L2 Chinese: Grammatical development and processing

Abstract

Two recent books (Jiang, 2014, Advances in Chinese as a second language; Wang, 2013, Grammatical development of Chinese among non-native speakers) provide new resources for exploring the role of processing in acquiring Chinese as a second language (L2). This review article summarizes, assesses and compares some of the findings in these books with reference to current interests in L2 sentence processing and other research findings in L2 Chinese. It is shown that there have been serious attempts to link acquisition and processing of L2 Chinese, and that Chinese offers rich materials to study how L2 learners develop sensitivity to semantic and discourse conditions on L2 structures and acquire new grammatical features associated with them through parsing the input. However, to reach this goal, a more fine-grained approach to L2 Chinese structures and their processability needs to be taken.

Keywords

grammar L2 Chinese processability processing

Jiang N (ed.) (2014) Advances in Chinese as a second language: Acquisition and processing. Newcastle: Cambridge Scholars Publishing. vii + 301 pp. UK £49.99. ISBN: 978-1-4438-5346-0 (hardback).

Wang X (2013) Grammatical development of Chinese among non-native speakers: From a processability account. Newcastle: Cambridge Scholars Publishing. xvi + 221 pp. UK £44.99. ISBN: 978-1-4438-4958-6.

I Introduction

Processing and acquisition used to be two independent themes in second language research. Conventionally, the former refers to the construction of structural representations of sentences and phrases in real-time language comprehension and production, and the latter the emergence of linguistic knowledge resulting from analysing linguistic information in the input (Dekydtspotter and Renaud, 2014). Now it has become increasingly difficult to discuss one without alluding to the other. Second language learners process to acquire and acquire to process (Carroll, 2001; Fodor, 1998, etc.). Nevertheless, how this is achieved is often the subject of much speculation and little empirical investigation (Dekydtspotter and Renaud, 2014; Roberts, 2013).

This article reviews two recently published books on second language (L2) Chinese. Wang’s (2013) dissertation tests Pienemann’s (1998a, 1998b, 2005, 2011) Processability Theory against the developmental sequence of a number of linguistically prominent and theoretically important structures in Chinese. Jiang (2014) features 13 independent articles reporting findings of small-to-medium-scale studies on various aspects of L2 Chinese, divided into two categories: acquisition and processing. As summarized in Y Zhao (2011), linguistic approaches to L2 Chinese have centred on a limited set of grammatical structures such as wh-interrogatives, restrictive relative clauses, the ba- and bei- constructions, and grammatical aspect marking. Some of them received further attention in these two books and constitute the overlapping area between them. For this reason, this article will focus on the acquisition and processing of those grammatical structures and leave aside that of other linguistic elements (e.g. phonological and orthographic information). Studies exploring non-linguistic factors such as motivation, attitude and learning strategies are also excluded.¹

II Wang, 2013: Processability and acquirability

Wang (2013) investigates factors constraining the emergence order of grammatical structures in L2 Chinese under the framework of Processability Theory (PT, Pienemann, 1998a, 1998b). A brief sketch of the theory is provided below. For full depictions and critiques of the theory, refer to Pienemann (1998a, 1998b), Pienemann and Keßler (2011) and a special issue in Bilingualism: Language and Cognition in 1998.

1 Overview of the Processability Theory (PT)

Processability Theory (PT) is a processing-oriented approach to language development. Following Levelt’s (1989) view of language production, PT posits that different processing units such as words, phrases and sentences feed into each other in language generation, and that exchange of grammatical information between these units ensures that the resulting linguistic production conforms to L2 grammatical rules. In this theory, the information exchange is formalized as the unification of lexical and grammatical features (e.g. [number], [gender], [tense] and [root]), based on the theoretical architecture of the Lexical Functional Grammar (LFG). According to PT, in language production, different processing units are activated in an incremental manner and in a set sequence (in temporal order):

word/lemma (accessing the word/lemma in the lexicon);

category procedure (assigning the grammatical category of the lemma and attaching lexical morphemes);

phrasal procedure (assigning the grammatical category of the phrasal head, and producing phrasal agreement);

S-procedure and Word Order Rules (determining the grammatical function of the phrase and producing inter-phrasal agreement) and, optionally,

matrix/subordinate clause (forming matrix and subordinate clauses and producing inter-clausal agreement).

This sequence is termed the ‘processability hierarchy’, which should be followed in both real-time language generation and acquisition. Assuming processing routines are language-specific and have to be acquired by L2 learners, the hierarchy generates many testable predictions. Since its genesis, PT has spurred many empirical studies and received support from a range of typologically different languages such as English, Japanese and Spanish (see Bonilla, 2015; Pienemann and Keßler, 2011). Wang (2013) provides the most recent evidence from Mandarin Chinese.

2 Wang’s study on PT stages in L2 Chinese

Table 1 synthesizes the key elements of Wang’s study. Previous studies have identified a PT-derived acquisition sequence (i.e. items without asterisks in Table 1; Gao, 2005; Zhang, 2001, 2008; cited in Wang, 2013). Wang’s research aims to test whether this sequence also applies to a group of learners who receive formal L2 instruction under a different syllabus, and to extend PT to additional syntactic structures (i.e. items with asterisks in Table 1) such that the acquisition sequence identified in previous studies can be enriched.

Table 1.

PT stages in Chinese.

