Age effects on L2 grammars: Evidence from Cantonese learners of Mandarin

Abstract

Based on the Critical Period Hypothesis (CPH), numerous studies have investigated the relationship between age of acquisition (AoA) and second language (L2) ultimate attainment. However, it is not yet clear when is the end of the critical period (CP) and how AoA affects L2 attainment. Different termini of CP have been identified in literature, and the realisation of AoA effects on L2 attainment is also under debate. This study critically reviewed the controversial issues concerning the CPH and analysed production data of Cantonese-speaking learners of Mandarin with piecewise regression modelling. The data were elicited production of Mandarin ba-sentences, which possess a pre-transitive construction with a unique syntactic structure and several semantic constraints. The data showed that regression models with various breakpoints were not better than a simple linear regression model, based on which we rejected the discontinuous decline hypothesis and provided evidence against the CPH. Also, our results indicated a tendency that with the increase of AoA, L2 ultimate attainment would become more divergent from target language grammar. A further comparison did not display qualitative differences in Cantonese learners’ acquisition of two types of ba-construction. Moreover, the Cantonese learners acquired the ba-construction more successfully than learners with other language backgrounds, which might be attributed to the similarities between Cantonese and Mandarin.

Full text

Yang, Y. (2020). Age effects on L2 grammars: Evidence from Cantonese learners of
Mandarin. In D. Chen & D. Bell (Eds.), Explorations of Chinese Theoretical and
Applied Linguistics (pp. 220-252). Newcastle upon Tyne: Cambridge Scholars
Publishing. Copyright belongs to Cambridge Scholars Publishing.
CHAPTER NINE
AGE EFFECTS ON L2 GRAMMARS:
EVIDENCE FROM CANTONESE LEARNERS OF MANDARIN
YIKE YANG
Abstract
Based on the Critical Period Hypothesis (CPH), numerous studies have investigated the relationship
between age of acquisition (AoA) and second language (L2) ultimate attainment. However, it is not
yet clear when is the end of the critical period (CP) and how AoA affects L2 attainment. Different
termini of CP have been identified in literature, and the realisation of AoA effects on L2 attainment
is also under debate. This study critically reviewed the controversial issues concerning the CPH and
analysed production data of Cantonese-speaking learners of Mandarin with piecewise regression
modelling. The data were elicited production of Mandarin ba-sentences, which possess a pre-
transitive construction with a unique syntactic structure and several semantic constraints. The data
showed that regression models with various breakpoints were not better than a simple linear
regression model, based on which we rejected the discontinuous decline hypothesis and provided
evidence against the CPH. Also, our results indicated a tendency that with the increase of AoA, L2
ultimate attainment would become more divergent from target language grammar. A further
comparison did not display qualitative differences in Cantonese learners’ acquisition of two types
of ba-construction. Moreover, the Cantonese learners acquired the ba-construction more
successfully than learners with other language backgrounds, which might be attributed to the
similarities between Cantonese and Mandarin.
Keywords: second language acquisition; age of acquisition; ultimate attainment; critical period
hypothesis; pre-transitive construction

1. Introduction
In the context of typical language development, every child can successfully acquire their first
language (L1) in a fast and seemingly effortless manner (Clark 2009; O’Grady and Cho 2001).
However, it has proved to be a painful journey for adolescents and adults to learn a second language
(L2), the outcome of which is always divergent in many aspects from L1 acquisition (Lardiere 2007;
Meisel 2011; Sorace 2003; White 2003; Yuan 2010). Tremendous research has thus focused on the
effects of age of acquisition (AoA) on L2 acquisition, particularly on L2 ultimate attainment,
because such evidence can provide “upper limits” of L2 acquisition (Birdsong 2006, 10). This
section starts with an introduction of related issues and the pre-transitive constructions in Mandarin
and Cantonese, and then raises our research questions based on the gaps in the field.
1.1 Age Effects on L2 Attainment
Ultimate attainment of an L2 refers to the “steady-state grammars of people who have completed
their L2 acquisition” (White 2003, 241). L2 learners are generally found different from native
speakers in ultimate attainment of a target language (Sorace and Filiaci 2006; Birdsong 2004; Moyer
1999; Johnson et al. 1996; Bley-Vroman 1990; Coppieters 1987). Extensive studies have
investigated the relationship between AoA and L2 ultimate attainment since the Critical Period
Hypothesis (CPH) was put forward decades ago (Lenneberg 1967; Penfield and Roberts 1959).
According to CPH, there exists a certain critical period (CP) for language acquisition, after which
the ability to acquire a language disappears or declines gradually. It is suggested that if L2
acquisition starts early (within the critical period), then the attainment of nativelikeness in an L2 is
possible (Birdsong and Vanhove 2016). There are, nonetheless, several controversial issues
concerning the CPH, as reviewed below.
Assuming the existence of CP, the first question one would ask is when is the end of it. Different
termini of CP have been proposed by researchers (See Singleton (2005) for a summary). In an earlier
review, Hyltenstam and Abrahamsson (2003) proposed that the ability to learn an L2 starts to decline
shortly after birth. Research on English-speaking infants’ perception of native and foreign speech
revealed that infants can discriminate phonetic contrasts across languages from six months onwards,
but their ability to discriminate phonetic contrasts in a foreign language declines by their first year
of life (Werker and Tees 1984; Werker et al. 1981). Similarly, even L2 speakers with AoA of one
year old were perceived differently from native speakers in pronunciation (Flege, Yeni-Komshian,
and Liu 1999). Cheng and Tang (2016) studied the acquisition of Mandarin semantic operators by
Cantonese learners, and their results showed that learners with AoA before age three outperformed
learners with AoA after age six. Another study on English syntactic knowledge found that

participants with AoA before age seven converged to native speakers in a test and those with AoA
after age seven performed poorly in the test (Johnson and Newport 1989). Oyama (1976) provided
evidence for successful acquisition of L2 English phonological system by children, and her data also
verified that those who arrived in an L2 environment after age 12 were perceived as foreign-accented.
Based on their findings, Granena and Long (2013) proposed multiple CPs for phonology, lexis-
collocation and morphosyntax. In addition to spoken languages, the AoA effects have also been
supported by data from sign languages. For instance, native signers of American Sign Language
(ASL) were fastest in processing ASL lexical items in a primed lexical decision task, and early
learners also had shorter reaction time than late learners (Mayberry and Witcher 2005).
Another issue under debate is the realisation of AoA effects on L2 ultimate attainment. Having taken
the characteristics of the CP into account (Bornstein 1989), Birdsong (2006, 2009) summarised three
patterns of possible AoA effects on L2 ultimate attainment. Figure (9-1a) represents a stretched “L”
shape, where the L2 attainment declines as AoA increases, and such decline ceases at the end of the
CP. The left portion of this pattern, namely the decline of L2 with the increase of AoA, has been
supported in previous studies (DeKeyser and Larson-Hall 2005), but the right portion, namely the
floor effect on L2, which exhibits continuous performance at the lowest limit (Hessling, Schmidt,
and Traxel 2014), has not been demonstrated. Contrary to Figure (9-1a), Figure (9-1b) indicates a
stretched “7” shape, which shows peak attainment within the CP and a gradual decline of attainment
after the CP. Such a ceiling effect (Hessling, Traxel, and Schmidt 2004) and a gradual decrease
afterwards were found in Veríssimo et al. (2017) where participants with AoA before age five
exhibited nativelike performance and participants with AoA after five displayed a decrease in the
performance with the increase of AoA. The “Z” shape in Figure (9-1c) is a combination of the
previous two shapes, which consists of three developmental stages. A ceiling effect is hypothesised
at the beginning, which reveals consistent peak attainment, followed by a gradual decrease of L2
attainment during the CP. When the CP ends, a floor effect appears, which is similar to the right
portion of the stretched “L” shape.
Figure 9-1. AoA effects on L2 attainment (Adapted from Birdsong 2009, 416)

