Characteristics of Children’s Oral Texts in terms of Coherence and Cohesion

The purpose of this study was to examine the characteristics of coherence and cohesion in the oral texts of children between the ages of 60 to 72 months. In the study, a survey method in a quantitative research approach was adopted, and a total of 110 preschoolers were included as participants. The data of the study were obtained from the oral texts taken from the conversations during the interviews with the children. Oral texts consisting of at least eight sentences were analyzed by using the Text Evaluation Form and a rubric. The coherenceand cohesion-related qualities of the spoken texts were evaluated by two experts. The quantitative data were analyzed through descriptive and inferential statistics. Descriptive statistics were calculated as frequencies and percentages to present the levels of the oral texts in terms of cohesion and coherence. In addition, independent sample t-test and one-way analysis of variance were computed as inferential statistics to determine whether scores for cohesion and coherence differed significantly based on the child’s gender, socioeconomic status, and the duration of the early childhood education. First, the results of the analyses revealed that the oral texts were acceptable and sufficient with respect to the means of cohesion, whereas they were inadequate and unacceptable in terms of coherence. Second, it was found that the cohesion device that children were the most successful was the ellipsis. Third, the results also indicated that the levels of cohesion of the oral texts displayed a significant difference regarding gender and SES, while they did not differ depending on the duration of preschool education. Moreover, the levels of coherence showed no significant difference in terms of gender, SES, and the duration of early childhood education. In conclusion, in the study, it was determined that the oral texts of the children were more appropriate to the standard language in terms of cohesion compared to coherence.


Introduction
Language acquisition takes place through distinct stages. The language acquisition process, which starts with the hearing of linguistic voices from the environment, continues until it reaches a level to communicate with other people. In fact, the primary function of language as a natural tool is to communicate (Clark, 2000;Owens, 2012). Since communication takes place mutually between at least two individuals, the language acquired through listening is expected to become operational through speech over time. When children produce a communicative value through language, it can be thought to have spoken. The communicative value of language is estimated through the texts produced in different lengths. As Beaugrande and Dressler (1987) stated, communication takes place with texts.
For this reason, to what extend the sentences produced by children in the process of first language acquisition have textual quality is an important aspect of language acquisition.
According to Halliday and Hasan (1976), a text is the collection of written or spoken sentences that forms a unified structure. However, not every collection of sentences forms a text. That is, the "stack" of sentences must meet certain criteria in order to be regarded as a text. These textuality criteria are cohesion, coherence, intentionality, acceptability, informativity, situationality, and intertextuality (Beaugrande & Dressler, 1981). While coherence and coherence refer to internal criteria, others are external criteria. The cohesion referring to the grammatical harmony between the elements of the text includes linguistic devices such as reference, substitution, ellipsis, and conjunction. Besides, coherence means the logical and semantic relations of large parts of the text. In this way, the text emerges as a uniformed structure reflecting a main idea.
Cohesion is the semantic relationship of one item in the text with another item that is important in the meaning-construction of a text. The appearance of semantic relationships among the sentences of the text in the grammatical plane reflects the cohesion (Beaugrande & Dressler, 1981).
The relations among the sentences constructed by the cohesion running on the grammatical plane form the discourse (Gudwinsky, 1976). Halliday and Hasan (1976) defined this term as the semantic relation between an item and another in the text that is important in the meaning of the text. The interpretation of an item in the text comes true by its relation to other items. Cohesion, which constructs the grammatical and lexical relations of the text, functions on grammatical and semantic planes (Bex, 1996).
The devices that provide the formation of cohesion establish some relations among the units of the text and semantically link the propositions to each other (Schiffrin, 1988). Cohesion includes five grammatical cohesion. On the other hand, conjunctions are considered as structures that combine grammatical and lexical features. In this study, the conjunctions are dealt with only the grammatical correlation dimension.
One of the grammatical cohesion devices is reference which is the re-expression of a unit elsewhere in one part of the text (Dilidüzgün, 2008). The use of the reference, which has two forms (i.e., exophoric and endorphoric reference), is provided with various pronouns, accusatives, possessives, and personal endings in Turkish (Keçik & Subaşı, 2004;Uzun, 1995;. The linearity and continuity of the text are maintained by re-expression of the information that constitutes the subject through the reference devices (Onursal, 2003). Substitution, which is another cohesion device, occurs when a unit used in the text is expressed with another linguistic unit in subsequent sentence(s).
