Exposure to Language in Video and its Impact on Linguistic Development in Children Aged 3–11: A Scoping Review

Do children learn language from video?

Some studies were specifically designed to investigate target word-learning following exposure to video stimuli. A number of studies compared pre- and post-test target vocabulary scores and report that children had significantly higher scores after exposure to video with the target words embedded (Perez, 1990; Rice et al., 1992; Rice et al., 1994; Neulight, 2012; Strouse et al., 2013; Gev et al., 2017; Hseuh et al., 2017; Neuman, Samudra, et al., 2020; Neuman et al., 2021). Significant differences in post-test target vocabulary scores were also reported between groups who viewed videos containing the target vocabulary words and groups who did not – either because they viewed an edited version of the video that did not contain the target words, or because they did not view any video stimuli (Rice & Woodsmall, 1988; Rice et al., 1992; Rice et al., 1994; Oetting et al., 1995; Schryer et al., 2015; Gev et al., 2017; Hsueh et al., 2017; Gaudreau, 2021). A number of studies also assessed programme specific target word-learning by comparing scores on vocabulary outcomes to chance level. Performance above chance on at least one post-viewing vocabulary measure was reported by Neuman, Flynn, et al. (2020); Samudra, Wong, et al. (2019); Samudra, Fynn, et al. (2019); and Nussenbaum and Amso, (2016). These findings were sometimes limited to particular contexts or measures. For example, the youngest children tested by Nussenbaum and Amso (2016) did not perform above chance in low-interactivity conditions, and Neuman, Flynn, et al. (2020) did not find above chance performance on a measure of understanding words in a new context. Taken together, the studies above suggest that children can learn target vocabulary words that are embedded in video stimuli.

Three studies tested for improvements in generalised vocabulary measures following video exposure. Schryer et al. (2015) report significantly improved scores on a generalised vocabulary measure for a group of children who were exposed to an animated book reading intervention but not for children who did not view the videos. In the remaining two studies, there were significant improvements from pre- to post-test in vocabulary scores but there were no group effects (Farah et al., 2019; Strouse et al., 2013). This makes it difficult to untangle the effects of the intervention from improvements due to time or repeated testing. More specifically, Farah et al., (2019) found significant improvements over time for both a group that received a shared book reading intervention and a control group who viewed videos of the same stories, while Strouse et al. (2013) report significant improvements over time for four groups of children who all viewed the same video stories with varying levels of additional scaffolding. These studies therefore do not provide clear evidence of improved general vocabulary scores following exposure to video.

Findings about the impact of video on narrative skills were less consistent. Linebarger & Piotrowski (2009) tested the effects of exposure to 40 episodes of television programmes in three different narrative styles on generalised measures of narrative skill and story knowledge. Narrative skills were assessed by asking children to retell a story from a wordless picture book and answer comprehension questions about it. Narrative production was assessed by asking children to arrange pictures in order and then tell a story using the pictures. These measures are therefore intended as indices of children’s knowledge of stories and narratives generally, rather than measures of their comprehension of specific video content. There is some evidence of improved narrative skills following viewing, with significant improvements from middle to post-test on explicit and implicit comprehension and story knowledge measures for children viewing traditional narratives. However, these results are complicated by the finding that the non-viewing group also improved significantly over time on a number of measures. Better performance was found on all measures for children who viewed either traditional or embedded narratives rather than expository narratives or no viewing, but no individual condition resulted in consistently better performance on all measures. It is therefore unclear from these results whether exposure to stories in video can improve children’s general narrative skills. The remaining studies that measured narrative outcomes are not designed to test for improvements in narrative production or understanding following exposure to video. Instead, these studies tested children’s recall or understanding of specific stories presented through video.

Do children learn better from video or other media types?

While the majority of experimental studies examined children’s responses to video media only (43, 69.4%), the remaining 19 studies (30.6%) compared outcomes following exposure to video and another media type such as books or radio. Most commonly, the same stories were presented in each format (e.g. by creating a storybook based on the video) to control for story effects (e.g. Neuman et al., 2021; Meringoff, 1980; and Terrell & Daniloff, 1996). In some cases, media type was manipulated as a within-subjects variable, usually by asking children to watch one story in video format and then listen to a second story in another media type (e.g. Neuman et al., 2017). In other cases, a between-subjects design was used in which some children watched a story in video format while others heard a story in a different medium (e.g. Meringoff, 1980). Both cases allow for an assessment of whether learning differs when stories are presented through video or other mediums.

