Development Sequence of Person Markers in Pre-school Children

This study investigates the development sequence of person markers in pre-school children. It observed participants’ accurate use of person markers through a picture-supported elicitation task. Ten children with normal language development, aged between 2.5 and 4 years, participated in the task. Eight different verbs with three person markers for progressive aspect in present tense were elicited from participants. The mean score and standard deviation for each person marker used indicated that participants faced difficulty in producing accurate second person marker. Each participant showed highest accuracy in producing third person marker for each verb form. An implication of the study can be to determine the accuracy order of verbal inflection based on person markers in the process of Bengali language acquisition in children. It can also be helpful in cross-linguistic studies on morphological development of young children.

One of the generally accepted notions regarding child language acquisition is that by age of 4, children master the basic structures of their first language (Lightbown & Spada, 2006). This mastery does not necessarily assure the accuracy of their performance. For instance, children having English as their first language can show an asymmetrical pattern in using verbal inflections (Buijs, Reijen & Weerman, 2013). However, children's grammar also develops gradually as they grow older (Ionin & Wexler, 2002). Several studies have tried to investigate the development sequence of linguistic items, with specific regard to morphological features of English. On the other hand, even though Bengali has richly inflected morphological features, it did not receive much attention (Chakraborty & Leonard, 2012).
In this study the researcher examined production and accuracy of three person markers with eight verb forms for progressive aspect in present tense. The purpose of the study is to find the development sequence of person markers in pre-school children. The researcher also took a brief look at how children can use other markers, such as simple aspect, even after getting prompts in progressive aspect. Thus the central research question of this study was: what is the development sequence, with regard to Bengali person markers, in children aged 2.5 to 4?
The study examines existing literature on the language acquisition process of Bangla. It could also contribute to the understanding of the difficulty of Bangla person marker acquisition among preschool children.

Literature Review
In Bangla language, a total of three types of verbs can be identified: simple, conjunct, and complex (Bhattacharyya, Chakrabarti & Sharma, 2006;Bhattacharja, 2010;Chatterjee, 2014). A detailed table demonstrating the verbal system of Bangla language is given in Appendix A. According to Cysouw (2001), there are three categories of linguistic coding that are used to give reference to speaker, addressee, and participant. These specialized linguistic elements that code for speaker is called "first person," the coding for addressee is "second person," and the coding for any other participant is "third person" (Cysouw, 2001, p. 6).
In Bangla, the agreement of verb does not change its form based on gender or number; rather it changes with respect to tense, aspect, modality, and person only (Bhattacharya, Choudhury, Sarkar, & Basu, 2005). This verb agreement feature is restricted to three forms of persons, i.e., 1 st person, 2 nd person, and 3 rd person (Mondal, 2014). In order to use these person markers separately based on different verb forms, a child needs to understand the differences among different persons. In this respect, research showed that one of the most important steps in cognitive development of children is to understand the difference between the "self " and the "rest of the world" (Schmalstieg, 1977, p. 72). Therefore, it might be reasonable to claim that the person is considered to be more "vulnerable" in terms of error production than tense and aspect (Chakraborty & Leonard, 2012, p. 50).
Researchers claimed that children acquire the knowledge of verbal agreement at a very early age (Buijs, Reijen & Weerman, 2013). However, Ionin and Wexler (2002) mentioned that while acquiring L1, children aged 2 to 4 years can show inconsistent behavior in using verb forms. In case of English language, a study also showed that development of "noun versus verb" agreement and "verbal inflection" in children is asymmetrical (Conti-Ramsden & Windfuhr, 2002). Furthermore, in case of Bengali, the development of verbal inflection for 1 st and 3 rd person emerges earlier than 2 nd person marker (Chakraborty & Leonard, 2012).
In the case of language itself, Bengali is considered to have "morphological richness" since a single verb root can take more than 50 different forms (Dasgupta & Ng, 2006, p. 312). Pienemann (1999, 2003 proposed processability theory that suggested that language acquisition deals with one's capacity to process. This processing capacity allows an individual to exhibit his/her existing knowledge in real life (as cited in Lightbown & Spada, 2006). Therefore, the diversity that Bengali morphology offers may cause the possible difficulty a child might have to process varied choices during language acquisition.
In order to observe grammatical production capacity of participants, elicitation tasks can be considered one of the effective instruments (Hesketh, 2004). However, it can also lead to some difficulties. Mackey and Gass (2005) mentioned that even after receiving prompts, participants can use different forms other than the target form. Hence, a researcher needs to design the study in such a way that it can ensure the target response through elicitation.
The present study used an elicitation task to investigate the development sequence of person markers in young children. A similar study by Chakraborty and Leonard (2012) used the elicitation task and demonstrated that Bengali-speaking children achieve high level of accuracy in verb inflection both in present and past tense (Chakraborty & Leonard, 2012). It also claimed that agreement paradigm of Bengali could be the reason for which children used accurate verbal inflection (Chakraborty & Leonard, 2012).

