flipping a coin. I = Intuition – you don’t know why you are right, it’s a ‘feeling’ that you have.R = Rule – you have learnt the rule before and you are knowingly applying it to the sentence, and you can say what the rule is.

This reporting mechanism is useful as it provides an avenue for learners to indicate their beliefs on how they made their judgements, which in turn helps the researcher draw conclusions on how the judgement was made. In sum, these measures when combined together provide a comprehensive picture of the participant’s anaphoric binding strategies.

Three different types of randomizations were used insuring that each participant received a unique test script. The first type is intersentential, where a question about a possible antecedent randomly appears. The second type is intrasentential, where the stimulus randomly appears. This was constrained following Cowart’s (1997) suggestion, block randomization. If participants forget which category meant what, every screen with the questions shows the definitions for the confidence and knowledge source scales. Regarding the learners, a vocabulary list is attached to each question to counteract any possible unknown vocabulary, as Birdsong (1989:86) indicated, “learners in judgment elicitation experiments are often confronted with unfamiliar sentence types and lexical items”.

This study targets the language learners’ implicit knowledge, as Ellis (2009:13) stated that “default L2 production relies on implicit knowledge”. However, it is not a clear-cut task in discerning whether a participant accessed their implicit or explicit knowledge in their judgement. Hence, to assist in the elicitation of implicit knowledge this study draws upon criteria set out by Ellis (2009a). Firstly, the focus is on meaning (judging anaphoric antecedence) which encourages implicit judgements; secondly, the participant’s awareness of the task is reduced by giving minimal


information about the experiment and including the use of distracter items; thirdly, the participant is encouraged to use ‘feel’ when answering; finally, the participant is encouraged to use their first impression, supported by time pressure given by setting a time limit (the maximum time is suggested), 45 minutes for the learners and 20 minutes for the native speakers. While it cannot be guaranteed that implicit knowledge has been tapped, the controls introduced here make it more likely than not.18

In the actual testing procedure, participants were placed in a lab/private room, and using a computer they first filled basic information about themselves, such as gender and age. After reading the instructions about the test, participants were presented with randomized stimuli and the accompanying question along with it. They first judged antecedence, and then selected the psycholinguistic measures. This continued until all questions were answered.

This study encompassed eight different groups totaling 100 participants, which were a mix of native control and learner experimental groups.19 There were four learner groups, split between English and Korean speakers, who were international students studying at a university in China (grouped according to proficiency obtained from their class levels). The English learners of Chinese group had intermediate (ELCi) and advanced (ELCa) participants, the intermediate group containing two males (from the USA) and one female (from New Zealand), their ages ranging between 17-21. The advanced group had two females (20 and 33), both from the USA. The Korean learners of Chinese group was split into an intermediate (KLCi) and an advanced group (KLCa). The intermediate group had two females, aged 20 and 21. The advanced group had five males and one female, their ages ranging from 22-28. Both groups were from South Korea.20

Native speaker groups were included to discover how they interpreted

18 Previous studies on the SLA of anaphora have paid little to no attention to the type of knowledge elicited.19 There was a near-native speaker of Chinese group that was part of the overall study, but their data analysis is not included in this article.20 While the number of participants is not large, the results found are significant, as discussed in the next section.

624 Darcy Sperlich

the experimental stimuli in their own language.21 This was done to verify the binding patterns described in the literature. Moreover, the L1 data provides useful information about theorizing what has been transferred in the learner groups. There were three native speaker control groups, Chinese native speakers (CNS), having 6 male and 19 female participants, whose ages ranged from 17-60; Korean native speakers (KNS) made up of 8 male and 11 female participants, whose ages ranged from 18-55; English native speakers (ENS) having 16 males and 20 females, whose ages ranged from 25-49.

4. Results

The analysis focuses on finding evidence of syntactic transfer from the English learners of Chinese, and pragmatic transfer from Korean learners of Chinese. To achieve this, the antecedent judgement data analyzed per group is paired against the psycholinguistic data gathered. Once transfer patterns have been established, cross linguistic comparisons are made between the two learner groups to highlight key differences that support the syntactic versus pragmatic transfer argument. Tables B.1-B.7 in Appendix B detail how each group judged each sentence type, including their psycholinguistic measures (example data are given in this section).Tables 1 and 2 below show how all the groups bound in the target sentences, R1 being the LD binder in both biclausal and triclausal sentences, while R2 is a local binder in the biclausal sentence; R2 is a LD binder in the triclausal sentence while R3 is the local binder.

Beginning the analysis with the antecedent judgement data, a 2-sided, 2x2 Fisher’s exact test shows that there are significant differences between LD and local binding in combined biased versus neutral conditions for advanced English learners of Chinese (ELCa) (p < .01), intermediate Korean learners of Chinese (KLCi) (p = .02) and advanced Korean learners of Chinese (KLCa) ( < .01); but not for intermediate English learners of Chinese (ELCi). Thus, there is a clear difference between the binding

21 In English himself was used, while in Korean casin was used as it shares more similarities with ziji than caki does, as discussed in Section 2.

626 Darcy Sperlich

patterns of the ELC and KLC groups. The KLC groups clearly follow the pragmatic bias when present (e.g. KLCi follow bias 100% of the time in P-F), and are split between the antecedents in the neutral conditions. ELCa also follow bias to a good degree (e.g. ELCa follow bias 83% of the time in PS-F), but there is still local binding intermingled, while in the neutral sentences they default to local binding. ELCi on the other hand do not treat biased or neutral sentences differently; they bind locally when they do bind, but they have difficulties in selecting antecedents overall (e.g. in P-LD1 the rejection rates for the antecedents range between 80-90%).

