1. Introduction
Imagine a leaf blown from a tree on a windy day. What causes the leaf, which cannot move on its own, to move? While adults can easily answer, can children do the same? What language structures help adults and children understand and explain such phenomena? According to the literature, we use causal language to explain events like these in our daily lives and include both cause and effect (Marini & Singer, Reference Marini and Singer1988). Causal language includes components such as causal connectives, lexical causatives, and morphological causatives, even though understanding, interpreting, and lexicalizing events vary by language (Ger et al., Reference Ger, Küntay, Goksun, Stoll and Daum2022). We also see different types of causal events when we observe daily events. On the one hand, causality can arise from person-to-person or person-to-object events, and on the other hand, it can occur in object-to-object events (Bonawitz et al., Reference Bonawitz, Griffiths, Schulz, Sun& and Miyake2006; Sanefuji & Haryu, Reference Sanefuji and Haryu2018; Schulz et al., Reference Schulz, Bonawitz and Griffiths2007). Causality is a wide range of phenomena, and both the production and comprehension of causal events differ regarding cognitive differences and language skills (Kupersmitt & Armon-Lotem, Reference Kupersmitt and Armon-Lotem2019; Williams, Reference Williams2020).
The purpose of this study is threefold. First, we are exploring the difference of paternal and maternal causal language input. The underlying reason is that, while some related studies emphasize that both mothers and fathers similarly influence their children’s language use, other studies suggest that mothers and fathers contribute differently (Leaper et al., Reference Leaper, Anderson and Sanders1998; McLaughlin et al., Reference McLaughlin, White, McDevitt and Raskin1983; Rowe et al., Reference Rowe, Coker and Pan2004; Tamis-LeMonda et al., Reference Tamis-LeMonda, Shannon, Cabrera and Lamb2004). For instance, Rowe et al. (Reference Rowe, Coker and Pan2004) showed that fathers and mothers use similar number of words, linguistic complexity, and their diversity of vocabulary did not differ. However, fathers used more-wh questions compared to mothers. On the other hand, Leaper et al. (Reference Leaper, Anderson and Sanders1998) mentioned different impacts of fathers and mothers in their meta-analyses study. According to results, mothers talked more and used less direct and informative language compared to fathers. To clarify these findings, the first focus of this study is to examine the separate relations of mothers and fathers on their children’s language production.
Secondly, we are focusing on children’s use of causal language in two distinct tasks, specifically examining which parent the child is interacting with during these tasks. In this study, the tasks involve two distinct types of play and a storytelling activity. Play and storytelling play a crucial role in parent–child interactions, with particular emphasis on two main types of play: Free play and guided play. Free play allows children to develop more creative thinking and language use, while guided play is more structured and goal-oriented. Research suggests that variations in play types offer distinct insights into parent–child interactions (Kwon et al., Reference Kwon, Bingham and Lewsader2013; Tamis-LeMonda et al., Reference Tamis-LeMonda, Kuchirko, Escobar, Bornstein and Bornstein2019). Different types of play allow for the diversity in parenting behaviours, the ways parents support their children’s play, the extent to which they provide space for their children’s independence, whether or not they effectively scaffold learning, the level of parental responsiveness, and the nature of the resulting narration. Thus, the second aim of this study is to examine how parental causal language input and children’s causal language production vary when interacting individually with mothers and fathers across three different tasks: Storytelling, free play, and guided play. The third and final aim is to explore the contribution of both maternal and paternal causal language input to children’s causal language production. For this purpose, we asked three research questions: (1) Does maternal and paternal causal language input differ? (2) Does children’s causal language production differ in terms of task types and which parent the child is interacting with? (3) Are maternal and paternal causal language inputs associated with children’s causal language production?
2. Children’s comprehension of causality and causal language production
Many studies in literature prove that the causal event comprehension skill emerges starting from the first year of life (Leslie, Reference Leslie1982; Leslie & Keeble, Reference Leslie and Keeble1987; Muentener & Carey, Reference Muentener and Carey2010). Starting from infancy, this ability develops over time and continues in childhood. Two different factors affect children’s comprehension of causality. Firstly, the spatial contiguity between cause and effect simplifies causal events for children to perceive (Lesser, Reference Lesser1977). In other words, children can comprehend the causality of objects physically close to each other more quickly than others. Nevertheless, it is not possible to generalize for every event because some causal events do not have spatial contiguity. To exemplify, a lamp in a room may illuminate the opposite wall spatially distant from it, and this distance does not inhibit causality. Secondly, it is necessary to mention another concept, temporal order, which is a powerful cue for children. Temporal order represents the order between cause and effect, and children benefit from this cue beginning from third year of life (Shultz et al., Reference Shultz, Altmann and Asselin1986). In addition to these two factors, children also use the explanations given to them as clues, and these clues can also affect their ability to make causal inferences. A study investigated the role of verbal framing in children extracting causal evidence from complex scenes (Butler & Markman, Reference Butler and Markman2012). The verbal framing mentioned refers to how adults, in their conversations with children, present and guide causal events in a way that is easily understandable for the child. In other words, verbal framing involves providing verbal cues that help direct the child’s attention to the relevant aspects of a situation, enabling them to make causal inferences. For example, asking to a child “Can you help me figure out what makes Catherine giggle?” as a verbal frame gives a verbal clue to understand causal process. Results indicate that verbal framing or verbal cues significantly improves preschooler children’s causal comprehension of complex scenes. Cues of context and semantic content are essential for comprehending causality. However, grasping and taking advantage of linguistic features, causal verbs, and causal connectors is a turning point for children’s causal language production.
The emergence of causal language typically begins during the preschool years, around ages 2–3 (Bowerman, Reference Bowerman1974), but it becomes more sophisticated between ages 3–5 (Trabasso & Magliano, Reference Trabasso and Magliano1996; Trabasso & Nickels, Reference Trabasso and Nickels1992). Preschool children can comprehend the process and correctly answer questions about causal events (Bullock, Reference Bullock1985; Kun, Reference Kun1978). However, children need to learn how to use conjunctions such as because and so to remark the cause-and-effect order (Brown et al., Reference Brown, Lile and Burns2011). For instance, if a boy hits the vase and the vase gets broken, we can explain this causal event in two ways. On the one hand, we can form a sentence with because and indicate the effect first (e.g., Vase was broken because he hit the vase). On the other hand, we can reframe our sentence so that causes come first (e.g., He hit the vase, so the vase broke). However, when considering their linguistic abilities, making a complete causal sentence with two clauses is difficult for children (French, Reference French1988). Besides, children have difficulties with comprehension “because” of a conjunction that reports a cause and effect until they are almost 7 or 8-year-olds, and they presume that this conjunction is like “and” or “then” (Kyrantzis et al., Reference Kyrantzis, Guo and Ervin-Tripp1990; McCabe & Peterson, Reference McCabe and Peterson1985).