A substitution device may substitute for nouns, verbs, or sentences which determines the type of the substitution (Halliday & Hasan, 1976;Uzun, 1995). Ellipsis is another cohesion device. Ellipsis, which occurs by omitting a unit from the text, are comprehended by inference. Ellipsis should not cause any change in meaning (Günay, 2003). When a person says, "Dün kütüphaneden aldım (I bought it from the library yesterday)" by referring to a book makes a certain language economy by using ellipsis (Uzun, 2013). Ellipsis also prevents unnecessary repetitions (Külebi, 1990). The types of ellipsis are nominal, verbal, and clausal. That is, ellipsis might be with the omission of noun, action, or clause (Halliday & Hasan, 1976). The last cohesion device that this study focused on is conjunctions which are linguistic tools that connect sentences and phrases to each other to provide certain semantic relationships. Conjunctions increasing the cohesion quality of the text, and making it more holistic, create a semantic quality among the linguistic units. Unlike other cohesion devices, conjunctions perform their functions indirectly (Halliday & Hasan, 1976). In Turkish, linguistic units used as conjunctions are connectives and gerunds.
In addition to the cohesion that characterizes grammatical relationships, another criterion which establishes semantic and logical relationships to make a collection of sentences into a text is coherence. Coherence is achieved through the relationships among sentences or larger parts of the text. By means of this criterion, harmony is achieved by creating semantic-logical relations among the concepts in the text (Beaugrande & Dressler, 1981;Keçik & Subaşı, 2004). This harmony ensures that sentences and large parts are patterned and repeated within each other. Thus, the receiver of the text can move from one sentence to another without any semantic conflict; moreover, s/he can perceive the sentences not as a stack that comes together randomly, but as parts of the whole (McCrimmon & Miller, 1973). That is, coherence is the result of the associations among concepts gathered around the main idea (Beaugrande & Dressler, 1981). The grammatical relationships established with cohesion are completed with the logical relations that coherence provides.
There are two types of coherence: local and global (van Dijk & Kintsch, 1983). Local coherence is related to the microstructure of the text, whereas global coherence is related to the macrostructure. Besides, while the microstructure refers to the consecutive sentences in the texts, the macrostructure refers to the parts of the text presented with the content scheme that presents the topics and sub-topics (Uzun, 2013). Local coherence connects successive sentences or smaller units semantically and logically while global coherence establishes the same relationship among the paragraphs, the chapters, and the larger parts of the text. In other words, global coherence functions in the broader area. A text in which global coherence is achieved, discourse referents, individuals and objects are lined up concerning some main referents, and the sequence of actions is organized according to the main actions (van Dijk & Kintsch, 1983). This arrangement makes it easier for the listener or reader to comprehend the text.
Cohesion and coherence are crucial in creating text that is the basic unit of communication. In the first language acquisition process, children are expected to use the language to produce texts that have communicative value. A healthy child needs to acquire the language system by the age of four (Lust, 2006). After the age of four, textual criteria that children's use of language should meet affect the commutation abilities of children directly. Therefore, the sufficiency level that language acquisition is expected to reach can be observed through the texts creation of children (Justice, et al., 2010;Pankratz et al., 2007). The level of language acquisition of children becomes more meaningful with their textual acquisitions. Furthermore, the level of textual acquisition in children is reflected in their narratives.
The ability of children to produce narratives and texts takes place through various stages. Oral texts with various deficiencies and errors, in the beginning, have problems in terms of cohesion and coherence. This situation is related to the nature of language acquisition. How or at what level children acquire the language can be monitored through narrative skills. Bloome et al. (2003) describe the development of narrative skills as creating a good story and conveying it, and gaining certain skills in telling experiences. The acquisition of these skills is significantly related to establishing relationships of cohesion and consistency. Accordingly, children who have reached the process of forming sentences are expected to use the linguistic devices that determine the criteria for textuality correctly and effectively while connecting the sentences they produce. Thus, they are expected to create cohesive and coherent oral texts.
