Sociolinguistic Competence by L2 Chinese Learners Through the Lens of Null Object Use

Abstract

Using the analytical tool Rbrul, this study explores the object expression variations in the speech of twenty CSL (Chinese as a Second Language) learners whose first languages (L1) were English, Russian, Korean, and Japanese, and compares them to native speaker (NS) patterns. Multivariate analysis revealed that the learner patterns closely resembled NSs’ in most dimensions explored, except the learners tended to overuse overt pronouns and underuse null forms. For both CSL learners and NSs, the general patterns in object use were as follows: (1) animate objects tended to favor overt objects, while inanimate objects favored null forms, (2) switch in referents favored overt forms, while no referent change favored null, (3) specific referents favored null forms, whereas nonspecific referents favored overt expressions, and (4) conversational contexts favored null forms, but elicited narratives favored explicit forms. As for the patterns specific to learners, the findings were as follows: (1) among the four L1s included in the analysis, Japanese and Russian speakers tended to use null objects more than Korean and English speakers, (2) a stay in China of one, two, or four years tended to favor null forms, but a three-year stay favored explicit forms, and (3) high-intermediate learners tended to use null forms more frequently than advanced learners. These results indicated that learners successfully acquired null object use patterns in spoken Chinese, but they still required further development in understanding the nuances between overt object forms and null object use to enhance their sociolinguistic competence.

1. Introduction

In recent decades, the variationist approach within the field of second language acquisition has witnessed rapid growth, manifesting itself in the burgeoning exploration of a multitude of language varieties and an expanded examination of diverse linguistic features. Originally, this approach primarily centered around the English language (e.g., Adamson, 1988; Dickerson, 1975), but it has since broadened its scope to encompass a range of other languages, including French (e.g., Dewaele, 2004; Gudmestad et al., 2020; Howard, 2006; Kennedy Terry, 2022; Mougeon et al., 2004; Regan, 2022; Rehner et al., 2022), Spanish (e.g., Escalante & Wright, 2022; Geeslin, 2003, 2013; Geeslin & Fafulas, 2022; Gudmestad, 2006, 2008; Pozzi, 2022), Chinese (e.g., Li, 2010, 2014; Li et al., 2022; Starr, 2022), Korean (Park, 2022), Italian (Di Salvo & Nagy, 2022), and Catalan (Davidson, 2022). The variationist paradigm has extended its inquiry to encompass nearly all levels of linguistic analysis, including phonological, semantic, syntactic, morphological, and pragmatic aspects.

The variationist approach acknowledges the interlanguage of language learners as a legitimate and independent linguistic system, characterized by its own distinct patterns. Furthermore, it places considerable emphasis on the examination of language use within authentic social contexts, often drawing upon natural speech data obtained from sociolinguistic interviews. Second language (L2) variation studies have yielded valuable insights into how L2 learners utilize their acquired language, the influencing factors at play, and, in turn, have enriched our understanding of the underlying principles and grammar in second language acquisition in general.

Within the realm of L2 variation studies, two prominent trends have emerged: Type 1 and Type 2 variation. The former primarily investigates the acquisition of obligatory linguistic forms, whereas the latter delves into instances where NSs exhibit variability and employ diverse linguistic styles. Optional use of language variables is one widely acknowledged testing ground for the acquisition of Type 2 variation, such as schwa deletion in French (Kennedy Terry, 2022) and LE use in Chinese (Li et al., 2022). This has been acknowledged to be a great indicator of how learners develop sociolinguistic competence, because the ability to style shift appropriately is an important aspect of sociolinguistic competence (Li, 2010; Pozzi, 2022). The null argument, such as null subject and null object, is another great feature for exploring the acquisition of stylistic variation by L2 learners. Learners need to learn when to use a noun phrase, a pronoun, or a null form to express subjects and/or objects. Subject pronominal use has been a widely explored sociolinguistic variable under the variationist paradigm in different second languages, including Spanish and Chinese (e.g., Abreu, 2009; Cameron & Flores-Ferrán, 2004; Flores-Ferrán, 2004, 2007; Li, 2014). Object expression, however, has not received as much investigation.

Mandarin Chinese allows both null subjects and null objects (example 1):

(1)	我	喜欢	看	电视，	但是	Ø	不	经常
	wŏ	xĭhuān	kàn	diànshì,	dànshì	Ø	bù	jīngcháng
	I	like	watch	TV	but	Ø	not	often
	看	Ø
	kàn	Ø
	watch	Ø
	I like watching TV, but (I) don’t watch (it) very often.

The initial occurrence of the word “I” is represented by a pronoun, and in the subsequent mention, it is left out. Similarly, the word “TV” is introduced as a noun phrase initially but is omitted at the second mention. This study focuses on null object use in L2 Chinese.

Previous research has examined the concept of null objects in Chinese from various angles, such as formal and functional linguistics (Huang, 1984; Li & Thompson, 1981; Sato, 2019; Wang et al., 1992; Wu, 1999; Xu, 2006; Zhao, 2012; Zhou, 2014; Zhu & Gavarró, 2019). Nevertheless, to my knowledge, there have been no variation studies addressing how null objects are used by CSL learners in oral discourse, nor have there been investigations into how Chinese NSs employ null objects orally from variationist perspective. This study intends to fill the gaps and serve a two-fold purpose: firstly, to explore the variations in the use of null objects in the oral discourse of CSL learners and, secondly, to compare these patterns with those found in the oral discourse of Chinese NSs. Therefore, the research questions explored are the following:

(1)

What is the variation in null object use by CSL learners? What are the factors that influence their object use?