Syntactic structures in causal language include causal verbs and conjunctions for producing and comprehending causal events. Studies show that children begin to produce and comprehend causal and transitive verbs from their second year of life (Corrigan & Stevenson, Reference Corrigan and Stevenson1994; Göksun et al., Reference Göksun, Küntay and Naigles2008; Naigles, Reference Naigles1990; Sexton, Reference Sexton1983). Even though infants comprehend causal verbs such as push or open during this period, producing these types of verbs is difficult for them when lexicalizing an event (Cohen et al., Reference Cohen, Amsel, Redford, Casasola and Slator1998). Children aged 3 or 4 can establish a cause-and-effect relationship and produce causal verbs (Bowerman, Reference Bowerman1974; Corrigan & Stevenson, Reference Corrigan and Stevenson1994). On the other hand, according to Kline et al. (Reference Kline, Muentener and Schulz2013), exposure to transitive verbs helps speakers to perceive and remember causal events. Children’s comprehension of causality from this period allows them to construct a new thinking ability by enabling them to use causal reasoning (Harris et al., Reference Harris, German and Mills1996).
Some studies focus on children’s causal production and comprehension from various perspectives. These are mostly types of tasks used depending on the context or involve various conditions. In research focused on language production, the tasks used are predominantly “play”-centred, with examples including free and guided play. Free play is a type of play that allows the child to freely and creatively express themselves without any commands or limitations (Ahmed et al., Reference Ahmed, Khan and Mehmood2023; Belsky & Most, Reference Belsky and Most1981). Guided play, on the other hand, encourages the child to play within a structured or semi-structured system and involves commands that create boundaries in the play (Weisberg et al., Reference Weisberg, Hirsh-Pasek and Golinkoff2013; Yu et al., Reference Yu, Shafto, Bonawitz, Yang, Golinkoff, Corriveau and Xu2018). In addition to these two different types of play, storytelling stands out as a method commonly used in language studies. Storytelling is a task where a story is generated sometimes by the parents and sometimes by the child, using a picture or sentence as a prompt, or sometimes without any prompts at all (Kandemir et al., Reference Kandemir, Özer and Aktan-Erciyes2024; Lucarevschi, Reference Lucarevschi2016; Stavans & Goldzweig, Reference Stavans and Goldzweig2008). Although these three tasks have similarities, their structures are quite different, and they reveal different findings when used for different purposes in research. For example, Kwon et al. (Reference Kwon, Bingham and Lewsader2013) conducted a study comparing mother–child and father–child interactions during free and guided play. The results indicated that in the child–parent interactions, parents provided more cognitive scaffolding, less negativity, more complex language, during free play than in structured, guided play. Additionally, another study revealed that during storytelling with their toddlers, mothers made more references to emotions, states of consciousness, causal explanations, and connections to the children’s lives compared to free play (Farkas et al., Reference Farkas, Del Real, Strasser, Álvarez, Santelices and Sieverson2018). Overall, research highlights the importance of incorporating diverse tasks into the design of language-focused studies, such as causal language input and production. Each task can yield unique insights using different measurement methods, and they also provide opportunities for cross-comparison. Making this comparison is highly important and informative for the current study and future studies examining differences in parental language input, as it provides a deeper perspective on the findings. Rather than comparing overall paternal and maternal language input, comparing them based on tasks that are inherently and structurally different may provide stronger evidence for understanding these differences.
The home environment and the child’s exposure to language at home are essential in language development. On the one hand, while the contribution of maternal speech in terms of the language development of the child is frequently investigated, paternal speech is neglected (Dieterich et al., Reference Dieterich, Assel, Swank, Smith and Landry2006; Tamis-LeMonda et al., Reference Tamis-LeMonda, Bornstein and Baumwell2001). On the other hand, some studies compare maternal and paternal speech and behaviour directed at children (Cabrera et al., Reference Cabrera, Karberg, Malin and Aldoney2017; Fagan et al., Reference Fagan, Day, Lamb and Cabrera2014; Golinkoff & Ames, Reference Golinkoff and Ames1979; Pancsofar & Vernon-Feagans, Reference Pancsofar and Vernon-Feagans2006; Teti et al., Reference Teti, Bond and Gibbs1988). Although research focusing on fathers is less compared to mothers, previous studies show that fathers’ contribution to children’s language development is as beneficial and necessary as mothers’ (Rowe et al., Reference Rowe, Coker and Pan2004; Tamis-LeMonda et al., Reference Tamis-LeMonda, Shannon, Cabrera and Lamb2004). According to Gleason’s (Reference Gleason1975) “bridge hypothesis,” fathers have more references and teaching to the outside world because paternal communication with children is cognitively demanding compared to mothers. In a more recent study, Rowe et al. (Reference Rowe, Leech and Cabrera2017) investigated the bridge hypothesis in low-income families and found that while fathers and mothers were similar in terms of the amount, diversity of vocabulary, and linguistic complexity, fathers tended to use more “wh” questions with their children. Even though significant changes in family dynamics and parental practices have emerged to date, studies focusing on paternal influences are still rare. For this reason, the overall language input the child receives from the fathers and mothers should be investigated and compared simultaneously.