Beginning with the studies that examined the effect of media type on targeted vocabulary, there were three studies that reported a main effect of media type. Firstly, Terrell and Daniloff (1996) asked children to either watch a short story in the form of an animated cartoon, listen to the same story while viewing a series of still images from the video on a computer screen, or listen to the same story being read to them by the researcher (again accompanied by the same still images from the cartoon). They found that children performed significantly better in the book-reading condition compared to the other two media conditions on targeted vocabulary measures including auditory-verbal recognition (identifying the target word that they heard in the story from a list of three words) and auditory-visual recognition (word-to-picture matching). No differences were reported between the video and still image conditions. The two other studies reporting a main effect of media type were both published in Neuman et al. (2021). While the first and second studies published in this paper find no significant differences in target word learning between a video condition and a live storybook reading condition, the second and third studies additionally report a media effect wherein groups who were exposed to the same medium twice (e.g. watching two videos or reading two books) had significantly lower scores on a receptive vocabulary measure compared to groups who were exposed to a story in each medium (e.g. a video followed by a book).

In contrast, six studies that compared vocabulary learning across different media types found no significant main effects or interactions involving different media conditions. These studies compared either pre-existing television clips (Goldstein et al., 2022; Neuman et al., 2021), short animated stories (Van Horn & Kan, 2016; Neuman et al., 2017), or videos of researchers reading books aloud (Gaudreau, 2021; Farah et al., 2019) to a range of alternate media types including live storybook reading (Neuman et al., 2021; Van Horn & Kan, 2016; Neuman et al., 2017; Gaudreau, 2021; Farah et al., 2019) or interactive play sessions (Goldstein et al., 2022). Most commonly researchers used targeted tests of vocabulary that had appeared in the stimuli, although Farah et al. (2019) instead report scores on a standardised test of vocabulary (the WPPSI). No significant differences between media conditions were reported for any of these tests. The media effects reported in this review therefore do not provide clear evidence that vocabulary learning is more or less successful in one medium over another, with only one study showing significantly worse performance in the video condition.

Media type was also manipulated by 13 studies that assessed narrative recall. These studies tended to compare the amount of story information that children were able to recall after exposure to stories in different media. Six of these studies found significant main effects of media type on narrative measures. In three cases these effects indicated a video advantage with children making fewer recall errors after watching a video vs listening to a radio clip (Hayes et al., 1986), and recalling more information when stories were presented as videos rather than audio clips (Gibbons et al., 1986) or storybooks read by the researcher (Beentjes & van der Voort, 1993). Two studies reported mixed results, with benefits for video for recall of actions (Meringoff, 1980) and story order and details (Beagles-Roos & Gat, 1983) but benefits for recall of figurative or expressive language from books (Meringoff, 1980) and radio (Beagles-Roos & Gat, 1983). Finally, Kendeou et al. (2008) report a significant effect of media type (books vs TV) on narrative recall, but do not specify the direction of the effect.

Seven studies reported no significant main effects or interactions involving media type for narrative recall outcomes. These studies compared results following exposure to video clips (either clips from TV shows or films (Neuman et al., 2021; Kim et al., 2008; Hayes & Kelly, 1985; Neuman, 1989; Neuman, 1992), animated storybooks (Neuman et al., 2017), or a video of the researcher reading a storybook (Gaudreau, 2021)) vs exposure to other media (either books read aloud by the researcher (Neuman et al., 2021; Neuman et al., 2017; Gaudreau, 2021; Neuman, 1989), audio clips of stories (Kim et al., 2008; Hayes & Kelly, 1985) or books read by the children themselves (Neuman, 1992)). None of these studies found significant condition differences for measures of recall of story content (Neuman et al., 2021, Neuman et al., 2017, Kim et al., 2008; Gaudreau, 2021; Hayes & Kelly, 1985; Neuman, 1989), story sequencing (Neuman et al., 2017) or inferencing (Gaudreau, 2021; Neuman, 1992). The findings relating media type to narrative measures are therefore more variable than those for vocabulary measures, with some evidence that children perform better on particular aspects of narrative recall depending on the media type. It is important here to bear in mind that it is very difficult to separate recall of linguistic and visual content in measures of narrative recall as children may be remembering visually presented story information. Therefore, even when comparing video to picture books, any observed differences in recall could be influenced by the differing amounts of visual information in each medium and aren’t necessarily due to better or worse processing of linguistic information. So, while some studies report ‘better recall’ of content presented via video, this doesn’t allow for strong conclusions about linguistic comprehension specifically.

