1. Introduction
A fundamental function of human language is to express temporal reference through tense and aspect (Comrie, Reference Comrie1976: 3). The acquisition of tense and aspect is one of the earliest tasks in language acquisition (Li & Shirai, Reference Li and Shirai2000), and its investigation yields insights into children’s consolidation of temporal reference and its corresponding syntactic structure (Weist, Reference Weist, Salaberry and Shirai2002).
Children with Developmental Language Disorder (DLD) demonstrate weak language ability resembling younger typically developing (TD) children (Sansavini et al., Reference Sansavini, Favilla, Guasti, Marini, Millepiedi and Di Martino2021), which is not caused by hearing impairment or neurological disorder (Leonard, Reference Leonard2014). Children with DLD show weakness in the use of grammatical morphemes that mark tense and aspect (Leonard et al., Reference Leonard, Lukács and Kas2012). However, while various studies have reported deficits in the production of tense by children with DLD (e.g., see Bishop, Reference Bishop2014; Chondrogianni & John, Reference Chondrogianni and John2018; Deevy & Leonard, Reference Deevy and Leonard2018; Hilvert et al., Reference Hilvert, Hoover, Sterling and Schroeder2020; Krok & Leonard, Reference Krok and Leonard2015; Oetting et al., Reference Oetting, McDonald and Vaughn2023; Rice et al., Reference Rice, Wexler and Cleave1995; Wexler et al., Reference Wexler, Schaeffer and Bol2004), very little work has been conducted on aspectual production by children with DLD. To address this lacuna in the literature, the production of grammatical aspect by preschool children with DLD who speak Mandarin Chinese (hereafter, Mandarin) is the focus of the current study.
1.1. Aspect
Aspect is the semantic domain of the temporal point of view in language (Smith, Reference Smith1994: 107). According to Smith’s two-component theory, aspect includes grammatical aspect (or viewpoint aspect) and lexical aspect (or situation types/Aktionsart) (Smith, Reference Smith1994).
Grammatical aspect, which is usually indicated by overt markers such as auxiliaries or inflections,Footnote 1 conveys a temporal perspective that focuses on all or part of a situation (Comrie, Reference Comrie1976; Gvozdanović, Reference Gvozdanović and Binnick2012; Smith, Reference Smith1997). Imperfective aspect and perfective aspect are the two main viewpoints of grammatical aspect, depending on whether an event is viewed in terms of its internal structure/from the inside (in the former case), or as a whole, from the outside (in the latter case). Put differently, imperfective aspect focuses on intervals that are neither initial nor final, thus excluding the starting and ending points, which are encompassed by perfective aspect. In English, imperfective aspect (and in particular progressive aspect) is formed by preceding the verb with the auxiliary “be” and following it with the suffix -ing (as in 1), and perfective aspect is conveyed through the simple past tense (as in 2).
-
(1) The workers were building a house last month.
-
(2) The workers built a house last month.
Lexical aspect refers to the inherent temporal properties of a verb or verb phrase (Filip, Reference Filip and Binnick2012; Smith, Reference Smith1994; van Hout, Reference van Hout, Lidz, Snyder and Pater2016). According to their inherent temporal meanings, English verbs are divided into four types, including Activity, State, Accomplishment, and Achievement verbs (Vendler, Reference Vendler1957, Reference Vendler1967).
1.2. Aspect in Mandarin
Mandarin is an aspect-prominent and tenseless language in which there is no systematic and overt tense marking on the verb (Li, Reference Li2012). Grammatical aspect in Mandarin is typically expressed by four overt markers. The progressive marker zai- and the durative marker -zhe represent imperfective aspect; the former shows the progressive/ongoing stage of an action/event, and the latter indicates the duration of a state or the resultative state of an action/event (Chao, Reference Chao1968; Xiao & McEnery, Reference Xiao and McEnery2004), as illustrated in sentences (3–4). Perfective aspect is expressed by the perfective marker -leFootnote 2 and the experiential marker -guo. -le indicates the completion of an action/event, and -guo is used to express that an action/event type has been experienced at least once in the past (Li & Thompson, Reference Li and Thompson1981; Zhu, Reference Zhu1982), as displayed in sentences (5–6).
As illustrated in examples (3–6), the progressive marker zai- is pre-verbal, while the other three aspect markers are post-verbal. Further, zai- is pronounced with the falling tone, while the three post-verbal markers are pronounced with the neutral tone and are stressless in a sentence.
In Mandarin, there are five typesFootnote 4 of verbs used to express lexical aspect, including Activity (e.g., youyong “swim” and changge “sing”), State (shuyu “belong” and zhidao “know”), Accomplishment (gai yizuo fangzi “build a house” and pao yibaimi “run one hundred meters”), Achievement (si “die” and dasui “break into pieces”), and Semelfactive verbs (e.g., qiaomen “knock the door” and diantou “nod head”) (Smith, Reference Smith1994).
1.3. Acquisition of grammatical aspect by TD children
Cross-linguistically, TD children start to use tense/aspect markers from 1;8 to 2;0 years old (Yang et al., Reference Yang, Shi and Xu2018), and there is asymmetrical development in aspectual acquisition. For instance, English-speaking TD children can understand perfective aspect at an early age, but they have difficulty comprehending or producing the past progressive (Stuart & van der Lely, Reference Stuart and van der Lely2015; Wagner, Reference Wagner2001); and Dutch, Italian, and Polish L1 learners have been observed to perform better on perfective aspect than on imperfective aspect in sentence-picture matching tasks (van Hout, Reference van Hout2008). No previous studies investigated whether Mandarin-speaking TD children show asymmetry in aspectual acquisition. However, studies on individual aspect markers found that these children acquire the perfective marker -le first, then the progressive marker zai- and the durative marker -zhe, and finally the experiential marker -guo (Erbaugh, Reference Erbaugh and Slobin1992; Li et al., Reference Li, Liang and Wu2022).
The acquisition of grammatical aspect could be impacted by factors such as lexical aspect, temporal context, or the morphological salience of aspect markers.
Lexical aspect plays a role in the acquisition of grammatical aspect by TD children (e.g., see Blank et al., Reference Blank, Holt and Wagner2020; Ryu & Shirai, Reference Ryu and Shirai2022; Wagner, Reference Wagner2002), and even second language learners (e.g., see Bardovi-Harlig, Reference Bardovi-Harlig2000; Bardovi-Harlig & Comajoan-Colomé, Reference Bardovi-Harlig and Comajoan-Colomé2020; González & Hernández, Reference González and Hernández2018; Nishi & Shirai, Reference Nishi and Shirai2021), as predicted by the Aspect Hypothesis (Andersen & Shirai, Reference Andersen and Shirai1994, Reference Andersen, Shirai, Ritchie and Bhatia1996). According to the Aspect Hypothesis, children or second language learners tend to combine aspect markers with their prototypical verbs, such as perfective markers with telic verbs (e.g., Achievement and Accomplishment verbs) and imperfective markers with atelic verbs (e.g., Activity verbs). In a later stage of aspectual development, children or second language learners begin to combine aspect markers with other types of verbs (i.e., their non-prototypical verbs). Studies on Mandarin-speaking TD children also reported that their acquisition of grammatical aspect is influenced by lexical aspect. For instance, they initially combine the progressive marker zai- with Activity verbs, the perfective marker -le with Achievement and Accomplishment verbs, the durative marker -zhe with Activity and State verbs, and the experiential marker -guo with Accomplishment and Activity verbs (Chen & Shirai, Reference Chen and Shirai2010; Li et al., Reference Li, Liang and Wu2022).
Besides the impact of lexical aspect, the acquisition of grammatical aspect appears to also be influenced by its interactions with the temporal context. Through analysing imperfective/perfective aspect and their inflections in the spontaneous production of three TD children from the onset of speech to the age of 3 years, Gagarina (Reference Gagarina2004) reported a strong preference for using perfective aspect for past events and imperfective aspect for present events in Russian-speaking children. Using a computer-mouse tracking task, Anderson et al. (Reference Anderson, Matlock and Spivey2013) also found that 64 undergraduate students showed better performance on perfective aspect in the “distant past” context and better performance on imperfective aspect in the “recent past” context.