Moving onto the confidence measures, a one-way ANOVA shows there are significant differences between the combined neutral and biased conditions [F(7,511) = 25.119, p < .001]. Conducting a post-hoc Tukey test shows significant differences between the ELC versus KLC groups in both biased and neutral conditions. Thus, the ELC groups have lower confidence than the KLC groups overall (e.g. N-F R2 confidence for ELCi/a, KLCi/a are 3, 3.2, 3.33 and 3.74 respectively). Moreover, the confidence levels between each learner level in English and in Korean do not differ significantly, showing that language proficiency within groups is not a factor here.

The next measure is reaction timing, a one-way ANOVA test reveals in the neutral and biased conditions significant differences between the groups [F(7,511) = 4.532, p < .001]. A Tukey post-hoc places these differences in the biased conditions between ELCi versus ELCa, and ELCa versus KLCa. Thus in the neutral conditions, the reaction time differences between the groups are not significant, which are at similar durations. However in the biased sentences ELCa perform quicker than ELCi, but much slower than KLCa.

Finally, in the knowledge source measures the proportion of Guess used in all groups is of the most interest here. Comparing the overall proportion shows ELCi are significantly different to ELCa (Z = 2.5709, p = .01), KLCi (Z = 7.147, p < .01) and KLCa (Z = 9.1527, p < .01). ELCa is significantly different from KLCi (Z = 4.5871, p < 0.01) and KLCa (Z = 4.7382, p < .01). The KLC groups do not differ from one another. Therefore, ELCi use Guess the most (31% of the time in Y answered in the biased sentences), ELCa have a much lower proportion (14% in Y answers for biased sentences), and


there is minimal use in KLCi/a (4% and 7% respectively in the Y answers for biased sentences).

In sum, the general picture established is that ELCi have trouble with binding, seen also in their poor timing, low confidence and high use of Guess. ELCa on the other hand have advanced over ELCi by responding more to pragmatic bias, albeit still with slow timing, and lower confidence than the KLC groups but with less use of Guess. The KLC groups overall are in an advantageous position performing well in their binding, also seen through strong performance in the psycholinguistic measures. We now move onto a more detailed analysis of the groups.

4.1 English Learners of Chinese, Intermediate and Advanced

Beginning with the neutral sentences (N-F/NF, as seen in tables B.1 and B.2), the ELC groups bind locally as expected. In other words, syntactic transfer is clearly evident in these neutral sentences given the local binding seen. An example of a neutral sentence is as below (11):

(11) Lao Wang renwei Lao Chen juede ziji hen liaobuqi. Lao Wang believe Lao Chen think self very great ‘Lao Wang knows that Lao Chen thinks highly of himself.’

Here, the ELC groups believe Lao Chen to be ‘great’, even though there is a clear allowance for Lao Wang to be considered ‘great’ as well. Both groups are confident, and in terms of timing, it is interesting to note that it takes longer for the groups to accept antecedents rather than to reject them. ELCi generally have difficulty in assessing antecedence given their poor timing; however ELCa’s slow timing (who are at a higher proficiency level) may be explained by competing L1 syntactic versus L2 pragmatic binding strategies present.

Turning to the biased sentences (P-F/NF, PS-F/NF), the patterns of binding for both groups are now dissimilar. That is, ELCi generally prefer to reject the LD antecedent than accept it, but they are receptive to the bias given the high rejection rate of the local antecedent. ELCa on the other hand perform well in selecting for the LD antecedent. For example, in P-F

628 Darcy Sperlich

they confidently select the LD antecedent with an overall 3 while an overall 1.5 is for accepting the local antecedent. Timing-wise, it is important to note that generally LD cases take longer to process in comparison to the neutral sentences. Overall, this shows the increased processing load incurred by both groups when considering pragmatic information, especially seen in ELCa who take longer than ELCi to arrive at their binding choices.

Moving to the triclausal sentences, beginning with T-N local binding is preferred by both groups. LD binding can be seen in ELCi, however incurring low confidence and more Guess use (indicating that triclausal sentences are difficult to process), which is not seen in ELCa.

The next is P/PS-LD1, ELCa select the LD antecedent to a high degree, quickly and confidently. ELCi on the other hand have a preference not to select antecedents (the majority answers are No), showing the difficulty of antecedent processing despite clear pragmatic bias. Finally looking at the P/PS-LD2 sentences, ELCa follow bias almost perfectly with the quickest response time to the second LD antecedent. ELCi again prefer to accept no antecedent, along with low confidence and a selection of Guess across the board.

To summarize these findings, ELCa have made good progress over ELCi in regard to LD binding. This can be seen by ELCi generally rejecting possible bindees, along with higher uncertainty and more guessing. ELCa and ELCi are only similar in that local binding is strong in the neutral sentences—clear transfer from English. Therefore, it can be surmised that ELCi have not yet obtained the pragmatic competence that ELCa displays (being closer to Chinese native speakers (CNS)).

4.2 Korean Learners of Chinese, Intermediate and Advanced

In the first sentence type N-F (seen in tables B.3 and B.4), both groups prefer the local antecedent. In N-NF, both groups prefer the LD antecedent. Put together, there is a rough 50/50 distribution between both antecedents. This implies that the majority of responses can be classified as ‘either or’, due to the neutral sentences’ ambiguous nature. Examining the psycholinguistic measures shows quick timing and high confidence (use of Guess is low) in their choices pointing to no processing difficulties. Overall,


selection of either antecedent is generally unconstrained.Turning to P-F/NF, both groups follow the bias well, albeit with KLCi

allowing some local binding, but with lower confidence. KLCa do an even better job at responding to bias while being highly confident. The patterns seen here are also mirrored by PS-F/NF. Both KLCi/a are highly confident, quick and Guess use is minimal.