What factors modulate children’s language learning from video?

Almost all experimental studies examine the factors that moderate learning by comparing outcomes between different participant groups, between manipulations of stimuli features, or between viewing environments. Figures 6 and 7 give a summary of the experimental factors included in different designs (along the y-axis) in different studies (along the x-axis), and whether these factors had a statistically significant effect on outcome measures (significant main effects on at least one outcome measure are indicated by a filled circle, interactions by an asterisk, empty circles indicate no reported effect). For example, Rice et al. (1994) find a main effect of amount of exposure to target words (with better performance for the group who had more exposures) but no main effect of target word class. Since there were a number of interactions involving target word class, this is marked on the graph with an asterisk. Figure 6 shows results for the studies that measured vocabulary while Figure 7 shows results for studies that measured narrative outcomes. Note that for each study, we only include those factors whose effect on the outcome was tested, and that an individual study may appear in both figures if both vocabulary and narrative measures were examined. Factors that were included as a covariate, but were not part of the research questions or hypotheses of a given study are not included in the figures. Additionally, secondary outcome measures which were used to predict language outcomes (e.g. visual attention while viewing) are not included in this analysis.

Figure 6

Figure 7

Studies with vocabulary outcomes

Beginning with the studies that measured vocabulary outcomes (Figure 6), we can see that it was most common for researchers to examine factors related to the video stimuli. Target word class (e.g. noun, verb, adjective, etc.) was manipulated by seven studies by embedding words of different classes within the scripts of short animated video clips (as in a paradigm developed by Rice & Woodsmall, 1988). While target word class does appear to impact learning in individual studies, the direction of reported effects in the literature is not consistent. For example, while two studies report an advantage for verbs (Perez, 1990; Rice et al., 1994), three studies report that performance on verbs was significantly lower than at least one other word class (Oetting et al. 1995; Terrell & Daniloff, 1996; Neuman et al., 2021). One study reports no effects of word class (Rice et al., 1992).

The effects of additional cues (e.g. text captions or pauses) or learning strategies (e.g. referencing prior knowledge) were investigated by five studies, again with mixed results. Captions were associated with better performance on a target vocabulary comprehension post-test (Linebarger et al., 2013) in one study, but another reports better scores in a low-SES group when text was absent (Linebarger et al., 2010). Adding a pause before target words had no effect on vocabulary learning (Rice et al., 1992). Other strategies such as adding attention-directing cues led to improved scores but not on all vocabulary outcomes measured (Neuman et al., 2019). It is not clear from this pattern of results that any specific cues or learning strategies are linked to consistently better outcomes.

Narrative content, structure, or style was manipulated by five studies in total. Again, no clear pattern of results emerges from these manipulations. One study embedded target vocabulary items in at least two different stories to check whether the learning of target words was dependent on them being embedded in a specific story or if learning would occur regardless of story context (Neuman et al., 2021; no effect of story). Another tested different target words embedded in four different stories that were presented either through video or live storybook reading to try and understand whether differences in learning related more to the story content or the medium of presentation (Neuman et al., 2017). There was a main effect of story suggesting that story content affected children’s ability to learn target words (although note that in this case, story was confounded with target word features as different words were embedded in each story). Both Neuman, Flynn, et al. (2020) and Linebarger and Piotrowski (2010) manipulated narrative style by comparing expositories (programmes that ‘used multiple vignettes and programme hosts to educate, inform, or describe a particular topic”, Linebarger & Piotrowski, 2010, p.1585) to narratives (programmes that “used a set of story events … to deliver prosocial or academic content”, Linebarger & Piotrowski, 2010, p.1585) and report opposite patterns of results.