3. Maternal and paternal causal language input and children’s outcomes
The language input of parents provide to their children is essential in terms of quantity and quality. Anderson et al., (Reference Anderson, Graham, Prime, Jenkins and Madigan2021) conducted a meta-analysis of parental linguistic input regarding quality and quantity. Results indicated that parents’ quality of linguistic input is more strongly associated with child language skills compared to quantity. However, they mentioned that there were many studies about maternal language input but limited studies which focused on paternal language input. Besides, limited studies show us that maternal and paternal speech have similar and different aspects. Golinkoff and Ames (Reference Golinkoff and Ames1979) conducted a study comparing maternal and paternal language input. They set a free play session for each parent and measured utterances, conversational turns, directives, questions, and repetitions of parents to their children. Results show that the language mothers and fathers use for their 19-month-old children does not differ quantitatively or qualitatively. However, fathers seem to talk less than mothers in terms of the number of utterances during triadic sessions. In another research conducted by Rowe et al. (Reference Rowe, Coker and Pan2004), there was no difference in vocabulary variety, linguistic complexity, or length of utterance in parents’ speech to their toddlers. A further study found that the mothers and fathers did not differ in terms of the words and questions they used, but only the different words used by the fathers predicted the child’s later language skills (Pancsofar & Vernon-Feagans, Reference Pancsofar and Vernon-Feagans2006).
Overall, some studies show that maternal and paternal speech are similar (Malone & Guy, Reference Malone and Guy1982; Rowe et al., Reference Rowe, Coker and Pan2004; Tamis-LeMonda et al., Reference Tamis-LeMonda, Shannon, Cabrera and Lamb2004), and some studies suggest that the language input of mothers and fathers is quite different and that mothers are more influential (Davidson & Snow, Reference Davidson and Snow1996; Leaper et al., Reference Leaper, Anderson and Sanders1998; McLaughlin et al., Reference McLaughlin, White, McDevitt and Raskin1983). For example, Endevelt-Shapira et al. (Reference Endevelt-Shapira, Bosseler, Mizrahi, Meltzoff and Kuhl2024) conducted a longitudinal study and showed that maternal interaction and directed speech with infant improve later language production of infant. In another research, they focused on maternal and paternal language input directed to their toddler (Quigley & Nixon, Reference Quigley and Nixon2024). Results indicated that maternal and paternal language input were similar in terms of complexity and verb production. However, maternal language input included more diverse noun types than paternal language input, and toddlers produced more noun types directed to their mothers compared to fathers. In sum, these contradictory findings may stem from the lack of systematic inclusion of both maternal and paternal input in study designs. Therefore, a more comprehensive approach is needed where maternal and paternal language input is considered.
Although maternal input has been examined regarding quality and quantity, a few studies focus on the causal language input used for the child and later child outcomes. For instance, some studies have focused on parental causal connectives and examined their impact on children’s causal connective production. Two different growth curve analyses found that as the causal connectives used by the parents increased, the child’s causal connective production also increased (Van Veen et al., Reference Van Veen, Evers-Vermeul, Sanders and Van Den Bergh2009, Reference Van Veen, Evers-Vermeul, Sanders and Van Den Bergh2013). In addition, parents used more causal connectives as their children got older (Van Veen, Reference Van Veen2011). On the other hand, Aktan-Erciyes and Göksun (Reference Aktan-Erciyes and Göksun2023) conducted a study to investigate how early parental causal input predicts children’s later causal verb comprehension. Results indicated that only parental morphological causative input in free-play but not goal-directed play sessions predict children’s later causal verb comprehension performance. In another study conducted by Dunn et al. (Reference Dunn, Brown and Beardsall1991), it was found that mothers’ causal and mental state conversations predicted later reasoning and emotional understanding skills in children. In addition to these findings, LaBounty et al. (Reference LaBounty, Wellman, Olson, Lagattuta and Liu2008) compared the mothers’ and fathers’ causal language inputs. They measured maternal and paternal causal term usage such as because, how, and why for coding causal references. They found that the causal references of both the mother and the father when talking about emotions predicted the later emotion understanding and theory of mind skills. In addition, this study shows that mothers make more causal inferences than fathers and emphasizes the need for detailed research into the causal input of the parents. Recent studies have found a relationship between mothers’ causal speech and children’s ability to comprehend causal stances (Alvarez & Booth, Reference Alvarez and Booth2016; Booth et al., Reference Booth, Shavlik and Haden2020). These results show that the causal input of the parents positively affects and improves the children’s reasoning ability.
4. The present study
Based on the research findings, the present study aims to achieve three key objectives: (1) examining differences in maternal and paternal language use, (2) assessing children’s causal language production across different tasks and interaction contexts, and (3) investigating the relationship between parental causal language input and children’s causal language development. Paternal input, which might have a crucial explanatory power for child outcomes, should be addressed in input studies. Research on parental language input often focuses on structural and semantic aspects, with causality receiving less attention (e.g., Aktan-Erciyes & Göksun, Reference Aktan-Erciyes and Göksun2023; Booth et al., Reference Booth, Shavlik and Haden2020; LaBounty et al., Reference LaBounty, Wellman, Olson, Lagattuta and Liu2008). In the present study, free and guided play sessions, along with a storytelling task, were used to assess maternal and paternal causal language inputs and children’s causal language production directed towards each parent. To address the research questions, we controlled for variables such as children’s age, expressive language skills, socioeconomic status (SES), and parental linguistic complexity. The study’s hypotheses are outlined below, with detailed procedures provided in the methods section.
H1: Mothers will use more causal language directed to their child compared to fathers.
H2: Children’s causal language production will differ in terms of task types. Children will use more causal language in story telling task compared to play tasks due to the story includes various causal actions.
H3: Children’s causal language production will differ in terms of which parent the child is interacting with. Children will use more causal language directed to mothers compared to fathers.
H4: Maternal causal language input will be positively associated with the child’s causal language directed to the mother.
H5: Paternal causal language input will be positively associated with the child’s causal language directed to the father.
5. Method
5.1. Participants
Sixty 4- to 5-year-old children (Mage = 56 months, SD = 6.5) and their parents (both mothers and fathers) participated. We used the G*Power 3.1.9.2 program to determine the required sample size (α = .05, power = .90). The effect size was .33 and drawn from the recent meta-analysis of parental linguistic input and children’s language outcomes (Anderson et al., Reference Anderson, Graham, Prime, Jenkins and Madigan2021). The target sample size was calculated as 60, and 60 families were recruited. A demographic form was used to determine the child’s age and gender as well as the parents’ years of education, and subjective ratings of socioeconomic status. Socioeconomic status, representing participants’ income level, was measured based on subjective rating using a 1–5 Likert scale (1 – low, 2 – low to middle, 3 – middle, 4 – middle to high, 5 – high). A composite SES score with Z scores of education levels and socioeconomic status was created, and the SES level was controlled for all analyses. The rationale behind adopting this approach is that SES is a multidimensional construct; therefore, creating a composite SES score that incorporates both objective (years of education) and subjective (self-reported socioeconomic status) components enhances the quality of the data for this variable. In this context, we combined maternal and paternal education years with the family’s perceived socioeconomic status in order to strengthen the results and included this composite score as a control variable in the analyses. We showed fathers’ and mother’s education year, and also information about socioeconomic status of families in Table 1.