When communicating, children first produce texts concerning the experienced events in their lives. Around the age of three, they have the ability to tell the experiences orally such as going to the supermarket, eating somewhere, or various similar events (Hudson & Saphiro, 1991). Around the age of four, it is observed that the events are transferred in the order of time (Nakano & Nagasaki, 2012).
This development can be considered as an increase in the level of cohesion and coherence of the texts.
Children between the ages of five and six begin to construct more abstract and complex narratives (Hudson & Saphiro, 1991). They can understand the causal and intentional relationships among events and actions (Nakano & Nagasaki, 2012). Mardell (1991) emphasizes that as children grow up and develop, they are more sensitive to the needs of the audience and make the beginning and ending parts of their narratives more explicit. The consciousness about the beginning and ending parts of the narration can be considered as a sign that the text production has started to become intuitively competent.
As children grow up, they begin to use references and conjunctions more frequently in their oral texts (Fivush et al., 1995;Özcan, 1993;Van Dam, 2010). For example, Saphiro (1990) reported that first graders create texts that include more cohesion devices than preschoolers. It was also found that children learning how to read and write use mostly reference devices and conjunctions in oral texts compared to written texts (Lee et al., 2013).
The fact that the texts created by children in oral communication reach a level similar to those by adults in terms of cohesion and coherence can be considered as an essential indicator of language acquisition level and quality. Acquiring the ability to create text at an early age can contribute to achieve linguistic competence in later periods of life. As a result, determining the level of the acquisition of preschoolers can provide more qualified planning of language activities in preschool environments. For this reason, determining children's competence and identifying the deficiencies in oral texts are important points. In this study, the oral texts of the children between 60-72 months who are in the process of the acquisition of Turkish as their first language were examined in terms of cohesion and coherence. Accordingly, this study sought answers to the following questions:

1.
What are the characteristics of the oral texts of the children in terms of coherence?

2.
What are the characteristics of the oral texts of the children in terms of cohesion?
3. Do the characteristics of the oral texts of the children in terms of coherence differ significantly based on gender, socioeconomic status (SES), and the duration of early childhood education?

4.
Do the characteristics of the oral texts of the children in terms of cohesion differ significantly based on gender, socioeconomic status (SES), and the duration of early childhood education?

Research Design
This study is descriptive in nature and adopts the survey method with which the characteristics of certain groups can be determined, and these properties can be compared, classified, and analyzed according to various variables (Cohen et al., 2007). Correspondingly, in this study, the coherence-and cohesion-related characteristics of children's oral texts were determined to present the textual quality.
In addition, it was also investigated that whether the quality of texts produced by children differs based on some child-related variables (i.e., gender, socioeconomic status (SES), and the duration of early childhood education).

Participants
The participants of the research consisted of 110 preschoolers enrolled in various preschools in Antalya, Turkey. The participants, whose age range was 60-72 months old, included 61 girls and 49 boys. Depending on the region that children live, three levels of socioeconomic status were determined. Accordingly, the 40 of the participants were in the low, 35 in the middle, and 35 in the high SES groups. The selection of the children participating in the research was made based on the stratified sampling technique. This sampling method was preferred to provide heterogeneity among the specified groups (Kumar, 2014). Thus, children with different characteristics could get involved in the study. Attention was paid to ensure that the gender and SES level distributions of the participants were close to each other.

Data Collection Instruments and Procedure
First, the Text Evaluation Form created in line with the expert opinions was used in the analysis of verbal texts obtained from children. With the help of this form, the oral texts of the children were evaluated in terms of cohesion and coherence. Cohesion-related part of the form included evaluation of reference, substitution, ellipsis, and conjunctions of the texts, whereas the coherence-related part included evaluation of local and global coherence of the texts.
Second, a rubric was developed to evaluate children's oral texts. A rubric is a measurement tool used for the evaluation of the criteria that a product or person is expected to meet through the predetermined levels or standard criteria (Lodico et al., 2006). The development process of this rubric began with the literature review to determine the textuality criteria. In this process, among the textuality criteria, the cohesion and coherence which are related to the internal features were focused.