Finally, three studies examined the impact of embedding participatory language or interactivity within the video stimuli themselves. This typically involved on-screen characters that would attempt to directly engage the audience by asking questions and by leaving pauses for responses. Of the three studies that looked at this factor, two report simple effects indicating better performance on targeted vocabulary measures when levels of interactivity were higher (Neuman, Flynn, et al., 2020; Nussenbaum & Amso, 2016). Finally, Strouse et al. (2013) compared a viewing condition in which an on-screen actress engaged in dialogic questioning techniques throughout the story to a series of viewing conditions where the clip was played without the on-screen actress (further condition differences are discussed in the ‘viewing environment’ section below). This dialogic actress condition was compared to a condition in which the video itself did not contain participatory language but parents were encouraged to use dialogic questioning techniques while their children were viewing. There was no significant difference between these conditions suggesting that children had a similar response to an interactive adult on video and live parent interaction. The paucity of evidence relating to stimuli interactivity again limits the conclusions that can be drawn.

Finally, two studies examined the effect of amount of exposure to target words. In both cases this involved comparing groups who saw versions of the same video with target words embedded a different number of times (Rice et al., 1994; Neuman, Samudra, et al., 2020); both of these studies report a main effect of repetition wherein more exposures lead to better learning.

The links between participant factors and vocabulary learning were examined by 16 studies in total, with more consistent results than were observed for stimuli factors. The most common participant variable was age, with four studies reporting that older children performed significantly better on at least one targeted vocabulary measure (Rice & Woodsmall, 1988; Hsueh et al., 2017; Samudra, Wong, et al., 2019; Goldstein et al., 2022; Nussenbaum & Amso, 2016). Neulight (2012) and Oetting et al. (1995) find no effect of age after controlling for pre-test scores suggesting that baseline knowledge of target vocabulary may be more important for learning than age. Similarly, five studies report better word learning abilities for children with higher initial vocabulary levels as tested with standardised measures (significant main effects: Samudra, Flynn et al., 2019; Samudra, Wong, et al., 2019; Neuman, Flynn, et al., 2020; significant simple effects within conditions: Nussenbaum & Amso, 2016; Neuman et al., 2019). Linebarger et al. (2010) was the only study to report no significant effect of reading ability on their target word comprehension measure.

Of the studies that included developmental diagnosis as a factor, all report better vocabulary learning in a typically developing control group than the group with a developmental diagnosis. Three studies compared children with and without a diagnosis of developmental language disorder (Rice et al., 1992; Rice et al., 1994; Oetting et al., 1995). In the fourth study, children with autism were shown to have lower initial performance on a test of emotion vocabulary than peers, but after viewing a programme that contained target emotion vocabulary, both groups performed similarly (Gev et al., 2017). Effects of age, language ability, and developmental diagnosis are therefore fairly consistent and in the expected direction.

The remaining participant factors (gender, L1 vs L2, geographical area, and SES) were only examined in a few studies. Two studies examined the impact of gender on targeted vocabulary measures with inconsistent results (Neulight, 2012; Hsueh et al., 2017). One study examined language background and found larger gain scores when children were tested in their second language (Van Horn & Kan, 2016). One study reported an effect of geographical area wherein children from an urban area improved more from pre- to post-test on a targeted measure of vocabulary than children from a rural area (Hsueh et al., 2017 – although note that geographical area may be confounded with SES here). One study reported an SES interaction whereby SES moderated the effect of on-screen print on vocabulary with the on-screen print leading to significantly worse performance for children in the lower SES group only (Linebarger et al., 2013).

The final category of experimental factors relates to the context in which children were exposed to video stimuli. For vocabulary outcome measures, this commonly involved a manipulation of the interactivity of the viewing environment (e.g. co-viewing with a researcher or parent) leading to inconsistent results. Three studies found no effect of interactivity on targeted vocabulary measures (Goldstein et al., 2022; Gaudreau, 2021; Gev et al., 2017) and three report significant interactions involving interactivity. In an examination of the effect of co-viewing, Samudra, Flynn, et al. (2019) found that only children with lower initial PPVT scores benefited from co-viewing an educational programme with the researcher whereas children with higher PPVT scores performed equally well with or without a co-viewer. Similarly, Neuman, Samudra, et al. (2020) found that co-viewing with a researcher only benefitted learning of expressive vocabulary in a low repetition condition in which children heard each target word only 3–4 times, but had no impact on learning when target words were repeated 8–11 times. Finally, Strouse et al. (2013) report that children performed better on a test of target vocabulary in a dialogic questioning condition where parents were instructed to use dialogic questioning techniques (e.g. prompting their children to talk about what was happening in the story), compared to a condition in which parents were instructed to direct their child’s attention to the screen without asking questions, or a condition in which parents were not instructed to interact with their child at all during viewing. These findings suggest that co-viewing and interactivity of the viewing context are not universally beneficial for target vocabulary learning.

