Eighty-two mothers and their monolingual English-reared infants between the ages of 15- to 21-months participated. However, 29 participants' word learning performance was excluded from analyses due to counterbalancing issues (n = 4), technological issues (n = 3), fussiness (n = 5), video quality (n = 8), side bias (i.e., the infant looked toward one side of the screen 70% of the time; n = 4), or low attention (i.e., the infant looked at the screen for less than 50% of the time, n = 5). The final sample included 53 mother-child dyads (23 males, M_age= 17.92, SD = 1.99). On a background questionnaire, participants self-reported that they identified as White (n = 39, 73.6%), Asian (n = 2, 3.8%), Hispanic (n = 2, 3.8%), African American (n = 1, 1.9%), and of mixed race (n = 8, 15.2 %). We used education as a proxy for SES, as education is the proximal component of SES that likely impacts children's language trajectories (Hoff, 2013). In our sample, one participant held only a high school degree (1.9%). Two had some college experience (3.8%). Fifteen held a bachelor's degree (28.3%). Twenty-two held a master's degree (41.5%). Thirteen received a doctorate (24.5%).

Participants were contacted through social media, recruitment platforms (i.e., Children Helping Science), and a database from a research laboratory located in the mid-Atlantic. Potential subjects were excluded if they were deaf or had a hearing impairment because the task required participants to listen to auditory stimuli. Children were also excluded if they did not hear English at least 70% of the time. Mothers indicated their child's hearing and language background when completing the demographic questionnaire. This project was approved by the University's Institutional Review Board (1548843-11).

A post-hoc power analysis was conducted for a multiple regression analysis with maternal beliefs as the independent variable and word learning as the dependent variable using the G^*Power 3.0 program (Faul et al., 2007). Children's age was entered as a control variable. Results from the power analysis (α = 0.05, b = 0.80) indicated that a sample of 68 mother-child dyads was needed to detect a medium effect (f²= 0.15). Thus, our sample was slightly underpowered with only 53 participants.

Visual stimuli included two novel household objects (see Figure 1) to ensure the child was unlikely to have seen either object prior to the study. Object A resembled a blue travel razor case (i.e., glorp). Object B was a gray spaghetti measurer with four holes (i.e., dax). Objects were labeled with nonsense words that followed English phonotactic constraints (Parish-Morris et al., 2007). The auditory stimuli were created by a female native monolingual speaker of American English who used infant-directed speech.

Before the one-time video-chat appointment, mothers were given a consent form and a socio-demographic information questionnaire electronically via Qualtrics. The video-chat appointment included 5 blocks: ADS elicitation, children's novel object exploration, infant teaching, testing, and assessments (Table 1). Mothers were then asked to complete the Parent Language Belief Questionnaire (PLBQ; Ramirez et al., 2023) and MacArthur Communicative Development Inventory (MCDI; Fenson et al., 2000). The independent variables were mothers' beliefs and change in speech register. The dependent variable was children's word learning performance.

The twelve items of the Parent Belief Questionnaire that related to assessing mothers' beliefs about IDS contained high item-reliability (Cronbach's α = 0.83). A variable (referenced as IDS Belief) was created by summing each mother's responses to IDS Belief items and dividing by 12, the number of IDS Belief questions. Some items were reversed-scored. Mothers' scores ranged from 0–3. A higher score indicated that mothers thought IDS was more valuable for their children's language learning while a lower score suggested that mothers did not think IDS would be helpful. Results revealed mixed responses to the IDS Belief items. Table 3 shows the percentage of mothers that often or always (scores two and three) supported the IDS item. Overall, mothers were neutral toward their use of IDS when interacting with their infant (M = 1.72, SD = 0.46, Range = 0.67–2.75). Mothers' beliefs did not differ according to the gender of the infant, children's age, nor language ability (as measured with the MCDI), all ps > 0.05.

To examine how mothers interacted in the infant-directed communication context, we investigated how mothers' acoustic features changed when speaking with an experimenter compared to speaking with their infant. To explore changes related to speech register, we extracted the average fundamental frequency, pitch minimum, and pitch maximum in both the ADS elicitation and infant teaching block using Praat software (Boersma and Weeninik, 2018). Both blocks were at least 45 seconds long. However, the time duration of the ADS elicitation (M = 39.68, SD = 3.10, Range = 24.68–45.48 seconds) and infant teaching (M = 37.95, SD = 5.50, Range = 15.00–45.84 seconds) block still significantly differed, t (52) = 2.145, p = 0.04, d = 0.30.

