Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 May 1.
Published in final edited form as: Infant Behav Dev. 2021 Apr 10;63:101560. doi: 10.1016/j.infbeh.2021.101560

Auditory Joint Attention Skills: Development and Diagnostic Differences During Infancy

Lauren B Adamson 1, Katharine Suma 1, Roger Bakeman 1, Ashleigh Kellerman 2, Diana L Robins 3
PMCID: PMC8172433  NIHMSID: NIHMS1692826  PMID: 33848771

Abstract

To date, joint attention skill assessments have focused on children’s responses to multimodal bids (RJA) and their initiation of bids (IJA) to multimodal spectacles. Here we gain a systematic view of auditory joint attention skills using a novel assessment that measures both auditory and multimodal RJA and IJA. In Study 1, 47 typically developing (TD) children were tested 5 times from 12 to 30 months to document auditory joint attention skill development. In Study 2, 113 toddlers (39 TD, 33 autism spectrum disorder [ASD], and 41 non-ASD developmental disorders [DD]; average age 22.4 months) were tested to discern the effects of ASD. Our findings fit well within the established depiction of joint attention skills with one important caveat: auditory items were far more difficult to execute than multimodal ones. By 24 months, TD children passed multimodal RJA items at the near-ceiling level, an accomplishment not reached even by 30 months for auditory RJA items. Intentional communicative IJA bids also emerged more slowly to auditory spectacles than to multimodal spectacles. Toddlers with DD outperformed toddlers with ASD on multimodal RJA items but toddlers in both groups rarely passed any auditory RJA items. Toddlers with ASD often monitored their partner’s attention during IJA items, albeit less often than toddlers with DD and TD toddlers, but they essentially never produced higher-level IJA bids, regardless of modality. Future studies should investigate further how variations in bids and targets affect auditory joint attention skills and probe the relation between these skills and language development.

Keywords: joint attention, infants, auditory, multimodal, skill development, autism spectrum disorder

The development of joint attention during infancy is a pivotal accomplishment that positions children to explore their surroundings while caregivers guide their engagement with objects and language (Adamson & Bakeman, 1991; Bruner, 1983). Using the skill of responding to joint attention (RJA), infants follow their partners’ bids towards new targets; using the skill of initiating joint attention (IJA), they attempt to direct their partners’ attention to something that already interests them (Mundy et al., 2007).

A vast literature now details the typical developmental trajectory of RJA and IJA (e.g., Mundy et al., 2007), their neurological underpinnings (Mundy, 2016), the role these skills play in early language acquisition (Adamson & Dimitrova, 2016), the impact of autism spectrum disorder (ASD) on these skills (Bottema-Beutel, 2016), and joint attention interventions for children with developmental challenges, including ASD (Murza et al., 2016). There are now also several assessments that measure joint attention skills (e.g., the Early Social Communication Scales [ESCS], Mundy et al., 2003; the Attention-Following and Initiating Joint Attention Protocol [JA Protocol], Watson et al., 2003) as well as several broader tests that generate joint attention scores (e.g., Communication and Symbolic Behavior Scales [CSBS], Wetherby & Prizant, 2002; Autism Diagnostic Observation Schedule [ADOS-2], Lord et al., 2012).

Given the breadth and rigor of existing research on early joint attention skills, it is surprising that auditory aspects of joint attention have rarely been considered. Instead, beginning with Butterworth’s seminal studies on visual joint attention (Butterworth, 1997; see also Scaife & Bruner, 1975), experimental work has focused primarily on how infants use gaze shifts and points to respond and initiate joint attention bids. Assessments of joint attention followed suit by emphasizing visible communicative acts towards visible targets. When auditory signals were also provided, they were used as adjuncts to draw attention to intended target (“Jonas, look [gaze towards and point]”) or as part of a visible object that an infant might share (e.g., a noisy toy robot).

Yet, there are several reasons to broaden the consideration of joint attention skills to explicitly encompass sounds. Infants, even before birth, are immersed in an auditory world; soon after birth, they orient towards sounds (Clifton, 1992). By the end of their first year, auditory attention skills of arousal, orientation, selective attention, and sustained attention are well established (Gomes et al., 2000). Moreover, by 24 months of age typically developing (TD) toddlers sustain periods of auditory joint engagement—the sharing of sounds with a partner—during interactions with their parents (Adamson et al., 2019a). Research on early social cognition also indicates that an understanding of sound as a shared topic begins to emerge during the second year as children take into consideration what a partner has previously heard (Moll et al., 2014) and start to understand that sounds may reveal concealed events (Melis et al., 2010) or disturb a sleeping baby (Williamson et al., 2015). In contrast, toddlers with ASD often display aberrant reactions to sounds that are not associated with peripheral hearing loss (Beers et al., 2014), and they are far less likely to be observed in periods of auditory joint engagement (Adamson et al., 2021).