Time since exposure was manipulated by six studies. In all cases this involved a comparison of an immediate post-test to a delayed retention test. In three studies, there was a significant main effect of testing time, with scores improving from immediate to delayed test on measures of target emotion vocabulary (Gev et al., 2017, three month delay) and video-specific vocabulary (Rice et al., 1994, one to three day delay, significant improvement only seen for age-matched typically developing control children; Samudra, Wong, et al., 2019, one week delay, significant improvement for expressive targeted vocabulary measure only). It is possible that these effects are due to a simple effect of time or repeat testing since the same items were used across all time points so children would be encountering the test words for the third time in the delayed test. In the final two studies, there was no main effect or interaction involving test time, indicating that scores did not differ significantly from post-test to delayed test (Strouse et al., 2013, two week delay; Goldstein et al., 2022, two week delay). No studies reported a significant drop in performance at delayed test, suggesting that children did not forget target words. This may be due to the relatively short amount of time between immediate and delayed test or intensive exposure to stimuli (10 minutes per day over eight weeks in the case of Gev et al., 2017).

Finally, four studies manipulated the amount of video that children are exposed to by comparing groups who watched different numbers of repetitions of video stimuli. Although the details of this manipulation differed across studies, there was a fairly consistent finding that children performed better when they had more exposure to video stimuli (e.g. compared to a group who had fewer repetitions as in Samudra, Wong, et al., 2019).

Studies with narrative outcomes

Figure 7 shows the experimental factors in studies that measured narrative outcomes. As with studies of vocabulary, the majority of these studies manipulated something about the video stimuli. Twenty-one studies manipulated something about the narrative content (e.g. testing recall for different stories), structure (e.g. comparing recall for events that differed in number of causal connections or hierarchical structure within the narrative), or style (e.g. comparing recall of entertainment programmes vs educational programmes). In all cases, narrative manipulations resulted in a significant main effect or interaction. While the details of the effects vary between studies, a number of generalisations can be made. Firstly, there is a consistent finding that children are better at recalling information that is central to the plot compared to peripheral information (Friedlander et al., 1974; Calvert et al., 1982; Lorch et al., 1987; van den Broek et al., 1996; Lorch et al., 1999; Kim et al., 2008; Lorch et al., 2010). Secondly, children appear to be sensitive to the style of the narrative, with better recall for videos that were child-directed and entertaining rather than educational (Campbell et al., 1987; Lorch et al., 2000). Differences in recall were also reported when comparing narratives vs expositories and when comparing different types of narratives such as traditional or embedded narratives (Linebarger & Piotrowski, 2009; Linebarger & Piotrowski, 2010), however, the direction of these effects was not consistent and varied for different types of outcome measure. Effects of narrative content, structure, or style were frequently moderated by participant factors such as age or developmental diagnosis, with older and typically developing children being more sensitive to narrative manipulations (Calvert et al., 1982; Lorch et al., 1987; Lorch et al., 2010).

A number of additional cues or learning strategies were found to have a positive impact on specific narrative outcomes. For example, Linebarger et al. (2010) find a positive effect of on-screen print for literal but not inferential comprehension, while Calvert et al. (1982) found that the salience of visual cues positively impacted recall of central, but not incidental, story content. Findings relating to participatory language or interactivity within clips were inconsistent, with one study finding no difference between a co-viewing condition and a participatory language condition (Strouse et al., 2013), while two others found improvements linked to participatory language only for subgroups of their samples (Calvert et al., 2007; Piotrowski, 2010). The small number of studies that explore the effects cues, learning strategies, and participatory language on comprehension, as well as inconsistencies between these studies, make it hard to draw strong conclusions about the benefits of these stimuli characteristics.