Three one-way ANOVAs were separately conducted to compare the effect of the block (ADS elicitation vs. infant teaching) on the three dependent variables, mothers' average fundamental frequency, maximum pitch, and minimum pitch. Results revealed a significant difference in mothers' average fundamental frequency between the ADS elicitation and infant teaching block, F_{(1, 104)} = 95.753, p < 0.001, η² = 0.48. However, there were no significant differences in mothers' maximum nor minimum pitch across these blocks, all ps > 0.05. Altogether, results revealed that mothers had a significantly higher average fundamental frequency during the infant teaching block in comparison to the ADS elicitation block (Figure 2), suggesting that they spoke differently to their child when teaching a novel word than when they spoke to an adult.

For subsequent analyses, we created a difference score for each participant between Time 2 (infant teaching) and Time 1 (ADS elicitation) to examine mothers' speech register adjustment, a technique used in past work (Thiessen et al., 2005; Singh et al., 2009; Ma et al., 2011). Mothers' average fundamental frequency at Time 1 was subtracted from their average fundamental frequency at Time 2, yielding a difference score that allows us to control for individual differences. A positive difference score signifies that a mother's average fundamental frequency was higher when teaching a novel word to their child than when talking to an adult. Additionally, a smaller score signifies less change when interacting with their child. We next averaged speech adjustment scores across mothers and found that there was an average 52.15 Hz increase in fundamental frequency between the infant teaching and ADS elicitation blocks.

To answer this question, we ran several correlations. First, we found that a correlation between mothers' IDS Belief scores and their speech register adjustment scores was significant (r = 0.28, p = 0.05). To explore further, we examined how mothers' speech register adjustment scores related to how they responded to different components of the 12 IDS- items from the PLBQ. Below, we present these results by theme: beliefs about changes in speech related to prosody, word and sentence choice, and purpose.

To assess mothers' beliefs about adjusting their prosody—the intonation of speech—when interacting with their child, a composite score was created by adding mothers' responses to the two items and dividing by three, the number of items in this cluster. Specifically, item 11 (“I use baby talk–my voice gets higher and more melodic when speaking to my child than when speaking to an adult”), item 1 (“I address my child and my adult friends with the same tone of voice”; scored in reverse), and item 16 (“I speak more slowly when I talk to my child than when I talk to my friends”) were included. Overall, mothers reported that they occasionally adjusted their prosody when interacting with their infant (M = 1.59, SD = 0.69). A correlation between mothers' beliefs about prosody and their speech adjustment score was significant (r = 0.32, p = 0.02). When looking at the items separately, only the relation between mothers' response to item 11 (melodic speaking) and their speech register adjustment score was significant (r = 0.38, p < 0.01), suggesting that mothers' change in speech register during the infant teaching block was aligned with their belief that they used a higher and more melodic voice when speaking with their child. These findings suggest that mothers may be aware that they adjust their pitch when interacting with their child but are incognizant of the extent in comparison to interacting with another adult.

The second theme included seven items that examined mothers' beliefs about how often they adjust their words and sentences when interacting with their infant. These items included item 20: “I repeat myself more when I talk to my child than when I talk to my friends,” item 36: “I change my words when my child does not understand me,” item 8: “During a typical conversation with my child, I ask lots of questions,” item 28: “I repeat what my child says, adding new words,” item 3: “I use shorter sentences when I talk to my child than I do when talking to my friends,” item 32: “I correct my child if s/he uses the wrong word” (reverse-scored), and item 15: “I use shorter words when I talk to my child than when I talk to my friends.” A composite score was created by adding mothers' responses to the seven items and dividing by seven, the number of items in this cluster. Mothers reported that they often adjusted their words and sentences when interacting with their infant (M = 1.90, SD = 0.36). There were no significant relations between mothers' responses to these items individually and their speech register adjustment score.

Lastly, our third theme is purpose, which includes item 6: “I speak to my child as I would speak to an adult; this is how they will learn to speak intelligently” and item 40: “I speak to my child as I speak to adults to help him or her learn proper language.” Both items were reversed-scored. A composite score was created by adding mothers' responses to the two items and dividing by two, the number of items in this cluster. The relation between mothers' beliefs about their purpose in using IDS (M = 1.41, SD = 0.80) and their speech adjustment score was significant (r = 0.31, p = 0.03). When exploring the items individually, only the relation between mothers' response to item 40 and their speech register adjustment score was significant (r = 0.33, p = 0.02). Together, results suggest that mothers may talk to their child differently to help them learn language.