In line with these two criteria, the draft form of the rubric was created. Six experts were consulted to ensure the content and face validity of the rubric. After the corrections, depending on the suggestions of experts, a pilot study was carried out. For the reliability of the instrument, the agreement between the two raters was examined. The reliability coefficient was calculated using the formula [the number of agreements / (the number of agreements + the number of disagreements) x 100] specified by Milles and Huberman (2002). According to this calculation, the intercoder reliability coefficient was found as .94. After these processes, the final form of the instrument was created.
The data of the research were collected through face to face interviews with the children.
Before the interviews, the consent forms were sent to the parents of children to inform them about the study and take their permission to interview with their children. During the interviews, the necessary conditions were created for children to express themselves comfortably, and they were allowed to speak the topics that make sense for them. No intervention was made during their speech. The interviews, which took place in the form of conversations, were recorded. After the interviews, the conversations were transcribed by the researchers to be analyzed. Rather than the whole oral texts of the children, their oral texts consisting of at least eight sentences and having integrity were analyzed by using the Text Evaluation Form and the rubric.

Data Analysis
After the data collection process, the number of data collected from 145 children decreased to 110 after the elimination of inappropriate ones. That is, the conversations that do not include the text or texts with at least eight sentences were excluded from the data set. The data decided to be included in the data analysis were analyzed terms of the assessment items the Text Evaluation Form. Besides, using the rubric, the narratives of the children were assessed whether they have textual quality. For this, the cohesion-and coherence-related characteristics of texts were examined. First, oral texts were examined in terms of reference, substitution, ellipsis, and conjunctions depending on the rubric.
Second, the texts were scored in terms of the coherence based on the criteria of coherence specified in the rubric. In addition, the frequency of appropriate use of reference, substitution, ellipsis, and conjunctions as in the standard language and deviations from standard language were determined.
Only pronouns considered in the evaluation of reference cohesion devices. References regarding conjugation and possessives suffixes were not included in the scope of evaluation in the study. The references made in Turkish with suffixes are the structures that appear unintentionally as a requirement of the language, and these structures are used substantially correct. As a result, because the examination of the suffixes as the reference devices might cause the uncertainty in terms of the evaluation of the results regarding the reference, the reference made with the suffixes were excluded from the study.
Following the initial analysis of cohesion, it was seen that the data did not distribute normally and revealed a negatively skewed distribution. For this reason, the data were re-evaluated over a quarter deviation by calculating the ratio of deviations from standard language to total usage. Whether the ratios obtained as a result of the evaluation were in the first and last quartiles were examined. Data in the first quartile were considered as sufficient, data in the last quartile were considered as insufficient and the data between these two groups were considered as acceptable. The normality test was conducted with the obtained depending on this classification. As a result of this process, it was seen that the data revealed a normal distribution. For this reason, parametric tests were conducted in the data analysis. Descriptive statistics were calculated as frequencies and percentages to present the levels of the oral texts in terms of cohesion and coherence while independent sample t-test and oneway analysis of variance (ANOVA) were computed as inferential statistics to determine whether scores for cohesion and coherence differed considerably based on the child's gender, SES, and the duration of the early childhood education.

Results
In the first research question, it was aimed to evaluate the oral texts in terms of global and local coherence. The descriptive statistics in relation to the global and local coherence were presented in Tables 1 and 2: As seen in Table 1, the oral texts of children were found insufficient at most (68.18%, f = 75), followed by acceptable (30.00%, f = 33) and sufficient (1.82%, f = 2). In relations to the second research questions, the results of the descriptive analyses regarding the frequency of cohesion devices that used in the texts (i.e., reference, substitution, ellipsis, and conjunction) were presented in Table 3: The results of evaluating the sufficiency in the use of devices that provide cohesion were presented in Table 4, Table 5, Table 6, and Table 7: As seen in Table 4, the use of reference in the oral texts of children was found sufficient at most (52.73%, f = 58), followed by acceptable (32.73%, f = 36) and insufficient (14.54%, f=16).  Table 5 indicates that the use of the ellipsis in the children's texts was sufficient to a large extent (71.82%), followed by acceptable (20.91%), and insufficient uses (7.27%).  Table 6 shows that the use of substitution in the oral texts of children was found sufficient at most (44.55%), followed by acceptable (35.45%) and insufficient (20.00%). As depicted in Table 7, the use of conjunctions in the oral texts of children was found sufficient at most (54.55%), followed by acceptable (34.55%) and insufficient (10.90%).