Stage	Processing procedures	Information exchange	Morpheme	Syntax
5	Subordinate clause-procedure	Inter-clausal	(not investigated)	bei construction*
5	Subordinate clause-procedure	Inter-clausal	(not investigated)	ba construction
4	S-procedure	Inter-phrasal	Relative clause de	Topicalization:
				T(=O)SV(COMP)*
				T(=O)(S)V(COMP)*
3	Phrasal procedure	Phrasal	Classifier	Topicalization:
3	Phrasal procedure	Phrasal	V-COMP marker -de	T(=Adjunct)SV(O)*
2	Category procedure	Lexical	Possessive -de	Canonical SV(O):
			Adjective -de	declaratives
			Attributive -de	interrogatives
			Progressive (zheng)zai-Experiential -guo	(yes/no questions, wh-questions, intonation)
				Topicalization:
				T(=S)VO*
1	Word/lemma access	No exchange	Single words and constituents	Formulaic expressions

Note. * newly added items in Wang (2013).

Source. Adapted from Table 7-11 in Wang (2013: 153), with modifications.

In common with many other studies testing PT, an LFG-based analysis of the target structures constitutes the syntactic foundation of the empirical study, providing justifications for a certain structure being placed into one of the five processing procedure categories (second column on the left in Table 1). Instead of synthesizing existing LFG analyses of Chinese syntax, Wang’s analysis of the target structures draws on the work from various theoretical frameworks and approaches including descriptive, functional, generative and pedagogical grammars (e.g. Chao, 1968; Chomsky, 2000; Li and Thompson, 1981; Norman, 1991; Yip and Rimmington, 1997, 2004, cited in Wang, 2013), as well as the working hypotheses in Zhang’s and Gao’s L2 studies, presumably because systematic LFG-based descriptions of Mandarin Chinese are still comparatively scarce. A number of -de structures are examined. The ‘possessive -de’, the ‘attributive -de’ and the ‘adjective -de’, illustrated in (1), (2) and (3) respectively, are taken as suffixes. As their suffixation requires ‘no information exchange between constituents’, they should be acquired very early at the lexical level (Stage 2 in Table 1). However, de in the ‘V-de-Complement structure’ as in (4) is analysed as ‘a marker for complement of the verb’, and information exchange between the verb zou (‘to walk’) and its complement hen man (‘very slowly’) within the VP is considered necessary, hence an instance of phrasal procedure (Stage 3 in Table 1). Additionally, the relative clause marker -de, as in (5), marks the relationship between the head noun and its modifier, and information exchange across these phrases renders this structure at the inter-phrasal level,² which is the most difficult morpheme among all -de structures (Stage 4 in Table 1). Also included in the target morphemes are two aspectual markers: the progressive (zheng)zai and the experiential -guo ((6) and (7)), which are put into the lexical category together with the suffix -de because the retrieval of these morphemes is, like that of the -de suffix, ‘directly triggered by conceptualization and requires no information exchange with other constituents’. Classifiers are another morpheme under investigation. Chinese possesses a large number of classifiers, as exemplified in (8). Successful acquisition of them requires matching the idiosyncratic features between the classifier and the noun. A phrasal procedure (Stage 3 in Table 1) is then implicated (Wang, 2013: 77, 8).

(1) Possessive -de³

wo -de yifu

1SG DE clothes

‘my clothes’

(2) Attributive -de

mutou -de fangzi

wood DE house

‘wooden house’

(3) Adjective -de

piaoliang -de qiqiu

pretty DE balloon

‘pretty balloon’

(4) V-de-Complement

zou -de hen man

walk DE very slowly

‘walk slowly’

(5) Relative clause

Ni gei de qian hen duo.

2SG give DE money very much

‘The money you gave (me) was quite a lot.’

(6) Progressive zhengzai

Tamen zhengzai chi fan.

3PL PROG eat meal

‘They are eating.’

(7) Experiential -guo

Tamen chi -guo niurou.

3PL eat EXP beef

‘They have had beef before.’

(8) Classifier

san zhang bing

three CL pancake

‘three pancakes’

(Examples from Wang, 2013, with minor modifications)

Turning to the syntactic structures, topicalization structures are divided into a number of sub-types according to the underlying position of the topic NP (i.e. whether it is the underlying subject, object or adjunct) and the presence/absence of a syntactic subject in addition to the topic NP (i.e. whether the surface structure appears in the Topic–Subject–Verb or Topic–Verb sequence). Four of these sub-types are examined in Wang’s study. Among them, topic structures with underlying objects as topics (e.g. (9) and (10)) are deemed to involve cross-phrasal information exchange as there is ‘object movement’ in these structures, whereas those with adjuncts as topics (e.g. (11)) involve information exchange at a more local level (i.e. phrasal) and should become processable earlier than the previous two sub-types. Finally, topic structures with the underlying subject as the topic (e.g. (12)) appear in the canonical SVO sequence and thus ‘can be processed at the early stage without unification and processing exchange’. Apart from the topic structures, the ba- and bei-constructions, as illustrated in (13) and (14), also appear in the non-canonical word order. In Wang’s study, both ba and bei are treated as the main verb sub-categorizing clausal complements, and the matrix subject (i.e. the pre-ba/bei NP) a preposed argument of the lexical verb (e.g. da, ‘to beat’, in both (13) and (14)) in the embedded clause. Therefore, the information exchange in both structures takes place across clauses, a category at the top level of the processability hierarchy (Stage 5). Note that Chinese interrogatives, regardless of their types (question particle, wh-words in (15) and (16)) are all predicted to emerge early because they ‘normally keep the original canonical word order’ and are characterized by ‘direct mapping without unification and processing exchange’ (Wang, 2013: 78).

(9) T(=O)SV(COMP)

Zhe ben shu, wo du-de hen lei.

this CL book 1SG read-DE very tired

‘This book, I had a hard time reading (it).’

(10) T(=O)(S)V(COMP)

Na ben shu jie zou le.

that CL book borrow away PERF

‘That book has been borrowed.’