Finally, the morphological salience of aspect markers also plays a role in aspectual acquisition. The overt markers that express grammatical aspect differ in salience both across and within languages. For instance, the morphological inflection for perfective aspect in both Dutch and Polish is represented with prefixes, while it is represented with suffixes in Italian. Van Hout (Reference van Hout2008) reported that 3-year-old Dutch-speaking and Polish-speaking TD children performed target-like on perfective aspect in their study, while Italian-speaking children of the same age showed weak performance on perfective aspect. The author argued that the better performance of Dutch-speaking and Polish-speaking TD children, compared to their Italian-speaking peers, in producing perfective aspect was caused by the different levels of morphological salience of inflections expressing grammatical aspect. Indeed, prefixes are arguably more salient than suffixes, and according to the Morphological Salience hypothesis, “the semantics of morphologically salient paradigms is acquired early” (Van Hout, Reference van Hout2008: 1753).
Pursuing this reasoning, the progressive marker zai- in Mandarin, which is pre-verbal and pronounced with the falling tone, would be more salient than the post-verbal markers -le, -zhe, and -guo. So far, however, no studies have investigated whether the saliency of zai- facilitates its acquisition in children with DLD as compared to the post-verbal markers, which will be explored in this study.
1.4. Acquisition of grammatical aspect by children with DLD
Similar to TD children, children with DLD reportedly show asymmetrical performance in the acquisition of grammatical aspect. For instance, 12- to 16-year-old English-speaking and six- to eight-year-old Greek-speaking children comprehend perfective aspect better than imperfective aspect (Dosi, Reference Dosi2019; Konstantzou, Reference Konstantzou2015; Stuart & van der Lely, Reference Stuart and van der Lely2015); five- to nine-year-old Turkish-speaking and four- to six-year-old Mandarin-speaking children with DLD displayed better performance in the comprehension of imperfective aspect than of perfective aspect (Chen & Durrleman, Reference Chen and Durrleman2022; Duman & Topbas, Reference Duman and Topbas2016).
Considering aspectual production, Leonard et al. (Reference Leonard, Deevy, Kurtz, Chorev, Owen and Polite2007) found that five-year-old English-speaking children with DLD performed significantly worse than three-year-old language-matched TD children in the production of the perfective marker -ed but displayed similar performance to their TD peers in the production of the progressive marker -ing in an acting-out elicitation task. In contrast, Konstantzou (Reference Konstantzou2015) reported that six-year-old Greek-speaking children with DLD performed similarly to their age-matched TD children in the production of perfective aspect, but showed worse performance than their TD peers in the production of imperfective aspect, based on an elicited production task that used video stimuli.
Because semantically more complex aspectual operations need more cognitive resources in order to be processed than semantically less complex ones (van Hout, Reference van Hout2008), the contradictory findings between Leonard et al. (Reference Leonard, Deevy, Kurtz, Chorev, Owen and Polite2007) and Konstantzou (Reference Konstantzou2015) might be attributed to cross-linguistic variation in the semantic complexity of aspect markers. To be specific, both past tense and perfective aspect are conflated in the inflection -ed in English, while the inflection -ing only represents progressive aspect. As a result, children need to incorporate past tense when they develop the notion of (perfective) grammatical aspect in their application of -ed, which makes the process of producing -ed more complex than (imperfective) -ing. In Greek, however, imperfective aspect can assign a habitual, a continuous, a progressive, an iterative, or a durative interpretation without morphological change (Xydopoulos, Reference Xydopoulos1996), which makes imperfective aspect in this language more complex than perfective aspect, since the latter only represents completed or finished actions.
However, the semantic complexity of grammatical aspect can also be caused by the interaction between grammatical aspect and the experiment’s temporal context. Better performance in perfective aspect in the “distant past” context and better performance in imperfective aspect in the “recent past/here and now” context have been confirmed by Gagarina (Reference Gagarina2004) and Anderson et al. (Reference Anderson, Matlock and Spivey2013). The acting-out task in Leonard et al. (Reference Leonard, Deevy, Kurtz, Chorev, Owen and Polite2007), which requires the participants to follow the experimenter’s instruction and do actions immediately, provided a “recent past” context that is conflated with imperfective aspect. However, the video stimuli in the elicitation production task of Konstantzou (Reference Konstantzou2015) were recorded in advance and offered a “distant past” context that is conflated with perfective aspect. Therefore, these experimental conditions could also explain the better performance on imperfective aspect reported by Leonard et al. (Reference Leonard, Deevy, Kurtz, Chorev, Owen and Polite2007) and the better performance on perfective aspect found by Konstantzou (Reference Konstantzou2015). In addition, the studies of Dosi et al. also demonstrated the influence of the interaction between grammatical aspect and the temporal context of experimental tasks on aspectual acquisition in children with DLD. The authors found that Greek-speaking children with DLD performed better on imperfective aspect than perfective aspect in an elicitation completion task; however, they showed better performance on perfective aspect than on imperfective aspect in a picture-based narrative task (Dosi, Reference Dosi2019; Dosi et al., Reference Dosi, Andreou and Peristeri2018).
Some studies have yielded a purer assessment of grammatical aspect in children with DLD independently of tense and temporal context. In Hungarian, tense is marked by the presence or absence of a verb inflection, distinguishing past and present tense, while perfective and imperfective aspect is distinguished by the presence or absence of a prefix (Kiefer, Reference Kiefer, Kiefer and Kiss1994). Therefore, it is possible to investigate the acquisition of grammatical aspect without the involvement of tense. Leonard et al. (Reference Leonard, Lukács and Kas2012) examined the production of grammatical aspect in five-year-old Hungarian-speaking children with DLD using a sentence completion task, and they found that children with DLD performed worse than age-matched as well as language-matched TD children in the production of both perfective and imperfective aspect. The authors contended that Hungarian-speaking children with DLD show weakness in aspectual production. In Cantonese (a Chinese language), there are no overt tense inflections, and the expression of tense rather relies on lexical and contextual cues (Matthews & Yip, Reference Matthews and Yip2010). Fletcher et al. (Reference Fletcher, Leonard, Stokes and Wong2005) investigated the production of grammatical aspect in five-year-old Cantonese-speaking children with DLD using a picture-description task. The authors found that children with DLD performed significantly worse than both age-matched and language-matched TD children in the production of both perfective and imperfective aspect.
Despite the worse performance of children with DLD than their TD peers in the production of grammatical aspect in tenseless conditions, it remains unknown whether the production of grammatical aspect in a tenseless language (such as Cantonese of the study by Fletcher et al. [Reference Fletcher, Leonard, Stokes and Wong2005]) indeed reveals the asymmetrical pattern attested in the acquisition of tense languages evoked above. In addition, quite high standard deviations (almost near the means in most conditions) were found in children with DLD in the study by Fletcher et al. (Reference Fletcher, Leonard, Stokes and Wong2005). The high standard deviations indicated a very heterogeneous performance among children in the DLD group in the production of grammatical aspect, in line with reports of their heterogeneous language development patterns in general (Bishop, Reference Bishop, Verhoeven and van Balkon2003; Kuiack & Archibald, Reference Kuiack and Archibald2024; Lancaster & Camarata, Reference Lancaster and Camarata2019; van der Lely, Reference van der Lely, Levy and Schaeffer2003; van Weerdenburg et al., Reference van Weerdenburg, Verhoeven and Van Balkom2006). Nonetheless, despite this considerable heterogeneity, the authors did not analyse the individual performance, nor in other studies that concerned the production of grammatical aspect by children with DLD, which will be considered in this study.
In order to address these issues, this study aims to investigate the production of perfective and imperfective aspect by children with DLD who speak Mandarin. Although no study considered whether Mandarin-speaking children with DLD show asymmetry in the production of grammatical aspect, a dearth of previous studies has investigated the production of individual aspect markers by Mandarin-speaking children with DLD and reported inconsistent findings. He et al. (Reference He, Sun and Tian2013) found that four- to six-year-old children with DLD were less likely to produce the perfective marker -le than age-matched as well as language-matched TD children, while they showed worse performance only than age-matched TD children in the production of the progressive marker zai-. Chen et al. (Reference Chen, An and He2021) reported that four- to six-year-old children with DLD produced significantly fewer sentences with the three post-verbal markers -le, -zhe, and -guo than TD children; however, they performed similarly to their TD peers in the production of sentences with the pre-verbal marker zai-. Chen et al. (Reference Chen, An, Dai and He2022) compared the production of Mandarin aspect markers zai-, -le, -zhe, and -guo in children with DLD, children with high-functioning autism plus language impairment, and age-matched TD children. The authors found that children with DLD, like those with high-functioning autism plus language impairment, produced fewer target responses for all four aspect markers than their TD peers in a priming picture-description task. Furthermore, the impact of lexical aspect on grammatical aspect was revealed in the three studies above. Namely, children with DLD tended to combine aspect markers with their prototypical verbs in their production.