(2)

Are the learner patterns of null object use different from NS patterns? If so, how?

In the next sections, relevant literature will be reviewed, followed by the methodology of the current study. After that, the results will be reported, followed by the discussion and conclusion.

2. Literature Review

The acquisition and use of null objects has been a subject of investigation across various languages, shedding light on the intricate dynamics of first and second language acquisition. This section first synthesizes findings from some studies on null objects in different languages, including both L1 and L2 acquisition. Subsequently, example studies of null subject and object use in Chinese are reviewed.

In the field of L1 acquisition, the exploration of object omission has revealed a consistent developmental pattern. Studies of L1 Spanish, for instance, have reported varying rates of null object use. Fujino and Sano (2002) observed high initial rates of object omission that decreased over time, while Lizsckowski (1999) reported significantly lower rates. Experimental studies by Wexler et al. (2004) found no null object use among 2–4-year-olds, contrasting with Castilla and Pérez-Leroux’s (2010) discovery of early null object use in 3-year-olds that disappeared by age 4–5. Kim (2000) and Pérez-Leroux et al. (2008) identified similar patterns in children aged 1–2 and 2–3, with high rates of object omission decreasing significantly as they approached ages 2–3 and 3–4. This pattern was also reflected in L1 bilingual acquisition, as seen in the study by Larrañaga and Guijarro-Fuentes (2012), where children initially exhibited high rates of clitic omission before converging with adult patterns over time. Despite differing results among studies, the literature on L1 Spanish acquisition of null objects appears to converge on a general trend of increased null object use in children at the outset, followed by a decline as they mature linguistically.

Another pattern that was found of null objects in different languages is that animacy and specificity are the two main factors that govern null object use. Schwenter (2006) reviewed studies of null direct object use in Brazilian Portuguese (e.g., Schwenter & Silva, 2002, 2003) and two Spanish dialects in South America—colloquial spoken Quiteño Spanish and Paraguayan Spanish (Choi, 1998, 2000; Morgan, 2004). A main pattern occurs when the referent of an anaphoric direct object is animate and specific; in such cases, an overt pronoun tends to be used, while the null form tends to be used when the referent is inanimate, mainly in the case of third-person anaphoric direct objects.

Null object use has also been explored in L2 learning. For example, Zyzik (2008) examined null object usage by 50 second-language (L2) Spanish learners of varying proficiency levels. In oral tasks, null objects were mainly found in conjoined clauses and interview responses where context had already established referents. Beginners were more accepting of null objects with specific referents in grammaticality judgments, while higher-level learners tended to reject them. Zyzik argued for a performance-based explanation, as null object use lacked a clear developmental pattern, often involved self-corrections to full noun phrases, and was context-specific. Factors such as verb type, lexical retrieval, and verb morphology influenced null object usage.

Exploring the impact of bilingualism on null object acquisition, Pirvulescu et al. (2014) observed that bilingual children exhibited extended periods of object omission in both French and English compared to their monolingual peers, with more omissions occurring in French. The frequency of omissions also depended on language dominance, with fewer omissions in the dominant language.

Furthermore, Zhou et al. (2021) studied Cantonese–English bilingual children’s object expression in different verb contexts. These bilingual children demonstrated similarities to their Cantonese monolingual peers in object omission, suggesting bidirectional cross-linguistic influence. Input quantity, structural frequencies, verb types, and cross-linguistic interactions were identified as factors influencing object realization in bilingual acquisition.

In summary, the investigation of object omission in L1 and L2 acquisition has revealed intriguing developmental patterns, with L1 acquisition showing more null object use, followed by a decrease over time, and L2 acquisition showing a lack of developmental patterns and a strong relationship between null object use and referent animacy and specificity. While there may be variations across languages and individual learners, the interplay between animacy, specificity, proficiency levels, and cross-linguistic interactions was shown to shape the acquisition and use of null objects.

As far as Chinese is concerned, a plethora of studies have explored null subject features in both L1 and L2 Chinese (e.g., Block, 1993; Casentini et al., 2023; Fang & Yin, 2013; Xu & Yuan, 2024) and have found that various factors affect the use and acquisition of null subjects in Chinese, such as proficiency level, native language, and pragmatic factors. Some other scholars have investigated subject pronoun use in L1 and L2 Chinese from a sociolinguistic variation perspective (e.g., Guy et al., forthcoming; Jia & Bayley, 2002; Li, 2014; Li et al., 2012; Li & Bayley, 2018; X. Zhang, 2021) and have found that a variety of linguistic and social constraints govern the use of subject pronouns, such as subject person and number, subject coreference, sentence type, referent specificity, native language, gender, and learner proficiency level. This work has piqued interest in investigating null object use in Chinese.