Other studies on the AoA effects, however, suggested that such discontinuity does not exist, and
there might be a linear relationship between AoA and L2 attainment in the acquisition process,
providing evidence against the CPH (Birdsong and Vanhove 2016). Hakuta, Bialystok and Wiley
(2003) analysed census data of more than two million immigrants in the United States to test the
CPH in L2 acquisition. They set ages 15 and 20 as two hypothesised cut-off points for CP, but they
did not find such discontinuity in their analysis. What their data showed was a more linear
relationship between AoA and degree of L2 success. Although the census-derived approach has been
criticised by linguists (Muñoz and Singleton 2011; Long 2013), there are recent empirical studies
that provide evidence against the discontinuous decline hypothesis. In a study on the L2 acquisition
of information structure in French, Reichle (2010) observed a gradual decrease in performance with
the increase of AoA till adulthood, suggesting that age effects exist even after puberty and the CPH
is not favoured. More recently, data from a study on Japanese learners’ English pronunciation also
revealed a negatively linear relationship between AoA and L2 pronunciation in a group of
participants with AoAs from 16 to 40 (Saito 2015).
1.2 Pre-transitive Constructions in Mandarin and Cantonese
The pre-transitive construction refers to the result of changing the subject–verb–object (SVO)
sequence into the subject–object–verb (SOV) sequence, and is regarded as a “language-specific but
dialect-universal feature in Chinese grammar” (Cheung 1992, 241). In Mandarin, the pre-transitive
ba-construction is widely used and has been extensively discussed in literature, but there is still no
consensus on the nature of ba-construction among linguists (Ding 2001; Zou 1993; Cheng 1988; Li
and Thompson 1981; Li 1924/2007; Teng 1975; Wang 1943/1985). The canonical word order in
Mandarin is SVO as is shown in (1a) (Sun and Givón 1985). The similar meaning can also be
expressed with the pre-transitive ba-construction in (1b), the syntactic order of which is different
from the canonical SVO order, and the object chuanghu “window” precedes the verb guanshang
“close”. The ba-construction provides an alternate form where the object of the main verb occurs
right after ba and before the verb (and hence the “pre-transitive”). In general, the word order of a
typical ba-sentence can be represented in (2), although there are variations. The ba-nominal phrase
(ba-NP) is the noun phrase immediately following ba, and the ba-verb phrase (ba-VP) refers to the
constituent of “verb + others”, the characteristics of which will be introduced shortly in this section.
Ba-construction was firstly analysed by Li (1924/2007) in his monograph, according to which the
function of ba is to move the object forward and place it in front of the verb. Henceforth, the ba-
construction has been studied from various perspectives (See Bender (2000) for a lexical functional
grammar approach, Huang, Li, and Li (2009) for a generative approach, Liu (1997) for an aspectual
approach, Lv (1955/2002) for a disposal approach, Mullie (1932) for an accusative approach,

Sybesma (1999) for a syntactic-semantic approach, Thompson (1973) for a transitivity approach,
and Tsao (1987) for a topic-comment approach).
(1) a. ta guanshang le chuanghu
He close ASP window
“He closed the window.”
b. ta ba chuanghu guanshang le
he ba window close ASP
“He closed the window.”
(2) Subject + ba + object + verb + others
Apart from the distinction in word order, there are some constraints on ba-construction as well. The
constraint on the ba-NP is that the NP must be either definite or generic (Mullie 1932; Liu 1997).
(3a) and (3b) are almost the same except for the determiners preceding the Classifier Phrases. (3a)
is not acceptable in Mandarin because yiliangche “one car” is an indefinite phrase, which refers to
an entity (a car) that the listener does not know about previously. Meanwhile, (3b) is fine because
zheliangche “this car” is a definite phrase referring to an entity (a certain car), the existence of which
is shared by both the speaker and the listener. This illustrates the constraint of definiteness on the
ba-NP. Unlike a canonical SVO sentence, an indefinite NP is not acceptable in ba-construction. The
object yan “salt” in (3c), however, is not a particular entity that the speaker and the listener share;
rather, it is a generic concept representing a class or a type of entities. The generic ba-NP may be
acceptable under some circumstances.
(3) a. *ta ba yi liang che mai le
he ba one CL car buy ASP
“He bought a car.”
b. ta ba zhe liang che mai le
he ba this CL car buy ASP
“He bought this car.”
c. ta youshihou ba yan dang tang chi
he sometimes ba salt take sugar eat
“He sometimes eats salt thinking it is sugar.”

(Adapted from Li and Thompson 1981, 464-466)
Wang (1943/1985) put forward the notion “disposal”, which means that the verb in a ba-sentence
must be dealing with something, and this view has been supported by many linguists. For example,
(4a) is not a good sentence because the verb kandao “see” does not make any change to or have any
effect on the object “Zhangsan”. The constraint on the ba-VP is that a bare verb is not allowed, and
the ba-VP needs to be morphologically complex, whether it is: a) a resultative verbal complement
(RVC) as in (4b), b) a verb plus an aspect marker as in (4c), or c) a verb plus a locative prepositional
phrase (PP) as in (4d). Moreover, Feng (2001) argued that ba-construction might be under some
kind of prosodic constraint, and Yang (1998a, 1998b) advocated that ba-construction must satisfy
the feature of [+telic] and [+perfective], so the VP needs to be complex to indicate the telicity of the
activity.
(4) a. *ta ba Zhangsan kandao le
He ba Zhangsan see ASP
“He saw Zhangsan.”
b. kuai yidian ba zhe kuai rou na zou
quick a little ba this CL meat take go
“Take this piece of meat away quickly!”
c. wo ba wode qiche mai le
I ba my car sell ASP
“I sold my car.”
d. ta ba biaoyu tie zai qiang shang
he ba slogan paste at wall on
“He posted the slogan on the wall.”
(Adapted from Li and Thompson 1981, 463-485)
Lv (1994) collected 1,094 ba-sentences from a pile of linguistic data containing 530,000 sentences,
classified them into six semantic types, and calculated the proportion of each type. The disposal and
the displacement types were found to be the most frequently used ba-sentences in native speakers’
daily lives, with the former constituting 49.8% of all the speakers’ sentences (546 out of 1094) and