Demographic information of families
5.2. Instruments of assessment
Both child–parent dyadic tasks and an expressive language test for children were used. Before the data collection process, parents fill out a demographic information form. After the demographic information form, there were three tasks for each parent and one task for the child. Each parent was alone with the child during their own session, and the other parent was not present or involved. Children completed child tasks independently, without the presence of their mothers or fathers, and without receiving any support.
5.2.1. Parent–child tasks: free/guided play sessions
The toy was used during mother–child and father–child play sessions. This toy has seven geometric shapes, and these shapes can be combined in different combinations to form different figures. The free play session began with the instruction for mothers and fathers to play with their children in any way they wished. In this context, they explored the tangram toy and created any figures they desired using its pieces. In contrast, the guided play session involved a structured format. The goal was to complete the two specific target figures based on the instructions (i.e., rabbit and cat figure was presented) provided to each parent–child dyad.
First, both mothers and fathers participated in the free play session with their child separately in a dyadic design. Each parent was given 5 minutes to play with their child using the tangram toy. Providing the same free play duration for each parent supports standardization and allows for comparisons between them. The determination of this duration was guided by children’s attention spans during play, similar studies in the literature, and the need for a sufficient amount of time for both language input and production. They were instructed that they have a limited amount of time and can play as they wish. After the free play session, parent–child dyads were asked to construct one target figure for each parent–child dyad using tangram shapes. These two figures were chosen based on their similarity in contours and close render times. Thus, parents are involved in both free play and guided play.
5.2.2. Parent–child tasks: parent’s storytelling task
The picture book named “A Boy, A Dog, and A Frog” was presented (Mayer, Reference Mayer1967). Both mothers and fathers were asked to narrate the story to their children during their turns. In the father–child session, after the free/guided play, fathers narrated the story to their child. Following this, the father–child session was completed, and the mother–child session began. Similarly, after completing the free/guided play, mothers also narrated the story to their child. We used nine images from this picture book, each depicting causal events. Unlike the materials used in free or guided play, this story features a range of causal actions within each image (Appendix A).
5.2.3. Child tasks: turkish expressive and receptive language test (TIFALDI) – expressive subtest
Children’s expressive language performance was measured with TIFALDI – Turkish Expressive and Receptive Language Test – Expressive subtest (Berument & Güven, Reference Berument and Güven2013). There are 80 pictures in this test. These pictures depicted animals, plants, objects, and foods from daily life, and they range from easy to more difficult to express for children (Appendix B). Children were instructed to name the picture they were presented with. They received 1 point for each correct answer, and the test ended when they made 8 incorrect responses out of 10. According to the reliability analysis conducted by Berument and Güven (Reference Berument and Güven2013), the Cronbach’s alpha reliability coefficient for the TIFALDI – Expressive subtest was found to be .95 for the 4–5 age group. In addition, test–retest reliability was also analysed, and the coefficients were found to be .89 and .94 for the same age group.
5.3. Procedure
Ethical approval was obtained from the [Blinded University] Human Research Ethics Committee prior to conducting this research (E-20562626-300-24656). After obtaining ethical approval, the present study was conducted online, and the data were collected via the Zoom program. To inform families about the study, announcements were made through various social media channels. Families first filled out demographic information forms including information regarding their children, the family’s socioeconomic status, the parents’ education level their mailing addresses. In the first stage, tangram toys were sent to families via shipping company to be able to conduct the research. After the toy was delivered to the families, arrangements were made to schedule a meeting with them. Until the meeting took place, the children were not allowed to see or play with the toy. All families played with the toy for the first time during the interview. Four- and five-year-olds children and their parents attended the study, and there was one session with three parts. In the first part, father–child dyads played with the tangram toy in the free and guided play game. After that, fathers narrated the picture-based story to their children. In the second part, mother–child dyads played the game and told the story similarly.
The first and second parts of the session took approximately 30 minutes for mother–child and father–child dyads. After that, in the third part of the session, the children attended the session independently. In the third part, the expressive subtest of TIFALDI was administered. We required mothers to remain with the child in case any technical issues could occur during the Zoom session, and they always attended the second part. Fathers were involved in the study in the first part. All three parts of the session were held on the same day for all participants, and sessions were recorded. Research assistants completed transcriptions for all sessions. The transcriptions were prepared to provide a verbatim written record of the utterances used by the mothers, fathers, and children throughout the video-recorded sessions, including both free/guided play and storytelling. Transcriptions from video recordings were prepared and documented in Excel files by four research assistants. Each transcription was then reviewed and verified by another researcher; there was 100% agreement. After finalizing the transcriptions, all data were coded.
5.4. Data coding
5.4.1. Causal language coding
After the data collection and transcription processes, the causal language coding was completed for the maternal and paternal play and storytelling tasks. Also, children’s causal language production directed to the mother and father regarding the parent–child dyadic part of the session was coded. In this coding, causal connectives (e.g., because so), lexical causal verbs (e.g., open, put), and morphological causal verbs (e.g., ye-dir, “make someone eat”) used by parents and children were coded (Table 2).
Causal language coding scheme
Two research assistants completed causal language coding. The main coder coded all the data, and the reliability coder coded 20% of the data. The intraclass correlations coefficients (ICC) analyses were conducted to investigate the reliability between coders. The two coders indicated high reliability for causal language coding (CI(95%) = .96 to .99, p < .01).