Much of the earlier literature on null object use in Chinese operated within the generative framework (e.g., Huang, 1984; Sato, 2019; Wang et al., 1992; Xu, 2006; Zhao, 2012; Zhou, 2014; Zhu & Gavarró, 2019). However, within the field of generative grammar, there has been a disagreement regarding the analysis of null objects. Some argue for their existence due to empty topics, while others posit a pro-drop parameter explanation. When analyzing null subjects and objects in generative grammar, Huang (1995) proposed a neo-Gricean pragmatic approach, suggesting that many instances of null objects may be better explained by pragmatic or discourse factors rather than purely syntactic parameters, and the factors licensing and identifying them were not fully understood.

Regarding the acquisition of null objects in L1 Chinese, studies indicated that this occurs at a very early age, typically around 1.8 or 2 years old (Wang et al., 1992; Wexler, 1994; Yuan, 1993; Zhu & Gavarró, 2019). Li and Thompson (1981) suggested that objects in Chinese are often omitted if they have been previously mentioned or are nonspecific referents. Relatively few studies have explored how CSL learners acquire this aspect of the language (e.g., Yuan, 1993; Zhao, 2009, 2012), with most focusing on the perception and interpretation of null object use. Chang and Zheng (2018) examined CSL learners’ compositions in HSK tests (an official examination to assess the Chinese language proficiency of non-NSs) and found that null objects were mainly used in inanimate situations. They also found that L1 did not play a significant role in learners’ acquisition of null arguments. In addition, both L1 English and L1 Japanese speakers used few null objects, even though Japanese is a null-object language. They contended that infrequent use of null objects is a universal phenomenon for L2 learners, and it is neither a developmental feature nor transferrable from L1 to L2.

As mentioned earlier, no studies have investigated how CSL learners use null objects in natural speech from a sociolinguistic variation perspective and how various factors influence their usage. Therefore, this study aims to address this gap in the literature.

3. Method

3.1. Data

The current study is a component of a larger research project that explores the diverse use of Chinese language forms among both Chinese NSs and CSL learners. Specifically, this study centers on the variability in the use of object expressions, especially null objects, by CSL learners, using Chinese NS data as a baseline.

To gather data for this investigation, two distinct discourse settings were examined. First, sociolinguistic interviews were conducted involving a total of thirty-three participants. These conversations took place over fifty-three sessions, lasting approximately 30 to 40 min for NSs who were interviewed once, and 45 to 60 min for each of the two interviews with the learners. These interviews took place in a comfortable lounge within a northeastern Chinese University, a convenient location for the participants.

These interviews had two parts. The initial segment involved informal discussions on topics of interest to the participants, such as their hobbies, experiences during middle/high school, memorable travel moments, favorite movies or novels, and beloved teachers. While a set of primary questions to guide the interview was set up, most conversations naturally evolved, often veering away from the prescribed protocol as new, engaging topics emerged. In the second part of the interview, the participants watched and retold “The Pear Stories” by Chafe (1980).

3.2. Participants

The participants of the study included 20 CSL learners (8 high-intermediate and 12 advanced) and 13 Chinese NS peers as the baseline. When recruiting the CSL participants, best efforts were made to ensure a balance in terms of gender, native language, and the representation from different classes. The demographic characteristics of these 20 CSL participants are summarized in

Table 1. CSL learner participants: demographic characteristics.

Pseudonyms	Age	Sex	Origin	Native Language	Prior Chinese Learning (Months)	Length of Stay (Months)	Chinese Proficiency
Ren	22	F	Korea	Korean	0	12	Advanced
Nan	24	F	Korea	Korean	6	18	Advanced
Zhu	24	M	Korea	Korean	0	48	Advanced
Xuan	25	M	Korea	Korean	0	11	Intermediate
Gui	31	M	Korea	Korean	18	8	Advanced
Suo	25	F	Russia	Russian	66	5	Advanced
Ye	25	F	Russia	Russian	0	48	Advanced
Ta	21	F	Russia	Russian	18	7	Advanced
Ya	22	F	Russia	Russian	24	6	Advanced
Ni	21	F	Russia	Russian	30	7	Advanced
Ann	30	F	Britain	English	0	25	Intermediate
Wen	40	M	Australia	English	30	30	Intermediate
Fen	29	M	Australia	English	18	30	Advanced
Long	29	M	USA	English	0	48	Advanced
Liang	25	M	Australia	English	36	6	Intermediate
Mei	22	F	Japan	Japanese	30	6	Advanced
Chi	23	F	Japan	Japanese	0	18	Intermediate
Zhuang	22	F	Japan	Japanese	36	3	Intermediate
Lu	33	M	Japan	Japanese	0	13	Intermediate
Cun	33	M	Japan	Japanese	0	21	Intermediate

.

All the NSs (seven males and six females) were college students, aged 18 to 20, from a northeastern university in China and were fluent in Mandarin Chinese.

3.3. Variables

Since this is the first study to investigate null object use in Chinese from a sociolinguistic perspective, insights were drawn from the previous literature on null object and sociolinguistic studies on subject use regarding what factors to explore, because Chinese is both a null-subject and a null-object language. Subject use has been extensively explored under the variationist paradigm, and coreference and subject person/number have been shown by various studies to be the two main constraints for subject pronominal expressions (e.g., Cameron, 1992, 1993; Flores-Ferrán, 2002, 2010). Li et al. (2012) and Li (2014) also explored null subject use in L1 and L2 Chinese and found that coreference and subject person/number were the two main constraints. Sentence type, age, gender, occupation, proficiency, and L1 also reached significance.