the latter 27.8% (305 out of 1094)
1
. Based on Lv’s classification, this study investigates these two
types of ba-construction: the disposal ba-construction as illustrated in (4c) and the displacement ba-
construction as illustrated in (4d). The ba-NPs in the disposal type is affected by the action as “being
sold” in (4c), and the ba-NPs in the displacement type undergo locational displacement as “being
posted on the wall” in (4d).
Although Mandarin and Cantonese belong to Sino-Tibetan languages and they share the same word
order of SVO, there are some differences between them. Unlike Mandarin, there is no ba-
construction in Cantonese, but there is zoeng-construction as exemplified in (5), the usage of which
is very limited.
(5) zoeng nei ze go bun syu bei ngo
zoeng you borrow DET CL book give me
“Give me the book that you borrowed.”
(Cheung 2006, 55)
Regarded as the counterpart of Mandarin ba-construction (Tang 2002), zoeng-construction shares
similar syntactic properties with ba-construction, but it can only be used for the disposal cases, and
occurs in more formal situations while rarely in colloquial contexts (Cheung 1992; Leung 2004). In
a comparative study of ba-construction and zoeng-construction, Li (1993) found several differences
between them. For instance, ba takes all kinds of nominals as subjects, but zeong can only take
[+animate] agents as subjects; zeong only allows [+definite] objects while ba also accepts [+generic]
objects like (3c). Still, the Cantonese zoeng-construction may have some effects on the interlanguage
of Cantonese learners when they learn Mandarin.
1.3 The Current Study
As reviewed in Section 1.1, controversies exist in the literature of age effects on L2 ultimate
attainment. CP was found in some studies but not in others. Within the studies that did support the
CPH, different offsets of CP were proposed. Meanwhile, it is not yet clear whether age would
influence language acquisition ability throughout the life span. To fill these gaps, the present study
attempts to address the following research questions:
1
The other four semantic types of ba-sentences and their frequency identified in Lv (1994) are: connectionalised ba-
sentences (8.4%), equative ba-sentences (6.3%), adverse ba-sentences (5.1%) and causative ba-sentences (1.5%).

1) Is a certain CP observable in our data? Following previous studies (e.g. Johnson and Newport
1989), we chose ages six, eight and nine as hypothesised offsets of CP to see if there exists a CP
for L2 acquisition.
a. if our data support CPH, which pattern of the CP effects (Figure 9-1) is supported?
b. if our data do not support CPH, is there a linear relationship between AoA and L2
attainment?
2) Is there any difference in age effects on the acquisition of two types of ba-construction?
3) What are the characteristics of the ba-sentences produced by the Cantonese speakers?
2. Methodology
The data were collected in a previous study on the L2 acquisition of the Mandarin ba-construction
(Yang 2013, 2020). Section 2.1 and Section 2.2 briefly describe the participants and the experiment,
while Section 2.3 reports how the data were analysed in the current study.
2.1 Participants
Fifty participants were recruited from the Chinese University of Hong Kong. Written informed
consent was obtained from all the participants prior to data collection. The target group consists of
30 Cantonese-speaking undergraduates who were born and brought up in Hong Kong. A background
questionnaire was completed before the experiment to assess the participants’ language background
and guarantee their proficiency in Mandarin. To test the proficiency level of Mandarin, an
independent test of Mandarin proficiency was administered at the end of each experiment. The
highest possible score for the independent test was 20, and participants’ scores ranged from 17 to
20 (M = 18.97), from which we can conclude that the Mandarin proficiency of the participants was
at the advanced level. The summary of the participants’ information has been listed in Table 9-1.
The Cantonese learners were aged from 18 to 22, and their AoA of Mandarin ranged from 3 to 18.
Table 9-1. Background information on the participants
Group
No of participants
(Male/Female)
Age
AoA
Place of birth
and growth
Score in the test of
Mandarin proficiency
Can
30 (14/16)
18 - 22 (M = 19.7)
3 - 18 (M = 8.67)
Hong Kong
17 - 20 (M = 18.97)
Man
20 (9/11)
20 - 25 (M = 23.4)
N/A
Northern China
N/A
Note: Can = Cantonese-speaking learner group; Man = Mandarin-speaking control group

Twenty native speakers of Mandarin aged from 20 to 25 were recruited to form a control group.
Data collected from them would provide a baseline for obligatory or preferred contexts of ba-
construction from the native speakers’ perspective. Because the Chinese spoken in Northern China
is closest to Standard Mandarin, only those who were born and brought up in Northern China were
invited to participate.
2.2 Materials and Procedure
A picture-elicited production test was designed to collect ba-sentences from both groups of
participants. Because ba-construction always describes the change of state, three graphs were
prepared as a set to elicit an utterance of ba-construction. There were 24 originally designed sets of
pictures in total. Among them, 12 describe one situation that involves either the case of disposal
(Figure 9-2) or the case of displacement (Figure 9-3) as target sentences, and the other 12 are fillers
(Figure 9-4).
Figure 9-2. An example of stimuli for target sentences (disposal)
Target sentence: ta ba heiban ca ganjing le “She cleaned the blackboard.”
Figure 9-3. An example of stimuli for target sentences (displacement)
Target sentence: ta ba shubao fang dao le zhuo shang “She put her schoolbag on the table.”

Figure 9-4. An example of fillers that cannot elicit ba-sentences
This filler sentence has no meaning related to disposal or displacement, and the main verb, mengjian “to
dream of”, rarely occurs in native speakers’ ba-sentences.
In the task, each participant was asked to describe the pictures with one sentence in Mandarin. If the
participants failed to express themselves clearly, the investigator would ask them to repeat their
sentence or describe the picture in another way. The entire task was audio-recorded to guarantee that
the answers were recorded accurately. The recordings were then manually transcribed to text for
further analysis. A practice session was conducted before the test to ensure that the participants fully
understood the instructions.
2.3 Data Analysis
Two general approaches have been the norm in the analysis of age effects on L2 attainment. One of
them is to divide the participants into two or more groups according to their AoAs and compare the
means and proportions of the groups, and sometimes the comparison of correlation coefficients is
also included (Johnson 1992; McDonald 2000). Another approach is to compute regression models
with AoA as the predictor variable and L2 attainment as the response variable (Birdsong and Molis
2001; Granena and Long 2013). Vanhove (2013) critically reviewed these approaches and
reanalysed the data from DeKeyser, Alfi-Shabtay, and Ravid (2010) using piecewise regression
modelling. Also known as segmented or multi-phase regression, piecewise regression is used when
the effect of the predictor variable on the response variable changes abruptly (Muggeo 2003), which
is an appropriate tool for testing CPH because it allows the slope of the predictor to change at a
certain point. The equation of a piecewise regression model can be found in Equation (1)
2
, where y
is the response variable, which is a function of an overall intercept α and two slope parameters β1
and β2, depending on whether xi lying below or above ψ, the breakpoint to be modelled (Baayen
2008; Vanhove 2013).
2
Equation (1):      
    