In the research reported here, we introduce a new assessment, the Joint Attention Skills Assessment: Auditory and Multimodal (JASA; Adamson et al., 2016) that measures children’s reaction to and production of auditory and multimodal joint attention bids. The JASA retains the format of earlier assessments where a friendly yet unfamiliar examiner presents the child with a series of items, some measuring RJA and some IJA. However, in addition to assessing multimodal RJA using items where the examiner says the child’s name and then gazes and points to a target, auditory RJA was also assessed using items where the examiner calls the child’s name and then uses a familiar word that is likely to be in the child’s vocabulary (e.g., “dog” or “ball”) to label a target without shifting gaze or pointing. Similarly, in addition to multimodal IJA items where the child is presented with an interesting visible, tangible, noisy toy, auditory IJA items were included where the interesting spectacle was a sound without a visible source. The JASA generates scores for RJA and IJA that decompose into multimodal and auditory sub-scores, that we used here to document developmental and diagnostic differences of auditory and multimodal joint attention skills during infancy.

1. Study 1: The Development of Auditory Joint Attention Skills

The aim of Study 1 was to document early trajectories of auditory and multimodal joint attention skills in TD children. To do this, we assessed joint attention skills longitudinally from 12 to 30 months of age, a period that typically begins with the emergence of multimodal joint attention skills and ends with their consolidation.

To assess multimodal RJA, the examiner attempted to direct a child’s attention to a target that is a familiar object or a picture of such an object by calling the child’s name and then gazing and pointing towards the target without naming it. In contrast, auditory RJA bids followed the child’s name with a label that is likely to be in the child’s vocabulary for the target (e.g., “dog”; “ball”), but without the visible deictic acts of gazing towards the target and pointing towards it. Consistent with previous reports (Beuker et al., 2013; Morales et al., 2000; Mundy et al., 2007), we anticipated that multimodal RJA bids would be increasingly successful during the second year and thereafter they would almost always be followed (Hypothesis 1). In contrast, because language is integral to auditory RJA bids, we expected that auditory RJA would be more difficult than multimodal RJA, emerging more slowly and not occurring in almost all items until 30 months (Hypothesis 2). However, because we selected RJA targets whose labels would likely be understood even at 12 and 15 months, we expected that early language comprehension would correlate weakly if at all with multimodal and auditory RJA scores (Hypothesis 3). This finding would support the argument that developmental changes in auditory RJA reflect not only comprehension of specific words but, more interestingly, also the development of the understanding that speech can provide crucial information about a partner’s interests and intentions (Nelson, 1996).

To document IJA, we assessed children’s bids to a partner to share an interesting spectacle using acts such as coordinated gaze shifts between the spectacle and partner and intentional communicative behaviors such as points, shows, and speech. We generated a composite IJA score that considered any type of bid to the partner. In addition, following the lead of Mundy and colleagues (2007), we generated two component scores—lower-level IJA bids that required only gaze shifts and higher-level IJA bids that included communicative gestures and speech— that may reflect different underlying processes and be associated with different developmental outcomes (Pickard & Ingersoll, 2015).

In line with previous studies, we anticipated that composite IJA—which included both lowerand higher-level bids—would occur often in both modalities even at 12 months of age and that there would be little change in performance over time (Hypothesis 4) as children readily display at least lower-level IJA. In contrast, we anticipated that higher-level IJA, which involves more complex communication skills and social motivation, would only gradually emerge from 12 to 30 months (Hypothesis 5a), so that by 18 months onward almost all children would pass at least one of the six IJA items (either multimodal or auditory) at the higher level (Hypothesis 5b). However, we also anticipated that even older children might not bid to share every interesting spectacle, especially auditory ones that lack the tangible presence of multimodal objects, and that even at older ages, children would be more likely to pass multimodal than auditory items at the higher level (Hypothesis 5c).

2. Method

2.1. Participants

This study included 47 TD children (51% boys) recruited from the greater Atlanta metropolitan area in 2014 through 2017 using strategies such as posts in online community forums seeking volunteers for an early communication development study. All children were full-term and without significant medical complications at birth or developmental problems since birth. Because we were interested in the child’s reaction to auditory stimuli, some of which were English words, we required parental confirmation that the child had significant daily exposure to English, had passed their most recent hearing screening and did not have an ear infection. As detailed in Table 1, the sample was moderately racially and ethnically diverse with mothers who were generally well-educated and relatively mature (mean age at the first visit was 34.8 years, range 26.7–47.6).

Table 1.

Characteristics of Children and Their Mothers

Characteristic Category Study 2
Study 1 ASD DD TD
Child sex Male 24 26 22 29
Female 23 7 19 10
Child race/ethnicity White non-Hispanic 32 10 13 8
Black non-Hispanic 5 14 26 20
Asian non-Hispanic — — 1 1
Multi-racial non-Hispanic 4 3 1 5
Hispanic White 5 2 — 5
Hispanic Black — 3 — —
Hispanic multi-racial 1 1 — —
Mother race/ethnicity White non-Hispanic 40 11 14 13
Black non-Hispanic 4 16 26 21
Asian non-Hispanic — — 1 2
Multi-racial non-Hispanic 1 2 — —
Hispanic White 2 2 — 3
Hispanic Black — 2 — —
Mother employment Full-time 17 13 13 24
Part-time 10 7 9 8
Not employed 20 13 19 7
Mother education Less than high school — 1 5 —
Only high school — 5 12 4
Some college 1 13 4 13
4-year degree 20 9 10 10
Some post college 26 5 10 12
Graduate degree — 1 5 —
Household income $30K–$50K 5 11 19 13
$50Kȁ$100k 17 8 7 12
$100K or more 24 9 9 14
None given 1 5 6 —
Open in a new tab

Note. Scores are numbers of children. ASD = autism spectrum disorder, DD = non-ASD developmental disorders, TD = typically developing. N = 47 for Study 1 and 33, 41, and 39 for ASD, DD, and TD groups, respectively, for Study 2.