(11) T(=Adjunct)SV(O)

Zuotian, xue xia-de hen da.

yesterday snow drop-DE very heavy

‘Yesterday it snowed heavily.’

(12) T(=S)VO

Mama xihuan youyong.

Mum like swim

‘Mum likes swimming.’

(13) Ba-structure

Zhangsan ba Lisi da le.

Zhangsan BA Lisi beat PERF

‘Zhangsan beat Lisi.’

(14) Bei-structure

Lisi bei Zhangsan da le.

Lisi BEI Zhangsan beat PERF

‘Lisi was beaten by Zhangsan.’

(15) Question particle

Ni shi laoshi ma?

2SG COP teacher Q

‘Are you a teacher?’

(16) Wh-words

Ni xihuan shenme?

2SG like what

‘What do you like?’

(Examples from Wang, 2013, with minor modifications)

Wang collected spontaneous and prompted oral data through semi-structured interviews at an interval of 2 or 3 weeks over 38 weeks from 8 undergraduate students (aged 19 to 22) enrolled in an intensive Chinese language beginners’ program at Newcastle University in the UK, who had diverse language learning experiences and backgrounds. The speech data were then transcribed into text, resulting in a 30,000-word corpus, of which 25,000 words were used in the study. A stringent ‘first emergence’ criterion was adopted in the data analysis: A structure is viewed as having emerged if a minimum of four tokens had been observed with lexically-varied contexts, as this is considered the point where the relevant linguistic knowledge has become operational in a learner’s interlanguage system. The eight participants generated eight individual data sets, and the data analysis was conducted within individuals, and then across them. The general finding is that the participants followed the hypothesized PT stages consistently in their L2 development, suggesting that processing constraints do play quite an important role in second language acquisition.

3 A more fine-grained approach to L2 Chinese structures and their processability

Wang’s study makes a contribution to the field by attempting to demonstrate the emergence sequence of a number of core structures in L2 Chinese. Existing studies on L2 Chinese, especially those taking formal linguistic approaches, often focus on a single structure or a very limited set of them. Wang’s study has a much wider scope and serves as a good base for future experimental designs aiming to carry out in-depth investigations of these structures. The semi-structured interviews and the subsequent transcription and coding were designed and carried out in a consistent and meticulous fashion. Although the resulting oral production dataset, not included in the original book, could become a rich corpus resource for future studies on various aspects of beginning learners’ L2 Chinese, a number of issues deserve some scrutiny, including the syntactic analysis, the data interpretation and the theoretical implication of the study.

While the majority of the syntactic analysis in Wang’s study is formalized in LFG terms, some analyses of the Chinese structures seem unnecessarily simplified. Take the grammatical aspect markers as an example. In Wang’s study, the progressive marker (zheng)zai and the experiential marker -guo are taken as lexical morphology marking devices because they ‘attach to verbs and mark their tense and aspect’. Consequently, they require no information exchange between constituents and should be acquired very early in Stage 2. This analysis, presumably under the influence of the analysis of English tense marking, is problematic in the present case on at least two levels. First, (zheng)zai, a progressive aspect marker, is head of an aspectual phrase, rather than an affix, whereas other aspect markers such as -guo, -zhe and -le behave in a way similar to suffixes in morphologically rich languages (Huang et al. 2009: 101). The categorial differences between the progressive (zheng)zai and the experiential -guo is unjustifiably dismissed. Second, Chinese aspect markers have telicity requirements on the verb or VP they select/attach to (Smith, 1994). The progressive (zheng)zai, for example, has a durative requirement on the VP it subcategorizes for and is thus incompatible with stative or achievement verbs. In feature unification terms, the progressive marker (zheng)zai and its complement should match for the feature [+/– durative]. Information exchange between the aspect marker and the verb (phrase) is unavoidable for full acquisition of Chinese aspect markers. A more fine-grained approach to these structures is called for. Similar problems can be found in the analysis of the yes/no question particles and wh-interrogatives (for detailed analysis of the complexities of wh-interrogatives in L2 Chinese, see Dugarova, 2014; Yuan, 2013; Yuan and Dugarova, 2013).

A second issue concerns data interpretation. In Wang’s study and many other PT-based studies, the emergence of the target structure in the learners’ production is taken as reliable indications of target-like processing procedures in the L2 grammars. This, again, needs to be examined more closely, given that producing the target structure four times (i.e. the emergence criterion adopted in Wang’s study) does not mean that the complex hierarchical structure (or in LFG terms, the complex feature unification procedures) as assumed in the native grammar is necessarily at work. As proposed by Clahsen and Felser (2006) and demonstrated in many processing studies, shallow structures predominate in L2 processing. Approaching L2 Chinese production from a processing perspective, PT-based L2 studies need to provide tangible evidence showing what exactly is being processed in language generation. In fact, there is evidence suggesting that the seemingly target-like surface forms produced by L2 Chinese learners may result from processing procedures considerably different from those of native speakers, a point to be returned to below.