5.4.2. Linguistic complexity coding
On the other hand, the linguistic complexity coding of the parents was made for use as a control variable. The reason for including linguistic complexity as a control variable was to account for its potential confounding effect when comparing maternal and paternal causal language input. Although some studies in the literature have found similar levels of linguistic complexity between mothers and fathers, it was important to control for the possibility that differences in linguistic complexity might influence the comparison. Both free/guided play and storytelling sessions were coded for each parent. First, we divided participants’ sentences clause by clause. A clause can be defined as “any unit that contains a unified predicate … expressing a single situation (activity, event, or state)” (Berman & Slobin, Reference Berman and Slobin1994, p. 660). After that, we detect clauses which include one predicate, and we coded them as simple clauses. On the other hand, we coded clauses which include conjunctions as infinitival, coordination, converb, or subordination clauses. The predicate of the sentence with a conjunction coded as main clause. The details about components of linguistic complexity coding, their definitions, and examples are shown in Table 3. When calculate linguistic complexity score, we divided number of complex clauses (infinitival, coordination, converb, subordination, and main clauses) to all clauses’ numbers (simple clauses and complex clauses). The main coder coded all the data regarding linguistic complexity, and the reliability coder coded 20% of the data. The intraclass correlations coefficients (ICC) analyses were conducted to investigate the reliability between coders. The two coders indicated high reliability for linguistic complexity coding (CI (95%) = .95 to .99, p < .01).
Linguistic complexity coding scheme
5.5. Data preparation
Paternal and maternal causal language inputs were extracted from free play, guided play, and storytelling tasks. Likewise, paternal and maternal linguistic complexity, number of clauses, number of words, and duration scores were obtained from free play, guided play, and storytelling tasks. Children’s causal language directed to fathers is determined by the father–child dyadic part. The mother–child dyadic part determines children’s causal language directed to mothers. Children had different causal language scores from three tasks and two parent–child dyadic sessions.
5.6. Analysis plan
Preliminary analyses were first conducted to rule out potential gender or order effects. Inter-rater reliability for the coded scores was calculated using intraclass correlation analysis. Independent t-tests were performed to assess whether children’s gender influenced maternal and paternal causal language input, as well as children’s causal language directed towards mothers and fathers. To examine potential order effects related to the inclusion of fathers first in the study, paired samples t-tests were used to compare linguistic complexity, number of clauses, number of words, and duration scores between mothers and fathers. Additionally, Pearson correlation analysis was conducted to measure the relationships among the independent, dependent, and control variables included in the analyses. Next, we tested our hypotheses. Paired samples t-tests were used to compare maternal and paternal causal language inputs. Repeated measures ANCOVA was then conducted to analyse children’s causal language production based on task types and the parent they interacted with. The dependent variable was children’s causal language production, while the independent variables were task types (free play, guided play, and storytelling) and the parent the child was interacting with (father or mother). SES composite scores, TIFALDI expressive language scores, and children’s age were included as covariates. Finally, hierarchical regression analyses were performed to examine the relationship between parental causal language input and children’s causal language outcomes. All analyses were conducted using IBM SPSS Statistics 26 and JAMOVI software.
6. Results
6.1. Descriptive statistics for all variables
The descriptive statistics of all variables are presented in Table 4. The skewness and kurtosis values of the variables were within the normal range.
Descriptive statistics across variables
6.2. Preliminary analyses
Firstly, gender differences among the children were compared. Then, differences in general language input of mothers and fathers, except for causal language measures, were examined. The linguistic complexity, number of words, clauses, and duration of mothers’ and fathers’ free/guided play and storytelling sessions were compared. Lastly, the relationship between the independent, dependent, and control variables was examined through a correlation analysis.
6.2.1. Gender differences across variables
Independent t-tests conducted to investigate gender differences on maternal/paternal causal language, child causal language directed to fathers/mothers, child causal language in the storytelling task, child counterfactual thinking, child causal verb production causality score, and accuracy scores were not significant (all p’s > .05). All analyses details are shown in Table 5.
Gender differences across variables
6.2.2. Comparison of maternal and paternal language input
To investigate whether there were any differences in the language input of mothers and fathers, paired samples t-tests were conducted. For this purpose, the linguistic complexity scores from mothers and fathers were compared, and a paired samples t-test was used for this comparison. The analysis revealed no significant difference between mothers and fathers regarding linguistic complexity, t(59) = −.306, p = .761. Besides, mothers and fathers were compared in terms of their different scores. In the procedure of the study, mothers and fathers were included in a certain order, and fathers always played with their children first. To control for the effect of this order, mothers and fathers were compared in terms of their different scores. To do this, the total number of clauses, total number of words, and duration (in seconds) from mothers and fathers were compared. The paired sample t-test revealed that there was no significant difference between mothers and fathers in terms of the total number of clauses, the total number of words, and duration, t(59) = .230, p = .819; t(59) = .308, p = .759; t(59) = .878, p = .383 respectively. Descriptive statistics and analyses are shown in Table 6.
Comparison of paternal and maternal language input
6.2.3. Relations between variables
We conducted zero-order Pearson correlation analyses to investigate the relation between all variables. There was a significant positive relationship between parental causal language and maternal causal language, r(58) = .305, p = .018. When we checked child-causal language directed to fathers and mothers separately, there was a positive and significant relationship between child-causal language directed to fathers and paternal causal language, r(58) = .358, p = .005. Also, child-causal language directed to mothers and maternal-causal language had a significant and positive relationship, r(58) = .372, p = .003. Likewise, there was also a positive association between child-causal language directed to fathers and maternal-causal language, r(58) = .264, p = .042. Children’s TIFALDI – expressive language scores and age were also significantly positively related, r(58) = .449, p = .000. Details of results are presented in Table 7.
Zero-order correlations between variables
Note: *p < .05, **p < .01, ***p < .001.
6.3. Comparison of maternal and paternal causal language inputs
Our first hypothesis was to investigate differences between maternal and paternal causal language. We conducted paired samples t-tests using Bonferroni correctionFootnote 1 and compared paternal and maternal overall use of causal language inputs and in different tasks (H1). The results showed a significant difference between mothers and fathers regarding the total causal language scores, t(59) = 2.710, p = .009. Fathers (M = .336, SD = .059) used more causal language than mothers (M = .314, SD = .049) towards their children. We conducted another paired samples t-test to investigate the source of this difference. Results indicated that maternal and paternal causal language input in storytelling and guided play tasks were not different, t(59) = .285, p = .777; t(59) = −.030, p = .976 respectively. There is a significant difference between fathers and mothers in causal language input only for free play task, t(59) = 3.139, p = .003. Fathers (M = .404, SD = .092) used more causal language in free play than mothers (M = .353, SD = .086). Details of the analyses are shown in Table 8.