Following Birdsong and Molis (2001) and Vanhove (2013), the current study fitted piecewise
regression models with R (R Core Team 2017) in R Studio (R Studio Team 2016) to tackle our
research questions. The R package “stats” (R Core Team 2017) was used for the modelling and the
package “ggplot2” (Wickham 2016) was used for the visualisation. We first fitted a regression model
with no breakpoint. We then fitted three piecewise regression models with ages six, eight and nine
as the breakpoints, and compared them with the previously fitted regression model. If none of these
breakpoints were better than the fitted model, we would compute models with varying breakpoints
to see if we could find an optimal breakpoint with our data. If we still could not find an optimal
breakpoint, we would then adopt the first linear regression model instead. To answer our second
research question, we further compared the acquisition of the two types of ba-construction with
regression models.
Furthermore, qualitative analysis was also included in this study, where two trained Mandarin-
speaking linguists (including the first author) manually coded the properties
3
of each ba-sentence
and the error type if it was ill-formed, and this would allow us to confirm whether the Cantonese
learners have difficulty acquiring the constraints on ba-construction. In addition, we also coded the
non-ba-sentences produced by both groups to figure out what other structures were adopted to
substitute for ba-construction. Both coders did the coding independently and then compared all the
judgements to ensure coder reliability (Révész 2012). When there was inconsistency between the
two coders, further discussion with other linguists was required to make the final decision.
3. Results and Discussion
An overview of the production data can be found in Figure 9-4. It is clear that both groups produced
a considerable amount of ba-sentences, accounting for the majority of all the sentences (252 out of
360 for the Cantonese learner group and 212 out of 240 for the Mandarin control group). An
independent-samples t-test was conducted to compare the production of ba-sentences and non-ba-
sentences by both groups of participants. There was a significant difference between the two groups
(t(596.486) = -5.751, p < .001), which suggested that the Mandarin speakers outperformed the
Cantonese learners in the production. This section reports the results in detail and raises issues for
further discussion.
3
Several properties were considered in our analysis: types of ba-NP and ba-VP, animacy of subject and ba-NP, the
telicity of the ba-sentence, etc.

Figure 9-4. An overview of the production
Note: Type 1 = the disposal type; Type 2 = the displacement type
3.1 Testing the CPH
A simple linear regression model with no breakpoint was fitted to predict the production of ba-
sentences based on the AoA. A significant regression equation was found (F(1, 28) =4.423, p = .045),
with an R2 of .136
4
. Participants’ expected production of ba-sentences is equal to 11.030 – 0.304
(AoA) when AoA is measured in years. The model was plotted in Figure 9-5. There is an upward
curve in the middle of the line at around age 11, which was difficult to explain. A possible reason is
that our sample size may not be sufficient, so with some outstanding performers at around age 11,
the curve exhibited something unusual. We replotted the regression model without the smoother in
Figure 9-6, from which a linear relationship can be observed.
Figure 9-5. AoA and production of ba-sentences
Note: The trend line is a non-parametric scatterplot smoother.
4
The raw data and R scripts have been uploaded to the author’s GitHub:
https://github.com/yikeyang/age_effect_piecewise_reg

Figure 9-6. AoA and production of ba-sentences (replotted without smoother)
For models with breakpoints, we first fitted a linear regression model with the breakpoint at age six.
No significance was found in the regression (F(2, 27) = 2.151, p = .136), with an R2 of .136.
Moreover, no significance was reached for each age range: p = .538 for AoA < 6 and p = .145 for
AoA > 6, respectively. The model was then plotted in Figure 9-7. The solid line represents the model
with the breakpoint and the dashed lines indicate its 95% confidence interval. The dot-dash line
represents the regression model without breakpoint, which falls well within the 95% confidence
interval of the model with the breakpoint. An analysis of variance (ANOVA) also suggested no
statistically significant differences between the two models (F(28, 27) =.033, p = .859). Thus, our
data do not support the breakpoint at age six.
Figure 9-7. Model with breakpoint at age six
Note: Solid line: regression with breakpoint at age six. Dashed lines: 95% confidence interval of the model.
Dot-dash line: regression without breakpoint.

We then fitted linear regression models with the breakpoints at ages eight and nine. Neither of them
proved better than the regression model without breakpoint: F(2, 27) = 2.280, p = .111, R2 = .150
for the regression with breakpoint at eight; and F(2, 27) = 2.251, p = .125, R2 = .142 for the regression
with breakpoint at nine. The models were plotted in Figures 9-8 and 9-9. Again, the regression
without breakpoint is within the 95% confidence intervals of both models.
Figure 9-8. Regression model with breakpoint at age eight
Note: Solid line: regression with breakpoint at age eight. Dashed lines: 95% confidence interval of the
model. Dot-dash line: regression without breakpoint.
Figure 9-9. Regression model with breakpoint at age nine
Note: Solid line: regression with breakpoint at age nine. Dashed lines: 95% confidence interval of the
model. Dot-dash line: regression without breakpoint.
So far, our results did not favour any of the hypothesised cut-off points for the CP. To exhaust all
the possibilities, we fitted regression models with varying breakpoints (from age three to age 18, the
range of our AoAs) to determine an optimal breakpoint. Following Vanhove (2013), we calculated

d2, namely, the sum of the squared differences between the actual data points and the values
predicted by the model, with Equation (2)
5
. The model with the smallest d2 would be the optimal
model. The breakpoints at ages 16 and 17 were found to have the smallest values. But there are two
reasons why we would not consider them as the cut-off points. First, we only have two participants
with AoAs above 16 (one aged 16 and another aged 18 at the time of the test), so it is impossible for
us to explore whether having age 16 or age 17 as a breakpoint would contribute to a better regression
model for AoA and L2 attainment with our data. More importantly, all theories of CPH assume the
CP ends by puberty. Failure to fit a model with a breakpoint age before puberty serves as
counterevidence against the CPH.
To sum up, regression models with various breakpoints were computed, and the optimal one was
the simple linear regression model in Figure 9-6, based on which we rejected the discontinuous
decline hypothesis and provided evidence against the CPH. As indicated by Figure 9-6, L2 ultimate
attainment and AoA are negatively correlated (p = .045), although the relationship is not very strong
(R2 = .136).
3.2 The Acquisition of Two Types of ba-construction
We then explored the acquisition of two types of ba-construction and age effects, if any. As
suggested in Figure 9-4, both groups uttered ba-sentences more frequently under the disposal
conditions. Paired-samples t-tests were employed to compare the two conditions for each group,
which revealed a significant difference for both groups: t(179) = -2.484, p = .014 for Cantonese
learners, and t(119) = -3.636, p < .001 for Mandarin speakers. The preference for the disposal ba-
construction might be explained by the disposal meaning’s being an obligatory factor for both the
Mandarin ba-construction and the Cantonese zoeng-construction.
5
Equation 2: 