Moving now to T-N, as in the biclausal neutral sentences there is no clear preference for any of the antecedents, and the psycholinguistic measures do not suggest otherwise. Turning to P/PS-LD1 the KLC groups follow the bias very well, being very confident and quick (as in the biclausal sentences). In P/PS-LD2 they follow the bias to a similar high degree with high confidence, quick timing and low Guess use.

To summarize, both KLCi and KLCa are highly sensitive to bias (reflected in the psycholinguistic measures), while accepting varied antecedents in the neutral sentences. Hence, the KLC groups have little difficulty interpreting Chinese ziji, due to pragmatic language transfer, and thus are closer to CNS.

4.3 Control Group Results

In order to support the theorized transfer occurring in the learner groups, the native speaker groups are investigated. Firstly, English native speakers (ENS) clearly prefer local binding in all sentence types. Even though they show a little inclination to LD bind, it can be inferred that syntactic anaphoric strategies are used, and in turn they are resistant to pragmatic bias. As discussed in Sperlich (2015), it is possible for ENS to LD bind from a non-argument position. In fact, there are a few tokens that place himself in a non-argument position, and ignoring these there is 45% LD binding in P-F (token 2) and 28% in PS-F (both tokens) from an argument position, which cannot be ignored. Moreover, confidence generally drops, timing slows and more Guess use is prevalent when selecting for LD antecedents compared to local antecedents.22 Chinese native speakers (CNS) and Korean native speakers (KNS) on the other hand show strong similarities to the

22 The fact that ENS can respond to pragmatic bias is not pursued here but can be read about in Sperlich (2013).

630 Darcy Sperlich

KLC groups because all groups follow bias to the LD antecedent without a problem, and do not have strong preferences in the neutral sentences. Moreover, the psycholinguistic measures do not indicate that there are any processing difficulties, as compared to ENS and ELC. In sum, the results obtained from these L1s clearly show the origins of language transfer observed in the learner groups and the pragmatic strategies needed for Chinese.

4.4 Cross Group Comparisons and Overall Summary

Comparing ELCi/a to KLCi/a allows several conclusions to be made regarding their binding differences through their antecedent, timing, confidence and knowledge source selection. Firstly, ELCi/a both have a tendency to bind locally in neutral sentences, overlooking the LD antecedent. KLCi/a on the other hand do not clearly prefer the local antecedent. ELCi/a differ in the sense that ELCa have progressed over ELCi in allowing LD binding when LD bias is present;23 KLCi/a exhibit no major differences in their LD binding behavior as they both readily follow LD bias. Therefore, the KLC groups are at a clear advantage over the ELC groups.

Confidence measures are extremely useful in corroborating the evidence. ELCi’s binding struggle is demonstrated by low confidence levels, which is not seen in ELCa or the KLC groups (who have the highest confidence). Moreover, while ELCa perform closer to the KLC groups in antecedent selection, their judgements are not very confident (compared to KLCi/a) due to residue syntactic transfer.

Moving onto knowledge source, the most useful measurement is participants’ use of Guess. It is clearly prevalent in ELCi, given their problems with interpretation (ELCa have a lesser use of Guess compared to ELCi but have more use than the KLC groups). Using Guess is also seen

23 It should be pointed out the ELCi do a worse job than ELCa in LD binding, even though the evidence seems to suggest that they allow LD binding in neutral sentences, when ELCa do not. If this were the case, then one would expect more LD binding in the LD biased sentences, which is not the case (and thus they do not treat ziji like a pronoun).


in areas where the antecedent is not clear (e.g. in neutral sentences for non-English speakers).24

Finally considering reaction timing, ELCa who bind closer to the KLC groups show that in order to reach KLC’s level they must expend more resources, slowing their processing. Finding a pattern for the fastest time is difficult to establish, as it is shared between ELCi and KLCa. However, these two groups are certainly not equivalent, but rather polar opposites of each other. Therefore, their quick responses are symptomatic of two different processes as evidenced by their Yes/No judgements, confidence and knowledge source selections. ELCi suffer from low confidence levels, use Guess often and have trouble finding antecedents, hence they judge antecedence quickly based on shallow considerations—KLCa work vice versa. Quick timing therefore represents these two realities, a group with low performance (difficultly with binding) to a group with high performance (little difficulty with binding), which is supported by the other psycholinguistic measures.

Summing up, results from both learner groups and native speaker groups support the argument regarding a syntactic versus pragmatic transfer paradigm.

5. Discussion

The aim of this section is two-fold, first to compare these results to the literature in order to support the pragmatic perspective about ziji, and second to discuss the implications of the data found here in terms of understanding the place of pragmatics and syntax in anaphora.

24 Commenting on Intuition and Rule use, these represent something different as there is a high use of Rule in the Korean groups when compared to the English groups who prefer Intuition. The use of Rule would imply that the participants are using explicit knowledge to answer their questions, while accessing implicit knowledge with Intuition. This however must be interpreted with caution, as it might be a cultural artefact, i.e. the ELC groups prefer to express confident judgements via Intuition and KLC by Rule. This then might point to how they interpret Intuition and Rule on a basic level, for if you intuitively know something is correct, choosing either category may express this. Moreover, the same pattern of Rule and Intuition use is seen in the native speaker groups.