Observations were scheduled to occur when children were aged approximately 12, 15, 18, 24, and 30 months; 40 children completed all five visits and 7 children completed four (one missed the 12-month, three the 24-month, and another three the 30-month visit). Mean ages at the five visits were 12.6, 15.2, 18.3, 24.2, and 30.1 months (ranges = 12.0–13.9, 14.7–15.8, 17.7–19.6, 23.5–25.1, and 29.2–31.4). Not included in the final sample were six children who missed two visits, four children who missed three, and one child whose parents and pediatrician were concerned about his lack of expressive language at 30 months.

2.2. Procedure

The research reported here was approved by Georgia State University’s Institutional Review Board, and parents provided written informed consent before the JASA assessment was done. It took place during the second hour of a two-hour long laboratory visit that began with observations of parent-child interactions (Adamson et al., 2019a; 2021). After these observations, the child and parent moved to a 3.0 m by 4.6 m playroom arranged for the JASA (Figure 1). Similar to other joint attention skills assessments, a trained examiner and the child faced each other across a small table, the parent was seated behind the child, and several posters and objects were visible on the walls or bookcase shelves. Seven members of our research staff (all female, 6 White and 1 Asian, 21 to 33 years old) participated in data collection. Video recording was done from behind a one-way mirror.

Figure 1:

Figure 1:

Open in a new tab

The JASA Assessment Room

2.3. The Auditory and Multimodal Joint Attention Skills Assessment (JASA)

The JASA (Adamson et al., 2016) is a structured assessment designed to measure RJA and IJA in multimodal (M) and auditory (A) contexts. The examiner presented the child with 12 items—three each that probe multimodal RJA, auditory RJA, multimodal IJA, and auditory IJA—ordered randomly with the constraint that RJA and IJA items alternate. Each item consisted of two presentations of the probe separated by a 5 sec pause. The JASA took about 15 min to complete.

There were six RJA targets, three objects (toy baby doll, large plastic bottle, beach ball) and three pictures (two dogs, car, banana). The targets’ labels were among the first words infants understand (see, e.g., http://wordbank.stanford.edu, which indicates that most participants at 12 months and almost all at 18 months could understand them). Two targets (an object and a picture) were located to the right, two behind, and two to the left of the child (see Figure 1). Prior to each JASA administration, one target at each location was randomly designated for a multimodal RJA item and one for an auditory RJA item with the constraint that at least one of the three items in each modality was a picture and one was an object.

During each RJA item, the examiner waited until the child was attending to a toy she placed on the table and then she called the child’s name. In the multimodal items, she then turned, looked towards the target, and pointed towards the target with an extended arm while saying the child’s name two additional times. In the auditory presentations, the examiner continued to look at the child while she labeled the target three times in quick succession (“Janelle, a dog; Janelle, there’s a dog; Janelle, <gasp>, a dog!”), placing emphasis on the object’s label and increasing her excitement across the three statements. The probes in both the multimodal and the auditory RJA presentations lasted about 10 sec and were followed by a 5 sec pause.

The IJA items presented the child with an interesting spectacle in order to observe if the child would attempt to share it with the examiner (or the parent). The multimodal spectacles were novel, visible, noisy toys (hopping bunny, walking robot, turtle pull toy) that moved across the table. Each IJA multimodal item began with the examiner clearing the table and then launching the novel toy so that it moved around on it. The auditory spectacles were sounds produced without a visible source (knocks on the door accompanied by someone asking “Is anyone in there?”, phone ringing, electronic buzzer that emitted a series of tones). Each IJA auditory items was initiated when the child was focused on an activity, such as playing with toys on the table or playing a tickle game with the examiner. Each IJA presentation lasted approximately 10 sec and was followed by a 5 sec period of silence.

2.3.1. Coding and Scoring the JASA

The examiner coded the JASA during the session and then, as needed, viewed the video to refine the coding after administration was completed. Each item was scored using the child’s best performance out of the two presentations. Each RJA item was coded yes if the child attempted to follow the bid, no otherwise. Each IJA item was coded yes higher level if the child made a bid using words, gestures, or both directed to the partner, but if not then yes lower level if the child made only a gaze shift between target and partner. Criteria for coding are detailed in Figure 2. Additionally, the administrative fidelity of each item’s two presentations was coded yes or no.

Figure 2:

Figure 2:

Open in a new tab

Coding JASA Items

We computed four summary scores for multimodal items and four for auditory items; scores could vary from 0 to 3. For RJA items, the score was the number of items passed. For IJA items, the scores were the number passed with either a lower- or a higher-level response (composite IJA), the number passed with a higher-level response (higher-level IJA), and the number passed with a lower-level response (lower-level IJA). Because a code of no for lower-level IJA score could occur when the infant made either no bid or a higher-level response, we also computed an exclusively lower-level summary score—the number of items passed with a lower-level response but with no item coded higher level—to be used instead of the lower-level IJA score for analysis.