Figure 9-10. Models for AoA and production of disposal type
Note: Solid line: regression with breakpoint at age eight. Dashed lines: 95% confidence interval of the
model. Dot-dash line: regression without breakpoint.
Similar procedures were used in analysing the data. We first fitted regression models with various
breakpoints (from age three to age 18, the range of our AoAs) to determine an optimal breakpoint
for the acquisition of the disposal type. According to the values of d2, age eight exhibited the least
variation. We then fitted a regression model with age eight as the breakpoint, but no significance
was found (F(2,27) = 1.662, p = .209, R2 = .110; p = .224 before age eight and p = .675 after age
eight). The models with and without a breakpoint were plotted in Figure 9-10, and the model without
any breakpoint fall within the 95% confidence intervals of the model with a breakpoint at age eight.
An ANOVA comparing two models also suggested no significance between them. For the
displacement case, the models with ages 15 and 16 showed the smallest differences. As previously
discussed, we would not consider having age16 as the optimal model. Nor did we have a participant
with an AoA at 15. So we adopted the linear regression model without any breakpoint as the optimal
one for displacement case, where F(1, 28) = 5.285, p = .029, R2 = .159. The optimal models for each
case were plotted in Figures 9-11 and 9-12.

Figure 9-11. AoA and ba-sentences of disposal
Figure 9-12. AoA and ba-sentences of displacement
To further compare the acquisition of the two types of ba-sentences, we employed a paired samples
t-test to examine the production data under two conditions. There was no significant difference in
the production for the disposal type (M = 4.3, SD = 1.745) or the displacement type (M = 4.1, SD =
1.470; t(29) =-1.030, p = .312). A very strong correlation was found between the two variables
(F(1,29) = 47.856, p < .001, R2 = .631), suggesting that the learners actually acquired the two types
of ba-construction without qualitative differences.
3.3 Qualitative Analysis
In this part, we first analysed the ba-NPs and ba-VPs to examine the production data in detail. Figure
9-13 presents the proportion of different ba-NPs produced by the two groups, from which we can
find a similar distribution pattern of the two groups’ data. More than half of the ba-NPs were bare

nouns as in (6a), which normally refer to the objects already known by both the speaker and the
listener, so bare nouns are well in line with the [+definite] constraint on the ba-NP. The two groups
also adopted locative DPs as in (6b), numerical phrases as in (6c) and relative clauses to modify the
head noun as in (6d). Besides, participants in both groups would replace the nouns by pronouns in
multiclausal ba-sentences as in (6e).
Figure 9-13. Types of ba-NPs in the ba-sentences
Note: Loc DP = locative DP; Poss DP = possessive DP; Pron = pronoun; Num + Cl + N = number +
classifier + noun; De + Cl + N = determiner + classifier + noun; RC = relative clause; A + N = adjective +
noun
(6) a. xiaonanhai ba maozi dai zai le tou shang
child ba hat put on at ASP head above
“The child put on his hat.”
b. xiaoming ba zhuo shang de pingguo chi diao le
Xiaoming ba table above de apple eat drop ASP
“Xiaoming ate up the apple on the table.”
c.mama ba yi fu hua gua zai le qiang shang
mother ba one Cl picture hang at ASP wall above
“The mother hung a picture on the wall.”
d. you ge ren ba ta kandao de tupian
have Cl person ba she see de picture

gua shang qiang shang
hang to wall above
“A girl hung the picture she saw on the wall.”
e. zhuo shang you yi ge pingguo, xiaonanhai
table above have one CL apple child
ba ta chidiao le
ba it eat drop ASP
“There was an apple on the table, and a child ate it up.”
Figure 9-14 shows the types of ba-VPs produced by the two groups. Of all the types, “RVC(+le)”
as in (7a) was the most frequently used type of VP, followed by “V+P(+le)+L”, “V+P(+le)+N” and
“V+le+(C)” as shown in (7b-d) respectively. These four types constitute 95.07% of all the ba-
sentences (95.26% for the L2 group, and 94.86% for the L1 group), as they clearly indicate the
results or changes of the objects, which fits the semantic and pragmatic components of ba-
construction. Thus, echoing with previous studies (Huang and Yang 2004, 2005), most of the ba-
sentences produced by our participants perfectly met the requirement of the [+telic] feature. A closer
look at the ba-NPs and ba-VPs suggested that the L2 learners had acquired the constraint on the ba-
NPs and ba-VPs and were able to produce the correct ba-sentences consciously or unconsciously.
Figure 9-14. Types of ba-VPs in the ba-sentences
Note: RVC = resultative verbal complement; V + P + L = verb + preposition + location; V + P + N = verb +
preposition + noun; V + le + C = verb + le + complement

(7) a.xiaonanhai ba pingguo chi diao le
child ba apple eat drop ASP
“The child ate the apple.”
b. wo ba beibao fang dao le zhuozi shang
I ba bag put arrive ASP table above
“I put the bag on the table.”
c. laoshi ba zuoye fagei le tongxuemen
teacher ba assignment send le students
“The teacher returned the assignments to the students.”
d. xiaopengyou ba yumaoqiu na le chulai
child ba shuttlecock take le out
“The child took out a shuttlecock.”
We also examined the production data to see if there were ill-formed sentences. There were 29
sentences identified as ill-formed, but only three of them failed to meet the requirements of ba-
construction. Specifically, these sentences were formed with bare verbs as in (8a), which did not
indicate the telicity of the action “discuss”, thus violating the constraints on the ba-VPs. We did find
a sentence with a wrong ba-NP, but the error did not concern the constraints on the ba-NPs. As can
be seen in (8b), the oddness was caused by the unusual noun shengliang “volume”, which is not
commonly used in Mandarin. The most frequently occurred error in our data was word collocation,
which accounted for the majority of the errors (19 out of 29). For instance, the verb zhan “cut” in
(8c) does not collocate with shu “tree” in Mandarin. Other errors included code-mixing as in (8d)
and missing elements as in (8e).
(8) a. *tongxuemen ba laoshi fa de zuoye taolun
students ba teacher return de assignment discuss
“The students discussed the assignment returned by theteacher.”

b. *ta ba ta de shengliang jiangdi
she ba it de volume turn down
“She turned down the volume.”
c. *you yi ge ren ba shu zhan le
have one CL man ba tree cut ASP
“A man cut down the tree.”
d. *ta ba zuoye delete diao le
she ba assignment delete fall ASP
“She deleted her assignment (on the blackboard).”
e. *ta ba yumaoqiu cong yi ge tong (li) na chulai
she ba shuttlecock from one CL barrel (in) pick out
“She took a shuttlecock out of the barrel.”
Of all the 252 ba-sentences, only three of them violate the constraints on ba-construction, from
which we can conclude that the Cantonese learners acquired ba-construction very well (although
they did not use it as frequently as the Mandarin control group). This finding contradicts previous
studies on the L2 acquisition of ba-construction, where L2 learners lagged far behind the native
speakers in production. For example, Wen (2010) tested English-speaking learners of Mandarin at
different proficiency levels. Even advanced learners in her study produced much fewer ba-sentences
than native speakers (43.1% vs. 93.5%). However, even though the L2 learners did not produce
many ba-sentences, their relatively good performance in the acceptability judgement tasks revealed
that they actually had acquired some knowledge of ba-construction (Du 2006; Xu 2012). A possible
explanation for such inconsistency is that the degree of distance between L1 and L2 may play a role
in L2 acquisition (Kellerman 1995). The pre-transitive constructions exist in both Cantonese and
Mandarin, and the similarities in linguistic features between them may have contributed to the more
successful acquisition of ba-construction by the Cantonese learners, whereas the absence of a
counterpart in the L1s may result in less successful acquisition of ba-construction by learners with
other language backgrounds (Mai 2016). Lacking any prior knowledge of such construction in their
L1s, learners would find it very difficult to fully acquire the complex structure and the constraints
of ba-construction, and they might consequently tend to avoid using it (Yu 2000).