The better performance on pre-verbal zai- than on post-verbal markers in He et al. (Reference He, Sun and Tian2013) and Chen et al. (Reference Chen, An and He2021) may stem from the influence of morphological salience on the production of grammatical aspect by children with DLD. In Mandarin, the aspect marker zai-, which is pre-verbal and pronounced with the falling tone, is more salient than the aspect markers -le, -zhe, and -guo. Those three aspect markers are post-verbal and pronounced with the neutral tone. Chen et al. (Reference Chen, An and He2021) attributed the difficulty of children with DLD in the production of post-verbal markers to their weak verbal working memory capacity (VWM). To be specific, children with DLD, who are reported to show deficits in their VWM (Archibald & Gathercole, Reference Archibald and Gathercole2007; Larson & Weismer, Reference Larson and Weismer2022; Vugs et al., Reference Vugs, Hendriks, Cuperus, Knoors and Verhoeven2017; Weismer et al., Reference Weismer, Evans and Hesketh1999), would not store the inconspicuous post-verbal aspect markers as efficiently as TD children in the process of sentence formation. As a result, they sometimes ignore the three post-verbal aspect markers and produce more sentences with bare verb forms than TD children. However, the VWM of children with DLD was not measured in Chen et al. (Reference Chen, An and He2021), which will be investigated in this study.
1.5. VWM in children with DLD
The VWM is the ability to temporarily store and manipulate verbal information, which is crucial for language comprehension, production, and other cognitive functions that rely on language processing (Baddeley, Reference Baddeley2003). Individuals with higher VWM tend to have larger vocabularies and better lexical access (Gupta & MacWhinney, Reference Gupta and MacWhinney1997) and show better performance in tasks involving the processing of complex syntactic structures (Caplan & Waters, Reference Caplan and Waters2013; Just & Carpenter, Reference Just and Carpenter1992). In contrast, individuals with weaker VWM would have more difficulty accessing and retrieving inconspicuous morphological forms during language production and comprehension because the VWM must compensate for weaker encoding by relying on memory traces that are less robust when individuals try to retrieve these inconspicuous morphological forms (Baddeley, Reference Baddeley2003; Schwering & MacDonald, Reference Schwering and MacDonald2020).
Children with DLD showed weaker VWM than their TD peers (Archibald & Gathercole, Reference Archibald and Gathercole2007; Larson & Weismer, Reference Larson and Weismer2022; Vugs et al., Reference Vugs, Hendriks, Cuperus, Knoors and Verhoeven2017; Weismer et al., Reference Weismer, Evans and Hesketh1999), and their weak VWM has been found to be linked to other aspects of grammatical competency, such as their weak performance in the production of morphosyntactic inflections that express tense (e.g., see Bishop et al., Reference Bishop, Adams and Norbury2006; Norbury et al., Reference Norbury, Bishop and Briscoe2001). However, no previous studies have examined whether the VWM of children with DLD also relates to their production of grammatical aspect, which will be considered in this study.
Forward digital recall, word repetition, and non-word repetition tasks are the classic paradigms used to measure the VWM in young children (Alloway et al., Reference Alloway, Gathercole and Pickering2006; Pickering & Gathercole, Reference Pickering and Gathercole2001). Since performance on word repetition tasks may potentially rely on the participant’s familiarity with the presented words (Gathercole, Reference Gathercole1995), the current study instead opts for the digital recall and non-word repetition tasks to measure participants’ VWM.
1.6. Research questions
The aims of this study will be achieved by answering three research questions through the investigation of aspectual production by preschool Mandarin-speaking children with DLD in a priming picture-description task, as well as a digital recall and a word repetition task, as compared to age-matched TD children:
-
(1) What is the performance of children with DLD in the production of the four typical Mandarin aspect markers, as compared to age-matched TD children? Since the pre-verbal marker zai- is more salient than the three post-verbal markers, we predict that children with DLD may perform better in the production of zai- than in other three post-verbal markers. Furthermore, we predict that children with DLD may produce fewer target sentences than their age-matched TD peers for the four aspect markers because of their lower language ability than age-matched TD children.
-
(2) Do children with DLD show asymmetrical performance in the production of perfective and imperfective aspect? We predict that Mandarin-speaking children with DLD would display asymmetrical performance in the production of grammatical aspect, like those in Leonard et al. (Reference Leonard, Deevy, Kurtz, Chorev, Owen and Polite2007), Konstantzou (Reference Konstantzou2015), Dosi et al. (Reference Dosi, Andreou and Peristeri2018), and Dosi (Reference Dosi2019). Since the priming picture-description task, which requires the participants to describe the actions in the pictures presented to them, provides a “here and now” context, in which children with DLD may demonstrate better performance in producing imperfective aspect than perfective aspect.
-
(3) Does the VWM of children with DLD correlate with their performance in the production of grammatical aspect? Children with DLD reportedly display deficits in their VWM (Archibald & Gathercole, Reference Archibald and Gathercole2007; Larson & Weismer, Reference Larson and Weismer2022; Vugs et al., Reference Vugs, Hendriks, Cuperus, Knoors and Verhoeven2017; Weismer et al., Reference Weismer, Evans and Hesketh1999), so they may consequently have more difficulty maintaining and selecting the appropriate morphological forms during language processing (Baddeley, Reference Baddeley2003), including aspect markers. Therefore, we predict that the VWM of children with DLD would be related to their performance on aspectual production.
2. Methodology
2.1. Participants
A total of 43 four- to six-year-old children participated in this study, including 18 children with DLD and 25 TD children. Children with DLD were recruited from special education schools, kindergartens, and hospitals, while TD children were recruited from kindergartens.
Participants’ non-verbal intelligence was assessed using the Chinese version of the Wechsler Preschool and Primary Scale of Intelligence–Fourth Edition (Li & Zhu, Reference Li and Zhu2014). All participants’ non-verbal intelligence was higher than 80.
The score of the Peabody Picture Vocabulary Test-Revised (PPVT; Sang & Miao, Reference Sang and Miao1990) and the scores of language comprehension and language production of the Rating Scale for Pre-school Children with Language Disorder-Revised (RSPCLD; for children who were younger than 71 months old; Lin, Reference Lin2008) or the Rating Scale for School Children with Language Disorder-Revised (RSSCLD; for children who were older than 71 months old; Lin, Reference Lin2009) were used to measure participants’ global language ability. The three scores for TD children were within the norms of their age, while at least two out of the three scores for children with DLD were 1.25 standard deviations below the norms (following Tomblin et al., Reference Tomblin, Records, Buckwalter, Zhang, Smith and O’Brien1997). Table 1 provides information on participants’ gender, age, non-verbal intelligence, and global language ability as represented by a composite score of the PPVT and RSPCLD/RSSCLD.
Table 1. Participants’ basic information
Abbreviations: GL = global language ability, NVI = non-verbal intelligence, SD = standard deviations.
The results of independent-sample t-tests showed that the DLD and TD groups were matched in age (t(41) = .030, p = .976). However, the TD group was significantly higher than the DLD group on non-verbal intelligence (t(41) = 4.566, p < .001) and global language ability (t (41) = 12.579, p < .001).
2.2. Aspectual production: Priming picture-description task
Since children with DLD showed similar structural priming effects to TD children (Miller & Deevy, Reference Miller and Deevy2006), we administered a priming picture-description task, including structural priming, to elicit the production of grammatical aspect in Mandarin. In this task, the experimenter describes a picture with a specific syntactic structure first (e.g., “The boy is playing basketball.”), and then a similar picture (i.e., with the same event structure) is displayed to the participant, and s/he is expected to describe this target picture using the same syntactic structure (e.g., “The girl is reading a book.”). If the participant uses the target structure in his/her description, we assume that s/he has mastered it; otherwise, we assume that s/he has not.