A multiple regression analysis was run to assess the contribution of mothers' beliefs about IDS and their speech register adjustment score on the dependent variable, children's expressive language skills measured as the number of words they produced on the MCDI. We inserted mothers' speech register adjustment score first, as mothers' speech register adjustment has been shown to be associated with children's language acquisition in past studies (Thiessen et al., 2005; Ma et al., 2011; Cristia, 2013). This model accounted for a non-significant 4.5% of the variance in children's expressive language skills, F_{(2, 52)} = 1.123, MSE = 0.025, p = 0.33.

The Parent Language Belief Questionnaire (PLBQ; Ramirez et al., 2023) is the first questionnaire designed to examine parents' beliefs about infant-directed speech. Past studies have examined parents' beliefs about “baby talk” and defined baby talk as mispronunciations and irregular words (Rodriguez and Olswang, 2003; Simmons and Johnston, 2007). However, this definition does not align with what research defines as infant-directed speech (e.g., Kaye, 1980). To address this, we designed 12 items on the PLBQ to assess mothers' beliefs about IDS based on how previous studies defined IDS (Henning et al., 2005; Song et al., 2010; Cristia, 2013; Narayan and McDermott, 2016). For example, we asked if mothers were likely to ask more questions (Luo et al., 2022) and use more repetition (McRoberts et al., 2009) when speaking with children than when speaking with adults. We intentionally did not include mispronunciations (e.g., “wawa” for water) in our definition of IDS.

Our first research question probed mothers' beliefs about infant-directed speech. We hypothesized that mothers would underestimate their IDS usage based on mothers' responses in past studies (Rodriguez and Olswang, 2003; Simmons and Johnston, 2007). We found that mothers had mixed responses to the IDS items. For example, only roughly half of mothers (50.9%) reported that they often or always “use baby talk–their voice gets higher and more melodic when speaking to their child than when speaking to an adult.” It seems that mothers might be unaware that they naturally adjust their speech register when interacting with children.

When further exploring mothers' perceptions of how they adjust their speech with their child, we found that mothers were more receptive to IDS-items that did not focus on prosody per se. For example, 81.1% of mothers reported that they often or always “change their words when their child does not understand them,” and 86.8% reported that they often or always “repeat themselves more when they talk to their child than when they talk to their friends.” Mothers are less likely to admit to adjusting their speech register in comparison to other behaviors such as simplifying vocabulary and repetition.

In past studies, parents' knowledge of child development was associated with parents' self-efficacy, child-rearing practices, variability in linguistic input, and children's language abilities (Bornstein and Cote, 2003; Albarran and Reich, 2014). For example, Rowe (2008) explored the factors contributing to parental language input variability. Exposure to child-directed speech at 24-months was correlated with children's language skills one year later. Differences in the proportion of parents' child-directed speech were related to parents' SES, as measured by income and education. Importantly, parents' knowledge of child development mediated the relationship between parents' use of child-directed speech and SES. This suggests that parents with more knowledge about children's development, as measured by parents' performance on the KIDI (MacPhee, 2002), were more likely to use child-directed speech with their children. These findings highlight how parent cognition is often connected with parent behavior (Huang et al., 2005) and, in turn, children's development (Bornstein et al., 2018; McKee et al., 2021). Therefore, understanding how parents conceptualize IDS may provide insight into why parents might not use IDS with their children.

We also considered whether mothers modified their speech register when interacting with their child compared to interacting with an adult. Variability in parents' speech register with adults and infants has been demonstrated in many studies and our findings are no different (Spinelli et al., 2017; Byers-Heinlein et al., 2021). There was a significant difference in mothers' average fundamental frequency between the ADS elicitation and infant teaching block. To further explore the extent of this difference, we created a speech register adjustment score for each participant by subtracting the average fundamental frequency during Time 1 (ADS elicitation) from Time 2 (infant teaching). An average difference of 52.15 Hz was found between the infant teaching and ADS elicitation block. These findings suggest that mothers adjusted their speech register depending on their communicative partner and align with past work in which mothers elevated their pitch when speaking with infants (Fernald and Simon, 1984; Song et al., 2010; Golinkoff et al., 2015).

Why do many parents use infant-directed speech with their young children? First, IDS may elicit and maintain children's attention (Fernald and Simon, 1984; Fernald, 1992). Indeed, Dunst et al. (2012) found that infants tend to prefer listening to IDS over ADS speech in their 34-study meta-analysis. Parents may notice that infants pay more attention to them when they use IDS.