The third research question of the study addressed whether the coherence levels of the oral texts of the children differed based on gender, SES, and the duration of early childhood education. The results of the independent sample t-test and ANOVAs were conducted to provide answers for this research question were depicted in Table 8, Table 9, and Table 10. As seen in Table 8, the levels of coherence for the oral texts of the children did not differ significantly by gender, t(108) = 1.08, p = .281.  Table 9 reveals that there was no significant difference in terms of SES between the mean scores for the coherence of the oral texts of the children, F(2, 107) = 1.60, p = .208.  Table 10 shows that there was no significant difference in terms of the duration of the early childhood education between the mean scores for the coherence of the oral texts of the children, F(2, 87)=1.688, p = .191.
The fourth research question of the study focused on determining whether the cohesion levels of the oral texts of the children differed based on gender, SES, and the duration of early childhood education. The results of the independent sample t-test and ANOVAs were conducted to provide answers for this research question were depicted in Table 11, Table 12, and Table 13. As seen in Table 11, the levels of cohesion for the oral texts of the children differed significantly based on gender, t(108) = 2.10, p = .038. Accordingly, the mean scores of girls (M = 6.16, SD = 1.31) was higher than the mean scores of boys (M = 5.63, SD = 1.33), revealing girls produced more sufficient texts than boys in terms of cohesion devices. As depicted in Table 12, the mean scores for cohesion devices revealed a significant difference in terms of the SES of the children, F(2, 107) = 4.766 p = .010. In order to determine the source of the difference, the Tukey test was conducted. As a result, it was observed that the mean scores of the children at the low SES (M = 6.23, SD = 1.17) and the average scores of the children at the high SES (M = 6.14, SD = 1.38) were higher than the average scores of the participants at the middle socioeconomic level (M = 5.37, SD = 1.35). The calculated effect size for this difference was determined as .08, revealing a small effect size.  Table 13 reveals that there was no significant difference in terms of the duration of the early childhood education between the mean scores for the cohesion of the oral texts of the children, F(2, 87)=1.871, p = .160.

Discussion, Conclusion, and Recommendations
In the study, the quality and the characteristics of the oral texts of children between 60-72 months were discussed in terms of coherence and cohesion. Considering the results regarding coherence, it was determined that there were some problems in the oral texts of the children. The global dimension of the coherence, which is defined as the logical associations among the large parts of the text (van Dijk, 1977), was evaluated on the basis of the presence of introduction, development, and conclusion sections in the oral texts produced by the children. Accordingly, it was seen that more than half of the participants' oral texts were problematic in terms of global coherence. In light of this result, it might be concluded that the majority of the children in the study could not convey a certain personal experience within the main idea, except few participants who met this criterion.
There are other studies reporting that textual acquisition, which is an advanced level of first language acquisition, is immature in terms of coherence. For instance, Bayat and Yurdakul (2014) found that introduction, development, and conclusion of the texts were not presented explicitly in the oral texts of the preschool children. Similarly, in the study of Silva et al. (2014), coherence was found to be problematic in children's stories. The results of the current study also provided similar evidence regarding the problematic use of coherence in the text production of children. On the other hand, some other research studies reported that older children are more competent to create coherent texts. To illustrate, Gwynn (2007) reported that eight-year-olds could create more coherent texts than four-and six-year-olds. In addition, Saphiro (1990) found that first-year students created texts including introduction and conclusion parts in their stories more than preschoolers. These results and the results of the current study may reveal that coherence is developed through the progress of language acquisition with age.
The conclusion reached in terms of global coherence is also valid for the local coherence. That is, half of the oral texts of participants did not meet the criteria of local coherence. In the current study, local coherence was evaluated, especially on the relations between consecutive sentences. The coherence of consecutive sentences is ensured through the continuity of an item of the previous proposition in the latter (van Dijk & Kintsch, 1983). The inability of the children to achieve coherence in their texts was based on the lack of correlation between the sentences they constructed consecutively. Despite various disconnections, the meaning can be understood to a certain extent in the texts of children, which were evaluated as acceptable. However, an exact meaning could not be reached in the oral text of half of the participants. This result is an indicator that the children in the study were not able to construct oral texts within the local coherence. Karmiloff-Smith (1985) stated that reference devices, which contribute relatively to the formation of local coherence, become active around the age of 6-7. Therefore, it might be proposed that the children in this study have not yet been able to overcome incompetence in constructing locally coherent texts.