Maternal and paternal causal language input differences
Note: *p < .05, **p < .01.
6.4. Comparison of children’s causal language directed to parents across different tasks
We investigated how children’s causal language production differed based on tasks and conditions to test H2 and H3. We conducted 3 (Task types: Free play, guided play, storytelling) x 2 (Parent: Father, mother) Repeated Measures ANCOVA. We used the SES composite score, TIFALDI – expressive language score, and children’s age as covariates. Results indicated a main effect of tasks, F (2, 86) = 19.12, p < .001, η2p = .30. Post hoc t-tests showed that children produced more causal language directed to father in free play task (M = .40, SD = .18) compared to storytelling task (M = .17, SD = .21), t (43) = 4.911, p < .001. However, we did not find significant main effect of which parent the child is interacting with on children’s causal language production, F (1, 43) = 1.51, p = .226, η2p = .03. We also did not find a significant interaction between task types and which parent the child is interacting with, F (2, 86) = 1.76, p = .177, η2p = .03). Overall depiction of scores can be seen in Figure 1.
Children’s causal language production scores across tasks and which parent the child is interacting with.
6.5. Relations between maternal causal language input and child causal language directed to the mother
The association between maternal causal language input and child causal language outcome directed to the mother was investigated. A hierarchical regression analysis was conducted for this aim, and child’s causal language directed to the mother was the outcome variable. The child’s age and the SES composite score were control variables included in the first step. In the second step, expressive language (TIFALDI) was included as another control variable. Maternal linguistic complexity was added in the third step as a control variable. Finally, in the fourth step, maternal causal language was included. According to the results, the first step only explained 1% of the total variance of the model and was not significant, F(2, 57) = .411, p = .665. In the second step, TIFALDI was introduced to the model. The second step explained 03% of the variance and was not significant, F(3, 56) = .593, p = .622. TIFALDI did not improve the model significantly, ΔR2 = .01, F(1, 56) = .958, p = .332. The third step explained 07% of the total variance and was not significant, F(4, 55) = 1.128, p = .353. Maternal linguistic complexity did not improve the model significantly, ΔR2 = .04, F(1, 55) = 2.681, p = .107. In the fourth step, maternal causal language input was introduced to the model, and the model was significant, F(5, 54) = 2.418, p = .047. Maternal causal language input improved the model significantly, ΔR2 = .10, F(1, 54) = 7.078, p = .010. The fourth step explained 18% of the total variance, and maternal causal language input was positively associated with child causal language directed to the mother, β = .340, p = .010 (see Table 9).
Relations between maternal causal language input and child causal language directed to mother
Note: *p < .05.
6.6. Relations between paternal causal language input and child causal language directed to the father
We conducted a hierarchical regression analysis to examine the relationship between paternal causal language input and child causal language outcome directed to the father. The outcome variable was the child’s causal language directed to the father. In the first step, age and SES composite were included as control variables. TIFALDI was introduced to the model as a control variable in the second step. Paternal linguistic complexity was added to the model as a control variable in the third step. Finally, paternal causal language is included in the model as a predictor in the fourth step. According to the results, the first step explained 03% of the variance and was not significant, F(2, 57) = .895, p = .414. The second step explained 03% of the variance and was not significant either, F(3, 56) = .646, p = .589. TIFALDI did not improve the model significantly, ΔR2 = .003, F(1, 56) = .174, p = .678. The third step explained 03% of the model with paternal linguistic complexity and was insignificant, F(4, 55) = .499, p = .736. Finally, the fourth step explained 14% of the total variance and was still insignificant, F(5, 54) = 1.794, p = .130. However, paternal causal language input significantly improved the fourth step, ΔR2 = .107, F(1, 54) = 6.763, p = .012. Paternal causal language input is positively associated with children’s causal language outcome directed to the father, β = .351, p = .011 (see Table 10).
Relations between paternal causal language input and child causal language directed to father
Note: *p < .05.
7. Discussion
The main objectives of this study were to examine the differences between paternal and maternal causal language, explore the relationship between paternal and maternal causal language input and children’s causal language production, and analyse the differences in children’s causal language production across different tasks and when interacting with each parent. Considering these purposes, we explored three research questions: (1) Do maternal and paternal causal language inputs differ? (2) Does children’s causal language production vary based on task types and the parent they are interacting with? (3) Are maternal and paternal causal language inputs linked to children’s causal language production? To address the research questions, we employed a design involving mothers, fathers, and their children. In order to minimize the bidirectional influence of parents on one another, separate sessions of father–child and mother–child were completed. On the other hand, children participated in the research both individually and alongside their parents in a specified order. This design allowed us to efficiently examine differences in paternal and maternal causal language, parents’ causal language input to their children, and children’s causal language production. Based on the participants’ data, we tested five hypotheses. The results showed that fathers used more causal language with their children during free play compared to mothers. We found a significant main effect of task type on children’s causal language directed at their parents, with children producing more causal language towards their fathers during the free play task compared to the storytelling task. Additionally, paternal causal language input showed a positive correlation with children’s causal language production directed at their fathers. Similarly, maternal causal language input was positively associated with children’s causal language production directed at their mothers.
7.1. Maternal and paternal causal language input differences
The first research question aimed to explore how maternal and paternal causal language input differs. To address this, we tested our first hypothesis, investigating the differences between maternal and paternal causal language inputs. The analysis revealed that fathers used more causal language structures than mothers, but only during free play. This finding was supported by the preliminary analysis, which examined maternal and paternal causal language differences. We compared mothers’ and fathers’ total word count, number of clauses, task duration, and linguistic complexity across all tasks. The comparison showed that mothers and fathers completed the tasks in a similar amount of time, with no significant difference in duration. Likewise, no differences were found in the total number of words or clauses used by mothers and fathers, and the complexity of the language used was also similar. These findings suggest that both mothers and fathers completed the tasks with comparable language complexity, word/clause count, and time spent.