A more general yet fundamental issue arising from Wang’s study is the theoretical contribution of processability in describing L2 developmental stages and acquirability problems. In the fast developing field of second language acquisition, acquirability issues in developmental grammars are central to the investigation of many modulating factors such as the quantity and quality of the input, first language (L1) transfer and the status and inherent complexity of the target structure/feature in both the L1 and the L2 (Hawkins and Chan, 1997; Lardiere, 2009; Sorace, 2011, etc.). Numerous studies have shown that one or a combination of two or more of these factors can lead to temporary or permanent breakdown in L2 grammars. Adding to the inventory a new processability factor, PT-based studies are expected to generate new acquisition data. However, the L2 Chinese acquisition sequence discovered in Wang’s study is less than surprising, as the relative ease or difficulty of these structures in L2 Chinese can or has been explained by independent reasons. For instance, it has been reported that aspect markers emerge early in English–Chinese interlanguages (Jin, 2004; Jin and Hendriks, 2003; Wen, 1995, 1997), because they are highly frequent in the L2 input, simple in their morpho-phonological form, and are taught explicitly and drilled intensively in L2 instruction. It is also shown that learners use fewer ba-constructions in obligatory contexts than native speakers do (Paul, 2014), probably because the construction is structurally complex and the conditions on its use in discourse contexts are unfamiliar to speakers who do not have a counterpart of ba-construction in their L1s.

It is not clear, at least in Wang’s study, how processability fits in with other well-established factors in predicting and explaining L2 developmental stages. Are they mutually exclusive? Or is processability a conceptual macro-factor? Take L1 transfer for an example. In Pienemann and Keßler (2011), the role of pre-existing processing routines on the processability of L2 structures is captured in the Developmental Moderated Transfer Hypothesis, which hypothesizes that the grammars of any existing languages cannot be transferred into the L2 until the L2 has developed to the point at which the L2 corresponding structures can be processed. However, this factor did not receive much attention in Wang’s study. Participants in Wang’s study were speakers of either English or other European languages, languages typologically distant from Mandarin Chinese. Those languages do not possess the L2-specific structures such as classifiers, grammatical aspect markers and prenominal relative clauses. Conceivably these learners have to establish processing routines of those structures from scratch. But will the same apply to speakers of languages that do have counterparts of these structures (e.g. Cantonese learners of Mandarin⁴)? In this sense, the acquisition sequence predicted in Table 1 is too strong and is subject to significant modification if factors such as L2 learners’ pre-existing linguistic knowledge are taken into account.

III Jiang, 2014: Acquisition and processing

Among the 13 studies included in Jiang (2014), four examine the acquisition and processing of grammatical structures by L2 learners from a linguistic perspective and contribute novel L2 data. Topics include the acquisition of the perfective marker le (Yang and Wu, 2014), the telicity requirement in the bei-construction (Yao, 2014), the production of caused motion event (Paul, 2014), and the processing of different types of relative clauses (Wang and Feng, 2014), which are subsumed under two categories in the following discussion.

1 Aspect marking, telicity requirement and event encoding

Chinese possesses a rich set of aspect markers. They are frequent in both the written and oral forms of the language, with their actual occurrence governed by many conditions including the lexical semantics of the verb and the syntactic and discourse environment surrounding it (Ljungqvist, 2007; Xiao and McEnery, 2004). Marking the verb with aspect markers is optional in some cases and obligatory in others, making the input rather ambiguous and acquisition additionally difficult. Yang and Wu (2014) investigated the use of the perfective aspect marker -le by American learners of Chinese who were at the post-beginning and mid-intermediate levels (n = 41) in different Chinese language programs including domestic formal instruction program (9 months), domestic immersion program (2 months) and study abroad program (2–3 months). The participants were given identical oral production tasks at the beginning and the end of the study programs (a pre/post-test design). The results showed that in both tests, learners from all three study programs produced -le significantly less frequently than the native controls (ps < 0.005), and that none of the learner groups performed meaningfully differently in this respect in the pre-test and the post-test. This is strong evidence that low-proficiency English-speaking learners of Chinese tend to undersupply the perfective -le in production, and that the quantity and quality of the input, operationalized as the three different study programs in the study, does not seem to increase -le production at this acquisition stage, the latter of which has not been documented in previous investigations of -le production.

Telicity, namely, whether an event has an inherent endpoint, is a universal notion found across languages. L2 learners do not have to acquire the notion of telicity from scratch, but they do have to learn whether a structure in the L2 is subject to telicity requirement. Many structures in Chinese are selective about the telicity of their predicates. Telic events are privileged in those structures, whereas atelic events are marginal, if acceptable at all. A prime example is the ba-construction, which expresses a strong disposal and causal-affected meaning (Liu, 1997). The bei-construction, which expresses a passive meaning, is subject to similar requirements. Yao (2014) presents a small-scale study testing whether learners from various L1 backgrounds with intermediate Chinese proficiency are sensitive to the telicity requirement of the bei-construction (n = 8). Three measures were used to elicit relevant data:

a grammaticality judgment task to test whether the learners were able to accept bei-constructions with achievement and accomplishment verbs and reject those with stative verbs and activity verbs;

a sentence-making task to test whether learners can create grammatical bei-constructions with verb roots given to them; and

analysis of corpus data to examine the telicity of the bei-constructions in spontaneous production.

The results of the three measures coincide. Learners demonstrated sensitivity to the telicity requirement across tasks, but had difficulty using L2-specific telicity encoding strategies to fulfil this requirement to different extents, a finding consistent with that of the ba-construction in Yang and Huang (2004).

In describing voluntary motion events in which Patient travels from one locative point to another while the Agent stays at the original place throughout the course (e.g. throwing a ball), both English and Chinese encode the trajectory, or Path, of the moving entity by prepositional phrases, as illustrated in (17). Paul (2014) studied whether English-speaking L2 Chinese learners describe these ‘caused motion events’ in Chinese in the native-like manner. Fifteen advanced English learners of Chinese watched 30 short videos containing caused motion events and described what happened in them. Results show that the learners encoded Path significantly less frequently than the native controls and, when they did, they tended to adhere to a preposition expressing static locative information (i.e. zai, at) rather than using prepositions indicating the trajectory of moving objects (e.g. xiang/wang, towards), and abide by the English surface word order (i.e. S-V-O-PP, as in ‘he kicked the ball into the swimming pool’), illustrated in (17). Another difference between the learners and the natives is the use of the disposal ba-construction (shown in (18) and (19)): the learners did not use it as frequently as the natives did.