2.2.1. Materials
The task includes 30 test items, with 3 test items for each grammatical combination of an aspect marker with one verb typeFootnote 5 (e.g., zai- + Activity verbs). Namely, one “aspect marker + verb” combination is used in the priming sentences of three test items to eliminate the influence of lexical aspect on the production of grammatical aspect. Specifically, there are six items for the progressive marker zai- (three are combined with Activity verbs, and three with Accomplishment verbs), nine test items for the perfective marker -le (three are combined with Activity verbs, three with Achievement verbs, and three with Accomplishment verbs), six test items for the durative marker -zhe (three are combined with Activity verbs, and three with State verbs), and nine test items for the experiential marker -guo (three are combined with Activity verbs, three with Achievement verbs, and three with Accomplishment verbs). Besides the test items, one practice item with the combination of the progressive marker zai- and the Activity verb is presented to familiarize the participant with the task.
The verbs and nouns used in the priming sentences are drawn from those used commonly in daily life and acquired by TD children before three years old (Li, Reference Li1995). Different verbs in the same verb type were used for different aspect markers, although four verbs (e.g., chi “eat,” he “drink,” hua “draw,” and xi “wash”) were used twice for two markers with different subjects and objects due to limited verbs that have been acquired by young children with DLD and can be depicted by pictures. A pilot study conducted on 10 TD children aged three years demonstrated that they could successfully understand these verbs and nouns, as well as the instructions of this task.
Each test item includes a priming part and a target part. The priming and target pictures for the progressive marker zai- demonstrate the ongoing part of an event; the priming and target parts for the perfective marker -le consist of two or three pictures displayed in quick succession to illustrate the process of an eventFootnote 6; the priming and target pictures for the durative marker -zhe represent the resultative state of an event; and for the experiential marker -guo, a photo behind the figure in the priming or target picture indicates that the figure has had the experience shown in the photo. The example pictures and picture descriptions of the four aspect markers are displayed in Table 2.
Table 2. Example pictures and sentences of the priming picture-description task
All pictures in this task are cartoons with white backgrounds that are easy to recognize, and they are presented with PowerPoint on a laptop or a tablet.
2.2.2. Procedures
Participants were tested by an experimenter and an assistant individually in a quiet room. Test items were presented in a certain order that was initiated with the progressive marker zai- first, then the perfective marker -le followed by the durative marker -zhe, and finally the experiential marker -guo. The target part of the first test item was used as the priming part of the subsequent test item in each “aspect marker + verb” combination. The overall procedure of this task included the following: general instruction → practice item → test items.
At the beginning of the task, an experimenter explained the task to each participant, “XXX, let’s play a game. You and I will describe the pictures in turn. I do it first. Are you ready?” Then, the practice item was played, and feedback was given to the participants to familiarize them with the task procedure. Specifically, the experimenter would praise the participant if s/he gave the expected answer; otherwise, the experimenter gave the target answer when the participant gave non-target responses or no response.
For each test item, the experimenter first described the priming partFootnote 7 with a sentence that contained one aspect marker, and then the participant was asked to describe the target part with the same aspect marker. The experimenter elicited the participants again by repeating the priming sentence if the participant did not give any response the first time, and the priming sentence was repeated no more than twice for a test item. The response time was not restricted, and the participants were not required to give a response within a certain time. The assistant audio-recorded the test process and wrote down the participants’ responses. Totally, the task took about 15 min.
2.3. VWM: Forward digital recall task and non-word repetition task
The forward digital recall task and the non-word repetition task were conducted to test participants’ VWM in this study. In the forward digital recall task, which is adopted from the Wechsler Intelligence Scale for Children–the fifth edition (Wechsler, Reference Wechsler2014), the participant is asked to repeat sequences of digits that increase in length in the same order. Therefore, when the experimenter says “2-8-3″, the participant needs to repeat “2-8-3.” In the non-word repetition task, which is adopted from Stokes et al. (Reference Stokes, Wong, Fletcher and Leonard2006), the participant is required to repeat nonsense syllables, varying in length. So, when the experimenter says ba-kao-si, the participant should repeat ba-kao-si.
2.3.1. Materials
There are 12 test items in the forward digital recall task. The 12 test items are in six levels, from three digits to eight digits, and each level has two sets of test items. Besides the test items, three practice items are included in the task, and each item contains two to three digits.
In the non-word repetition task, there are a total of eight test items of nonsense syllables that follow Mandarin phonotactic constraints. The length of the syllables in this task ranges from three to five, and there are two test items in each syllable length. Before the test items, there are two practice items, and two syllables are included in each practice item.
2.3.2. Procedures
Participants were tested individually by an experimenter and an assistant in a quiet room. The forward digital recall task was conducted before the non-word repetition task on a laptop or tablet with PowerPoint. The sounds of the numbers or syllablesFootnote 8 were recorded in advance.
A cartoon boy holding a loudspeaker displayed in the middle of the screen was used to present the sound. At the beginning of the task, the experimenter gave the participant the following instructions: “Look, the boy is Xiaobao. Xiaobao will say some numbers or strange words. You should repeat what Xiaobao says when he finishes.” After the instruction, the experimenter slid the PPT to play the practice items first for each task to familiarize the participant with the procedure of the task.
Feedback was given to the participants for the practice items by praising them when they repeated correctly and requiring them to repeat the numbers or syllables once more when they repeated incorrectly. For the test items, no feedback was given to the participants. The assistant audio-recorded the test process and wrote down the participants’ responses. It took about five min for participants to complete the two tasks.
2.4. Data treatment
2.4.1. Aspectual production
Responses transcribed by the assistant were cross-checked by the first author with the recordings before conducting the analysis. The data were treated in four ways. First, the number and proportion of target responses (i.e., sentences with the target aspect marker) for each aspect marker per group were calculated. Second, the mean of the target responses for zai- and -zhe was used to represent imperfective aspect, and that of the -le and -guo was used to represent perfective aspect. Third, the number and proportion of target responses for each “aspect marker + verb” combination per group were counted. Fourth, every participant’s proportion of target responses for each aspect marker was calculated.
2.4.2. VWM
In both the forward digital recall task and the non-word repetition task, the repetition span reported was the maximum number (of digits or non-syllables) repeated correctly. The cutoff point was the number level where a participant responded incorrectly (i.e., when s/he repeated wrongly, or in the wrong order) to both items. For instance, the span of one participant was three if s/he repeated incorrectly both numbers/syllables in the fourth level. Finally, the composite score of the forward digital recall and non-word repetition spans (i.e., the mean score of the two spans) was used to represent each participant’s VWM.
3. Results
All analyses of the results of this study were run in R (R Core Team, 2024) using RStudio, an application for data analysis (https://www.rstudio.com/).
3.1. Performance on each aspect marker
In total, 776 target responses were produced for the four aspect markers by children with DLD and their TD peers. The number and proportion of target responses for each aspect marker per group are presented in Table 3.
Table 3. Number (N) and proportion (%) of target responses for each aspect marker
Note: There were 108 responses produced by the DLD group for zai- and -zhe, respectively, and 162 responses were produced for -le and -guo, respectively. For the TD group, 150 responses were produced for zai- and -zhe, respectively, and 225 responses were produced for -le and -guo, respectively.
A generalized linear mixed model (GLMM) was employed to analyse the participants’ response patterns (target or non-target), with fixed effects for group (TD vs. DLD) and aspect markers (zai-, -zhe, -le, and -guo), alongside random intercepts for subjects and items.
The results showed significant main effects of group (χ 2(1) = 115.1, p < .001) and aspect markers (χ 2(3) = 24.3, p < .001), as well as a significant interaction between them (χ 2(3) = 10.5, p = .015). Post-hoc comparisons indicated that children with DLD produced significantly fewer target responses across all aspect markers compared to their TD peers (zai-: β = −2.26, z = −4.414, p < .001; -zhe: β = −1.28, z = −4.109, p < .001; -le: β = −1.82, z = −6.929, p < .001; -guo: β = −2.56, z = −9.269, p < .001). Within the DLD group, there were significantly more target responses for zai- than for -zhe (β = 2.18, z = 3.362, p = .004), -le (β = 2.18, z = 3.673, p = .001), and -guo (β = 2.73, z = 4.556, p < .001), while there were no significant differences between -zhe, −le, and -guo. Within the TD group, there were significantly more target responses for zai- than for -zhe (β = 3.17, z = 4.256, p < .001), -le (β = 2.62, z = 3.733, p = .001), and -guo (β = 2.43, z = 3.460, p = .003), while there were no significant differences between -zhe, -le, and -guo. The significant interaction effect between group and aspect markers suggests that the performance difference between the TD and DLD groups varied across different aspect markers. Specifically, the DLD group produced the least target responses for -guo among the four aspect markers, whereas the target responses for -guo produced by the TD group ranked second among the four aspect markers, as displayed in Figure 1.