Contrary to the findings of the present study, previous research suggests significant differences between maternal and paternal language input, with mothers typically using more words and more complex language when interacting with their children (Davidson & Snow, Reference Davidson and Snow1996; Leaper et al., Reference Leaper, Anderson and Sanders1998; McLaughlin et al., Reference McLaughlin, White, McDevitt and Raskin1983; Shapiro et al., Reference Shapiro, Hippe and Ramírez2021). However, the results from the current study, which found no significant differences between mothers and fathers in the number of words or linguistic complexity, diverge from this finding. On the other hand, Rowe et al. (Reference Rowe, Coker and Pan2004) reported no differences between maternal and paternal language input in terms of linguistic complexity and utterance length. Consistent with their findings, the present study also shows no difference between mothers and fathers regarding linguistic complexity, total words/clauses, and task duration. This could be attributed to the educational backgrounds of the parents in our sample, as most mothers and fathers had high levels of education. This suggests that, contrary to traditional norms, fathers may recognize the importance of playing an equally active role in child-rearing alongside mothers. Fathers’ active involvement and their ability to exhibit parenting effectiveness equivalent to mothers could explain the lack of significant differences observed between them in our findings. On the other hand, the findings reveal that paternal causal language input exceeded maternal input only during free play. When comparing our results with previous studies on maternal and paternal language input, we see both similarities and differences. Although studies in this area are limited, most findings suggest that maternal and paternal language inputs are similar (Malone & Guy, Reference Malone and Guy1982; Rowe et al., Reference Rowe, Coker and Pan2004; Tamis-LeMonda et al., Reference Tamis-LeMonda, Shannon, Cabrera and Lamb2004).
Additionally, while some research shows that fathers tend to speak less than mothers in triadic sessions, the current study found no difference in the total language input between mothers and fathers, however, in a dyadic session. A more recent study by LaBounty et al. (Reference LaBounty, Wellman, Olson, Lagattuta and Liu2008) identified a difference between mothers and fathers in making causal inferences. This study, which involved preschoolers and used separate mother–child and father–child sessions, explored the relationship between children’s emotion understanding and theory of mind skills. It focused on maternal and paternal differences in causal explanatory language, specifically relating to emotions and desires. The results indicated that mothers provided more emotional causal explanations than fathers. The authors suggested that because mothers tend to make more emotional inferences in daily interactions with their children, their causal explanations in emotional narratives were more effective. Likewise, Dunn et al. (Reference Dunn, Brown and Beardsall1991) found that only mothers’ causal and mental state conversations predicted children’s later reasoning and emotional understanding skills. In contrast, the current study found opposing results, showing that fathers provided more causal language input than mothers in a non-emotional design. One reason might be that the current study does not address research questions related to emotional understanding or mental state talk, nor does it utilize measurement tools for this purpose. Literature shows that mothers make valuable contributions to their children’s language development through emotional language or mental state talk. However, parental language can vary depending on the context and tools used. In particular, it could be argued that fathers may provide more analytical and logically focused language input than mothers, while mothers tend to rely more heavily on emotional inferences. To understand these differences, longitudinal studies examining the language input of mothers and fathers, as well as their long-term effects on children’s cognitive and social development, are needed.
7.2. Children’s causal language production in different tasks and parental interactions
The second research question aimed to examine whether children’s causal language production varies based on the type of task (free play, guided play, storytelling) and the parent they are interacting with (mother or father). The findings revealed a significant main effect of task type on children’s causal language production, indicating that the context in which a child is engaged plays a critical role in their use of causal language. Specifically, children produced more causal language during the free play task when interacting with their fathers compared to the storytelling task. This suggests that the open-ended nature of free play may provide a more conducive environment for children to express causal thoughts, particularly in father–child interactions. Contrary to our hypothesis that children would use more causal language during the storytelling task, due to its inclusion of numerous causal actions, the children actually produced more causal language during free play. Although this result was unexpected, it may be explained by the nature of the free play task, which involved a tangram toy, requiring children to use more lexical causal verbs. This necessity might account for the increased use of causal language during free play.
These findings are important as they suggest that the task context may be more influential in eliciting causal language from children than the specific parent–child dynamic. This aligns with previous research that underscores the role of play in supporting language development while adding new insights by showing that different task types can impact causal language use in distinct ways. On the other hand, no significant difference was found between the causal language used by children toward their mothers and fathers. Although we hypothesized that children would use more causal language with their mothers, the results showed that children produced similar amounts of causal language with both parents, despite using more causal language with their fathers during free play. This might indicate that both mothers and fathers provide equally supportive environments for language development during these tasks, or it could point to a limitation in the study’s design. The fact that children played with their fathers first may account for their greater use of causal language directed toward fathers. When they later interacted with their mothers, the second interaction may have involved less causal language. However, keeping the order fixed (with fathers first) did not have any effect on our language measures, namely, word count, number of clauses, and linguistic complexity.
The finding that children produced more causal language during free play was unexpected; however, it aligns with a similar study by Kwon et al. (Reference Kwon, Bingham and Lewsader2013). In their study, the free play task elicited more cognitive scaffolding compared to the structured guided play task. When we consider the storytelling as a semi-structured task due to the pictures that create a story structure, the current study’s findings are consistent with the Kwon et al. (Reference Kwon, Bingham and Lewsader2013) study. On the other hand, we did not observe the anticipated result that the storytelling task would elicit more causal language which is consistent with Farkas et al. (Reference Farkas, Del Real, Strasser, Álvarez, Santelices and Sieverson2018). In this aspect, our findings diverged from their study. This divergence may be mainly due to the differences in task design (unfinished story vignette versus our procedure) as well the fact that the children in our sample were of older age.
7.3. Maternal and paternal causal language inputs and children’s causal language production
The relationship between maternal causal language input and child causal language production directed to the mother was examined, and a positive association between them was assumed (H4). According to the results, maternal causal language input was positively associated with child causal language production directed to the mother. Thus, this hypothesis was confirmed. Similar studies in literature support this finding. Van Veen et al. (Reference Van Veen, Evers-Vermeul, Sanders and Van Den Bergh2009, Reference Van Veen, Evers-Vermeul, Sanders and Van Den Bergh2013) implemented two growth curve analyses in order to investigate the association between parental wh-questions and children’s causal connective production. Results indicated that the child’s causal connective production increased as parents used more causal connectives. Also, some studies focusing on parents’ wh-question input indicated that it is positively associated with child wh-question production (Hood et al., Reference Hood, Bloom and Brainerd1979; McCabe & Peterson, Reference McCabe and Peterson1985). These similar studies’ results support this finding. Although the current study focused on child causal language production, studies show that as maternal causal language input increases, so does a child’s causal verb comprehension (Aktan-Erciyes & Göksun, Reference Aktan-Erciyes and Göksun2023; Alvarez & Booth, Reference Alvarez and Booth2016; Booth et al., Reference Booth, Shavlik and Haden2020). Based on these findings, it can be concluded that maternal causal language input plays a significant role and is closely associated with children’s causal language production.