632 Darcy Sperlich

Firstly, the observations united from the biclausal and triclausal sentences for ELC and KLC yield clear patterns of acquisition. Regarding the ELC groups, ELCa are at a clear advantage over ELCi as they have progressed well in acquiring the binding rules surrounding ziji. Conversely ELCi are either stuck in local binding, or have difficultly processing antecedence, seen through their low confidence and spread of Guess. ELCa perform close to par with the KLC groups in following bias, however they take longer to process these judgements with lower confidence. The KLC groups are at a clear advantage over their ELC counterparts as they readily follow bias, and process these judgements quickly with high confidence. The learners’ patterns of binding are clearly products of transfer as seen in how ENS and KNS bind, compared to CNS. ENS favor local binding, and when they do respond to bias processing time increases, confidence drops, and an increase of Guess occurs.25 This is unmistakably evident in ELCi and to a lesser extent in ELCa. KNS respond well to bias, with quick processing and high confidence; the KLC groups share similar binding and psycholinguistic trends. Thus overall, KLC, KNS and CNS are more alike than ELC and ENS.

In comparing these results to what has been found in the literature (Yuan 1997, 1998; Dugarova 2007, 2008; Kong 2009, 2011), ELCi act roughly in line with the literature, showing strong local binding and low LD binding. The differences however are between Yuan’s (1997, 1998) learners who opted for local antecedents, while Dugarova’s (2007, 2008) and Kong’s (2009, 2011) learners favored LD antecedents (for different reasons however). ELCa follow LD bias with little difficulty, whereas ELCa in the other studies had a weaker selection of LD binding. ELCa’s stronger LD binding here can be considered as a result of adopting the new methodology; participants were given a full opportunity to consider each antecedent on their own merits, contrasted against selecting from a list, allowing preference rather than full grammatical considerations to enter judgements. Considering the KLC groups, as this is the first study that

25 It should be pointed out that these results are similar to Lee (2008), who with a discourse text managed to induce LD binding in English speakers. Here however, the same effect was achieved with single sentences only.


includes such a group, relevant comparisons cannot be made.26

Thus, these anaphoric binding observations can be surmised into three patterns of acquisition:

i) Beginning transition from syntactic to pragmatic anaphoric strategies

L2 learners have difficulty in finding a suitable antecedent in an LD environment. This results in more guessing, lower confidence and quick binding choices due to syntactic transfer.

ii) Advanced transition from syntactic to pragmatic anaphoric strategies

L2 learners have mostly acquired the pragmatic strategies but still have syntactic transfer residue. This interfacing causes an increased processing time in an LD environment, but with less guessing and higher confidence.

iii) Transition from pragmatic to pragmatic anaphoric strategiesL2 learners allow binding by a suitable antecedent quickly (driven by pragmatic considerations) with high confidence and minimal guessing. In a pragmatically impoverished environment there is freedom to choose any antecedent.

The transition from pattern i) to ii) is clearly evident in the English learner group, while pattern iii) is apparent in the Korean learner group.27 ELCa has progressed over ELCi because there is a shift from first language syntactic

26 As a brief aside, there are only two anaphoric studies that make use of confidence levels, Christie (1992) and Li and Kaiser (2009). Christie’s ELC and CNS group had lower confidence in sentences with ambiguity (neutral sentences) and Li and Kaiser (2009) found CNS were more confident in biased sentences than neutral. Both results are similar to what was found here.27 It is of interest to see if the reverse of these results hold true in transitioning from pragmatic to syntactic binding strategies in, for example, Chinese learners of English. While there is little room for exposition, Chinese learners of English have been shown to LD bind as beginners, to improve markedly in the advanced stages (e.g. see Wells 1998). This idea needs to be further investigated from a pragmatic perspective as studies in this area approach this from a pure syntactic perspective.

634 Darcy Sperlich

reflexive pronoun resolution strategies to that of Chinese pragmatic ones. The minor differences between KLCi to KLCa are due to the presence of pragmatic reflexive pronoun resolution strategies in their L1, which are adapted for Chinese. A syntax-only perspective cannot account for what has been found here. It is the presence of syntactic and pragmatic anaphoric strategies which makes ENS so different to CNS and KNS, while the latter two are more alike. Thus, these patterns collaborate with Huang’s (2000) suggestion that languages can be divided along syntactic and pragmatic lines in terms of anaphora.

Shifting the discussion to theoretical concerns, how does the data inform theory on anaphora? This study has used SLA as a tool to understand what is transferred in the interlanguage, being the integration of mother tongue transfer, target language anaphoric strategies and learner transitions. Thus, the guiding thoughts are 1) why is there progression evident from ELCi to ELCa; 2) why is there little difference between KLCi and KLCa; 3) why do ELC and KLC differ?

As alluded to throughout this study, both syntactic and pragmatics strategies must be considered for transfer possibilities, dependent on language. There are three hypotheses considered; first, a simple one which states reflexive pronoun regulation is purely syntactic in nature regardless of language; second, reflexive pronoun regulation is purely pragmatic in nature regardless of language; third, pragmatics and syntax are both important in reflexive pronoun regulation, but this is language dependent. As discussed previously, it has been through the first hypothesis that many SLA studies have been placed within, found to be incorrect. Firstly, in terms of Occam’s Razor the simplest explanation is preferred, and theorizing reflexive pronouns are purely syntactic seems simple, however the amount of syntactic theorizing required to capture languages’ reflexive pronoun distribution is extremely complex. This could apply to the second hypothesis as well. Therefore, this leaves the third hypothesis as the most likely one.