(17) Caused motion event expressions

wang louti shang reng qiu

towards stairs up throw ball

‘throw the ball towards the stairs’

(18) Learners’ production

* Ta ti qiu zai youyongchi.

3SG kick ball at swimming-pool

‘He kicked the ball into the swimming pool.’

(19) Encoding the event in ba-construction

ba zhe ge bao reng dao louti xia

BA this CL bag throw arrive stair down

‘throw the bag to the bottom of the stairs’

(Examples from Paul, 2014, with minor modifications)

These findings, taken together with previous findings and those in Wang (2013), give rise to intriguing questions for future research. With the aspect marker -le, since the learners did produce grammatical sentences with the perfective -le in the study, some kind of mental representation of this aspect marker is clearly present in the L2 grammars. As mentioned earlier, the actual use of aspect markers is conditioned by many factors, some of them highly complex and subtle. Supplying the marker in real-time production necessitates both target-like representation of these complex conditions and accurate analyses and fast integration of the syntactic and discourse information of the outputting utterances, which can be cognitively demanding for low-proficiency L2 learners in on-line production tasks. Recall that in Wang (2013), aspect markers are taken as a lexical morphology marking device, and thus producing aspect markers does not require information exchange between constituents. The behaviours of low-proficiency learners in Yang and Wu (2014) and in many previous studies (e.g. Jin, 2004; Wen, 1995, 1997), in fact, indicate that learners may not have established native-like processing routines for aspect marking at the post-beginner and lower-intermediate stages.

In L1 processing, a number of studies have used event-related potentials (ERPs) to record Mandarin native speakers’ brain responses to agreement violations between grammatical aspect markers and other constituents in the sentence (for a review, see Zhou et al., 2009). P600, an ERP component indicating detection of morphosyntactic violations and widely observed in the processing of tense violations in Indo-European languages, is also found in Chinese speakers’ responses to mismatches between future/present-oriented temporal adverbials and verbs with perfective suffixes such as -le (Zhang and Zhang, 2008) and -guo (Qiu and Zhou, 2012). This provides an exciting starting point for parallel research on L2 Chinese. Do English-speaking learners, like native speakers, process Chinese aspect markers in on-line comprehension tasks? In other words, can disagreement between aspect markers and other tense/aspect-bearing constituents elicit the P600 effects from L2 learners? If so, is the learners’ ability to process L2 aspect markers an approximation of L1 tense processing strategies (transfer) or a product of substantial L2 exposure (grammatical and processing restructuring)? In the latter case, how much exposure is needed for the learner to adopt this L2-specific processing strategy? Do learners of different L1s pattern together or differ from each other depending on the typological properties of tense and aspect marking in their L1? Answers to these questions will not only pave our path towards a better transition theory of L2 acquisition, but are also appealing to educators who strive to facilitate L2 acquisition by providing the right kind of instruction on processing the input, e.g. Processing Instruction (VanPatten and Cadierno, 1993).

Turning to the ba- and bei-constructions, how do learners become sensitive to the telicity requirement in those constructions? Undoubtedly, teaching can trigger learners’ awareness of it, as current L2 Chinese teaching materials do introduce properties of the events in which the ba- and bei-constructions are preferred over canonical SVO sentences, although the notions of aspect and telicity may not be explicitly taught. Alternatively, learners may achieve this goal by keeping track of the respective frequency of telic and atelic predicates in ba- and bei-sentences in the L2 input implicitly. In artificial language learning, it has been found that speakers of ‘gender languages’ (languages with grammatical gender marking) are subconsciously more sensitive to animacy information encoded in the determiner system of the target language, suggesting the role of prior linguistic knowledge in implicit learning (Williams, 2005). In the case of ba- and bei-constructions, will L2 learners speaking ‘aspect languages’ attend to telicity properties of the predicates in these constructions earlier than others? A related, but different property of ba- and bei-constructions is that both structures carry with them certain construction-specific pragmatic function. The bei-construction, for instance, is associated with an ‘adversative’ reading that one of the participants of the event is adversely affected (Li and Thompson, 1981). It has been reported that native speakers anticipate this adversative reading when they process bei-sentences in real time (Philipp et al., 2008). Will L2 learners be able to generate the same anticipation when they process bei-sentences? Complex structures as such provide an excellent opportunity for future investigations examining how various properties of those constructions are (not) parsed during L2 development, and most importantly how parsing the input helps to develop linguistic knowledge in the L2, an increasingly important topic in L2 sentence processing (Roberts, 2013). Future studies should attend to the structural, semantic and contextual features of such complex structures and design different experimental tasks targeting different features in order to improve our understanding of the differences between native and non-native speakers in the integration of multi-layered information in real-time comprehension and production.

2 Relative clauses

Relative clauses (RCs) have always been a topic of intense interest in psycholinguistics, as the psychological reality of syntactic gaps, the potential role of working memory capacity and frequency, among many others, can all be tested based on materials developed from the RCs (Gibson and Wu, 2013; Juffs and Rodríguez, 2015). Chinese is particularly fascinating in this area in that it is an SVO language with prenominal RCs, rather than postnominal ones.