The incompetence of the children in the first language acquisition process in terms of local coherence means that they create a collection of sentences without making connections among these sentences. Lack of this competence inhibits from reaching the main idea (Grabe, 1984). In a text, each sentence should contain information from previous sentences (van Dijk, 1977). The continuity of information contributes to the construction of the main idea. For this, linguistic production has to be purposeful and conscious. The insufficient levels of local coherence might result from not determining the main idea. In addition, if it was determined, insufficiency might be due to losing it in the flow of speech or not being able to monitor its linguistic production consciously. The incapability of participants to achieve local coherence may be related to the language acquisition processes that had not reached to this level.
The cohesion devices used for grammatical relations presented a different profile for the oral texts of the participants. Although there were inappropriate uses of the reference device, this cohesion device was used appropriately to a large extent in the oral texts of children. The reference means the repetition of a unit mentioned in a previous part of the text through the text (Dilidüzgün, 2008;Halliday & Hasan, 1976). Reference is made with pronouns and some affixes in the Turkish language (Keçik & Subaşı, 2004). In particular, the pronouns should be used properly, and what they refer to should be clear. In this research, the reference was evaluated only on the basis of pronouns. Accordingly, the children were able to use two-thirds of the references appropriately. However, in some studies, it is stated that problematic uses related to reference decrease at the ages between 9 and 12 (Ripich & Griffin, 1988;Bamberg, 1987). The participants of the current study can be considered to be at an expected level for their age group.
References that develop the patterns of text can not be considered detached from context (Dilidüzgün, 2008). The reference is interpreted through the item that is referred to in the context. This requires context sensitivity. Children's problematic uses of reference devices may also be related to the lack of contextual information. That is, contextual and grammatical information is effective in the use of reference devices. In a study related to the reference, it was found that children at the age of 5 years had 39 % proper use, while children at the age of 6 years had 50 % proper use (Demir, 2009).
This level is lower than the level of appropriate use in this study. Therefore, it can be said that the appropriate usage level of the reference device may differ in different age groups. In addition, Özcan's (1993) study determined that three-and five-year-olds were able to use reference devices significantly.
Based on the results of these types of research studies, Dodsworth (1990) emphasizes that problematic uses of cohesion devices decrease with age.
Considering the use of ellipsis in the oral texts of children in this study, it is obvious that children are more successful at the use of ellipsis in their texts. The proportion of proper use of ellipsis was found at 80 %. The ellipsis occurs when an item is not told in the context known to those involved in communication (Günay, 2003). The ellipsis enables individuals to use language more economically (Uzun, 2013). Because unnecessary repetition of the same units in the text can make the meaning of narrative ambitious, the ellipsis can enhance the transparency of the text. In Turkish, ellipsis structures are created with finite verbs that function as a predicate, possessive elements, predicative elements in transitive verbs, and the adverbs that require predicative element (Uzun, 2013). The first two of these are made with affixes. For example, in a sentence like "Bir bisiklet aldım. (I bought a bicycle)" the subject of "Ben (I)" was omitted because there is a personal ending that refers to the subject of the sentence in the predicate. Since Turkish has a language structure that works with suffixes, ellipsis structures appear naturally in some cases. In other words, if the personal ending (i.e., -m) at the end of the predicate in the sentence presented at the example is not used, the sentence becomes grammatically incorrect. Therefore, the use of ellipsis in Turkish is observed in the sentences as a requirement of the language system to some extent.