Informed by the past studies, in this study, dependent variables included how objects were expressed, namely, full noun phrase (NP), pronoun, and null. Independent variables included linguistic, social, and developmental constraints. Linguistic factor groups were coreference, object person/number/animacy, referent specificity, sentence type, and native language. Social factor groups included gender and discourse/speech style. Developmental factor groups included length of stay in China and proficiency level.

3.3.1. Linguistic Constraints

Coreference

Based on previous variationist studies of subject use in Chinese (Jia & Bayley, 2002; Li, 2014; Li et al., 2012), coreference was defined by whether the subject in one clause referred to the same entity as the subject in the clause just before it. However, Jia and Bayley (2002), Li et al. (2012), as well as Li (2014), pointed out that this approach to operationalizing coreference in data coding encountered a few challenges. The first challenge was in instances where the subject in one clause did not exactly match the subject in the preceding clause but instead had some overlap. The second challenge was in cases where there was a change in the surface form but the underlying referents of the two subjects were the same. A final challenge occurred when the subject was mentioned at locations in the preceding clause other than the subject position. This is related to a prominent feature of Chinese, the topic chain, which refers to “a chain of clauses sharing a single topic” (W. Li, 2004, p. 26). As W. Li (2004) claimed, “much of the use of zero NPs in Chinese can be accounted for by topic chains”. The position of a topic and/or a null NP is rather flexible. Therefore, a subject coreference may occur in the preceding clause as a fronted topic, an object, or in another position. This is also true for an object coreference. Therefore, these cases need to be separately coded. This study followed these categorizations of coreference in object use, and the factors were as follows (examples 2–8). Transcription abbreviations in the English glossary are shown in Appendix A.

Switch: when an object’s referent is different from the one in the previous clause,

(2)	她	去年	在	香港	写	论文，	我	在那	等	她。
	tā	qùnián	zài	xiānggǎng	xiě	lùnwén,	wǒ	zàinà	děng	tā
	she	last:year	at	Hong Kong	write	thesis,	I	there	wait	her
	She was doing her thesis in Hong Kong last year and I was there waiting for her.

No switch: when an object’s referent is the same as the one in the previous clause,

(3)	我们	可以	帮助	他们。	我们	两	个	小时	一直	帮助	他们。
	wǒmen	kěyǐ	bāngzhù	tāmen.	wǒmen	liǎng	gè	xiǎoshí	yìzhí	bāngzhù	tāmen
	we	can	help	them.	we	two	CL	hour	always	help	them
	We could help them. We had been helping them for two hours.

Same underlying: when an object’s referent is the same as the one in the previous clause, but the object surface forms are different,

(4)	A:	她	请	你？
		tā	qǐng	nǐ？
		she	invite	you
		She invited you?
	B:	对，	请	我。
		duì,	qǐng	wǒ
		yes,	invite	me
		Yeah, (she) invited me.

Partial overlap: when an object’s referent partially overlaps with the object in the previous clause,

(5)	他们	看见	我们，	然后	他们	帮	我。
	tāmen	kànjiàn	wǒmen,	ránhòu	tāmen	bāng	wǒ
	they	see	us,	then	they	help	me
	They saw us, and then they helped me.

Subject–object: when an object is the same as the subject in the previous clause,

(6)	我	去	上海，	在那	有	人	接	我。
	wǒ	qù	shànghǎi,	zàinà	yǒu	rén	jiē	wǒ
	I	go	Shanghai,	there	there:is	someone	receive	me
	I went to Shanghai, and somebody was there to meet me.

Fronted topic–object: when an object is the same as the fronted topic,

(7)	类似	这样	的	(电影)，	你	没	看	过	Ø	吗？
	lèisì	zhèyàng	de	(diànyǐng),	nǐ	méi	kàn	guò	Ø	ma
	similar	this: kind	ATT	movie,	you	not	see	EXP	Ø	Q
	Have you not seen this kind of movie?

Other—object: when an object occurs at other locations in the previous clause,

(8)	他	和	我	一起	学习，	他	教	我。
	tā	hé	wǒ	yìqǐ	xuéxí,	tā	jiāo	wǒ
	he	with	me	together	study,	he	teach	me
	He and I studied together. He taught me.

Object Person/Number/Animacy

Due to the characteristics of the data, the differentiation of object animacy in Chinese was restricted to third-person singular and plural subjects exclusively. In the case of first- and second-person singular and plural objects, all of them were considered as animate subjects. Regarding the concept of person, a Chinese object can be classified into first person (represented by “我, wŏ”), second person (indicated by “你, nĭ”), and third person (expressed through “她/他/它, tā”). In terms of number, each of these persons can be further classified as singular or plural, which is denoted by the addition of “们 (men)” to the singular forms. Additionally, third-person objects in both singular and plural forms were subcategorized as either animate or inanimate.

Specificity

The tokens were categorized based on whether the objects they referred to were specific or nonspecific. A specific object denoted a particular person or thing that both participants in the conversation were aware of. Conversely, a nonspecific object referred to something with a more general or broad meaning.