Wang and Feng (2014) investigated whether L2 learners of Chinese comprehend and produce three types of relative clauses with equal difficulty: direct object-extracted RCs (DOs), subject-extracted RCs (SUs), and object of preposition-extracted RCs (OPREPs), as illustrated in (20), (21) and (22) respectively. In Chinese DOs, the subject, verb and object appear in the SVO order, perfectly aligned with the word order of canonical simple sentences in the language, whereas those elements in the SUs appear in the VOS order, radically different from the canonical word order of the language. It is predicted that Chinese DOs should have superiority over SUs in L2 comprehension and production due to the isomorphism between the DOs and the canonical word order of the language. As for the OPREPs, a fundamental difference between the Chinese and English counterparts is that while Chinese fills the gap of the relativized head noun with a resumptive pronoun (e.g. ta in (22)), English does not allow this mechanism (e.g. * the man whom I made a bed for him). The prediction is that the cross-linguistic difference may lead to acquisition difficulty. Thirty intermediate learners of Chinese who were native speakers of English participated in their study and completed two oral tasks in which they were asked to translate Chinese RCs into English (comprehension) and English RCs into Chinese (production) respectively. Accuracy rates of each type of RCs were calculated. In both tasks, the learners demonstrated the highest accuracy rate with DOs, a lower accuracy rate with SUs, and the poorest performance with OPREPs, as predicted by the cross-linguistic comparison mentioned above.⁵ Erroneous sentences produced by the learners in the OPREPs category revealed that the problem lies in the filling pronoun (i.e. ta in (22)). Learners failed to produce either the pronoun (i.e. (23)) or the entire prepositional phrase and (i.e. (24)) in OPREPs. It remains to be discovered how learners parse grammatical OPREPs in the input and, in particular, how they achieve the pronoun resolution in those sentences (filler-gap dependencies).

(20) Direct object-extracted RC (DOs)

ta xihuan de nanren

3SG like DE man

‘the man who she likes’

(21) Subject-extracted RC (SUs)

xihuan ta de nanren

like 3SG DE man

‘the man who likes her’

(22) Object of preposition-extracted RC (OPREPs)

wo wei ta puchuang de laoren

1SG for 3SG make bed DE old man

‘the man whom I made a bed for’

(23) Sample error with object of preposition-extracted RC (OPREPs)

* wo wei _ kai men de laoshi

1SG for open door DE teacher

‘the teacher whom I opened the door for’

(24) Sample error with object of preposition-extracted RC (OPREPs)

* wo _ _ kai men de laoshi

1SG open door DE teacher

‘the teacher whom I opened the door for’

(Examples from Wang and Feng, 2014, with minor modifications)

Research in the processing of RCs in L2 Chinese is a welcome endeavour. Going beyond the articles included in the target books in this review article, there are other studies exploring factors affecting processing difficulties in different types of Chinese RCs. For instance, Chinese RCs can have demonstrative-classifier (DCl) strings in two different positions, as shown in (25) and (26). Xu (2014) administered a timed grammaticality judgement test to L2 Chinese learners (L1 English) and found that DCl-first RCs (e.g. (25)) is easier for the L2 learners than DCl-second RCs (e.g. (26)) as measured by reaction times and error rates, probably because the demonstrative-classifier string in DCl-second RCs is embedded within the NP that contains the relativized head noun, which can increase the processing load when one attempts to interpret the content of the gap, a result consistent with the filler-gap domain theory (Hawkins, 2004). Taken together, the above studies indicate that some Chinese RCs are indeed difficult to process by L2 learners whose L1s are typologically different. The late emergence of RCs in L2 Chinese, as demonstrated in Wang (2013), can be attributed to processability problems. If this is indeed the case, sentences where the subject of ba/bei-constructions is relativized, as in (27), should be even more problematic for L2 learners. In fact, in L1 processing studies, ba/bei-constructions and relative clauses, as well as combinations of these structures, have been used extensively to investigate the independent and interactive effects of animacy, word order and world knowledge (plausibility) in sentence comprehension, yielding many meaningful results (e.g. Chow and Phillips, 2013; Ng and Wicha, 2014; Philipp et al., 2008). Complicated sentence structures like (27) will provide useful materials and new insights in further examination of processing difficulty in the L2.

(25) Demonstrative-classifier-first RCs

zhe ge _ xihuan Xiao Lin de ren hen shuai.

this CL like Xiao Lin DE person very cute

‘The person that likes Xiao Lin is very cute.’

(26) Demonstrative-classifier-second RCs

_ xihuan Xiao Lin de zhe ge ren hen shuai.

like Xiao Lin DE this CL person very cute

‘The person that likes Xiao Lin is very cute.’

(Examples from Xu, 2014, with minor modifications)

(27) Sentences where the subject of ba/bei-constructions is relativized

Ba shengzi qie-duan le de xiaodao shengxiu le.

BA cord cut-break PERF DE knife rust PERF

‘The knife which cut off the cord is rusted.’

(Example from Phillipp et al., 2008, with minor modifications)

IV Concluding remarks

This review article has discussed the merits and limitations of two recently available resources in L2 Chinese and the theoretical and empirical issues arising from them. It is commendable that serious attempts are made to link acquisition and processing of L2 Chinese or to account for acquisition failures from a processing perspective. However, the current focus of the studies is still on the representation, rather than the parsing/processing side of L2 grammars. Existing studies typically tap into the property of L2 grammars using behavioural methods such as grammaticality judgments, translation, and structured oral production tasks. Research using time-course sensitive on-line measures such as eye-tracking and self-paced reading is still rare, and neurolinguistic and brain imaging tools have not been employed to explore the activation of the brain in processing Chinese as a second language. Despite the remarkable efforts to describe the processability constraints on L2 grammatical development, there is little direct and tangible evidence showing how processing abilities develop over time among L2 Chinese learners.