In comparison with the ellipsis, children showed less success in the substitutions. Only half of the oral texts had appropriate use of the substitutions. Approximately 20 % of the participants had significant problems with the substitutions. The substitutions are highly functional in ensuring the fluency of the text. The deterioration of fluency usually results from the unnecessary repetition of various words or phrases (Kuru, 2013;Sucuoğlu & Kargın, 2010). The substitution avoids this problem since linguistic units because of expressing a unit with other words in the flowing parts of the text. According to Peterson and Dodsworth (1991), children begin to use substitutions and ellipsis in their speech, starting from the age of 2-3. Baltaxe and D'Angiola (1992), on the other hand, stated that preschool children with normal development have the most problematic use in reference and substitution devices. In this research, the problems seen in the use of substitutions are consistent with the results of the aforementioned research studies.
The last device of correlation focused in the study is the conjunctions. The children's use of the conjunctions was examined on the situations that they did not use where conjunctions are necessary. Accordingly, it was determined that almost half of the participants produced problematic texts in terms of conjunctions, meaning relations between propositions in significant number of the texts produced by the children could not be established. It was observed that participants are more successful in temporal and causality conjunctions, whereas they had some problems in comparative and additive conjunctions. In a study conducted by Bayat et al. (2015), it was also found that children at the age group of five had problems in comparative and additive conjunctions. According to Ege (2006), the use of conjunctions between the ages of 3 and 7 is constantly increasing, and this increase is significantly apparent between the ages of 5 and 7. In contrast, only half of the participants of this study were able to use conjunctions explicably.
In the study, it was also examined whether the coherence and cohesion levels of oral texts of the children differ based on gender, SES, and the duration of early childhood education. First, the results of the study revealed that the levels of coherence in the oral texts of the participants did not differ statistically and significantly based on gender, SES, and the duration of the early childhood education. Second, the results also indicated that the participants' mean scores differed significantly in terms of gender and socioeconomic level, but there was no statistically significant difference in terms of the duration of early childhood education.
Previous studies found that gender is an effective variable in language acquisition. In his research, Öztürk (1995) determined that girls are more successful than boys at the same age in language acquisition. Similarly, girls were found to be more successful than boys in terms of speech length and grammatical appropriateness of their sentences (Seçmiş, 1996;Taner, 2003). Contrary to these research studies, there were also other studies indicating that there is no relationship between gender and first tongue acquisition (Temel, 2000;Yıldırım, 2008). The result obtained from this research revealed that the gender of the participants did not produce a significant effect on the levels of coherence. On the other hand, the current study found that girls were more successful than boys in the use of cohesion devices. The contradiction between results of coherence and cohesion in terms of the effect of the gender might result from the grammar-based construction of cohesion, i.e., the relevancy of cohesion with the grammatical characteristics of the text. In other words, cohesion functions at the surface structure of the text and easier, whereas coherence requires deeper and more difficult to be constructed in the text. Therefore, considering the developmental characteristics of the participants, it can be said that the coherence could not sufficiently be provided by both genders, but girls have successfully acquired and used the cohesion devices that can be acquired and used more quickly.
The SES is also a variable that might affect first language acquisition in various aspects. There are studies reporting that competence in language acquisition increases as the socioeconomic level increases (Ünal, 2007;Erkan, 1990;İpek, 2006). However, different results were obtained in terms of the effect of SES on coherence and cohesion in the current research. SES did not make a difference in the mean scores of the participants in terms of coherence. The reason for this may be attributed to an advanced level of coherence in the first acquisition process. The 60-72 month-old participants of this research are not yet competent in terms of coherence. On the other hand, the children at high and lower SES were found to be more successful in using cohesion devices than those at the middle level.
This result may be related to lower and upper-level children getting more qualified language inputs.
This situation is considered to suppress the effect of socioeconomic levels.
Lastly, it was found that the duration of the early childhood education did not affect the levels of coherence and cohesion. However, previous research studies reported that children who had early childhood education are more successful in language than those who did not have this education.
Moreover, some other studies also indicated that children who had early childhood education for longer are more competent in language than those had shorter (Erdoğan et al., 2005;Öztürk, 1995;Taner, 2003;Taner & Başal, 2005;Şengül, 2007). However, these studies focused on non-textual units of language use. The difference in the results obtained might result from this.
Depending on the results of the study, it was concluded that the oral texts of the children were closer to the standard language in terms of cohesion, and they had some problems in terms of coherence. It is recommended to repeat similar studies to confirm these results. In addition, various studies can be conducted to determine the levels of lexical cohesion that were not addressed in this study.