Methodology
This study was a quantitative research on the development sequence of person markers in children aged 2.5 to 4 years. There are 4 kinds of aspects in Bengali: simple, progressive, perfect, and habitual (Mondal, 2014). This study focused on participants' correct use of three person markers for progressive aspect in present tense.
Participants: The participants were 10 children from Old Dhaka in Bangladesh. There was no particular rationale behind choosing this place, however. The researcher simply had an acquaintance through whom she could get access to those children. This particular aspect also helped the researcher to control some of the variables, i.e., socio-economic status, overall language input, etc. of the participants. They all belonged to middle-class families.
The age range of the participants was from 2.5 to 4 years. The group of participants comprised five boys and five girls. All of them had Bengali as their first language. They were selected based on their age. The researcher also made sure that they have typical language development by asking their parents. In order to preserve anonymity, the researcher deliberately did not provide the names of the participants.

Materials
Elicitation task. Elicitation tasks make it easy not only to achieve the target grammatical construction but also to score quickly (Hesketh, 2004). The researcher attempted to elicit person markers use for eight different verbs in Bengali. The participants had to respond in present progressive tense. The researcher herself did not perform the task, however. Rather, another person, who is well acquainted with the child, participated as an elicitor. The elicitor was well informed beforehand regarding her role. She was also strictly forbidden to help or correct the language production of any of the participants. The role of the elicitor was to elicit answers from the participants by showing pictures, and also through enacting different gestures. Before starting the data collection process, the elicitor practiced her role with the help of the researcher. It is to be mentioned that the siblings of the participants also participated in the enactment and were trained beforehand. Verb forms. While choosing the verb forms, the researcher was cautious to select those verbs which were familiar to Bengali children. Being a Bengali speaker herself, the researcher's instinctive tendency was helpful to choose frequently-used verbs. Besides, since this study used enactment, the researcher chose those verbs which were possible to be acted out. The eight verbs were: / jharu deowa/ ("sweep"), /ghumano/ ("sleep"), /hasha/ ("laugh"), /laphano/ ("jump"), /chul achrano/ ("comb"), /ranna kora/ ("cook"), /boi pora/ ("read"), /nacha/ ("dance").
Pictures. Eliciting through pictures or models makes it possible to ensure the use of target grammatical structure (Hesketh, 2003). Hence, the researcher used eight cliparts from different websites that matched the focused eight verb forms. The colorful cartoon images were also helpful in grabbing the attention of the participants. All eight pictures can be found in Appendix B.