2.3.2. Observer Reliability

To provide feedback throughout coding and guard against observer drift, we double coded approximately every fifth session (21% of the combined corpus for this study and Study 2 reported below) randomly determined with the constraint that all ages in Study 1 and all diagnoses in Study 2 be about equally represented. We assessed reliability for the combined corpus because coding for both studies proceeded simultaneously. Examiners did not know which sessions were to be double coded and whether the child was participating in Study 1 or Study 2. A master coder scored the selected sessions and discussed discrepancies with the examiner. To assess observer reliability, master coder and examiner scores for the variables analyzed here were compared for the 72 double-coded sessions (52 from Study 1, 20 from Study 2). Intraclass correlation coefficients (ICCs—two-way mixed model, absolute agreement, single measures; see Cicchetti, 1994), weighted kappas (counting one-point disagreements as agreements), and estimated observed accuracies (see Bakeman, 2018; Gardner, 1995) are given in Table 2; all indicate acceptable reliability.

Table 2.

Observer Reliability

Variable Modality ICC Weighted κ Estimated % accuracy
RJA Multimodal .81 .87 95
Auditory .66 .62 85
Composite IJA Multimodal .73 1.00 99
Auditory .70 .83 97
Higher-level IJA Multimodal .86 .90 96
Auditory .88 .95 98
Lower-level IJA Multimodal .81 .86 94
Auditory .71 .77 92
Open in a new tab

Note. N = 72 double-coded sessions. ICC = intraclass correlation coefficient, RJA = responds to joint attention, IJA = initiates joint attention. For details on weighted kappa and estimated percentage accuracy, see text.

2.4. The MacArthur Communicative Development Inventory: Words and Gestures

Children’s comprehension of words was assessed using The MacArthur Communicative Development Inventory: Words and Gestures (CDI; Fenson et al., 2007), a parent-report vocabulary checklist appropriate for typically developing children aged 8 to 18 months. Before the 12- and 15-month visits, parents indicated which of the 386 words their child understood.

2.5. Data Analysis

Primary data analysis for Study 1 considered the mean number of items passed for four dependent variables—RJA, composite IJA, higher-level IJA, and exclusively lower-level IJA—separately for age (five visits) and modality (multimodal and auditory).

For age, we decomposed effects over the five visits into linear, quadratic, cubic, and quartic trends. Linear trends indicate steady increases over visits and quadratic trends indicate either acceleration (a U curve) or deceleration (an inverted U). Effect sizes for more than one trend can be noteworthy and statistically significant. We assessed effect sizes with generalized eta squared (η2G) as is appropriate for effects that include repeated measures (Bakeman, 2005; Olejnik & Algina, 2003). Cohen’s thresholds for small, medium, and large effects are .01, .06, and .14 (Cohen, 1988). We determined statistical significance with repeated measures analyses of variance, two for each dependent variable (separate analyses for each modality). As noted earlier, seven children were missing one visit. For these repeated-measures analyses, we substituted sample means for the missing scores; as Widaman (2006) noted, this is straightforward and leaves sample means of the non-missing values unchanged.

For modality, we assessed effect sizes with dz, Cohen’s d for repeated measures. Cohen’s thresholds for small, medium, and large effects are 0.20, 0.50, and 0.80 (Cohen, 1988). We determined statistical significance with repeated-measures t tests performed separately for each visit.

In addition to our primary analyses, two hypotheses concerned the number of children who passed a specific number of items. These hypotheses were tested with binomial z scores computed for each age.

3. Results

3.1. Trajectories for Responding to Joint Attention

Trajectories for the development from 12 to 30 months of age for multimodal and auditory RJA and IJA are shown in Figure 3 and results for the trend analyses of these trajectories are given in Table 3. As expected and as seen in Figure 3, multimodal RJA increased linearly from 12 to 24 months before leveling off (Hypothesis 1; linear η2G = .45, inverted U quadratic η2G = .046).

Figure 3: Trajectories for Mean Number of Items Passed (Study 1).

Figure 3:

Open in a new tab

Note. RJA = responds to joint attention, IJA = initiates joint attention. The RJA bars represent the mean number of items coded RJA. The full extent of the bars in the middle two series represents the mean number coded either lower- or higher-level IJA (composite IJA); their bottom portion represents the mean number coded higher-level IJA. The full extent of the bars in the two series on the right represents the mean number of items coded lower-level IJA; the bottom portion represents the mean number of items coded exclusively lower-level IJA. See text for details. N = 46, 47, 47, 43, and 44 for the 12-, 15-, 18-, 24-, and 30-month visit, respectively. Error bars are 95% CIs, for simplicity not shown for lower-level IJA but only for exclusively lower-level IJA.

Table 3.