Sentence Type

The tokens were also coded according to the type of sentences they were in. The factors were statement, question, and imperative.

Native Language

Native language has been shown to influence learners’ choice of stylistic variables (Li, 2010, 2014). In this study, learners’ native languages were Russian, English, Japanese, and Korean.

3.3.2. Social Constraints

Gender

Gender has been a compelling area of investigation for scholars, especially in the context of language use (Eckert, 1989, 2000; Eckert & McConnell-Ginet, 1992, 2003; Labov, 1990; Major, 2004; Zhang, 2001, 2005). Major’s research in 2004, for instance, offered compelling evidence suggesting that women tended to exhibit standard or formal language forms. In contrast, men were inclined to employ more informal and less prestigious language forms, often to assert their masculinity.

Furthermore, studies on subject pronominal expression have shed light on intriguing differences (Bayley & Pease-Alvarez, 1996, 1997; Cameron, 1992; Wulf et al., 2002), demonstrating that women were more likely than men to employ subject pronouns in their speech. Drawing on these insights, the hypothesis regarding the role of gender in this study is that female participants are expected to use overt object forms more frequently than their male counterparts.

Discourse/Speech Style

In previous research, it was observed that subject pronouns tended to be more prevalent in informal speech, as opposed to formal speech (Ávila-Jiménez, 1996; Hochberg, 1986). Furthermore, the usage of subject pronouns can vary depending on the context of the conversation, as demonstrated by studies such as those by Jia and Bayley (2002) and Travis (2007).

For the current study, data were gathered from informal conversations and the retelling of The Pear Stories (Chafe, 1980). Although the data from the retelling of The Pear Stories constituted only a small part of our dataset, they served as a valuable source for understanding how speakers used objects in their narratives.

3.3.3. Developmental Constraints

Length of Stay in China

Previous studies showed that the length of time in the target language environment significantly affected learners’ development of sociolinguistic competence, manifested by their acquisition of target stylistic variation (e.g., Li, 2010, 2014; Howard et al., 2006). The findings generally demonstrated that the longer learners stayed in the target language environment, the better they acquired stylistic language features. In this study, learners were categorized into 1-, 2-, 3-, and 4-year stays in China, which respectively represent equal to or below 12, 24, 36, and 48 months.

Proficiency Level

These learners were categorized into high-intermediate and advanced levels based on their class placement and researcher observations, in addition to the discussions with their Chinese language instructors.

In sum, the dependent variable in this study was object forms (full noun phrase, pronoun, and null). The independent variables were object person/number/animacy (i.e., 1st sg and pl, 2nd sg and pl, 3rd sg and pl +animate, 3rd sg and pl -animate), coreference (i.e., switch, no switch, same underlying, partial overlap, fronted topic → object, subject → object, or other → object), specificity (i.e., specific or nonspecific), sentence type (i.e., declarative, question, or imperative), native language (Russian, English, Korean, or Japanese), speaker gender (i.e., male or female), discourse context (i.e., conversation and elicited narrative), proficiency level (high-intermediate or advanced), and length of stay in China (1, 2, 3, or 4 years).

3.4. Coding Exclusions

The coding exclusions comprised the following criteria: (a) counting repeated object pronouns only once, (b) elimination of object pronouns in false starts, and (c) excluding cases of formulaic expressions (example 9) or pivotal pronouns in serial verbal constructions (example 10):

(9)	谢谢	(你)。
	Xièxiè	nǐ
	thank	(you)
	“Thank you”.

(10)	他们	让	我	选择	中文。
	Tāmen	ràng	wŏ	xuǎnzé	zhōngwén
	They	let	me	choose	Chinese
	“They wanted me to choose Chinese”.

3.5. Coding and Analytical Tool

All the objects and the dependent variables in the two datasets were coded into three groups: full noun phrase, pronoun, and null form. The coding was performed by a Chinese NS and checked by another NS, both of whom were experienced in coding Chinese interlanguage data. Disagreements were resolved through discussion, and the interrater reliability for both NS data and learner data was 0.95.

To determine the significance of the factors examined and their relative weights, a specialized application of logistic regression, Rbrul (Johnson, 2009), was used to analyze the data. Rbrul is a program specifically designed to handle natural speech data and is the most extensively used tool in the field of sociolinguistic variation. It has the ability to simultaneously examine and take into consideration the influence of several linguistic and social constraints on the use of the linguistic variation being studied. In addition to frequency data, the results include factor weights that indicate each factor’s relative effect strength compared to other factors in the same factor group, as well as the likelihood that the variant will be used. The factor favors the use of the variant if the factor weight is greater than or equal to 0.5, while a weight below 0.5 is unfavorable.

4. Results

Rbrul analyses of 4059 learner tokens (site for a null or overt object) and 1848 NS tokens were conducted to investigate patterns of object use, comparing learner patterns with the NS baseline pattern. Both full noun phrases and pronouns were included for overt object expressions because, in many cases, the null form was used to indicate the omission of a full noun phrase. Therefore, excluding full noun phrases might not reflect the complete picture. In the analyses, the application value (i.e., the default dependent variable against which all the factors were measured) was set as the null object, a setting used in all further analyses, unless stated otherwise. A factor with a weight of 0.50 or above tended to favor null object, while a weight below 0.50 disfavored null objects (i.e., favored overt forms).