For our fifth hypothesis, we expected a positive association between paternal causal language input and child causal language production directed to the father (H5). According to the results, paternal causal language input was positively associated with child causal language production directed to fathers. Thus, this hypothesis was confirmed. Although studies conducted for similar purposes are limited in number, it is possible to refer to a few studies that support this finding. In the study by Rowe et al. (Reference Rowe, Leech and Cabrera2017), only wh-questions used by fathers were related to children’s vocabulary and reasoning outcomes. Similarly, in another study focusing on wh-questions used by fathers, it was found that these questions, including causal expressions in themselves, were related to children’s vocabulary (Leech et al., Reference Leech, Salo, Rowe and Cabrera2013). Similarly, two growth curve analyses focusing on parental causal connective inputs showed that children’s causal connective production increases with parents using more causal connectives (Van Veen et al., Reference Van Veen, Evers-Vermeul, Sanders and Van Den Bergh2009, Reference Van Veen, Evers-Vermeul, Sanders and Van Den Bergh2013). These findings indicated that both maternal and paternal causal connective input is associated with children’s causal connective production.
A review of the literature reveals that studies specifically investigating the connection between causal language input and production are relatively limited. Therefore, it is important to interpret these findings in a broader context. For instance, some studies have shown that parental causal connective input is positively associated with children’s production and acquisition of causal connectives (Van Veen, Reference Van Veen2011; Van Veen et al., Reference Van Veen, Evers-Vermeul, Sanders and Van Den Bergh2009, Reference Van Veen, Evers-Vermeul, Sanders and Van Den Bergh2013). Similarly, parental use of wh-questions has been linked to increased wh-question production in children (Hood et al., Reference Hood, Bloom and Brainerd1979; McCabe & Peterson, Reference McCabe and Peterson1985). Since wh-questions often involve causal explanations, such as using terms like “because,” they may also foster greater causal language production in children.
These findings highlight the critical role of both maternal and paternal causal language inputs in shaping and enhancing children’s causal language production, emphasizing the need for further research to explore these dynamics in greater depth. While input–output models are supported by the positive associations between parent and child causal language, the results also highlight the scaffolding role of parents, especially in open-ended tasks like free play. Additionally, Kwon et al. (Reference Kwon, Bingham and Lewsader2013) study shows that task context often outweighs whether paternal or maternal input is more influential in shaping children’s language use, emphasizing the dynamic, reciprocal nature of language development. Consistent with these findings, the current study’s results refine assumptions about mother versus father roles in language development by showing that differences are less pronounced than traditionally assumed. This indicates that both parents contribute meaningfully to children’s causal language development, but their impact may vary depending on the interactional context. Thus, the study challenges assumptions about maternal versus paternal input and highlights the importance of considering contextual factors assumed. While prior research often emphasized maternal input as the primary linguistic model, our findings indicate that paternal input also provide rich and influential input, particularly in free-play contexts. This suggests that both parents contribute meaningfully to children’s causal language development, but their impact may vary depending on the interactional setting. In this way, the study challenges assumptions of maternal versus paternal input and highlights the importance of considering contextual factors.
7.4. Limitations and future directions
The present study has some limitations. For instance, fathers always participated first, followed by mothers. Although we examined linguistic differences between mothers and fathers, counterbalancing the order of participation would have been beneficial and should be considered in future studies. Additionally, future research could focus on parents’ use of wh-questions and analyse children’s causal language production through specific components, such as lexical choices, morphological structures, and causal connectives, to gain a more detailed understanding. Incorporating measures of causal language comprehension alongside production would also enhance the robustness of the findings.
The current study findings reveal that paternal causal language input may play a key role in children’s causal language production, specifically in free play settings. Focusing on paternal language inputs would broaden the language literature in terms of parent–child interaction. To understand why fathers used more causal language in free play compared to structural or semi-structured guided play and storytelling tasks, future research may provide a more distinct answer by examining data from different contexts in greater depth. Additionally, it is essential to comprehend the various aspects of maternal and paternal language inputs that are beneficial for children’s cognitive skills, particularly in terms of causal language usage and understanding or producing references to causality. Additionally, measuring the different contributions of the mother and father to the child’s output will specifically identify which inputs the child should receive from the mother or the father, providing a broader framework. After determining the most powerful and beneficial aspects of maternal and paternal language input on children’s language production, it would be possible to make suggestions for educational interventions to enhance children’s academic and social development.
8. Conclusion
Overall, this study makes significant contributions to the existing literature by addressing gaps in our understanding of paternal language input and providing fresh insights into parent–child interactions. It underscores the previously underexplored role of fathers in language development, highlighting the importance of including both parents in research on language acquisition. By incorporating data from diverse tasks and contexts, the study offers a comprehensive and nuanced perspective. Importantly, the research fills critical gaps in the literature, where much of the focus has traditionally been on mothers. By emphasizing the role of fathers and providing a comparative analysis of maternal and paternal language inputs, the study demonstrates that both parents contribute simultaneously to children’s language development. It not only explores the interplay between maternal and paternal causal language input and children’s causal language production but also reveals how different task types and interaction contexts influence children’s language use. Through the inclusion of varied tasks such as free play, guided play, and storytelling, this study enriches our understanding of the multifaceted ways in which parental input and task context shape causal language production in children. These findings also offer practical implications for child development. Specifically, encouraging both mothers and fathers to engage in causal language during interactions can support children’s causal reasoning, complex language development, and early cognitive skills. As causal language is linked to understanding cause–effect relationships, scaffolding this through everyday conversations and play may contribute to children’s complex thinking. Future interventions and parent-training programs might benefit from emphasizing the role of both parents in fostering such development through varied interactions.
Appendix A. A Boy, A Dog, and A Frog
Appendix B. Flower Story
Open access