Trend Analysis Statistics and Percentage of Children who Passed at Least One Item Each Visit (Study 1)

Variable Modality ANOVA statistics
 % of children who passed at least
one item at each age specified
Linear trend Quadratic trend
η2G p η2G p 12 mo 15 mo 18 mo 24 mo 30 mo
RJA Multimodal .45 <.001 .046 .006 87 96 98 100 100
Auditory .52 <.001 .068 .011 46 64 66 84 95
Composite IJA Multimodal .035 .023 ~0 .88 100 98 100 100 100
Auditory .004 .44 .002 .54 100 100 98 100 100
 Higher-level IJA Multimodal .44 <.001 .024 .14 30 57 81 77 93
Auditory .36 <.001 .049 .018 17 32 45 61 75
 Lower-level IJAa Multimodal .38 <.001 .064 .01 70 40 19 23 7
Auditory .42 <.001 .001 .80 83 68 53 39 25
Open in a new tab

Note. RJA = responds to joint attention, IJA = initiates joint attention. N = 47. Generalized eta squared (η2G) and p values are from a repeated measures analysis of variance with infant age (12, 15, 18, 24, and 30 months) as the repeated measure. No cubic or quartic trends were statistically significant (median η2G = .005 and .010, range = .002–.010 and ~0–.045, and median p value = .50 and .27, range = .092–.69 and .056–.84, respectively).

a

Exclusively lower-level IJA

In contrast, we had expected that auditory RJA would emerge more slowly than multimodal RJA but occur in almost all items at 30 months (Hypothesis 2). As seen in Figure 3, like multimodal RJA, it increased linearly from 18 to 30 months (linear η2G = .52) and even accelerated somewhat towards 30 months (U quadratic η2G = .068). As hypothesized, it did emerge more slowly than multimodal RJA. The multimodal mean was greater than the auditory RJA mean, strongly so, at each age except 30 months: dz = 0.93, 1.19, 1.08, 1.01, and 0.48 at months 12 through 30, respectively, with p < .001 for all except = .003 at 30 months (as determined by repeated-measures t tests at each visit). Data related to the performance of individual children provides additional support for Hypothesis 1 (see Table 3). As predicted, by 30 months, most children were able to pass auditory RJA items. Defining most children as 90%, the binomial test z scores for months 12 through 30 were −10.0, −5.98, −5.49, −1.31 and 1.21, showing no statistically significant difference from 90% for months 24 and 30.

3.2. Correlations of RJA with Receptive Language

The number of words parents reported on the CDI that their child comprehended exceeded 10 for all but four children at 12 months and for all children at 15 months. At 12 months, the median was 38, the mean 51 (standardized skew = 5.51), with a range from 2–152 excluding one extreme score of 233 (defining extreme as more than 1.5 times the interquartile range above the 75th percentile; Tukey, 1977). At 15 months, the median and mean were both 111 (no skew) and the range was 12–217 with no extreme scores.

Correlations of CDI comprehension scores were near zero at 12 months—ρ = .05 and .08; p = .72 and .61—for multimodal and auditory RJA, respectively. But at 15 months correlations were medium for multimodal and small for auditory RJA—ρ = .34 and .21, p = .019 and .16 (Spearman correlations due to skewed distributions of CDI scores). Thus Hypothesis 3, which predicted little correlation for both multimodal and auditory RJA, was supported at 12 months but was only partially supported at 15 months.

3.3. Trajectories for Initiating Joint Attention

We had expected that composite IJA—which, because it includes both lower- and higher-level bids—would occur often in both modalities even at 12 months with little change in performance from 12 to 30 months (Hypothesis 4). Generally, this hypothesis was supported (see Figure 3; Table 3). The mean composite multimodal IJA score was high, increasing from 2.57 to 2.89 (a small linear effect, η2G = .035) and the mean composite auditory IJA score was likewise high, varying from 2.62 to 2.80 with essentially no linear effect. At all ages, all children passed at least one multimodal and one auditory IJA item with just two exceptions—one child passed no multimodal items at 15 months and another child passed no auditory items at 18 months. Unlike RJA, differences at each age were minimal: dz varied from .013 to .018 and p values from .22 to .31 (as determined by repeated-measures t tests at each visit).

In contrast, we expected that higher-level IJA—which must include intentional communication—would only gradually emerge from 12 to 30 months in both multimodal and auditory modalities (Hypothesis 5a). Again, generally this was so (see Figure 3). Higher-level multimodal IJA (like multimodal RJA) increased linearly (linear η2G = .44) and auditory higher-level IJA (like auditory RJA) increased linearly (linear η2G = .36) and even accelerated somewhat towards 30 months (U quadratic η2G = .049). At the same time, exclusively lower-level IJA decreased strongly with age in both modalities (see Table 3). Additionally, we expected that by 18 months (and likewise at 24 and 30 months) almost all children would pass at least one of the six IJA items at the higher level (Hypothesis 5b). At months 12 through 30, respectively, 35%, 70%, 83%, 86%, and 93% of the children passed at least one IJA items at the higher level. Defining almost all children as 90%, the binomial test z scores for months 12 through 30 were −12.5, −4.52, −1.60, −0.80, and 0.70—which, showing no statistically significant difference from 90% for months 18 through 30, would support the hypothesis. We also expected that, even at older ages, children would be more likely to pass multimodal than auditory items at the higher level (Hypothesis 5c), which was likewise supported— dz comparing multimodal to auditory higher-level IJA were 0.34, 0.54, 0.84, 0.61, and 0.50; p = .026, .001, < .001, < .001, and .002; at months 12 through 30, respectively (as determined by repeated-measures t tests at each visit).