Figure 1. Proportion of target responses for each aspect marker.
Analyses in this section indicated that children with DLD showed significantly worse performance in the production of the four aspect markers than their TD peers. However, like their TD peers, children with DLD showed better performance on the pre-verbal marker zai- than on the post-verbal markers -le, -zhe, and -guo, noting that while all three are post-verbal, only -le and -guo mark perfective aspect, whereas -zhe marks imperfective aspect. Furthermore, the significant interaction effect between the group and aspect markers showed that children with DLD and their TD peers had the largest differences observed for the experiential marker -guo compared to the other three markers, which indicated that children with DLD faced particular difficulties in the production of -guo.
3.2. Performance on perfective and imperfective aspect
The numbers and proportions of target responses for perfective and imperfective aspect are presented in Table 4.
Table 4. Number (N) and proportion (%) of target responses for perfective and imperfective aspect
Note: There were 216 responses produced by the DLD group for imperfective aspect, and 324 responses were produced for perfective aspect. For the TD group, 300 responses were produced for imperfective aspect, and 450 responses were produced for perfective aspect.
A GLMM was performed to analyse the participants’ response patterns (target or non-target). The model included fixed effects for the group (TD vs. DLD) and aspect (perfective vs. imperfective), as well as random intercepts for subjects and items.
There were significant main effects of group (χ 2(1) = 126.0, p < .001) and aspect (χ 2(1) = 4.2, p = .041), as well as a significant interaction between them (χ 2(1) = 5.11, p = .024). Post-hoc comparisons indicated that children with DLD produced significantly fewer target responses for both perfective and imperfective aspect compared to their TD peers (perfective aspect: β = −2.23, z = −11.305, p < .001; imperfective: β = −1.52, z = −5.932, p < .001). Within the DLD group, there were significantly more target responses for imperfective aspect than for perfective aspect (β = 1.36, z = 2.628, p = .009). Within the TD group, however, no significant difference was observed between perfective and imperfective aspect (β = 0.65, z = 1.263, p = .207). The significant interaction effect between group and aspect suggests that the DLD group showed a marked difficulty with perfective aspect compared to imperfective aspect, while the TD group showed a strong performance on both perfective and imperfective aspect, as displayed in Figure 2.
Figure 2. Proportions of target responses for perfective and imperfective aspect.
Analyses in this section demonstrated that children with DLD showed significantly worse performance than TD children on both perfective and imperfective aspect, while they performed better on imperfective aspect than on perfective aspect. However, one should be cautious in concluding that children with DLD showed asymmetrical performance on aspectual production, because the morphological salience of the four aspect markers is different. Specifically, the imperfective marker zai- is morphologically more salient than the other three markers, and children in both groups displayed better performance on zai- than on the other three markers, as analysed in the last section. Further, there were no significant differences between the three post-verbal markers. Hence, further analyses are necessary to determine the asymmetry between perfective and imperfective aspect by excluding the impact of lexical aspect and morphological salience.
3.3. Impact of lexical aspect on aspectual production
In the priming picture-description task, the combinations of aspect markers with their prototypical and non-prototypical verbs were controlled in the priming sentences. To know how lexical aspect impacts aspectual production, the target responses of each aspect marker were compared between its combinations with their prototypical verbs and non-prototypical verbs on the one hand. On the other hand, the four aspect markers were compared when they were combined with their prototypical verbs. The distribution of the target responses for each “aspect marker + verb” combination is displayed in Figure 3.
Figure 3. Distribution of target responses for each “aspect marker + verb” combination.
3.3.1. Comparisons between the combinations of aspect markers with their prototypical and non-prototypical verbs
GLMMs were employed respectively for each aspect marker to examine participants’ response patterns (target or non-target), incorporating fixed effects for the group (TD vs. DLD) and verb types (zai-: Activity vs. Accomplishment; -zhe: Activity vs. State; -le: Achievement vs. Accomplishment vs. Activity; -guo: Achievement vs. Accomplishment vs. Activity), as well as random intercepts for subjects and items. The results are the following:
-
- zai-: No significant main effects were revealed in group (χ 2(1) = 0.019, p = .891), verb types (χ 2(1) = 0.014, p = .905), or interaction between them (χ 2(1) = 0.014, p = .907), which indicated that children in both groups performed similarly when zai- was combined with Activity and Accomplishment verbs.
-
- zhe: Significant main effects were found in both groups (χ 2(1) =17.71, p < .001), verb types (χ 2(1) = 28.65, p < .001). However, there were no significant interaction effects between them (χ 2(1) = 1.17, p = .280). Post-hoc comparisons indicated that children with DLD produced significantly fewer target responses compared to their TD peers when -zhe was either combined with Activity verbs or State verbs (Activity: β = −1.14, z = −2.566, p = .010; State: β = −1.86, z = −3.513, p < .001). Within the DLD group, there were significantly more target responses when -zhe was combined with Activity verbs than with State verbs (β = 2.86, z = 4.537, p < .001). The same holds for the TD group (β = 2.14, z = 4.176, p < .001).
-
- le: There was a significant main effect in group (χ 2(1) = 45.198, p < .001) rather than in verb types (χ 2(2) = 4.250, p = .119) or the interaction between them (χ 2(2) = 0.363, p = .834). Post-hoc comparisons indicated that children with DLD produced significantly fewer target responses compared to their TD peers when -le was either combined with Achievement, Accomplishment, or Activity verbs (Achievement: β = −1.83, z = −3.476, p < .001; Accomplishment: β = −2.19, z = −4.909, p < .001; Activity: β = −1.89, z = −4.059, p < .001).
-
- guo: Significant main effects were found in both groups (χ 2(1) =71.31, p < .001) and verb types (χ2(2) = 17.66, p < .001) rather than their interaction (χ 2(2) = 1.97, p = .374). Post-hoc comparisons indicated that children with DLD produced significantly fewer target responses compared to their TD peers when -guo was either combined with Activity, Accomplishment, or Achievement verbs (Activity: β = −3.43, z = −5.966, p < .001; Accomplishment: β = −2.54, z = −5.151, p < .001; Achievement: β = −3.37, z = −5.172, p < .001). Within the DLD group, there were significantly more target responses when -guo was combined with Activity and Accomplishment verbs than with Achievement verbs (Activity vs. Achievement: β = 2.348, z = 3.124, p = .005; Accomplishment vs. Achievement: β = 1.893, z = 2.515, p = .032), while no significant difference was observed between Activity and Accomplishment verbs. Within the TD group, significantly more target responses were produced when -guo was combined with Activity verbs than with Achievement verbs (Activity vs. Achievement: β = 2.409, z = 3.723, p < .001), while there were no significant differences between Activity and Accomplishment verbs, as well as Accomplishment and Achievement verbs.
The findings in this section showed that children with DLD performed similarly with TD children on zai-, while displaying significantly worse performance on -zhe, −le, and -guo than their TD peers. In addition, the impact of lexical aspect on aspectual production was confirmed in aspect markers -zhe and -guo.
3.3.2. Comparisons between the four aspect markers of their combinations with prototypical verbs
To further investigate the impact of lexical aspect on aspectual production, the four aspect markers were compared when they were combined with their prototypical verbs using the GLMM, which analysed the participants’ response patterns (target or non-target). The model incorporated fixed effects for group (TD vs. DLD) and aspect markers (zai-, -zhe, -le, and -guo), along with random intercepts for subjects and items.
There were no significant main effects of group (χ 2(1) = 0.011, p = .917) or aspect markers (χ 2(3) = 0.111, p = .991). However, a significant interaction effect was found between them (χ 2(3) = 8.587, p = .035). Post-hoc comparisons indicated that children with DLD produced significantly fewer target responses for the perfective markers -le and -guo compared to their TD peers (-le: β = −1.61, z = −3.327, p < .001; -guo: β = −2.94, z = −5.896, p < .001). However, the two groups indicated similar performance for the imperfective markers zai- and -zhe, as presented in Figure 4.