Procedures
The study was conducted over a period of two weeks. Each session had different time durations based on participants' elicitation. Allowing time to think can change the quantity and quality of one's language production (Mackey & Gass, 2005). Nonetheless, the researcher considered the young age of the participants and allowed time, provided that they delivered answers. Through elicitation tasks, all the three person markers were being elicited. In order to test the person marker contrasts, the researcher kept the verbs consistent for each person. In other words, each participant used three person markers for a single verb. In this way, they applied eight different verbs, which gave 24 responses from each participant.
The 1 st person marker use was elicited with the help of picture-supported elicitation task. Eight pictures were shown where each carried a particular verb form. While showing the picture, the elicitor asked, /tumi ekhane ki korcho?/ Zy wg GLv ‡b wK Ki ‡Qv? [What are you doing here?].
The participants had to imagine themselves as the child in the pictures. Then, in case of the 2 nd person marker use, the elicitor enacted all the eight verb forms. While enacting each verb, she again asked the participant, /ami akhon ki korchi? /Avwg GLb wK KiwQ? [What am I doing now?]. 3 rd person marker was elicited through the question, /apu/babuta akhon ki korche? / Avcy /evey Uv GLb wK Ki ‡Q? [What is sister/the baby doing?]. In this case, the participant's sibling enacted the verbs and elicitor asked the question. Only in case of participant B the elicitor asked him/her to describe what the baby in the picture was doing.
Responses from the participants were recorded during the elicitation task. Later, the researcher transcribed only those features which were of interest to the study. Apart from the researcher, two listeners were involved in transcribing the data so that there were no discrepancies between participants' responses and transcription. Throughout this paper, the researcher used the Romanized transliteration for Bengali terms along with Bengali transcription. Only the raw data presented in Appendix C was kept in Bengali.
Scoring procedure. The results were analyzed using the quantitative method. A score of 1 was assigned for each accurate person marker use. The standard use of person markers in Bengali determined the accuracy. A score of 0 was assigned to each inaccurate use of person marker. Some participants did not respond to some questions. Those were also marked as 0. However, if a participant used accurate person marker but failed to recognize the correct verb form, the response was counted as correct. For example, in case of the verb /chul achrano / Py j AvPov ‡bv, Participant A said /halo kortase/ n¨v ‡jv KiZv ‡m. The researcher counted it as accurate since s/he used the correct person marker, i.e., 3 rd person. It is to be noted that there were some exceptions in terms of responses from paricipants. To calculate those, the researcher solely focused on the person marker use; not on the correct verb use. For instance, in case of the verb /nacha/bvPv, participant C responded /eta ammu/ GUv Av¤ §y . This response was completely irrelevant and did not carry any person marker, and therefore, was considered inaccurate.
After scoring each participant, the researcher presented the central tendency for each person marker by calculating mean score and standard deviation.

Results
The responses of each participant for each person marker are provided in Appendix C. The participants are listed from A to J chronologically based on their age. Besides, the inaccurate responses are underlined. In cases where the participants did not respond, the spaces are kept blank.
Each participant was expected to provide 24 verb forms. Some of the participants did not give any answer in certain cases, probably due to their disinterest or shyness. For this reason, the researcher got 206 responses instead of 240 responses from participants.
Each participant was scored out of eight for each person marker using the eight verb forms. Therefore, the score could range from 0 (lowest) to 8 (highest) for each person marker.