4. Study 1 Discussion

As we anticipated, auditory RJA gradually increased from 12 to 30 months. However, it was consistently more difficult than multimodal RJA and, unlike multimodal RJA which reached ceiling level around 24 months of age, even at 30 months only about half the children followed all three auditory RJA bids. Comprehension of verbal labels likely does not account for the challenge of following an auditory prompt given that receptive vocabularies at 12 and 15 months were large enough to encompass the names of the targets we selected and did not correlate significantly with the number of auditory items passed. Rather, early difficulties understanding that a partner’s word might refer to an invisible entity more likely indicates that early in their second year, children may not yet fully appreciate the communicative thrust of absent reference (Ganea & Saylor, 2013; Nelson, 1996; Saylor, 2004; Werner & Kaplan, 1963).

Similarly, and as we expected, from 12 to 30 months of age children became more skilled at issuing higher-level communicative bids to share sounds. Yet, actively inviting a partner to share a sound was more difficult than bidding for shared attention to a multimodal object. Strikingly few 12-montholds passed even one auditory item at the higher-level, and higher-level IJA scores even at 30 months were well below ceiling. The difficulty posed by auditory IJA items was not likely due to lower saliency of sounds. Even at 12 months, composite IJA scores, which indicate that either a lower-level or a higher-level bid was produced, were near ceiling in both modalities, indicating that children almost always alerted to the sound and glanced directly at the partner even when they did not actively invite joint attention to the spectacle.

5. Study 2: The Effect of ASD on Early Auditory Joint Attention Skills

In Study 2 we used the JASA to examine the effect of ASD on early auditory and multimodal joint attention skills in samples of toddlers with ASD, toddlers with non-ASD developmental disorders (DD), and TD toddlers. Our aim was to broaden the already burgeoning literature on joint attention skills difficulties in ASD (Bottema-Beutel, 2016; Mundy, 2016) so that it encompasses auditory as well as visual modalities, providing novel information about autism’s effect on how skillfully a toddler might be able to share the auditory world with social partners.

Our hypotheses related to multimodal RJA were readily shaped by previous findings that document how ASD severely impacts RJA in very young children. Thus, we anticipated that multimodal RJA scores would be markedly lower in the ASD group compared to both the TD and DD groups, with the DD group scoring between the ASD and TD groups (Hypothesis 1a). Moreover, we anticipated that all the TD toddlers would pass at least one multimodal RJA, but that many of the toddlers with ASD and some toddlers with DD would not pass even one item (Hypothesis 1b).

In line with Study 1’s findings, we anticipated here that auditory RJA would be more difficult than multimodal RJA regardless of diagnostic group (Hypothesis 2a). Nevertheless, we also expected that toddlers with ASD would score lower on auditory RJA than toddlers with DD who in turn would score lower than TD toddlers (Hypothesis 2b) and that, although almost all TD toddlers would pass at least some auditory RJA item, a substantial number of toddlers with ASD and with DD might not pass any items (Hypothesis 2c).

We had different expectations about diagnostic effects for IJA skills depending on which IJA score we were considering. As in Study 1, we expected that almost all toddlers—regardless of diagnostic group—would pass at least one IJA item at some level in each modality (Hypothesis 3a), although toddlers in the ASD group might pass fewer auditory and multimodal items than toddlers in the other two groups (Hypothesis 3b).

In contrast, given the pervasive difficulty toddlers with ASD display in initiating intentional communicative bids, we anticipated that toddlers with ASD would be more likely than toddlers in the other two groups to not pass any IJA items at the higher level in either modality (Hypothesis 4a) and would score lower on higher-level IJA in both modalities than toddlers in the other two groups (Hypothesis 4b). Finally, in line with Study 1’s findings, we anticipated that TD toddlers would be more likely to pass multimodal than auditory items at the higher level. However, because we anticipated that ASD would severely impact higher-level IJA, we predicted that modality might not affect their IJA skills whether measured by the composite or higher-level IJA scores (Hypothesis 5).

6. Method

6.1. Participants

Study 2 included 33 toddlers with ASD, 41 with DD, and 39 TD toddlers (79%, 54%, and 74% boys, respectively). We recruited the sample from the greater Atlanta metropolitan area in 2013 through 2017 for an early ASD detection project (Robins et al., 2014). For this project, parents completed the Modified Checklist for Autism in Toddlers, Revised, with Follow-Up (M-CHAT-R/F; Robins et al., 2009) during the child’s 15, 18, or 24 months well-baby check-up. If scores indicated ASD risk and the child was less than 30 months old, we invited parents to participate. After the JASA session, each screen-positive child received a free diagnostic evaluation conducted by research-reliable licensed psychologists and other clinical personnel in a university clinic. The evaluation included standardized skill assessments including the Mullen Scales of Early Learning (Mullen, 1995) and gold standard ASDspecific assessments; final diagnosis was made by clinical best estimate using all available information. Of the 74 at-risk toddlers, 33 were diagnosed with ASD and 41 with another developmental disorder (including 20 with global developmental delay and 12 with developmental language disorder). We recruited the comparison TD group by asking parents of toddlers who screened negative on the MCHAT-R/F to participate, attempting to maintain comparability to the toddlers who screened at-risk for ASD on recruitment site, child age, sex, and minority status (see Table 3). For the ASD, DD, and TD groups, respectively, the toddlers’ mean ages were 22.0, 21.9, and 23.2 months (ranges = 19–29, 15–30, and 17–27), their mean composite scores on the Mullen Scales of Early Learning (Mullen, 1995) were 66, 71, and 100 (ranges = 49–117, 49–137, and 72–142), and their mothers’ mean ages were 33.3, 31.2, and 32.8 years (range = 20–42, 18–46, and 19–44).