Firstly, as it is argued by Huang (2000), among other works, it is simpler to move complex syntactic operations to pragmatics where the phenomenon appears to be pragmatically regulated. In that sense, Binding Condition A holds true for English but cannot be extended to


Chinese, as Chinese (and Korean) reflexive pronouns are pragmatic in nature. Secondly, this theoretical shift to cooperation between syntactic and pragmatic mechanisms (as discussed previously) is accepted by both syntactic and pragmatic fields; thus adopting a pragmatic solution for Chinese and Korean, and a syntactic solution for English, is in-line with current theoretical trends. However, there is a second possibility to consider in the third hypothesis, C.-T.J. Huang and Liu’s (2001) dual syntactic-pragmatic solution. This faces theory-internal difficulties as Minimalism does not make use of Binding theory which their theory relies on (among other mechanisms), and again on the principle of Occam’s Razor that a language would have two different strategies working side-by-side with a homonymous ‘syntactic ziji’ and ‘pragmatic ziji’ unnecessarily complicates theory to accommodate a partial syntactic solution where a full pragmatic one explains the data.

We conclude that the observed progression in the ELC groups, uniformness of the KLC groups, and the differences between the ELC and KLC groups can be pinpointed to English reflexive pronouns being syntactic in nature, while Chinese and Korean reflexive pronouns are pragmatic in nature.

6. Conclusion

This study has found two different routes regarding English and Korean learners of Chinese in terms of their L2 reflexive pronoun acquisition. This division is attributed to the basic divide between syntactic and pragmatic anaphoric processes used in these languages. Thus, the data gathered here supports a syntactic explanation regarding English reflexive pronouns, and a pragmatic one for Chinese and Korean reflexive pronouns, supporting Huang (2000). Extending these implications to SLA, either strategy is a target of transfer. In other words, one can take Corder’s (1992:21) observation that “the more similar the mother tongue and the target language the greater help the mother tongue can give in acquiring the second language” and paraphrase it as thus: “The more similar the mother and target languages’ anaphoric resolution strategies are, the greater help the mother tongue can give in acquiring anaphora in the second language”. This

636 Darcy Sperlich

then can be reformulated into a testable hypothesis (where the reverse may also hold true); if two languages both use pragmatic strategies to regulate intersentential reflexive pronouns, then it will be easier for a learner from one of those languages to acquire reflexive pronouns in the other, rather than for a learner from a language which uses syntactic strategies instead. The data gathered from this study and the theoretical analysis undertaken supports such a hypothesis, opening up a fresh perspective on binding.

Acknowledgements

I would like to thank the reviewers for all their helpful comments. All errors remain my own. Data presented here is drawn from Sperlich (2013).

References

Baker, C. L. 1995. Contrast, discourse prominence, and intensification, with special reference to locally free reflexives in British English. Language 71.1, 63–101.

Battistella, E. 1989. Chinese reflexivization: A movement to INFL approach. Linguistics 27, 987–1012.

Battistella, E. & Xu, Yonghui. 1990. Remarks on the reflexive in Chinese. Linguistics 28, 205–240.

Birdsong, D. 1989. Meta Linguistic Performance and Interlinguistic Competence. Berlin: Springer-Verlag.

Chomsky, N. 1981. Lectures on Government and Binding. Dordrecht: Foris.Chomsky, N. 1995. The Minimalist Program. Cambridge, MA: MIT Press.Christie, K. 1992. Universal Grammar in the Second Language: An Experimental

Study of the Cross Linguistic Properties of Reflexives in English, Spanish and Chinese. Newark: University of Delaware dissertation.

Clements, G. 1975. The logophoric pronoun in Ewe: Its role in discourse. Journal of West African Languages 2, 141–177.

Cole, P., Hermon, G. & Huang, C.-T. J. 2006. Long-distance binding in Asian languages. In M. Everaert, H. Riemsdijk, R. Goedemans & B. Hollebrandse (Eds.), The Blackwell Companion to Syntax, Volume 3, 21–84. Malden, MA: Blackwell.

Cole, P., Hermon, G., & Sung, Li-May. 1990. Principles and parameters of long-distance reflexives. Linguistic Inquiry 21.1, 1–22.

Cole, P., & Sung, Li-May. 1994. Head movement and long-distance reflexives.


Linguistic Inquiry 25.3, 355–406. Cole, P., & Sung, Li-May. 1997. Long distance reflexives and islandhood in Chinese

head movement versus adjunction to IP. Journal of Chinese Linguistics 25.2, 177–192.

Cole, P., & Wang, Chengchi. 1996. Antecedents and blockers of long-distance reflexives: The case of Chinese ziji. Linguistic Inquiry 27.3, 357–390.

Corder, S. 1992. A role for the mother tongue. In S. Gass & L. Selinker (Eds.), Language Transfer in Language Learning, 18–31. Amsterdam: John Benjamins.

Cowart, W. 1997. Experimental Syntax: Applying Objective Methods to Sentence Judgements. Thousand Oaks, CA: Sage Publications.

Culicover, P., & Jackendoff, R. 1995. Something Else for the Binding theory. Linguistic Inquiry 26.2, 249–275.

Demirci, M. 1997. The Role of Pragmatics in the Acquisition of Reflexive Binding in L2. East Lansing: Michigan State University dissertation.

Dienes, Z. 2008. Subjective measures of unconscious knowledge. Progress in Brain Research 168, 49–64.