Figure 4. Proportion of target responses for each “aspect marker + prototypical verb” combination.
The results of GLMM in this section showed that children with DLD underperformed their TD peers in the production of the perfective markers -le and -guo, which contrast with the imperfective markers zai- (pre-verbal) and -zhe (post-verbal), when these markers were combined with their prototypical verbs. Nonetheless, there were similar patterns of aspectual production, no matter whether these markers were combined with their prototypical verbs, as displayed in Figures 1 and 4.
3.4. VWM
Table 5 displays the mean score of the forward digital recall and non-word repetition tasks, as well as their composite score for the VWM.
Table 5. Participants’ mean spans of digital recall, non-word repetition, and VWM
The generalized linear models (GLMs), which were selected for their flexibility in handling both continuous and discrete outcome variables while accommodating potential non-normal distributions through appropriate link functions, were employed to examine whether children with DLD and their TD peers had different digital recall span, non-word repetition span, and the VWM span.
The results of the GLMs revealed statistically significant differences in the digital recall span (β = 1.06, z = 3.13, p = .002), non-word repetition span (β = 0.536, z = 2.53, p = .011), and the VWM span (β = 0.796, z = 3.23, p = .001) between children with DLD and their TD peers. TD children showed a higher digital recall span (with an odds ratio of 2.87), non-word repetition span (with an odds ratio of 1.71), and VWM span (with an odds ratio of 2.22) than the DLD group, which indicated that children with DLD showed significantly weaker VWM than TD children.
3.5. Correlation analysis
In this section, correlation analyses were employed to examine the relationships between the VWM of children with DLD and their performance on aspectual production. The non-verbal intelligence and global language ability of children with DLD were treated as controlled variables because these children demonstrated significantly worse non-verbal intelligence and global language ability than their TD peers. Every participant’s proportion of target responses for each aspect marker was used in the correlation analysis. Since the proportions of aspect markers were represented in percentage forms, Kendall’s partial tau correlations were conducted.
After accounting for the effects of non-verbal intelligence and global language ability, the analysis revealed non-significant correlations between the VWM of children with DLD and their performance on aspect markers zai- (τ = −0.078, p = .673), -zhe (τ = 0.132, p = .476), and -le (τ = 0.304, p = .100). However, a significant correlation was observed between the VWM of children with DLD and their performance on -guo (τ = 0.391, p = .034), indicating the significant association between their VWM and the performance in the production of -guo.
Furthermore, the relationships between the VWM of children with DLD and their performance on perfective aspect (the mean proportion of the target responses for -le and -guo) and imperfective aspect (the mean proportion of the target responses for zai- and -zhe) were examined with Kendall’s partial tau correlations while controlling their non-verbal intelligence and global language ability in two ways. First, the correlations between the VWM of children with DLD and their performance on perfective and imperfective aspect were examined, respectively. The results revealed a marginal correlation between the VWM and their perfective aspect (τ = 0.356, p = .055) rather than imperfective aspect (τ = 0.059, p = .748). Second, the correlation between their VWM and the difference between the proportion of perfective and imperfective aspectFootnote 9 was examined to see whether children with DLD with varying VWMs exhibited differential sensitivity to the aspectual differences. The results did not reveal a significant correlation between their VWM and the aspectual differences (τ = −0.266, p = .151).
The analyses in this section indicated that the VWM of children with DLD only significantly correlated with their performance in the production of the experiential marker -guo and perfective aspect.
3.6. Regression analysis
To further determine the extent to which the VWM of children with DLD could explain the variance in their production of each aspect marker, as well as perfective and imperfective aspect, the mixed effects logistic regressions were conducted to model the variation in aspectual production by children with DLD as a function of their VWM, respectively, while controlling for their non-verbal intelligence and global language ability. The results indicated that the VWM of children with DLD did not have a significant effect on their production of zai- (β = −0.094, z = −0.124, p = .832), -zhe (β = 0.477, z = −0.954, p = .340), -le (β = 0.521, z = 1.234, p = .217), or imperfective aspect (β = 0.121, z = 0.389, p = .697). However, significant effects of the VWM were revealed on -guo (β = 5.135, z = 2.46, p = .014) and perfective aspect (β = 1.246, z = 3.029, p = .002), indicating that children with DLD who owned higher VWM were more likely to produce the target responses for the experiential marker -guo as well as perfective aspect.
3.7. Individual analysis
Due to their heterogeneous language development patterns in general (Bishop, Reference Bishop, Verhoeven and van Balkon2003; Kuiack & Archibald, Reference Kuiack and Archibald2024; Lancaster & Camarata, Reference Lancaster and Camarata2019; van der Lely, Reference van der Lely, Levy and Schaeffer2003; van Weerdenburg et al., Reference van Weerdenburg, Verhoeven and Van Balkom2006), the performance of children (i.e., the proportion of target responses) in the DLD group on aspectual production was presented individually in Figures 5–8.
Figure 5. Individual performance of children with DLD on each aspect marker.
As shown in Figure 5, children with DLD demonstrated heterogeneous performance in the production of Mandarin aspect markers. Specifically, some children with DLD did not produce any -zhe (DLD2, DLD3, and DLD6) or -guo at all (DLD2, DLD3, DLD4, DLD6, DLD9, DLD13, DLD14, and DLD16), while some of them performed better on -guo than on other aspect markers (DLD1 and DLD18).
To better display the individual performance of the 18 children in the DLD group, we divided the proportions of target responses into four ranges (0–24%, 25–49%, 50–74%, and 75–100%) and counted the number of children in the DLD group within each range for one individual aspect markerFootnote 10. The distribution of children in the DLD group for each aspect marker is displayed in Figure 6.
Figure 6. Distribution of children with DLD for each aspect marker.
Figures 5 and 6 demonstrated that there was heterogeneous performance in the production of individual aspect markers within the DLD group, on the one hand. On the other hand, the individual analysis on specific Mandarin aspect markers confirmed the above findings that children with DLD performed better in the production of the pre-verbal marker zai- than in the other three post-verbal markers.
Individual performance in the production of perfective and imperfective aspect was presented with Figures 7 and 8 to better understand the heterogeneous performance of children with DLD on aspectual production.
Figure 7. Individual performance of children with DLD on perfective and imperfective aspect.
From Figure 7, it can be observed that three children with DLD demonstrated better performance on perfective aspect than on imperfective aspect (DLD1, DLD2, and DLD18), although they performed better on imperfective aspect than perfective aspect as a group. Furthermore, like the analyses on individual aspect markers, the number of children in the four ranges of proportions in the DLD group on perfective and imperfective aspect was countedFootnote 11, and the results are shown in Figure 8.
Figure 8. Distribution of children with DLD for perfective and imperfective aspect.
As shown in Figure 8, most children in the DLD group produced >50% of target responses for imperfective aspect, while the majority of them produced <50% target responses for perfective aspect.
4. Discussion
This study investigated the performance of 18 four- to six-year-old Mandarin-speaking children with DLD in the production of the four typical Mandarin aspect markers, as well as on perfective and imperfective aspect, and whether their performance on aspectual production would relate to their VWM. This was accomplished by administering a priming picture-description task alongside a forward digital recall and a non-word repetition task to both children with DLD and their age-matched TD peers. In this section, the results are discussed according to the three research questions.
4.1. What is the performance of children with DLD in the production of the four typical Mandarin aspect markers, as compared to age-matched TD children?
The results of this study indicated that preschool Mandarin-speaking children with DLD produced significantly fewer target responses than age-matched TD children on all four aspect markers, especially on the experiential marker -guo. This finding is consistent with He et al. (Reference He, Sun and Tian2013) and Chen et al. (Reference Chen, An, Dai and He2022), but it is different from Chen et al. (Reference Chen, An and He2021). The latter study found that preschool children with DLD performed similarly to their age-matched TD peers in producing sentences with the pre-verbal marker zai- while demonstrating worse performance in the production of the three post-verbal markers.