The results showed that the overall rates of null object use, as well as the likelihood of its use (input probability), were lower among CSL learners (10.5%, 0.06) compared to NSs (15%, 0.26;

Table 2. Null object use by CSL learners vs. Chinese NSs.

	CSL Learners				NSs
Factor Group	Factors	Weight	% Null	N	Factors	Weight	% Null	N
Coreference	Fronted topic No switch Other ⟶ Object Subject ⟶ Object Partial overlap Switch Same underlying	0.98 0.65 0.49 0.43 0.16 0.08 0	91.7 37 25 8.33 4.55 2.63 0	181 359 72 240 22 3121 18	Fronted topic No switch Subject ⟶ Object Other ⟶ Object Same underlying Switch Partial overlap	0.93 0.63 0.54 0.51 0.35 0.06 0	85.6 42.9 27.5 33.3 16.7 3.16 0	111 203 120 39 12 1331 29
Person/ number/ animacy	3rd pl (−animate) 3rd sg (−animate) 3rd pl (+animate) 2nd sg 3rd sg (+animate) 1st sg 1st pl 2nd pl	0.85 0.74 0.47 0.38 0.30 0.21 0 0	19.1 13 4.49 2.9 4.07 2.99 0 0	225 2532 468 69 467 234 40 4	3rd sg (−animate) 3rd pl (−animate) 1st pl 1st sg 3rd pl (+animate) 3rd sg (+animate) 2nd sg 2nd pl	0.70 0.66 0.62 0.50 0.42 0.40 0.23 0	15.8 18.8 18.8 20.4 9.82 11.3 5.26 0	1265 85 16 54 112 265 19 2
Specificity	Specific Nonspecific	0.63 0.37	16.3 6.92	1523 2472	Specific Nonspecific	0.56 0.44	23.2 11.1	587 1229
Discourse context	Conversation Narrative	0.68 0.32	11 2.58	3762 233	Not significant
Length of stay in China	4 years 1 year 2 years 3 years	0.61 0.55 0.54 0.30	9.86 9.89 17 5.92	690 2185 630 490	Not applicable
Native language	Japanese Russian English Korean	0.60 0.55 0.42 0.42	19.6 8.85 9.73 7.69	647 1039 1151 1158	Not applicable
Proficiency level	High-intermediate Advanced	0.57 0.43	15.2 8.53	1193 2802	Not applicable
Total		0.06	10.5	4059		0.26	15	1848

Notes. Application value: Null form; p < 0.05. N = Number of tokens. All the factor groups were significant unless noted “not significant” or “not applicable”.

). The distribution (

Table 3. Distribution of object use (NS vs. CSL).

	NS		CSL
	N	% Null	N	% Null
Full NP	1368	74.02%	2789	68.7%
Pronoun	208	11.26%	850	20.9%
Null	272	14.72%	420	10.3%

) indicated that while the overall rate of full noun phrase use by NSs and learners was relatively similar, learners used pronouns at nearly twice the rate of NSs and used the null form less frequently than NSs. Thus, compared with their NS peers, learners tended to overuse pronominal object expressions and underuse null objects.

Object coreference and object person/number/animacy were the two main constraints on null object use by CSL learners. This finding corroborates previous research on subject pronoun use (Li, 2014). Specifically, for object use, the effect of animacy was stronger than that of object person/number. Referent specificity also significantly influenced object use, followed by discourse context, length of stay in China, native language, and proficiency. Sentence type and gender did not prove significant (Table 2).

The general patterns were as follow: (1) In terms of coreference, a fronted topic favored the null form the most, and under other circumstances, if there was some kind of change in object referent, position, or object surface form, overt objects tended to be used. (2) Inanimate objects were more likely to be used with null forms than animate objects. (3) Objects with specific referents favored null, while those with nonspecific referents favored overt forms. (4) Null objects were more likely to be used in conversations than in narratives. (5) The effect of length of stay in China on null object use showed a U-shape pattern, with 1–2 years slightly favoring null, 3 years disfavoring null, and 4 years favoring null again. (6) Japanese speakers favored null objects, followed by Russian speakers, while English and Korean speakers favored overt forms. (7) High-intermediate learners tended to use null objects more than advanced learners.

5. Discussion

5.1. Coreference and Object Person/Number/Animacy

As mentioned earlier, this is the first variationist study examining null object use by L1 and L2 Chinese speakers. Therefore, insights were drawn from previous studies of subject pronoun use. Coreference and subject person/number have been found to be the two main constraints for subject pronoun use in both L1 and L2 (e.g., Jia & Bayley, 2002; Li, 2014, 2017; Li & Bayley, 2018; Li et al., 2012; Guy et al., forthcoming). This study showed that they are also the two main constraints for object use in L1 and L2 Chinese.