6.2. Procedures

Procedures related to the JASA were the same as in Study 1. Prior to the JASA, a researcher with a graduate degree in educational psychology administered the Mullen Scales of Early Learning (Mullen, 1995) to the participants in the TD group.

6.3. Data Analysis

Primary data analysis for Study 2 considered the mean number of items passed for four dependent variables—RJA, composite IJA, higher-level IJA, and exclusively lower-level IJA—separately for modality (multimodal and auditory) and diagnostic group.

For modality, we assessed effect sizes with dz, Cohen’s d for repeated measures. We determined statistical significance with repeated-measures t tests performed separately for each diagnostic group.

For diagnostic group, we assessed effect sizes with Cohen’s d for independent groups, comparing ASD to DD and ASD to TD group means. Cohen’s thresholds for small, medium, and large effects are 0.20, 0.50, and 0.80 (Cohen, 1988). We determined statistical significance with independent-groups analyses of variance, two for each dependent variable (separate analyses for each modality).

In addition to our primary analyses, hypotheses concerned with the number of children who passed a specific number of items or with modality differences in those numbers were tested with binomial z scores, whereas hypotheses concerned with diagnostic group differences in those numbers were tested with odds ratios.

7. Results

7.1. Diagnosis and Responding to Joint Attention

As expected, multimodal RJA was lowest for the ASD group, intermediate for the DD group, and highest for the TD group (Hypothesis 1a; see Figure 4 and Table 4). Also as expected, all TD toddlers passed at least one multimodal RJA item, but a substantial number (58% and 78%) of toddlers with ASD and with DD did not (Hypothesis 1b). Binomial test z scores indicating differences from 90% (our definition here for all) were −6.21, −2.55, and +2.08 for the ASD, DD, and TD groups, respectively.

Figure 4: Mean Number of Items Passed by Diagnostic Group (Study 2).

Figure 4:

Open in a new tab

Note. RJA = responds to joint attention, IJA = initiates joint attention. The RJA bars represent the mean number coded RJA. The full extent of the IJA bars represents the mean number coded either lower- or higher-level IJA (composite IJA); their bottom portion represents the mean number coded higher-level IJA. The full extent of the lower-level IJA bars represents the mean number coded lower-level IJA; the bottom portion represents the mean number coded exclusively lower-level IJA. See text for details. N = 46, 47, 47, 43, and 44 for the 12-, 15-, 18-, 24-, and 30-month visit, respectively. Error bars are 95% CIs, for simplicity not shown for lower-level IJA but only for exclusively lower-level IJA.

Table 4.

Diagnostic Group Statistics and Percentage of Toddlers who Passed at Least One Item (Study 2)

Variable Modality Mean # of items passed by diagnostic group Group statistics % of toddlers who passed at least
one item for each group
ASD DD TD ASD-DD d ASD–TD d p ASD DD TD
RJA Multimodal 0.91a 1.78b 2.62c 0.82 2.11 .000 58 78 100
Auditory 0.52a 0.63a 1.21b 0.15 0.68 .005 39 44 59
Composite IJA Multimodal 1.61a 2.56b 2.64b 1.00 1.04 .000 73 100 100
Auditory 1.85a 2.27ab 2.41b 0.42 0.64 .024 85 93 100
Higher-level IJA Multimodal 0.15a 0.54a 1.18b 0.50 1.24 .000 15 29 67
Auditory 0.18a 0.56ab 0.87b 0.49 0.90 .003 15 32 56
Lower-level IJAa Multimodal 1.24ab 1.76 b 0.79 a 0.40 0.35 .004 58 71 33
Auditory 1.42 1.39 0.92 0.03 0.44 .12 70 61 44
Open in a new tab

Note. RJA = responds to joint attention, IJA = initiates joint attention, ASD = autism spectrum disorder, DD = non-ASD developmental disorders, TD = typically developing. N = 33, 41, and 39 for ASD, DD, and TD groups, respectively. Cohen’s ds are standardized mean differences between the ASD-DD and ASD–TD groups. p values are from an independent-groups analysis of variance. Means that do not differ p < .05 per a Tukey HSD post hoc test share a common subscript.

a

Exclusively lower-level IJA

As we expected, auditory RJA items were more difficult than multimodal RJA items regardless of diagnostic group (Hypothesis 2a). The auditory RJA mean was less than the multimodal RJA mean, weakly for the ASD group and strongly for the DD and TD groups: dz = 0.38, 1.01, and 1.21; p = 0.35, < .001, and < .001, respectively (as determined by repeated-measures t tests for each group; see Table 4 for mean values). With respect to auditory RJA specifically, we expected that toddlers with ASD would score lower than toddlers with DD who in turn would score lower than TD toddlers (Hypothesis 2b). Toddlers with ASD did score lower than TD toddlers (d = 0.68, see Table 4), but the difference between toddlers with ASD and with DD did not even meet the threshold for a weak effect (d = 0.15; see Table 4). We also expected that almost all TD toddlers would pass at least some auditory RJA items, but that a substantial number of toddlers with ASD and with DD might not pass any (Hypothesis 2c). In fact, the percentage of toddlers who passed at least one item (see Table 4) did not differ significantly from 50% for any of the diagnostic groups: the binomial test z scores were −1.22, −0.78, and 1.21 for the ASD, DD, and TD groups, respectively.