The inconsistent findings between this study, He et al. (Reference He, Sun and Tian2013) and Chen et al. (Reference Chen, An, Dai and He2022) on the one hand, and Chen et al. (Reference Chen, An and He2021) on the other hand, are arguably attributable to the interaction between lexical aspect and grammatical aspect. To illustrate, the verbs used in the priming sentences of the Chen et al. (Reference Chen, An and He2021) task were all Activity verbs, which are the prototypical verbs of the progressive marker zai-. In the current study, as well as in He et al. (Reference He, Sun and Tian2013) and Chen et al. (Reference Chen, An, Dai and He2022), however, the progressive marker zai- was combined with both Activity and Accomplishment verbs (i.e., the non-prototypical verbs) in the priming sentences. According to the Aspect Hypothesis, children tend to combine aspect markers with their prototypical verbs at the initial stage of aspectual development, and the tendency weakens over time (Andersen & Shirai, Reference Andersen and Shirai1994, Reference Andersen, Shirai, Ritchie and Bhatia1996). Therefore, preschool children with DLD, who showed lower language ability than age-matched TD children, may struggle less combining zai- with prototypical verbs and thus performed better than other aspect markers in Chen et al. (Reference Chen, An and He2021).
The results of this study also demonstrated that children with DLD, like their TD peers, performed better on the pre-verbal marker zai- than on the three post-verbal markers -le, -zhe, and -guo. This pattern was not influenced by the impact of lexical aspect. Specifically, as displayed in Figures 1 and 4, children in both groups showed similar distributions of the four markers, no matter whether these markers were combined with their prototypical verbs. He et al. (Reference He, Sun and Tian2013) and Chen et al. (Reference Chen, An and He2021) also reported better performance on the pre-verbal marker zai- than on the post-verbal markers in children with DLD. In Chen et al. (Reference Chen, An, Dai and He2022), although intra-group comparisons were not conducted, data in their study indicated that the mean proportion of the target responses produced by children with DLD for the pre-verbal marker zai- (73.3%) was much higher than that of the post-verbal markers (-le: 30.5%, -zhe: 35.0%, and -guo: 21.7%)Footnote 12. Better performance by children with DLD on the pre-verbal marker zai- than on the post-verbal markers might be attributed to its morphologically more salient forms than the other three markers, as well as the impact of temporal context.
The impact of the morphologically more salient form of zai- than the other three markers was predicted by van Hout’s Morphological Salience hypothesis (van Hout, Reference van Hout2008). According to this hypothesis, “the semantics of morphologically salient paradigms is acquired early” (van Hout, Reference van Hout2008: 1753). In Mandarin, the progressive marker zai- precedes the predicate verb. It is pronounced with stress and in the falling tone. In contrast, the aspect markers -le, -zhe, and -guo follow the predicate verb and are pronounced in the neutral tone without stress. Therefore, zai- is indeed more salient and easier to produce than the other three post-verbal markers. As a result, children with DLD and their TD peers produced more target responses for the pre-verbal marker zai- than on the post-verbal -le, -zhe, and -guo.
Besides the impact of morphological salience, the temporal context of the priming picture-description task might also contribute to the better performance on zai- than on the other three markers. Because zai- is an imperfective marker, and children demonstrated better performance on imperfective aspect in the “here and now/recent past” context (Anderson et al., Reference Anderson, Matlock and Spivey2013; Dosi, Reference Dosi2019; Dosi et al., Reference Dosi, Andreou and Peristeri2018; Gagarina, Reference Gagarina2004). The priming picture-description task in this study provided a “here and now” context in that children were required to describe pictures that had no cue of past or future time. As a result, children with DLD performed similarly to their TD peers on the imperfective markers zai- and -zhe, while demonstrating worse performance on the perfective markers -le and -guo when these markers were combined with their prototypical verbs.
Individual performance of children with DLD in the production of each aspect marker was analysed in this study for the first time, and the variation within the DLD group was observed. To be specific, there were children in the DLD group who showed ceiling performance, while other children performed rather poorly for each aspect marker. Through the presentation of individual performance on each individual aspect marker in Figures 5 and 6, this study found that some children with DLD displayed weak performance on zai- and some of them showed even better performance on -le, -zhe, or -guo than on zai-, although the DLD group as a whole performed better on zai- than on -le, -zhe, and -guo. Furthermore, while some children with DLD performed better on -guo than on other aspect markers, some of them did not produce any -guo at all.
To sum up, preschool Mandarin-speaking children with DLD displayed worse performance in the production of all the Mandarin aspect markers than their age-matched TD peers. In addition, children with DLD, like TD children, performed better on the pre-verbal marker zai- than on the post-verbal markers -le, -zhe, and -guo, which could be attributed to the morphologically more salient form of zai- compared to other markers and the “here and now” context of the priming picture-description task. The individual analysis demonstrated the heterogeneous performance of children with DLD in the production of grammatical aspect.
4.2. Do children with DLD show asymmetrical performance in the production of perfective and imperfective aspect?
In this study, the number of the target responses for zai- and -zhe was used to represent imperfective aspect, and that of -le and -guo was used to represent perfective aspect. The results showed that children with DLD performed worse on both perfective and imperfective aspect than their age-matched TD peers, which is consistent with previous studies that did not intertwine the assessment of aspect with tense. By excluding the involvement of tense, Leonard et al. (Reference Leonard, Lukács and Kas2012) found that Hungarian-speaking children with DLD performed worse than age-matched as well as language-matched TD children in the production of both perfective and imperfective aspect in an elicited production task. Similarly, the findings of Fletcher et al. (Reference Fletcher, Leonard, Stokes and Wong2005) showed that children with DLD who spoke Cantonese (a language without grammatical tense) performed worse than both age-matched and language-matched TD children in the production of Cantonese aspect markers that indicate both perfective and imperfective aspect. However, neither study conducted intra-group comparisons to examine whether children with DLD demonstrate asymmetrical performance in the production of different grammatical aspect types, namely perfective versus imperfective.
In this study, we found that children with DLD demonstrated better performance on imperfective aspect than on perfective aspect as a group in the production task, although some individuals demonstrated better performance on perfective aspect than on imperfective aspect.
This finding was consistent with Leonard et al. (Reference Leonard, Deevy, Kurtz, Chorev, Owen and Polite2007). Leonard et al. (Reference Leonard, Deevy, Kurtz, Chorev, Owen and Polite2007) found that English-speaking children with DLD showed worse performance than language-matched TD children in the production of the perfective marker -ed but displayed similar performance to their language-matched TD peers in the production of the progressive marker -ing in an acting-out elicitation task. Unlike this study, however, Konstantzou (Reference Konstantzou2015) showed that 6-year-old Greek-speaking children with DLD performed similarly to their age-matched TD children in the production of perfective aspect, but showed worse performance in the production of imperfective aspect than their TD peers in an elicitation production task that consisted of video stimuli. The different findings between Leonard et al. (Reference Leonard, Deevy, Kurtz, Chorev, Owen and Polite2007) and Konstantzou (Reference Konstantzou2015) could be explained by the semantical complexity of grammatical aspect. In English, tense and aspect are confounded in the inflection -ed, while -ing only represents progressive aspect. In Greek, imperfective aspect can assign an imperfective, a habitual, a continuous, a progressive, an iterative, or a durative interpretation without morphological change, while perfective aspect assigns a perfective, semelfactive, or punctual interpretation (Xydopoulos, Reference Xydopoulos1996). As common sense, the morphological inflection with more complex semantics would be more difficult to process than the one with simpler semantics. Nonetheless, the Mandarin aspect markers represent both perfective and imperfective aspect have simple meanings (as shown in the Introduction part). Therefore, the semantic complexity of grammatical aspect could not explain the asymmetrical performance of children with DLD in the production of grammatical aspect.
We argued that the interaction between grammatical aspect and temporal context may contribute to the conflicting reports of previous studies. Previous studies have indicated the impact of temporal context on the acquisition of grammatical aspect (Anderson et al., Reference Anderson, Matlock and Spivey2013; Gagarina, Reference Gagarina2004). More specifically, young children have a strong preference for using perfective aspect for the past (or distant past) events and imperfective aspect for the “here and now/recent past” events. The interaction between grammatical aspect and temporal context was also reported in aspectual production by children with DLD. Dosi et al. (Reference Dosi, Andreou and Peristeri2018) found that Greek-speaking children with DLD performed better on imperfective aspect than on perfective aspect in an elicitation completion task that provides the “here and now” context; however, they showed better performance on perfective aspect than on imperfective aspect in a picture-based narrative task that cues a past tense context. Furthermore, like the production of grammatical aspect, in fact, children with DLD also demonstrated asymmetrical performance in the comprehension of grammatical aspect. For instance, English-speaking children and Greek-speaking children perform better in comprehending perfective aspect than in imperfective aspect in the past context (e.g., Dosi, Reference Dosi2019; Konstantzou, Reference Konstantzou2015; Stuart & van der Lely, Reference Stuart and van der Lely2015), while Turkish-speaking children and Mandarin-speaking children with DLD displayed better performance in the comprehension of imperfective aspect than in perfective aspect in the “here and now” context (e.g., Chen & Durrleman, Reference Chen and Durrleman2022; Duman & Topbas, Reference Duman and Topbas2016).