In terms of the coreference effect, a fronted topic favored a null object the most for both learners and their NS peers. Fronting constructions, especially object fronting, are a prominent feature of Chinese (e.g., M. Li, 2018; Li & Thompson, 1981). When an object is fronted to the topic position, there is usually a null form in the object position. Therefore, it is not surprising that the fronted topic favored the null form the most. Under other circumstances, if there was some kind of repetition of the object referent, a null object was more likely to be used. If there was a change in the referent, position, or surface expression, an overt form was more likely to be used. For example, when an object occurred in the subject or other positions in the preceding clause, a null form was less likely to be used for CSL learners. In cases of “same underlying”, where the object surface expression was different from the one in the preceding clause, but the referents were the same, “partial overlap”, where two object referents partially overlapped, and “switch”, where two objects were completely different, the null form was least likely to be used. The learner pattern aligned with the NS pattern, indicating that the learners have grasped the overall grammar of null object use under various coreference conditions.

As for the other main constraint, object person/number/animacy, the results showed that animacy had a stronger effect on null object use than object person/number. Inanimate third-person objects, singular or plural, were more likely to be used with the null form by both learners and NSs. This finding echoes studies of subject use in L1 and L2 Chinese (Li, 2014; Li et al., 2012), where inanimate subjects were more likely to be associated with the null form. Therefore, what Li and Thompson (1981) stated about the tendency of third-person subject omission for nonhuman referents is also true for objects. Learners in this study also successfully grasped this pattern. On the other hand, the effect of person/number on null object use was mixed. Relatively speaking, learners tended to use the null form for third-person plural objects more than for third-person singular ones, but they used overt forms for both first- and second-person objects, singular and plural. NSs tended to use the null form for third-person inanimate objects and first-person objects, but they used overt forms for third-person animate objects and second-person objects. Comparing learner and NS patterns, the biggest difference lay in the use of null object for first-person objects, which favored the null form for NSs but the overt form for learners. Rechecking the data showed that learners tended to produce more utterances with simple structures and more rephrased sentences, explicitly stating the first-person objects, whereas NSs tended to use more complicated sentence structures with more subordinate clauses and topic chains, providing more opportunities for object omission.

5.2. Specificity

As mentioned earlier, studies have shown that objects with nonspecific referents are more likely to be used with null forms (e.g., Chang & Zheng, 2018; Choi, 1998, 2000; Morgan, 2004; Schwenter & Silva, 2002, 2003). Contrary to what the literature suggests, the results of this study showed that specific referents were more likely to be used with null objects than nonspecific referents, especially for learners. Looking further into the data (

Table 4. Distribution of object use in regard to referent specificity.

		NS		CSL
Factor		N	%	N	%
Specific	Full NP	301	49.9%	745	47.5%
	Pronoun	166	27.5%	573	36.6%
	Null	136	22.6%	249	15.9%
Nonspecific	Full NP	1067	85.7%	2044	82%
	Pronoun	42	3.4%	277	11.1%
	Null	136	10.9%	171	6.9%

), I found that learners and their NS peers had similar rates of using full NPs in both specific and nonspecific situations. However, learners used more pronouns and fewer null forms than NSs in both situations. In particular, NSs used more null objects in nonspecific referents than overt pronouns, but CSL learners showed the opposite pattern. Reexamining the data revealed the same patterns mentioned above: compared with NSs, learners tended to use more sentences with simple structures, rephrasing, and overt object pronouns.

5.3. Discourse Context

There were two discourse contexts in the data, which were interview conversations and elicited narratives. The results (Table 2) showed that conversation tended to favor null objects, but narrative favored overt forms.

Table 5. Distribution of object use in regard to discourse context.

	Conversation						Narrative
	N	Full NP	N	Pronoun	N	Null	N	Full NP	N	Pronoun	N	Null
NS	1071	72.6%	170	11.5%	235	15.9%	297	79.8%	38	10.2%	37	9.9%
CSL	2641	69%	771	20.2%	414	10.8%	148	63.5%	79	33.9%	6	2.6%

indicates that compared with NSs, learners’ full NP use was roughly similar to NSs in conversations, but less so in narratives. In both discourses, learners used more pronouns, especially in narratives, but fewer null forms.

Rechecking the data showed that NSs tended to use more full noun phrases in narratives to clearly describe who they were talking about and more complicated sentence structures with more topic chains that provided more opportunities to omit the objects. In contrast, learners seemed to have a smaller repertoire of full noun phrases for description in narratives, and they tended to use simpler sentence structures with more rephrasing and thus more explicit use of pronouns.

5.4. Effect of Length of Stay, L1, and Proficiency (CSL)

As to the effect of length of stay, the results (Table 2) showed that there was a curve: learners started out with more null object use, and if they stayed for about three years, they tended to use more overt forms, but if they stayed for four years, they went back to more null object use. In terms of the native language effect, Japanese and Russian speakers favored null object use, while English and Korean speakers favored overt forms. Proficiency was also shown to have a significant effect on learners’ object use. High-intermediate learners favored null objects, while advanced learners favored overt forms.