As for research participants, current studies rarely go beyond late L2 Chinese learners who come to the classroom without prior exposure to Chinese and acquire the language through formal L2 instruction. Processing abilities of learners who are at or near the lowest or top ends of the proficiency scale (i.e. true beginners and near-native learners of Chinese) remain comparatively unknown, except for those reported in Han and Liu (2013) and LX Zhao (2014). On another dimension, bilingual speakers of Chinese other than late L2 Chinese learners are still under-represented. Data from these speakers will disentangle the effects of different variables. For instance, Cantonese speakers, who are widely available in southern provinces of China, Hong Kong and many overseas communities can serve as a good control group, as the acquisition of Mandarin by speakers of a typologically very close language can help to rule out confounding factors such as L1–L2 differences. Heritage speakers of Mandarin are another potentially useful control group when the onset age of acquisition is relevant (Mai, 2012; Montrul, 2008).

Chinese is a language with very limited inflectional morphology and many in-situ structures. In view of these typological properties, Chinese may be less suitable than its European peers to explore the acquisition or processing of overt morphological and syntactic movements. Nevertheless, grammatical relationships between content words in Chinese are predominately marked by functional words and word order variations, which are in many cases sensitive to both event structure (tense, aspect and telicity) and information structure (topic, focus and presupposition). Hence, Chinese offers exceptionally rich materials for studies venturing into the acquisition of new combinations of features and conditions in new functional categories, or more specifically how learners register those new feature sets and de-register L1-based feature sets in response to input in the course of L2 development, which is an emerging research theme in the field (Dekydtspotter and Renaud, 2014; Lardiere, 2009). While parallel syntactic, semantic, pragmatic and prosodic computations are generally necessary in parsing the input, a body of research has begun to show that L2 Chinese learners, like those of other L2s, do not master every detail about a structure at the same time (Mai, 2012; Yuan, 2010, 2013), possibly because they develop the ability to parse different aspects of a structure rather unevenly (Mai and Yuan, in press). Studying which features/aspects of a linguistic structure become processable or accessible for which type of L2 Chinese learner at which acquisition stage is undoubtedly an extremely promising topic for future research on L2 Chinese processing and on L2 sentence processing in general.

Footnotes

Acknowledgements

I would like to thank Boping Yuan, Margaret Thomas, Virginia Yip, Xiangjun Deng and Jiangling Zhou for helpful discussions and suggestions, and Shanrong Xie for technical assistance. All remaining errors are my own.

Declaration of conflicting interest

The author declares that there is no conflict of interest.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

References

Bonilla

(2015) From number agreement to the subjunctive: Evidence for processability theory in L2 Spanish. Second Language Research 31: 53–74.

Carroll

(2001) Input and evidence: The raw material of second language acquisition. Amsterdam: John Benjamins.

Chao

(1968) A grammar of spoken Chinese. Berkeley, CA: University of California Press.

Chomsky

(2000) New horizons in the study of language and mind. Cambridge: Cambridge University Press.

Chow

W-Y

Phillips

(2013) No semantic illusions in the ‘Semantic P600’ phenomenon: ERP evidence from Mandarin Chinese. Brain Research 1506: 76–93.

Clahsen

Felser

(2006) Grammatical processing in language learners. Applied Psycholinguistics 27: 3–42.

Dekydtspotter

Renaud

(2014) On second language processing and grammatical development: The parser in second language acquisition. Linguistic Approaches to Bilingualism 4: 131–65.

Dugarova

(2014) Russian speakers’ L2 Chinese acquisition of wh-topicalization at the syntax–discourse interface. Second Language Research 30: 411–37.

Fodor

(1998) Learning to parse? Journal of Psycholinguistic Research 27: 285–319.

10.

Gao

(2005) Noun phrase morphemes and topic development in L2 Mandarin Chinese: A processability perspective. Unpublished PhD thesis, Victoria University of Wellington, New Zealand.

11.

Gibson

HHI

(2013) Processing Chinese relative clauses in context. Language and Cognitive Processes 28: 125–55.

12.

Grosjean

(2008) Studying bilinguals. Oxford: Oxford University Press.

13.

Han

Liu

(2013) Input processing of Chinese by ab initio learners. Second Language Research 29: 145–64.

14.

Hawkins

(2004) Efficiency and complexity in grammars. Oxford: Oxford University Press.

15.

Hawkins

Chan

(1997) The partial availability of universal grammar in second language acquisition: The ‘failed’ functional features hypothesis. Second Language Research 13: 187–226.

16.

Huang

YHA

(2009) The syntax of Chinese. Cambridge: Cambridge University Press.

17.

Jin

(2004) The development of aspect marking in L2 Chinese by English native speakers. Unpublished PhD thesis, University of Cambridge, Cambridge, UK.

18.

Jin

Hendriks

(2003) The development of aspect marking in L1 and L2 Chinese. In: Working papers on applied linguistics. Cambridge: Research Centre for English and Applied Linguistics (RCEAL), University of Cambridge, 69–100.

19.

Juffs

Rodríguez

(2015) Second Language sentence processing. New York: Routledge.

20.

Lardiere

(2009) Some thoughts on the contrastive analysis of features in second language acquisition. Second Language Research 25: 173–227.

21.

Levelt

WJM

(1989) Speaking: From intention to articulation. Cambridge, MA: The MIT Press

22.

Thompson

(1981) Mandarin Chinese: A functional reference grammar. Berkeley, CA: University of California Press.

23.

Liu

(1997) An aspectual analysis of ba. Journal of East Asian Linguistics 6: 51–99.

24.

Ljungqvist

(2007) Le, guo and zhe in Mandarin Chinese: A relevance-theoretic account. Journal of East Asian Linguistics 16: 193–235.

25.