Lastly, we identified the patterns of the sentences produced by the Cantonese learners when ba-
construction was not adopted. The most frequent pattern (77/108, 71.30%) was a simple declarative
sentence which follows the canonical SVO order in Mandarin as in (9a). Some participants produced
multiclausal sentences, where the first clause described the particular entity and in the second clause
the object was dropped (19/108, 17.59%) as in (9b). There were also several passive sentences which
adopted different passive markers as in (9c) and (9d) (8/108, 7.41%). The learners’ patterns were
very similar to the native speakers who also adopted the SVO sentences for most of the non-ba-
sentences (24/28, 85.71%).
(9) a. wo na chu le yi ge yumaoqiu
I take out ASP one CL shuttlecock
“I took out a shuttlecock.”
b. zhe ge maozi hen piaoliang, wo shishi kan
this CL hat very smart I try see
“This hat is very smart. Let me try it.”
c. shu gei kan diao le
tree gei cut drop ASP
“The tree was cut down (by someone).”
d. heiban shang de zi bei zhirisheng
blackboard above de character bei student_on_duty
ca diao le
wipe drop ASP
“The characters on the blackboard were wiped off by the student on duty.”
4. Conclusions
In this study, we fitted regression models with various breakpoints to test the CPH in L2 acquisition,
but none of the models were better than a simple linear regression model in predicting the L2
attainment based on AoA. Our data thus did not support the claim that there exists a cut-off point
for the CP. However, we did provide evidence for age effects on L2 acquisition. There is a tendency
that with the increase of AoA, L2 ultimate attainment becomes more divergent from the target
language grammar. We also compared the acquisition of two types of ba-construction, i.e. the

disposal type and the displacement type. No qualitative difference was found in the acquisition of
the two types by Cantonese learners. Finally, the Cantonese learners were demonstrated to have
more successfully acquired the earlier-noted types of ba-construction than learners with other
language backgrounds, and this might be the consequence of the similarities between Cantonese and
Mandarin.
Acknowledgements
The author would like to acknowledge the input from Scholastica Lam and Si Chen at various stages
of this project. We thank Shihui Liu, Chencui Ma, Can Wang and Xuemei Zhou for their help with
the stimuli, Yaxue Wang for her support with the experiments, Yanli Du and Longxing Li for their
contribution to the data annotation, and all the informants for their participation in this study. We
are grateful to the organisers and audience of the International Symposium on Chinese Theoretical
and Applied Linguistics. We are also deeply indebted to the editors and reviewers for their insightful
suggestions which have significantly improved the quality of this manuscript.
References
Baayen, R. Harald. 2008. Analyzing Linguistic Data: A Practical Introduction to Statistics Using R. New
York: Cambridge University Press. doi:10.1558/sols.v2i3.471.
Bender, Emily. 2000. “The Syntax of Mandarin Bă: Reconsidering the Verbal Analysis.” Journal of East
Asian Linguistics 9: 105–45. http://link.springer.com/article/10.1023/A:1008348224800.
Birdsong, David. 2004. “Second Language and Ultimate Attainment.” In The Handbook of Applied
Linguistics, edited by Alan Davies and Catherine Elder, 82–105. Oxford: Blackwell Publishing Ltd.
doi:10.1002/9780470757000.ch3.
Birdsong, David. 2006. “Age and Second Language Acquisition and Processing: A Selective Overview.”
Language Learning 56, S1: 9–49.
Birdsong, David. 2009. “Age and the End State of Second Language Acquisition.” In The New Handbook of
Second Language Acquisition, edited by William C. Ritchie and Tej K. Bhatia, 401–24. Bingley: Emerald
Group Publishing Limited. doi:10.1017/S0272263100000589.
Birdsong, David, and Michelle Molis. 2001. “On the Evidence for Maturational Constraints in Second-
Language Acquisition.” Journal of Memory and Language 44, 2: 235–49. doi:10.1006/jmla.2000.2750.
Birdsong, David, and Jan Vanhove. 2016. “Age of Second Language Acquisition: Critical Periods and Social
Concerns.” In Bilingualism across the Lifespan: Factors Moderating Language Proficiency, edited by
Elena Nicoladis and Simona Montanari, 163–82. Berlin: De Gruyter Mouton. doi:10.1037/14939-010.
Bley-Vroman, Robert. 1990. “The Logical Problem of Foreign Language Learning.” Linguistic Analysis 20,
1–2: 3–49.
Bornstein, Marc H. 1989. “Sensitive Periods in Development: Structural Characteristics and Causal