The effects of participant variables on narrative outcomes generally mirrored results for vocabulary. All significant effects of age indicate that older children perform better than younger children (although note that Friedlander et al. (1974) report no difference between four- and five-year-olds). Where there were significant effects of language or reading ability, these indicated that children with higher initial vocabulary or story knowledge performed better overall on measures of recall (Kendeou et al., 2008; Samudra et al., 2020; Piotrowski, 2010) and inferencing (Mares, 2006). Although, note that three studies reported no effects of vocabulary level (Kendeou et al., 2008; Neuman, 1992) or reading risk status (Linebarger et al., 2010) on inferencing skills. All studies examining effects of developmental diagnosis compared children with ADHD to peers without ADHD. While four studies report overall better recall performance in the comparison group, two find that there are group differences only in certain conditions (e.g. when distracting toys are present, Lorch et al., 2000), and one reports equal performance between groups on a measure of free recall (Landau et al., 1992). Only one study examined gender and ethnicity as part of the main research questions (Calvert et al., 2007). They report that Caucasian children did better than Hispanic children on central content recall, but that Hispanic girls benefited more from viewing interactive media than other children.

A common manipulation of context involved comparing narrative scores when viewing a video in the presence or absence of distracting toys. Significant main effects indicate better performance by children who viewed in the absence of toys (e.g. Lorch et al., 2000), while significant interactions generally indicate that the presence of toys increased differences in performance between children with and without ADHD (e.g. Bailey et al., 2009). The effects of amount of exposure to video were inconsistent: positive effects of video repetition were found for story comprehension (Mares, 2006, study 2) and close inferences (Mares, 2006, study 1) but had no reported effect for far inferences (Mares, 2006, study 1) and Linebarger and Piotrowski (2009) report improved performance with more exposure on their implicit comprehension measure only and only for children viewing narratives that followed a traditional narrative structure.

Similar to the results for vocabulary, the effects of interactivity of the viewing context were inconsistent for narrative outcomes. Strouse et al. (2013) report simple effects showing that children whose parents engaged in dialogic questioning while viewing had higher comprehension scores, while Gaudreau (2021) finds no difference in comprehension between a live storybook reading group and a group who watched a pre-recorded video, and both Calvert et al. (2007) and Samudra et al. (2020) report significant effects of co-viewing only for children certain characteristics such as high verbal ability or high attention.

Few studies explored the effects of time since exposure (Beentjes & van der Voort, 1993; Friedlander et al., 1974) or task instructions (Piotrowski, 2010; Field & Anderson, 1985) and none reported significant main effects on narrative measures, making it difficult to draw conclusions about the impact of these variables.

Vocabulary

As with experimental studies, vocabulary was the most commonly assessed outcome measure in observational studies (18 studies, 58.1%). These studies used a wide range of standardised vocabulary measures (e.g. the Peabody Picture Vocabulary Test (PPVT)), or vocabulary subtests from broader measures (e.g. the Wechsler Preschool and Primary Scale of Intelligence (WPPSI)). There were no positive associations reported between measures of overall video screen time and generalised vocabulary outcomes. Studies which categorised participants according to their television viewing profile reported either that children in high exposure categories had lower scores on a particular vocabulary test (Plancoulaine et al., 2017; Kim & Chung, 2021; Stockdale et al., 2022) or that there was no significant difference between the different exposure groups (Plancoulaine et al., 2017; Barr et al., 2010; Bittman et al., 2011; Kim & Chung, 2021; Bosma & Blom, 2020; Stockdale et al., 2022). Of the studies that measured overall screen time as a continuous variable, two reported negative associations with generalised vocabulary (Pagani et al., 2013; Carson et al., 2017), while six reported no significant association after controlling for demographic factors (Zhang, Adamo, et al., 2022; Kühhirt & Klein, 2020; Hu et al., 2020; Blankson et al., 2015; Schmidt et al., 2009; Allen et al, 1992). In studies where some or all of the viewing time data was collected before the age of three, one reports a negative relationship between viewing and vocabulary (Pagani et al., 2013), four find no effect (Schmidt et al., 2009; Barr et al., 2010; Bittman et al., 2011; Kühhirt & Klein, 2020), and three find negative effects on some vocabulary measures and no effects on others (Plancoulaine et al., 2017; Kim & Chung, 2021; Stockdale et al., 2022). None of these studies reported positive associations between viewing before the age of three and vocabulary.

Where studies examined subcategories of viewing (e.g. viewing particular genres), results were more inconsistent. For example, a number of studies reported mixed results with positive associations between generalised vocabulary and viewing educational programmes (Wright et al., 2001) and positive effects of co-viewing (Bittman et al., 2011), alongside negative effects for cartoons (Wright et al., 2001) and background television (Bittman et al., 2011). Farangi and Mehrpour (2022a) report that background television exposure is negatively associated with vocabulary for children from high SES backgrounds but that the opposite pattern is observed for children from low SES backgrounds. Rice et al. (1990) report that viewing Sesame Street was significantly associated with increased PPVT score but that there was no relationship between vocabulary and viewing non-informative programmes. Finally, Barr et al. (2010) report a negative association between exposure to adult-directed television and vocabulary but no effect of child-directed television.