In this study, the priming picture-description task offered a “here and now” context in which the participants needed to describe the picture(s) that had no cue of the past or future time, and the acting-out elicitation task in Leonard et al. (Reference Leonard, Deevy, Kurtz, Chorev, Owen and Polite2007) offered a “recent past” context in which the participants should act immediately after the experimenter’s instruction. Therefore, children with DLD in this study and Leonard et al. (Reference Leonard, Deevy, Kurtz, Chorev, Owen and Polite2007) performed better in the production of imperfective aspect, which requires simpler aspectual operations than perfective aspect in the “here and now” context. In contrast, the elicitation production task that used video stimuli in Konstantzou (Reference Konstantzou2015) provided a “distant past” context since the videos were recorded in advance. Therefore, better performance was observed in perfective aspect rather than in imperfective aspect.
It might be argued, however, that the prominent performance on zai- than on other markers in children with DLD would contribute to their better performance observed on imperfective aspect compared to perfective aspect. Specifically, both children with DLD and their TD peers produced significantly more target responses for the morphologically more salient zai- than for the other three post-verbal markers. Therefore, better performance of children with DLD on imperfective aspect than on perfective aspect might be the result of the morphological salience effect rather than aspect itself. Nonetheless, the results of the comparison between the four aspect markers when they were combined with their prototypical verbs showed that children with DLD performed similarly to their TD peers on the imperfective markers zai- and -zhe, while displaying significantly worse performance than TD children on the perfective markers -le and -guo, which may support our claim that there was indeed an asymmetrical performance between perfective and imperfective aspect in Mandarin-speaking children with DLD.
4.3. Does the VWM of children with DLD correlate with their performance in the production of grammatical aspect?
Previous studies reported the weaker VWM in children with DLD compared to their TD peers (Archibald & Gathercole, Reference Archibald and Gathercole2007; Larson & Weismer, Reference Larson and Weismer2022; Vugs et al., Reference Vugs, Hendriks, Cuperus, Knoors and Verhoeven2017; Weismer et al., Reference Weismer, Evans and Hesketh1999), and their weak VWM has been found to be linked to other aspects of their grammatical competency, such as their weak performance in the production of morphosyntactic inflections that express tense (e.g., see Bishop et al., Reference Bishop, Adams and Norbury2006; Norbury et al., Reference Norbury, Bishop and Briscoe2001). However, no previous studies have explored the link between these children’s VWM and their production of grammatical aspect, although it was hypothesized in Chen et al. (Reference Chen, An and He2021) that the difficulty of Mandarin-speaking children with DLD in the production of the three post-verbal markers may be associated with weaknesses in their VWM.
In Mandarin, the three post-verbal markers -le, -zhe, and -guo are stressless with the neutral tone. The pronunciation features of the post-verbal aspect markers make them less conspicuous than the pre-verbal marker zai- in sentence production. Accessing and retrieving inconspicuous morphological forms during language production is difficult for individuals with weak VWM, because the VWM must compensate for weaker encoding by relying on memory traces that are less robust when they try to retrieve inconspicuous morphological forms (Baddeley, Reference Baddeley2003; Schwering & MacDonald, Reference Schwering and MacDonald2020). As a result, Chen et al. (Reference Chen, An and He2021) proposed that children with DLD, who are reported to show deficits in their VWM, would not access and retrieve the inconspicuous post-verbal aspect markers as efficiently as TD children in the process of sentence formation.
Nonetheless, Chen et al. (Reference Chen, An and He2021) did not measure the VWM of children with DLD. Using a forward digital recall task and a non-word repetition task, this study examined the VWM of children with DLD. The results confirmed the findings of previous studies (e.g., see Archibald & Gathercole, Reference Archibald and Gathercole2007; Larson & Weismer, Reference Larson and Weismer2022; Vugs et al., Reference Vugs, Hendriks, Cuperus, Knoors and Verhoeven2017; Weismer et al., Reference Weismer, Evans and Hesketh1999) that children with DLD demonstrate weaker VWM than their age-matched TD peers. The results of correlation analyses and regression analyses, which excluded the potential effect of non-verbal intelligence and global language ability, found that the VWM of Mandarin-speaking children with DLD is significantly correlated with their performance on -guo, and children with DLD who had higher VWM were more likely to produce the target responses for the experiential marker -guo.
Then, why did the VWM of Mandarin-speaking children with DLD have a significantly positive effect on their production of -guo rather than on other markers? It could be explained by the delay in aspectual acquisition by children with DLD, as proposed by Chen and Durrleman (Reference Chen and Durrleman2022). Specifically, the experiential marker -guo is the latest aspect marker acquired by TD children (Erbaugh, Reference Erbaugh and Slobin1992; Li et al., Reference Li, Liang and Wu2022). We believed that 4- to 6-year-old children with DLD, who exhibited language abilities with those of younger TD children (Sansavini et al., Reference Sansavini, Favilla, Guasti, Marini, Millepiedi and Di Martino2021), might have acquired the aspect markers zai-, -zhe, and -le, and can process these markers automatically. For the experiential marker -guo, they would not acquire it completely yet. As a result, when they tried to retrieve the inconspicuous -guo in the production task, they had to use their VWM to compensate for the retrieval.
In addition, the strong prediction of the VWM of children with DLD on their production of perfective aspect was also observed in this study. Since perfective aspect was represented by -le and -guo, and no significant correlation was found between the VWM of children with DLD and their production of -le, we assumed the significant effect of the VWM on perfective aspect could be attributed to the significant correlation between the VWM and -guo.
To sum up, this study investigated the production of grammatical aspect and the role of the VWM on aspectual production by four- to six-year-old Mandarin-speaking children with DLD using a priming picture-description task, as well as a digital recall and a non-word repetition task, as compared to age-matched TD children. The results of this study are the following: first, children with DLD produced significantly fewer target responses for all four typical Mandarin aspect markers than age-matched TD children as a group, and both groups demonstrated better performance on the pre-verbal zai- than on the three post-verbal markers. Second, comparisons on perfective and imperfective aspect demonstrated that children with DLD performed better on imperfective aspect than on perfective aspect, although they displayed worse performance on both than TD children. Third, the VWM of children with DLD, which was worse than age-matched TD children, had a significant effect on the production of the experiential marker -guo and perfective aspect. In addition, individual analyses demonstrated a variation within the DLD group on the production of individual aspect markers, as well as perfective and imperfective aspect.
The findings of this study proved that the asymmetrical performance in the production of grammatical aspect arises not only in languages that grammatically encode tense but also in those that are tenseless, both in TD children as well as in children with DLD. The significantly positive effect of the VWM of children with DLD on their production of the experiential marker -guo and perfective aspect means that training on the VWM of children with DLD may improve their ability in the production of -guo and perfective aspect. Furthermore, the heterogeneous performance observed in the DLD group calls for cautious consideration in the diagnosis and intervention of grammatical aspect problems in each child with DLD. To facilitate the acquisition of grammatical aspect by children with DLD, each aspect marker should be carefully assessed and, when appropriate, addressed in speech and language therapy. Nevertheless, a few verbs were used with different subjects and objects for different aspect markers due to the limited availability of verbs that are both clearly acquired by young children with DLD and depictable in pictures, which might have a potential impact on the results of this study. Future studies can use more verbs to test older children with DLD to exclude this potential impact.
Data availability statement
The data that support the findings of this study are available from the corresponding author upon request. The data are not publicly available due to privacy or ethical restrictions.
Acknowledgements
The authors would like to express their gratitude to the children, parents, teachers, and therapists who participated in the study and to their research team in Guangdong University of Foreign Studies, whose collaboration made the study possible.
Funding statement
This study was supported by the Swiss National Science Foundation (grant number PR00P1_193104/1) and Guangdong Philosophy and Social Science Fund Project (grant number GD25CWY12).
Competing interests
The authors declare none.
Open access