As mentioned earlier, L2 acquisition studies reported no developmental patterns for null object acquisition (Zyzik, 2008), and infrequent use of null objects is not a developmental feature (Chang & Zheng, 2018). However, L1 acquisition literature reported that children start out using more null objects, which later decreases over time (Larrañaga & Guijarro-Fuentes, 2012; Pérez-Leroux et al., 2008). Learners in this study showed a U-shaped pattern in regard to null object use. Learners who stayed in China for one or two years used more null objects, followed by a decrease in null object use among those who stayed for three years, with usage increasing again for learners who stayed for four years. This finding somewhat echoes L1 acquisition reports. However, it is important to note that the distribution of participants was not very balanced. Third-year participants were all English learners, while Russian and Korean learners were mostly first-year students. Japanese and Korean are null-object languages, and Russian also allows null objects in certain circumstances. Therefore, to tease out the nuances of the effects of length of stay, native language, and proficiency, and to gain a clearer picture of how they affect null object use, future studies with a more balanced, larger body of participants are needed.

5.5. NS vs. Learners

Generally speaking, learners’ patterns of object use, in terms of the relative likelihood of using explicit versus null forms in each factor group, aligned with NS patterns. This indicated that learners have acquired the underlying grammar of object use. For example, null objects tend to be associated with fronted topics, objects without a switch in referents, inanimate objects, objects with specific referents, and in conversations.

What is noticeable, though, is that the overall patterns of object use by NSs and CSL learners, as well as the results of each factor group, showed that learners, in general, used relatively fewer full NPs in object positions than NSs. In particular, compared with their NS peers, learners overused overt pronouns but underused null forms. This was true in almost all factor groups.

The primary reason for this, as mentioned earlier, is that, compared with NSs, learners’ utterances tended to have simple sentence structures with more rephrasing. They were also more likely to use more overt forms of pronouns in objects, which indicated that they are more explicit in terms of what they talk about and to whom they refer. This can be demonstrated by cases where NSs exhibited more variation, while learners used explicit forms obligatorily. For example, in the factor group of coreference, NSs used null forms 16.7% of the time in “same underlying” situations, where the referents of adjacent objects were the same, but the surface forms were different, whereas learners used explicit forms obligatorily. Another example is the first-person plural form. NSs had almost 19% null object use, but learners used explicit forms exclusively for all 40 tokens of first-person plural cases.

This finding corroborates previous literature that showed learners’ tendency to overuse explicit forms of optional language features. Li (2010, 2014) reported that CSL learners used the morphosyntactic particle DE and subject pronouns much more frequently than NSs. Li suggested two reasons for learners’ overuse of target language forms. One is that learners tend to land on the side of clarity in optional cases of language use. The other is the influence of the input they received from their teachers and textbooks.

According to Rehner’s (2005) proposal of three criteria for learners’ success in acquiring Type 2 variation, i.e., same expressions, similar frequency, and similar constraints, these learners satisfied two of them. They were able to use the same expressions with similar constraints as NSs, but the learners’ frequencies of the use of different forms did not align with those of NSs. Therefore, the results of this study agree with Li (2010) in that learners tended to prioritize clarity when it came to object use and were cautious about using null forms. This suggests that learners need time to become more proficient in the nuances of object use and to further develop their ability to adapt their style in object omission. In addition, learners need time to develop more advanced proficiency in managing more complicated sentence structures that provide more opportunities for topic chains and object omission.

6. Conclusions

In general, object use in L2 Chinese is systematic and subject to different linguistic and social constraints. A multivariate analysis of NS and CSL learner natural speech data showed that learners’ patterns were similar to those of their NS peers in most dimensions explored, except that they tended to overuse overt personal pronouns and underuse null objects. Specifically, coreference and object person/number/animacy were the two main constraints of object use. Other significant factors included referent specificity, discourse context, length of stay in China, learners’ L1, and proficiency levels. The results indicated that the learners had acquired the object use pattern in Chinese rather successfully, but still needed more work on acquiring more complex sentence structures and the nuances of using pronouns and null forms to further develop their sociolinguistic competence.

Bridging significant gaps in the literature, this study contributes to the fields of sociolinguistic variation, SLA, and CSL acquisition. The study extended variationist sociolinguistics to an underexplored linguistic feature—null object use in Chinese—where prior research was largely absent. Additionally, it adds to the growing body of work on how non-native speakers acquire and use stylistic variation, deepening the understanding of learner grammar in interlanguage variation. By demonstrating the interplay of various linguistic and social constraints, the study provided insights into how CSL learners acquire the subtle nuances of object omission, which is a challenging aspect of learning and using Chinese. These findings not only advance knowledge within the variationist paradigm and SLA but also offer practical insights into the challenges learners face in acquiring sociolinguistic competence in a L2. Future research can build on these findings to further explore the acquisition of stylistic variation in other underexplored linguistic features and languages.

However, this study is not without limitations. As mentioned earlier, the learner participant group was not balanced in terms of their native language, length of stay, and proficiency level. Future studies should aim to include a larger and more balanced participant group, accounting for these three factors. For example, each group of learners should have a balanced number of participants with varying lengths of stay and proficiency levels. It should also be noted that the length of stay and proficiency might interact. Sometimes, a learner with a shorter stay has a higher proficiency level than one with a longer stay. Therefore, careful planning is needed, although finding enough participants with all the needed variables might be a challenge.

Since this is the first variation study of object use in L2 Chinese, further studies along this line are certainly needed for a deeper understanding of this topic. Possible future directions include exploring more linguistic and social constraints in L2 object use, investigating object use by L3 learners, and comparing subject and object use by L2 learners.