Dienes, Z., & Scott, R. 2005. Measuring unconscious knowledge: Distinguishing structural knowledge and judgment knowledge. Psychological Research 69.5-6, 338–351.

Dugarova, E. 2007. Acquisition of the Chinese reflexive 'ziji' by Russian and English speakers. In N. Hilton, R. Arscott, K. Barden, A. Krishna, S. Shah & M. Zellers (Eds.), CamLing: Proceedings of the Fifth University of Cambridge Postgraduate Conference in Language Research, 48–55. Cambridge: Cambridge Institute of Language Research.

Dugarova, E. 2008. Interpretation of the Chinese reflexive 'ziji' by English speakers. In Boping Yuan (Ed.), Theoretical and Empirical Approach to Applied Chinese Language Studies, 7–16. London: Cypress Book Company.

Ellis, R. 2009. Implicit and explicit learning, knowledge and instruction. In R. Ellis, S. Loewen, C. Elder, R. Erlam, J. Philp & H. Reinders (Eds.), Implicit and Explicit Knowledge in Second Language Learning, Testing and Teaching, 3–25. Bristol: Multilingual Matters.

Ellis, R. 2009a. Measuring implicit and explicit knowledge of a second language. In R. Ellis, S. Loewen, C. Elder, R. Erlam, J. Philp & H. Reinders (Eds.), Implicit and Explicit Knowledge in Second Language Learning, Testing and Teaching, 31–64. Bristol: Multilingual Matters.

Heinat, F. 2008. Probes, Pronouns and Binding in the Minimalist Program. Saarbrücken: VDM Verlag.

Hicks, G. 2009. The Derivation of Anaphoric Relations. Amsterdam: John Benjamins.

638 Darcy Sperlich

Hu, Jianhua, Pan, Haihua & Xu, Liejiong. 2001. Is there a finite vs. nonfinite distinction in Chinese? Linguistics 39.6, 1117–1148.

Huang, C.-T. J. 1982. Logical Relations in Chinese and the Theory of Grammar. New York: Garland Publishing.

Huang, C.-T. J., Li, Y.-H. A., & Li, Yafei. 2009. The Syntax of Chinese. Cambridge: Cambridge University Press.

Huang, C.-T. J., & Liu, Chen-Sheng L. 2001. Logophoricity, attitudes and ziji at the interface. In P. Cole, G. Hermon & C.-T. J. Huang (Eds.), Syntax and Semantics 33: Long-Distance Reflexives, 141–195. New York: Academic Press.

Huang, C.-T. J., & Tang, Chih-Chen J. 1991. The local nature of the long-distance reflexive in Chinese. In J. Koster & E. Reuland (Eds.), Long Distance Anaphora, 263–281. Cambridge: Cambridge University Press.

Huang, Yan. 1991. A neo-Gricean pragmatic theory of anaphora. Journal of Linguistics 27.1, 301–335.

Huang, Yan. 1994. The Syntax and Pragmatics of Anaphora: A Study with Special Reference to Chinese. Cambridge: Cambridge University Press.

Huang, Yan. 2000. Anaphora: A Cross-linguistic Study. Oxford: Oxford University Press.

Huang, Yan. 2000a. Discourse anaphora: four theoretical models. Journal of Pragmatics 32, 151-176.

Huang, Yan. 2002. Logophoric marking in East Asian languages. In T. Güldemann & M. Roncador (Eds.), Reported Discourse: A Meeting Ground for Different Linguistic Domains, 211–224. Amsterdam: John Benjamins.

Huang, Yan. 2004. Anaphora and the pragmatics-syntax interface. In L. Horn & G. Ward (Eds.), The Handbook of Pragmatics, 288–314. Oxford: Blackwell.

Huang, Yan. 2006. Anaphora, cataphora, exophora and logophoricity. In K. Brown (Ed.), The Encyclopedia of Languages and Linguistics 2, Volume 8, 586–590. New York: Elsevier.

Huang, Yan. 2014. Pragmatics (2nd Ed.). Oxford: Oxford University Press.Iida, Masayo. 1996. Context and Binding in Japanese. Chicago, IL: CSLI

Publications.Kim ,Ilkyu. 2013. On blocking effects in Chinese: Syntactic, pragmatic, or both?.

Korean Journal of Linguistics 38.2, 305–325 Kim, Hye-Ryun. 1993. First and second language acquisition of binding: An

investigation of Korean-speaking children and Korean adult learners of English. Gainesville, FL: University of Florida dissertation.

Kim, Ji-Hye, & Yoon, James H. 2009. Long-distance bound local anaphors in Korean–an empirical study of the Korean anaphor caki-casin. Lingua 119, 733–755.

Kim, Sun-Hee. 1993. Division of Labor between Grammar and Pragmatics: The


Distribution and Interpretation of Anaphora. New Haven: Yale University dissertation.

Kong, S. 2009. The interpretation of Chinese monomorphemic and polymorphemic anaphors by adult English speakers. Hong Kong Journal of Applied Linguistics 12.1, 45–62.

Kong, S. 2011. Monomorphemic and polymorphemic anaphors in L2 Chinese. Journal of Chinese Linguistics 39.2, 451–477.

Lee, Kum-Young. 2008. The role of pragmatics in reflexive interpretation by Korean learners of English. In M. Bowles (Ed.), Selected Proceedings of the 2007 Second Language Research Forum, 97–112. Somerville, MA: Cascadilla Proceedings Project.

Lee, Kum-Young. 2009. Finite control in Korean. Iowa City: University of Iowa dissertation.