In summary, where measures of total video screen time were used, these had null or negative associations with vocabulary level. However, results were more variable when looking at specific aspects of video viewing, with some activities (e.g. viewing educational media or co-viewing with a parent) associating with higher vocabulary scores.

General language ability

Nine observational studies (29%) opted to use measures which focus on the general language level of a child. This was almost always assessed using standardised language screening tools such as the Test of Early Language Development (TELD, or the verbal IQ composite from the Wechsler Preschool and Primary Scale of Intelligence (WPPSI-III). Other general language measures included expressive and receptive language questions developed for the Kids in Taiwan longitudinal study (Yu et al., 2022), and measures of the quantity of verbal utterances during a language sample (Farangi & Mehrpour, 2022b).

Of studies that measured total screen time, one reported a positive association between language ability and overall television viewing measured both at age three and at age five (Law et al., 2018). Across five studies, there were seven negative associations reported between overall video exposure and measures of general language abilities including two that tested associations with receptive and expressive language separately (Plancoulaine et al., 2017; McKean et al., 2015; expressive language only: Yu et al., 2022; separate expressive and receptive language measures: Ozyurt & Dinsever, 2017; Hill et al., 2020). Three studies report a null association between overall video exposure and general language measures (Zhang, Weibe, et al., 2022; Arraf, 1990; receptive language only: Yu et al., 2022).

Results for studies that measured subcategories of viewing were again mixed. Farangi and Mehrpour (2022b) divided children according to their family SES and their exposure to background television and found that general language scores were best in the high SES, high background television group, but only for a subset of their measures. Arraf (1990) reported mixed effects of different genres with general language scores increasing with exposure to educational television and decreasing with exposure to commercial television. Finally, Zhang, Weibe, et al. (2022) used a binary variable that captured whether or not children were meeting screen-time recommendations and found better performance in children who did meet the recommendations.

Results for general language measures are therefore more inconsistent than for vocabulary outcomes. While overall screen time is most commonly negatively associated with general language measures, results are more varied for subcategories of viewing and there are some reported positive associations between viewing and language.

Syntax or morphology

Six observational studies (19.4%) used measures relating to syntactic or morphological knowledge. In four cases this involved collecting samples of children’s spoken language and analysing them for syntactic complexity (Selnow, 1978; Selnow & Bettinghaus, 1982) or diversity/number of particular syntactic categories such as path verbs (e.g. ‘come’ or ‘go’, Wu et al. 2022) or interrogatives (e.g. questions and question words such as ‘what’, Li et al., 2017). The only other syntactic measure used was a Sentence Repetition Test in which children are asked to repeat sentences of increasing grammatical complexity (i.e. longer sentences that contained more clauses, Oryadi-Zanjani et al., 2021). Bosma & Blom (2020) measured morphological development (the only study to do so) using a test in which children had to produce plural and past participle forms of test items. Of the studies that measured overall video exposure, one reported a positive effect of viewing on the number of interrogative forms used (Li et al., 2017), one reported a negative association between viewing and a sentence repetition test (Oryadi-Zanjani,et al., 2021), and four report null associations with syntactic or morphological measures (syntactic complexity: Selnow, 1978; Selnow & Bettinghaus, 1982; diversity of path verb types: Wu et al., 2022; morphology: Bosma & Blom, 2020). Both Selnow (1978) and Selnow and Bettinhaus (1982) also broke down viewing time into five genre categories and reported non-significant associations between viewing and syntax for all categories (note that this excludes cartoon viewing time in Selnow and Bettinghaus (1982) where a significant association is reported, but this is inconsistently presented as positive and negative in the text). Finally, Wu et al. (2022) report a positive association between parents’ viewing time with children and the quantity of path verb types that children produce. Once again, these findings are inconsistent and do not reveal a clear picture of an association between video exposure and syntax or morphology.

Conceptual knowledge

Wu et al. (2022) measured the semantic density of utterances produced by children during the study (i.e. the amount of semantic information contained within the verb phrases) and found no association with total video exposure.