Data Analysis
From Table 1, the mean score for each person marker can be seen. The mean score increases in this manner: 2 nd person< 1 st person< 3 rd person, and the standard deviation decreases in such manner: 2 nd person> 1 st person> 3 rd person. This particular phenomenon indicates that participants performed accurately mostly in using 3 rd person marker. Besides, participants showed more homogeneity in terms of accurate use of 3 rd person marker than the rest two person markers.
A further analysis was also done to investigate which person marker is easy to acquire than others. In order to do so, the researcher divided the participants into four groups based on their age. The groups were named Group I (2.5-2.9 years), Group II (2.10-3.2 years), Group III (3.3-3.7 years), and Group IV (3.8-4 years). The age difference was consistent for each group, which was four months.
After dividing the participants into four groups, the researcher determined the ratio of mean scores for three person markers for each group. The mean scores of Group I, II, III, and IV for 1 st person marker use were 2, 5.5, 8, and 8 respectively. For the 2 nd person marker use, the groups got 0.33, 4.5, 7, and 8 respectively. The mean scores of groups for 3 rd person marker use were 5.67, 7, 8, and 8 respectively. These findings are presented in Figure 1. Figure 1 illustrates that the participants aged 2.5 to around 3 years struggled with the use of 2 nd person marker. Nonetheless, all the participants of groups performed comparatively well in using 3 rd person marker. There could be several reasons behind such a performance.
Firstly, for participants of Group I, it could be difficult for them to imagine themselves as the child in the picture and respond to the questions on 1 st person marker use. Hence, their performance in this particular case may be due to their young age. Moreover, one of the studies stated that children can get confused while using correct person markers at an early age, especially between 2 nd and 3 rd person markers (Schmalstieg, 1977). This phenomenon is particularly visible when participants C and D expressed 2 nd person by using 3 rd person marker, in case of some verb forms.
Besides, the 3 rd person is frequently used in regular conversation in any language. For this reason, participants might make fewer errors in 3 rd person marker use. Another reason could be the imitating tendency of young children. To illustrate, the focus needs to be given on the questions that were used to elicit responses. The 1 st person marker question was /tumi ekhane ki korcho?/ ÒZy wg GLv ‡b wK Ki ‡Qv?Ó, which expected an answer carrying /korchi/ '. . . KiwQ. Again, 2 nd person marker use was elicited with the question /ami akhon ki korchi?/ "ÒAvwg GLb wK KiwQ?", where the answer was supposed to have /korcho/ '... Ki ‡Qv'. In case of 3 rd person marker use, the question was /apu/babuta akhon ki korche?/ ÒAvcy /evey Uv GLb wK Ki ‡Q?Ó that should have been answered with /korche/ '. . . Ki ‡Q'.
It can be seen that only in the case of 3 rd person, the person marker remains the same in the question and in the answer. Therefore, the question itself might have guided the participants to use third person marker accurately. Only 2 participants, D and G, showed evidence of imitating 2nd person marker from the question in case of 2 nd person. Having said that, the researcher would also like to point out that none of the participants imitated person markers directly from the question for 1st person marker.
That language production itself is very complex process needs to be taken into account here. For the competing demands of language production, it can become difficult for children to exhibit their potential entirely in using any individual linguistic item (Hesketh, 2004). Besides, Meisel, Clahsen and Pienemann (1981) proposed that the development sequence of linguistic features depends on how easy the feature is to process (as cited in Lightbown & Spada, 2006). From Figure  1, we therefore can consider that 3 rd person marker is most likely to be the easiest person marker to process for young children. Furthermore, all 3 lines in the graph increase in a positive manner, i.e., there is a positive relationship between participants' age and their mean score. This can indicate that the use of inaccurate verb agreement by young children can eventually resemble the target accurate form along with their maturity (Ionin & Wexler, 2002).

Some General Observations
While analyzing the data, the researcher observed an interesting phenomenon, i.e., the use of other markers by participants. Even though the researcher asked questions in progressive aspect, some of the participants responded in the simple aspect. One example can illustrate this point. Participant C, while responding to a 2 nd person verb form /chul achrano/ responded as "/chul achray/" instead of saying /chul achrachche/. Even though around 14% of total responses was in simple aspect, considering the research question, the researcher counted those responses correct which used accurate person marker.
Through the elicitation task, this study attempted to focus on the performance of participants. However, if a participant did not produce the target construction that will not necessarily indicate his/her incompetence in that particular construction (Hesketh, 2004), as the cognitive development of children plays some role in first language acquisition (Lightbown & Spada, 2006). There might be other reasons which probably were behind the findings of this study.
Firstly, the language itself is highly inflectional in nature (Dasgupta & Ng, 2006), and that can confuse young children. Besides, the addition of certain suffixes changes the root of some verb forms that lead towards the nonlinear behaviour of Bengali (Bhattacharya et al., 2005). Hence, the complexity and irregularity of Bengali verb morphology can make it difficult to process. Even though children might have the knowledge of the inflectional system, processing factors may obstruct their performance (Buijs, Reijen & Weerman, 2013).

Conclusion
Different sorts of processing abilities may determine the development sequence of different linguistic features (Lightbown & Spada, 2006). Research on English morphology showed that children tend to produce verbs in present progressive at a high percentage (Conti-Ramsden & Windfuhr, 2002). This study also showed similar findings for Bengali verbs. However, since verbs are changeable along with changing person, the process of acquiring different person markers asks for more processing capacity.
The participants did show productiveness in using person markers at around 2.5 years of age. Nonetheless, children who are one year older than them, aged around 3.5 years of age, achieved accuracy. By the end of 4 years, children show evidence of using three person markers accurately in present progressive form.
To conclude, this study does carry certain limitations. The researcher would like to acknowledge that the current findings are based on a specific elicitation task, i.e., question-answer form. Future studies can include other examples of elicitation tasks to further strengthen the findings. Besides, the present sample size was small which could limit its generalizability. A larger sample can give comparable data allowing generalization.