Interpretations.” Psychological Bulletin 105, 2: 179–97. doi:10.1037/0033-2909.105.2.179.
Cheng, Lisa Lai-shen. 1988. “Aspects of the Ba-Construction.” In Studies in Generative Approaches to
Aspect, edited by Carol Tenny, 73–84. Cambridge: MIT Press.
Cheng, Qi, and Gladys Tang. 2016. “On the L2 ultimate attainment of Mandarin additive and distributive
operators by Cantonese learners.” In Proceedings of the 13th Generative Approaches to Second Language
Acquisition Conference (GASLA 2015), edited by David Stringer, Jordan Garrett, Becky Halloran, and
Sabrina Mossman, 31–44. Somerville, MA: Cascadilla Proceedings Project.
Cheung, Antonio. 2006. “Processing Factors in Language Comprehension and Production: The Case of
Cantonese Dative Constructions.” MPhil thesis, University of Hong Kong.
Cheung, Samuel Hung-nin. 1992. “The Pretransitive in Cantonese.” In Proceedings of the International
Symposium on Chinese Languages and Linguistics, 241–303. Taipei: Institute of History and Philology,
Academia Sinica.
Clark, Eve V. 2009. First Language Acquisition. New York: Cambridge University Press.
Coppieters, René. 1987. “Competence Differences between Native and near-Native Speakers.” Language 63,
3: 544–73.
http://www.jstor.org/stable/415005.
DeKeyser, Robert, Iris Alfi-Shabtay, and Dorit Ravid. 2010. “Cross-Linguistic Evidence for the Nature of
Age Effects in Second Language Acquisition.” Applied Psycholinguistics 31, 3: 413–38.
doi:10.1017/S0142716410000056.
DeKeyser, Robert, and Jenifer Larson-Hall. 2005. “What Does the Critical Period Really Mean?” In
Handbook of Bilingualism: Psycholinguistic Approaches, edited by Judith F. Kroll and Annette M.B.
DeGroot, 88–108. New York: Oxford University Press.
Ding, Picus Sizhi. 2001. “Semantic Change versus Categorical Change: A Study of the Development of Ba
in Mandarin.” Journal of Chinese Linguistics 29, 1: 102–28.
Du, Hang. 2006. Acquisition of the Chinese Ba-Construction. Munich: Lincom Europa.
Feng, Shengli. 2001. “Prosodically Constrained Bare-Verb in Ba Constructions.” Journal of Chinese Linguistics
29, 2: 243–80.
Flege, James Emil, Grace H. Yeni-Komshian, and Serena Liu. 1999. “Age Constraints on Second-Language
Acquisition.” Journal of Memory and Language 41, 1: 78–104. doi:10.1006/jmla.1999.2638.
Granena, Gisela, and Michael H Long. 2013. “Age of Onset, Length of Residence, Language Aptitude, and
Ultimate L2 Attainment in Three Linguistic Domains.” Second Language Research 29, 3: 311–43.
doi:10.1177/0267658312461497.
Hakuta, Kenji, Ellen Bialystok, and Edward Wiley. 2003. “Acquisition, Critical Evidence: A Test of the
Critical Period Hypothesis for Second Language.” Psychological Science 41, 1: 31–38.
doi:10.1017/CBO9781107415324.004.
Hessling, Robert M., Tara J. Schmidt, and Nicole M. Traxel. 2014. “Floor Effect.” In The SAGE Encyclopedia
of Social Science Research Methods, edited by Michael S. Lewis-Beck, Alan Bryman, and Tim Futing
Liao, 391. Thousand Oaks: SAGE Publications, Inc.

doi:10.4135/9781483346304.n16.
Hessling, Robert M., Nicole M. Traxel, and Tara J. Schmidt. 2004. “Ceiling Effect.” In The SAGE
Encyclopedia of Social Science Research Methods, edited by Michael S. Lewis-Beck, Alan Bryman, and
Tim Futing Liao, 106. Thousand Oaks: SAGE Publications, Inc.
Huang, C.-T. James, Yen-Hui Audrey Li, and Yafei Li. 2009. “The Ba Construction.” In The Syntax of
Chinese, 153–96. New York: Cambridge University Press.
Huang, Yue Yuan, and Suying Yang. 2004. “The L2 Acquisition of the Chinese Ba-Construction.” Chinese
Teaching in the World 67, 1: 49–59.
Huang, Yue Yuan, and Suying Yang. 2005. “Telicity in L2 Chinese Acquisition.” In Proceedings of the 7th
Generative Approaches to Second Language Acquisition Conference (GASLA 2004), edited by Laurent
Dekydtspotter, Rex A. Sprouse, and Audrey Liljestrand, 150–62. Somerville: Cascadilla Press.
Hyltenstam, Kenneth, and Niclas Abrahamsson. 2003. “Maturational Constraints in SLA.” In The Handbook
of Second Language Acquisition, edited by Catherine Doughty and Michael H. Long, 539–88. Malden:
Blackwell. doi:10.1002/9780470756492.ch17.
Johnson, Jacqueline S. 1992. “Critical Period Effects in Second Language Acquisition: The Effect of Written
versus Auditory Materials on the Assessment of Grammatical Competence.” Language Learning 42, 2:
217–48. doi:10.1111/J.1467-1770.1992.TB00708.X.
Johnson, Jacqueline S., and Elissa L. Newport. 1989. “Critical Period Effects in Second Language Learning:
The Influence of Maturational State on the Acquisition of English as a Second Language.” Cognitive
Psychology 21, 1: 60–99. doi:10.1016/0010-0285(89)90003-0.
Johnson, Jacqueline S., Kenneth D. Shenkman, Elissa L. Newport, and Douglas L. Medin. 1996.
“Indeterminacy in the Grammar of Adult Language Learners.” Journal of Memory and Language 35, 3:
335–52. doi:10.1006/jmla.1996.0019.
Kellerman, Eric. 1995. “Age before Beauty: Johnson and Newport Revisited.” In The Current State of
Interlanguage: Studies in Honor of William E. Rutherford, edited by Lynn Eubank, Larry Selinker, and
Michael Sharwood Smith, 219–31. Amsterdam/Philadelphia: John Benjamins Publishing Company.
Lardiere, Donna. 2007. Ultimate Attainment in Second Language Acquisition: A Case Study. New York:
Routledge.
Lenneberg, Eric H. 1967. Biological Foundations of Language. New York: Wiley.
Leung, Yin-bing. 2004. “The Pretransitive Construction in Modern Standard Chinese and Cantonese.” In
Yuyan Jiechu Lunji, edited by Jiayan Zou and Rujie You, 375–94. Shanghai : Shanghai Education Press.
Li, Charles N., and Sandra A. Thompson. 1981. Mandarin Chinese: A Functional Reference Grammar.
Oakland: University of California Press.
Li, Jinxi. 1924/2007. “The Variant of Object Position.” In Xinzhu Guoyu Wenfa (The New Grammar of
Chinese), 41–43. Changsha: Hunan Education Press.
Li, Wei. 1993. “The Zoeng-Construction and the Ba-Construction.” In Guangzhouhua Jiaoxue Yu Yanjiu
(Research and Teaching of Cantonese), edited by Ding'ou Zheng and Xiaobing Zhou, 51–63. Guangzhou:
Sun Yat-sen University Press.