The second possibility is that parents might use IDS to reassure their children when their physical proximity decreases (Katz et al., 1996). Adults tend to display exaggerated positive affect by smiling, widening their eyes, and raising their eyebrows when interacting with infants (Scherer et al., 1991; Swerts and Krahmer, 2010). These exaggerated facial expressions often accompany infant-directed speech (Chong et al., 2003). However, there are times when infants cannot access their parents' facial signals. For example, a parent might wash dishes or run a load of laundry while their infant plays in the next room. Infants are more likely to decipher the affectionate intention of caregivers' messages when parents use IDS than ADS (Fernald et al., 1989; Moore et al., 1997; Spence and Moore, 2003).

Third, parents may use IDS to encourage infants to participate in the conversation. The slower tempo in IDS may create an opening for infants to respond, either through gestures, babbling, or talking (Ferjan Ramírez et al., 2019). Inviting children to engage in conversational turn-taking is a key component for building language, social skills, and cognitive development (Donnelly and Kidd, 2021). For example, Hirsh-Pasek et al. (2015) examined the contributions of the quantity and quality of input in 60 parent-child dyads on children's later language development. Results revealed that the turn-taking between the mother-child dyad (i.e., “fluency and connectedness”) at 24-months correlated with children's expressive language abilities one year later. Through turn-taking, parents can encourage children to take an active role in conversations, which is associated with stronger language skills (Tamis-LeMonda et al., 2014; Romeo et al., 2018).

The fourth possibility is that parents may use IDS with their children with an eye to facilitating language development (Saint-Georges et al., 2013). Parents may modify their talk according to their children's current language skills (Bruner, 1984). For example, when Bergelson et al. (2019) observed children's linguistic environments between the ages of 3- to 20-months, results revealed that the proportion of child-directed speech increased with age as adult-directed speech decreased. One hypothesis for this decrease in ADS is that parents are detecting that their children are attuned to language and need good language models. Indeed, parents talk more to infants who have begun to talk than infants who are not talking yet (Dailey and Bergelson, 2022).

Findings from our study are consistent with several of these possibilities. Specifically, mothers used IDS to direct infants' attention toward the screen. They also invited infants to engage in conversation by encouraging infants to pronounce the target word. Lastly, we gave mothers the goal of teaching their child a novel word, allowing mothers to facilitate word learning in whatever way they chose. Many mothers adjusted their speech register to teach their children the novel word. However, we did not directly ask mothers why they used IDS when they taught their children a novel word.

The second research question probed whether mothers' beliefs would align with their use of IDS when interacting with their child. We hypothesized that there would be a misalignment between mothers' beliefs about IDS and their speech registers, especially for those mothers who indicated that they did not use IDS. To address this research question, we averaged speech adjustment scores across mothers and observed an average increase of 52.15 Hz in fundamental frequency between the infant teaching and ADS elicitation blocks. Findings suggest that most mothers adjusted their speech register when interacting with children regardless of their beliefs about IDS. Although some literature observes a correlation between parental knowledge about child development and parenting practices (Huang et al., 2005), we do not see this pattern when analyzing the relation between IDS beliefs and speech behavior. In contrast, our findings revealed that mothers' beliefs about IDS do not correlate with their speech register adjustment.

Next, we broke down the IDS composite score to investigate whether mothers' speech register adjustment scores were aligned with beliefs regarding components of IDS: prosody, word and sentence choice, and purpose. Notably, mothers' speech register adjustment scores were related to their beliefs about prosody and purpose of using IDS. For example, mothers who had a higher speech register adjustment score were more likely to report that they use “baby talk–their voice gets higher and more melodic when speaking to my child than when speaking to an adult” on the questionnaire (p = 0.02). Interestingly, there was no relationship between mothers' speech register adjustment score and their response to other items related to word and sentence choice. There are two possibilities for why mothers responded differently to items related to prosody and word and sentence choice.

The first possibility could be that mothers believe that they should talk to children in the same way that they speak to adults in terms of word and sentence choice. Alternatively, it could be that mothers are aware that they adjust their speech when interacting with their children. However, they might be less conscious of the extent to which their speech to their child differs from the way they talk with adults, thinking the difference is small. Altogether, it seems that mothers have misconceptions about IDS. This could be due to inconsistent messaging about the use of baby talk with their children. A quick Google search of “should mothers use baby talk with kids” on March 20, 2023, yields 882,000,000 results with conflicting article titles, such as “Why baby talk is good and bad for kids,” “Why should one avoid using baby talk,” or “Using baby talk isn't just cute: it could help them learn to make words.” Additionally, early writers framed baby talk as harmful (McCarthy, 1954), potentially creating a stigma around its use. Mothers are more likely to turn to Google for their source of information than research articles. Therefore, defining infant-directed speech and clarifying that it is beneficial for babies may be an initial step toward addressing mothers' misconceptions about baby talk.