Mai

(2012) Properties of the (shi)…de focus construction in adult L2 acquisition and heritage language acquisition of Mandarin Chinese. Unpublished PhD thesis, University of Cambridge, Cambridge, UK.

26.

Mai

Yuan

(in press) Uneven reassembly of tense, telicity and discourse features in L2 acquisition of the Chinese shì … de cleft construction by adult English speakers. Second Language Research.

27.

Matthews

Yip

(2011) Cantonese: A comprehensive grammar. 2nd edition. London: Routledge.

28.

Montrul

(2008) Incomplete acquisition in bilingualism: Re-examining the age factor. Amsterdam: John Benjamins.

29.

Wicha

(2014) Processing gap-filler dependencies in Chinese: What does it tell us about semantic processing? Journal of Memory and Language 74: 16–35.

30.

Paul

(2014) Expressing caused motion events in L2 Chinese: The case of learning a language that is typologically similar to the learners’ L1. In: Jiang

(ed.) Advances in Chinese a second language: Acquisition and processing. Newcastle: Cambridge Scholar Publishers, 271–98.

31.

Philipp

Bornkessel-Schlesewsky

Bisang

Schlesewsky

(2008) The role of animacy in the real time comprehension of Mandarin Chinese: Evidence from auditory event-related brain potentials. Brain and Language 105: 112–33.

32.

Pienemann

(1998a) Developmental dynamics in L1 and L2 acquisition: Processability theory and generative entrenchment. Bilingualism: Language and Cognition 1: 1–20.

33.

Pienemann

(1998b) Language processing and second language development: Processability theory. Amsterdam/Philadelphia, PA: John Benjamins.

34.

Pienemann

Keßler

(2011) Studying processability theory. Amsterdam/Philadelphia, PA: John Benjamins.

35.

Qiu

Zhou

(2012) Processing temporal agreement in a tenseless language: An ERP study of Mandarin Chinese. Brain Research 1446: 91–108.

36.

Roberts

(2013) Sentence processing in bilinguals. In: van Gompel

RPG

(eds) Sentence processing. New York: Psychology Press, 221–46.

37.

Smith

(1994) Aspectual viewpoint and situation type in Mandarin Chinese. Journal of East Asian Linguistics 3: 107–46.

38.

Sorace

(2011) Pinning down the concept of ‘interface’ in bilingualism. Linguistic Approaches to Bilingualism 1: 1–33.

39.

VanPatten

Cadierno

(1993) Explicit instruction and input processing. Studies in Second Language Acquisition 15: 225–43.

40.

Wang

(2013) Grammatical development of Chinese among non-native speakers: From a processability account. Newcastle: Cambridge Scholars Publishing.

41.

Wang

Feng

(2014) Research on acquisition difficulty of relative clauses in Chinese as a second language. In: Jiang

(ed.) Advances in Chinese a second language: Acquisition and processing. Newcastle: Cambridge Scholar Publishers, 254–70.

42.

Wen

(1995) Second language acquisition of the Chinese particle le. International Journal of Applied Linguistics 5: 45–62.

43.

Wen

(1997) Acquisition of Chinese aspect: an analysis of the interlanguage of learners of Chinese as a foreign language. Review of Applied Linguistics 117–118: 1–26.

44.

Williams

(2005) Learning without awareness. Studies in Second Language Acquisition 27: 269–304.

45.

Xiao

McEnery

(2004) Aspect in Mandarin Chinese: A corpus-based study. Amsterdam/Philadelphia, PA: John Benjamins.

46.

(2014) Processing relative clauses in Chinese as a second language. Second Language Research 30: 439–61.

47.

Yang

(2014) Acquiring the perfective marker le in different learning contexts. In: Jiang

(ed.) Advances in Chinese a second language: Acquisition and processing. Newcastle: Cambridge Scholar Publishers, 10–32.

48.

Yang

Huang

(2004) The L2 acquisition of the Chinese ba-construction (in Chinese). Chinese teaching in the world 67: 49–59.

49.

Yao

(2014) The second-language acquisition of aspect in Chinese bei-constructions. In: Jiang

(ed.) Advances in Chinese a second language: Acquisition and processing. Newcastle: Cambridge Scholar Publishers, 54–76.

50.

Yip

Rimmington

(1997) Chinese: An essential grammar. London and New York: Routledge.

51.

Yip

Rimmington

(2004) Chinese: A comprehensive grammar. London and New York: Routledge.

52.

Yuan

(2010) Domain-wide or variable-dependent vulnerability of the semantics–syntax interface in L2 acquisition? Evidence from wh-words used as existential polarity words in L2 Chinese grammars. Second Language Research 26: 219–60.

53.

Yuan

(2013) Is Chinese daodi the hell in English speakers’ L2 acquisition of Chinese daodi … wh … questions? Effects and recoverability of L1 transfer at L2 interfaces. International Journal of Bilingualism 17: 403–30.

54.

Yuan

Dugarova

(2013) Wh-topicalization at the syntax–discourse interface in English speakers’ L2 Chinese grammars. Studies in Second Language Acquisition 34: 533–60.

55.

Zhang

(2008) Brain responses to agreement violations of Chinese grammatical aspect. Cognitive Neuroscience and Neuropsychology 19: 1039–43.

56.

Zhao

(2014) Ultimate attainment of anaphora resolution in L2 Chinese. Second Language Research 30: 381–407.

57.

Zhao

(2011) A tree in the wood: a review of research on L2 Chinese acquisition. Second Language Research 27: 559–72.

58.

Zhou

Cheung

Chen

H-C

(2009) Processing the Chinese language: An introduction. Language and Cognitive Processes 24: 929–46.