Liu, Feng-Hsi. 1997. “An Aspectual Analysis of BA.” Journal of East Asian Linguistics 6, 1: 51–99.
Long, Michael H. 2013. “Maturational Constraints in Child and Adult SLA.” In Sensitive Periods, Language
Aptitude, and Ultimate L2 Attainment, edited by Gisela Granena and Michael H. Long, 3–42. Amsterdam:
John Benjamins Publishing Company.
Lv, Shuxiang. 1955/2002. “The Study of the Usage of Ba.” In Hanyu Yufa Lunwenji (Papers on Chinese
Grammar), 169–91. Shenyang: Liaoning Education Press.
Lv, Wenhua. 1994. “The Syntactic Semantics of Ba-Construction.” Chinese Language Learning 82, 4: 26–
28.
Mai, Ziyin. 2016. “L2 Chinese: Grammatical Development and Processing.” Second Language Research 32,
1: 123–41.
doi:10.1177/0267658315597737.
Mayberry, Rachel I., and Pamela Witcher. 2005. “What Age of Acquisition Effects Reveal about the Nature
of Phonological Processing.” CRL Technical Reports 17, 3: 3–9.
McDonald, Janet L. 2000. “Grammaticality Judgments in a Second Language: Influences of Age of
Acquisition and Native Language.” Applied Psycholinguistics 21, 3: 395–423.
doi:10.1017/S0142716400003064.
Meisel, Jürgen M. 2011. First and Second Language Acquisition: Parallels and Differences. New York:
Cambridge University Press.
doi:10.1017/CBO9780511862694.
Moyer, Alene. 1999. “Ultimate Attainment Il L2 Phonology: The Critical Factors of Age, Motivation and
Instruction.” Studies in Second Language Acquisition 21: 81–108.
Muggeo, Vito M. R. 2003. “Estimating Regression Models with Unknown Break-Points.” Statistics in
Medicine 22, 19: 3055–71.
doi:10.1002/sim.1545.
Mullie, Jos. 1932. “Use of the Accusative.” In The Structural Principles of the Chinese Language, 178–85.
Peiping: Bureau of Engraving and Printing.
Muñoz, Carmen, and David Singleton. 2011. “A Critical Review of Age-Related Research on L2 Ultimate
Attainment.” Language Teaching 44, 1: 1–35. doi:10.1017/S0261444810000327.
O’Grady, William, and Sook Whan Cho. 2001. “First Language Acquisition.” In Contemporary Linguistics:
An Introduction, edited by William O”Grady and John Archibald, 326–62. Bedford: St. Martin's Press.
Oyama, Susan. 1976. “A Sensitive Period for the Acquisition of a Nonnative Phonological System.” Journal
of Psycholinguistic Research 5, 3: 261–83.
Penfield, Wilder, and Lamar Roberts. 1959. Speech and BrainMechanisms. Princeton, NJ: Princeton
University Press.
R Core Team. 2017. “R: A Language and Environment for Statistical Computing.” Vienna, Austria: R
Foundation for Statistical Computing. https://www.r-project.org.
Reichle, Robert V. 2010. “Judgments of Information Structure in L2 French: Nativelike Performance and the
Critical Period Hypothesis.” IRAL - International Review of Applied Linguistics in Language Teaching

48, 1: 53–85. doi:10.1515/iral.2010.003.
Révész, Andrea. 2012. “Coding Second Language Data Validly and Reliably.” In Research Methods in
Second Language Acquisition: A Practical Guide, edited by Alison Mackey and Susan M. Gass, 201–21.
Blackwell Publishing Ltd. doi:10.1002/9781444347340.ch11.
RStudio Team. 2016. “RStudio: Integrated Development for R.” Boston, MA: RStudio, Inc.
http://www.rstudio.com/.
Saito, Kazuya. 2015. “The Role of Age of Acquisition in Late Second Language Oral Proficiency Attainment.”
Studies in Second Language Acquisition 37, 4: 713–43. doi:10.1017/S0272263115000248.
Singleton, David. 2005. “The Critical Period Hypothesis: A Coat of Many Colours.” IRAL - International
Review of Applied Linguistics in Language Teaching 43, 4: 269–85. doi:10.1515/iral.2005.43.4.269.
Sorace, Antonella. 2003. “Near-Nativeness.” In The Handbook of Second Language Acquisition, edited by
Catherine Doughty and Michael Long, 130–51. Malden: Blackwell.
Sorace, Antonella, and Francesca Filiaci. 2006. “Anaphora Resolution in near-Native Speakers of Italian.”
Second Language Research 22, 3: 339–68. doi:10.1191/0267658306sr271oa.
Sun, Chao-Fen, and Talmy Givón. 1985. “On the so-Called SOV Word Order in Mandarin Chinese: A
Quantified Text Study and Its Implications.” Language 61, 2: 329–51.
Sybesma, Rint. 1999. “The Ba-Construction.” In The Mandarin VP, 131–82. Dordrecht: Springer
Netherlands.
Tang, Sze-Wing. 2002. “Focus and Dak in Cantonese.” Journal of Chinese Linguistics 30, 2: 266–309.
Teng, Shou-hsin. 1975. A Semantic Study of Transitivity Relations in Chinese. Berkeley: University of
California Press.
Thompson, Sandra A. 1973. “Transitivity and Some Problems with the BǍ Construction in Mandarin
Chinese.” Journal of Chinese Linguistics 1, 2: 208–21.
Tsao, Feng-fu. 1987. “A Topic-Comment Approach to the Ba Construction.” Journal of Chinese Linguistics
15, 2: 1–54.
Vanhove, Jan. 2013. “The Critical Period Hypothesis in Second Language Acquisition: A Statistical Critique
and a Reanalysis.” PLoS ONE 8, 7. doi:10.1371/journal.pone.0069172.
Veríssimo, João, Vera Heyer, Gunnar Jacob, and Harald Clahsen. 2017. “Selective Effects of Age of
Acquisition on Morphological Priming: Evidence for a Sensitive Period.” Language Acquisition, 25, 3:
1–12. doi:10.1080/10489223.2017.1346104.
Wang, Li. 1943/1985. “Disposal.” In Zhongguo Xiandai Yufa (A Modern Grammar of Chinese), 124–30.
Ji'nan: Shandong Education Press.
Wen, Xiaohong. 2010. “Acquisition of the Displacement Ba-Construction by English-Speaking Learners of
Chinese.” Journal of Chinese Teachers Association 45, 2: 73–99.
Werker, Janet F., John HV Gilbert, Keith Humphrey, and Richard C. Tees. 1981. “Developmental Aspects
of Cross-Language Speech Perception.” Child Development 52, 1: 349–55.
Werker, Janet F, and Richard C Tees. 1984. “Cross-Language Speech Perception: Evidence for Perceputal
Reorganization during the First Year of Life.” Infant Behavior and Development 7, 1: 49–63.

doi:10.1016/S0163-6383(84)80022-3.
White, Lydia. 2003. Second Language Acquisition and Universal Grammar. New York: Cambridge
University Press.
Wickham, Hadley. 2016. ggplot2: Elegant Graphics for Data Analysis. Cham: Springer.
Xu, Hongying. 2012. “The Acquisition of the BA Construction by English-Speaking Learners of Chinese.”
PhD diss., University of Kansas.
Yang, Suying. 1998a. “The Analysis of the Ba-Construction from the Viewpoint of Aspectuality (1).”
Chinese Langauge Learning 2: 10–13.
Yang, Suying. 1998b. “The Analysis of the Ba-Construction from the Viewpoint of Aspectuality (2).”
Chinese Language Learning 3: 10–12.
Yang, Yike. 2013. “Optionality in the Acquisition of Mandarin Ba-Construction by Cantonese Learners of
Mandarin.” MA thesis. The Chinese University of Hong Kong.
Yang, Yike. 2020. “Acquisition of the Mandarin ba-construction by Cantonese Learners.” Macrolinguistics
8, 1: 88-104. doi: 10.26478/ja2020.8.12.6.
Yu, Wenqing. 2000. “A Report of an Investigation of CSL Learner's Using of Ba Sentence.” Chinese
Language Learning 5: 49–54.
Yuan, Boping. 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, 2: 219–60.
doi:10.1177/0267658309349421.
Zou, Ke. 1993. “The Syntax of the Chinese BA Construction.” Linguistics 31, 4: 715–36.
doi:10.1515/ling.1993.31.4.715.