Levinson, S. 1987. Pragmatics and the grammar of anaphora: A partial pragmatic reduction of binding and control phenomena. Journal of Linguistics 23.2, 379–434.

Levinson, S. 1991. Pragmatic reduction of the binding conditions revisited. Journal of Linguistics 27.1, 107–161.

Levinson, S. 2000. Presumptive Meanings: The Theory o f Generalized Conversational Implicature. Cambridge, MA: MIT Press.

Li, D. Cheng-Huan, & Kaiser, E. 2009. Overcoming structural preference: Effects of context on the interpretation of the Chinese reflexive ziji. In L. Sobha, A. Branco & R. Mitkov (Eds.), Proceedings of the 7th Discourse Anaphora and Anaphor Resolution Colloquium, 64–72. Chennai: Computational Linguistics Research Group.

Liu, Chen-Sheng L. 1999. Anaphora in Mandarin Chinese and Binding at the Interface. Irvine: University of California at Irvine dissertation.

Liu, Chen-Sheng L. 2003. Pure reflexivity, pure identity, focus and Chinese ziji-benshen. Journal of East Asian Linguistics 12.1, 19–58.

Madigan, S. 2015. Anaphora and binding. In L. Brown & Jaehoon Yeon (Eds.), The Handbook of Korean Linguistics, 137-154. Malden: MA: Wiley-Blackwell.

Oshima, D. 2007. On empathic and logophoric binding. Research on Language and Computation 5.1, 9–35.

Oshima, D. 2011. Perspectives in Reported Discourse: The De Re/De Dicto Distinction, Indexicality, and Presupposition. Saarbrücken: VDM Verlag.

O'Grady, W. 2013.Processing and language acquisition: Reflexive pronouns in English and Korean. Enewa cengpo [Language and Information] 19, 33-59.

Pan, Haihua. 1997. Constraints on Reflexivization in Mandarin Chinese. New York: Garland Publishing.

Pan, Haihua. 2001. Why the blocking effect? In P. Cole, G. Hermon & C.-T. J.

640 Darcy Sperlich

Huang (Eds.), Syntax and Semantics 33: Long-distance Reflexives, 279–316. San Diego, CA: Academic Press.

Pearson, H. 2015. The interpretation of the logophoric pronoun in Ewe. Natural Language Semantics 23.2, 77–118.

Reuland, E. 2011. Anaphora and Language Design. Cambridge, MA: MIT Press.Rooryck, J., & Wyngaerd, G. 2011. Dissolving Binding Theory. Oxford: Oxford

University Press.Safir, K. 2004. The Syntax of Anaphora. Oxford: Oxford University Press.Safir, K. 2005. Abandoning Coreference. In J. Bermudez (Ed.), Thought, Reference,

and Experience: Themes from the Philosophy of Gareth Evans, 124–163. Oxford: Oxford University Press.

Safir, K. 2013. Syntax, binding, and patterns of anaphora. In M. Dikken (Ed.), The Cambridge Handbook of Generative Syntax, 515–576. Cambridge: Cambridge University Press.

Schwartz, D., & Sprouse, R. 1996. L2 cognitive states and the Full Transfer/Full Access model. Second Language Research 12.1, 40–72.

Sohng, Hong-Ki. 2004. A minimalist analysis of Xo reflexivization in Chinese and Korean. Studies in Generative Grammar 14.3, 375–396.

Sperlich, D. 2013. The Acquisition of Long-Distance Reflexives in Chinese as an Interlanguage: An Experimental Study. Auckland: The University of Auckland dissertation.

Sperlich, D. 2015. Assessing anaphoric relations via the phased choice methodology. International Review of Applied Linguistics in Language Teaching 53.4, 355–388.

Sung, Li-May, & Cole, P. 1991. The effect of morphology on long-distance morphology. Journal of Chinese Linguistics 19.1, 42–61.

Tang, Chih-Chen J. 1989. Chinese reflexives. Natural Language and Linguistic Theory 7, 93–121.

Tang, Chih-Chen J. 2000. Finite and nonfinite clauses in Chinese. Yuyan ji Yuyanxue 1.1, 191–214.

Wang, Yingying, & Pan, Haihua. 2014. A note on the interpretation of non-de se attitude report. Language 90, 746–754.

Wang, Yingying, & Pan, Haihua. 2015. Empathy and Chinese long distance reflexive ziji: Remarks on Giorgi (2006, 2007). Natural Language and Linguistic Theory 33, 307–322.

Wells, T. 1998. L2 acquisition of English binding domains. In M. Beck (Ed.), Morphology and its Interfaces in Second Language Knowledge, 227–256. Amsterdam: John Benjamins.

White, L., Joyce, B., Kawasaki, Takako, Pater, J., & Prévost, P. 1997. The researcher gave the subject a test about himself: Problems of ambiguity and preference in


the investigation of reflexive binding. Language Learning 47.1, 145–172. Xue, Ping, Pollard, C., & Sag, I. 1995. A new perspective on Chinese reflexive ziji.

In R. Aranovich (Ed.), Proceedings of the 13th West Coast Conference on Formal Linguistics, 432–447. Stanford, CA: CLSI Publications.

Yu, W. X. F. 2000. Chinese Reflexives. Levven: Peeters.Yuan, Boping. 1997. Acquisition of the Chinese reflexive ziji by English- and

Japanese-speaking learners. McGill Working Papers in Linguistics 13, 209–220.

Yuan, Boping. 1998. Interpretation of binding and orientation of the Chinese reflexive ziji by English and Japanese speakers. Second Language Research 14.4, 324–340.