Our final question assessed whether mothers' beliefs about IDS are associated with children's expressive language scores (MCDI) or children's performance on a word learning task. We hypothesized that mothers' beliefs about IDS would not relate to any of these outcome variables. Results were in alignment with our hypothesis, in that mothers' beliefs around IDS did not associate with children's expressive language scores nor children's performance on a word learning task. It seems that mothers' beliefs do not reflect what they do in real-time. In real-time, mothers engage in child-directed behaviors and elevate their speech register when talking with their children.

Most mothers in our sample used IDS when they taught a novel word to their child, even though mothers may have conceptualized IDS differently from the literature. From an evolutionary lens, these results suggest that many parents have a predisposition to use IDS with their children (Hilton et al., 2022; Schick et al., 2022). Still, the frequency of infant-directed speech varies widely in monolingual and bilingual families (Ramírez-Esparza et al., 2014; Ferjan Ramírez et al., 2022). In fact, infant-directed speech is rare in some communities (see Ochs and Schieffelin, 1984; Broesch and Bryant, 2018; Cristia et al., 2019). Thus, we cannot claim that IDS is necessary for children's language learning (Ma et al., 2011). Indeed, children in these communities develop language at the same rate as children from the United States (Casillas et al., 2020, 2021), a country that commonly uses IDS when interacting with infants (ManyBabies Consortium, 2020). Floor and Akhtar (2006) found that 18-month-old children could learn a novel word when stimuli were presented indirectly (i.e., overheard speech), presumably in ADS. However, exposure to IDS has many benefits for language acquisition, such as facilitating word segmentation (Thiessen et al., 2005), word recognition (Singh et al., 2009), and speech discrimination (Liu et al., 2003).

Additionally, infant-directed speech can be beneficial in communities where interactions with children are rare. For example, Shneidman and Goldin-Meadow (2012) explored sources of children's language input in a Mayan community and with families in the United States. Though interactions occurred far less in the Mayan community than in the U.S. sample, children's exposure to direct input by adults at 24-months was the most robust predictor of children's later vocabulary skills in the Mayan sample. Nearly all children become competent speakers in their respective communities. Therefore, although IDS is not necessary for word learning, we argue that IDS, among other aspects of the linguistic environment (e.g., eye-gaze, perceptual salience, turn-taking; Hollich et al., 2000), is a valuable tool to facilitate language development.

Our findings may be useful in existing interventions that address how parents interact with their language-learning children. For example, Ferjan Ramírez et al. (2019) examined whether coaching parents at 6- and 10-months would improve children's language skills at 14-months. Coaching included providing feedback on how parents interacted with their infant by sharing written reports and listening to recordings. Coaches also discussed the benefits of using IDS, turn-taking exchanges, and overall talk on children's language development with parents. Results revealed that parents who received coaching were more likely to increase their use of infant-directed speech than those who did not receive coaching. Furthermore, children whose parents received coaching had higher language gains than children whose parents did not receive coaching. Our findings suggest that coaches also address potential misconceptions about infant-directed speech. Addressing misconceptions may contribute to clarifying the messaging about baby talk and removing any stigma around its use.

Although these findings have implications for adding to current interventions that build children's language skills, there are several limitations to consider. First, mothers were asked to teach a novel word to their child to prompt IDS. Mothers may exhibit more IDS than they would in an everyday context because the task was goal oriented. An avenue for future work would be to see how results align with day-to-day interactions in the home, although numerous observational studies have shown that parents use IDS (e.g., Englund and Behne, 2005; Narayan and McDermott, 2016; Kalashnikova and Burnham, 2018).

Another weakness is that our sample was largely homogeneous. As beliefs about child-rearing practices differ across cultures (Johnston and Wong, 2002; Simmons and Johnston, 2007; Mancilla-Martinez and Lesaux, 2013) and the use of IDS is not present in all societies (e.g., Broesch and Bryant, 2018), a more diverse sample might differ in their beliefs about IDS. Additionally, our sample consisted of only mothers. Further research should explore whether fathers have a misalignment between their beliefs about IDS and their behavior.

Of course, it is always possible that our questionnaire did not tap into items that would show us a relationship between beliefs about IDS and its use. Thus, future work should examine whether there are other ways to ask the questions we posed that might yield more of a relation with mothers' actual behaviors. Nonetheless, it is fascinating that mothers do not seem to report they utilize some aspects of IDS.