642 Darcy Sperlich

Appendix A

N-F/NF

Tom knows that Dick thinks highly of himself.老王认为老陈觉得自己很了不起。철이는 민수가 자신을 높이 평가한다는 것을 알고 있다.Tom thinks that Dick saw a photo of himself in the newspaper.老王认为老陈看见了自己在报纸上的照片。민수는 철이가 신문에서 자신의 사진을 봤다고 생각한다.Tom wants Dick to give himself more time.老王要老陈给自己多一点时间。민수는 철이가 자신에게 더 많은 시간을 줄 것을 원한다.Tom wants Dick to post himself a letter.老王要老陈寄一封信给自己。 민수는 철이가 자신에게 편지를 부치기를 원한다.

P-F/NF

Confucius believed that the new disciple had not yet studied the writings about himself.孔子相信新的弟子还没有读过关于自己的文章。공자는 새로운 제자가 자신에 대한 글을 아직 공부하지 않았다고

믿었다.The teacher believes that the student will give himself an apple in every class.老师相信学生每一堂课都会给自己一颗苹果。선생님은 학생이 매 수업중 자신에게 사과를 줄 것을 믿는다.The Emperor commissioned the artist to paint a portrait of himself.皇帝派画家画一幅自己的画像。황제는 화가에게 자신의 초상화를 그릴 것을 위임했다.The Emperor commissioned the architect to design a new royal capital for himself.皇帝派设计师设计一座新的皇都给自己。황제는 건축가에게 자신을 위한 새로운 황도를 설계할 것을

위임했다.


PS-F/NF

The poisoned hero doesn’t know that the devious villain poisoned himself.中毒的英雄不知道是坏人毒死自己的。 중독된 영웅은 교활한 악한이 자신에게 독을 쓴 것을 알지

못한다.The injured policeman said that the drunken motorist assaulted himself.受伤的警察说酒醉的驾驶攻击自己。부상 당한 경찰은 음주 운전자가 자신을 폭행했다고 말했다.The old soldier wants to tell the young man the terrible war stories about himself.老兵想要跟年轻人说自己可怕的打仗的故事。 늙은 군인은 젊은 남자에게 자신에 대한 끔찍한 전쟁 이야기를

말하기를 원한다.The arrogant movie star wants to show the weak director the rewritten parts about himself.自大的电影明星给软弱的导演看自己角色重改的部分。 거만한 배우는 나약한 감독에게 자신에 대해 재작성된 부분을

보여주기를 원한다.

T-N

Tom heard that Dick thinks that Harry had hurt himself.老王听说老陈认为老李伤害了自己。철이는 민수가 동수는 자신을 해쳤다고 생각한다고 들었다.Tom thought that Dick said that Harry hates himself.老王以为老陈说老李讨厌自己。철이는 민수가 동수는 자신을 미워한다고 말했다고 생각했다.

P-LD1

The visiting president knows that the school teacher said that the student had met himself once before.来参观的总统知道学校老师说学生有见过自己。 방문한 대통령은 선생님이 학생은 자신을 전해 한번 만났다고

말했다는 것을 안다.

644 Darcy Sperlich

The school boy hopes that the teacher will guarantee that the school bully won’t tease himself in class tomorrow.小学生希望老师会保证校园流氓明天上课不会欺负自己。남학생은 선생님이 학교에서 따돌림이 내일 교실에서 자신을

괴롭히지 않을 것을 보장해 줄 것을 바란다.

P-LD2

The doctor found out that the patient doesn’t know that the chemist prescribed himself with the wrong medication.医生发现病人不知道药剂师开错药给自己。 의사는 환자가 약사는 자신에게 잘못된 약을 처방했다는 것을

알지 못 한다는 것을 간파했다.The cook heard that the customer complained that the waiter served himself with the wrong main.厨师听见客人抱怨服务生上错菜给自己。 조리사는 고객이 웨이터는 자신에게 잘못된 식사를 제공했다고

불평하는 것을 들었다.

PS-LD1

The famous scientist thinks that a colleague knows that the journalist will interview himself tomorrow.有名的科学家认为同事知道明天会有记者来采访自己。 유명한 과학자는 동료가 신문기자는 내일 자신을 면담 한다는

것을 안다고 생각한다.The imprisoned president suspected that the colonel knew that the general would overthrow himself.被囚禁的总统怀疑中校知道将军会推翻自己。 수감된 대통령은 중령이 장군은 자신을 전복 시킨다는 것을 알고

있다고 의심했다.


PS-LD2

The student heard that the lazy teacher knows that the angry principal will fire himself tomorrow.学生听说懒惰的老师知道生气的校长明天会解聘自己。 학생은 게으른 선생님이 화난 교장은 내일 자신을 해고 시킨다는

것을 안다고 들었다.The happy teacher believes that the bad student knows that the principal will expel himself tomorrow.快乐的老师相信坏学生知道校长明天会开除自己。 행복한 선생님은 불량 학생이 교장은 내일 자신을 쫓아 낸다는

것을 안다고 믿는다.

646 Darcy Sperlich

Appendix B28

28 Key to tables: Rx refers to the antecedent position in the sentence. R1 is an LD antecedent, while R2/3 depends on the sentence being bi/triclausal. Recorded time is in milliseconds, average shown here. Confidence average shown here. G = Guess, I = Intuition, R = Rule.

648 Darcy Sperlich

650 Darcy Sperlich

652 Darcy Sperlich

pragmatic or syntactic ziji? evidence from language...

Documents