7.2. Diagnosis and Initiating Joint Attention

We expected that almost all toddlers regardless of diagnostic group would pass at least one of the IJA items in each modality at some level (Hypothesis 3a). This was generally supported. Percentages exceeded 90%—our definition of almost all—for the DD and TD groups in both modalities (see Table 4). For the ASD group, percentages were slightly less than 90% for the auditory modality and significantly less for the multimodal modality (85% and 73%, binomial test z score = −0.99 and −3.31, respectively). We also expected that toddlers with ASD would pass fewer multimodal and auditory items than toddlers in the other two groups (i.e., have lower composite IJA scores; Hypothesis 3b). Again, this was generally supported. For both multimodal and auditory items, the mean number of items passed in the ASD group was lower than in the other two groups, although not significantly lower than the DD group mean for the auditory modality (for statistics see Table 4, composite IJA).

In contrast, we expected that toddlers with ASD would be more likely than toddlers in the other two groups to pass no IJA items at the higher level in either modality (Hypothesis 4a). This was generally supported. For multimodal and auditory items, respectively, toddlers with ASD were more likely than TD toddlers to pass no items at the higher level (85% vs. 33% and 85% vs. 44%, odds ratios = 11.2 and 7.25, 95% CI = [3.51, 35.8] and [2.31, 22.7], p < .001 for both), but the contrast with toddlers with DD was less marked (85% vs. 71% and 85% vs. 68%, odds ratio = 2.32 and 2.60, 95% CI = [0.72, 7.43] and 0.81, 8.26], p = .16 and .11). We also expected that toddlers with ASD would score lower on higher-level IJA in both modalities than toddlers in the other two groups (Hypothesis 4b). Again this was generally supported. For both multimodal and auditory items, the mean number of items passed in the ASD group was lower than in the other two groups, although not significantly lower than the DD group mean in both modalities (for statistics see Table 4, higher-level IJA).

Finally, we expected that TD toddlers would be more likely to pass multimodal than auditory IJA items at the higher level (Hypothesis 5), which they did: M = 1.18 vs. 0.87, ds = .031, p = .057—albeit at a lower level than in Study 1 (ds = .061, p < .001, for the 24-month visit). We also expected that, for toddlers with ASD, modality would have little effect on either composite or higher-level IJA, respectively (because they would rarely pass items), and it did not: M = 1.61 vs. 1.85 and 0.15 vs. 0.18, ds = −0.16 and −0.08, p = .36 and .66—less than the threshold for a small effect.

8. Study 2 Discussion

ASD severely impacted multimodal RJA skills. As anticipated, TD toddlers readily followed a partner’s head turns and points to locate targets and many toddlers with ASD did not do so even once. Moreover, multimodal RJA skills were more compromised in the ASD group than in the DD group. Although this finding is consistent with previous studies, its large effect size is striking given that all toddlers in both groups had screened at-risk for ASD prior to the JASA assessment and most displayed significant cognitive delays. Also, regardless of group, auditory RJA items were far more difficult than multimodal RJA, so much so that mean scores were near floor level for both ASD and DD groups and quite low even for toddlers in the TD group. Thus, words were not as effective as pointing when used as a bid for joint attention, even when, as in the ASD group, pointing itself was a relatively weak directive. Yet words were not completely ineffective bids: some toddlers—around 60% in the TD group and 40% in the ASD and DD groups—responded at least once to a label by seeking its referent, suggesting that at the end of their second year some toddlers, even those diagnosed with ASD, had begun to appreciate the deictic force of words.

Our findings related to IJA demonstrate toddlers often found sounds to be attractive spectacles. Analyses of a composite IJA score revealed that they usually monitored their partner’s attention during a spectacle, albeit toddlers with ASD were less likely to do so than TD toddlers and, for multimodal IJA, less likely to do so than toddlers with DD. In contrast, analyses of higher-level IJA scores indicated that even TD toddlers often did not actively try to entice their partner to share the spectacles, especially the auditory ones. More striking still is how severely ASD affected IJA. Although toddlers with ASD often monitored their partner’s attention, they almost never issued an intentional bid for joint attention to any of the JASA’s six spectacles.

9. General Discussion

The two complementary studies reported here expand our view of early joint attention skills by highlighting their auditory aspects using a systematic laboratory-based assessment. The JASA protocol shares several features with the widely used ESCS (Mundy et al., 2003). Like the ESCS, it measures RJA and IJA as separate skills, a division that reflects the conceptual distinction between responding to and initiating actions (Mundy, 2016). Our fundamental modification was to introduce auditory items so that we could contrast responses to a partner’s verbal bids for joint attention to bids that included visual cues of head turning and pointing as well as initiations for a partner’s joint attention to sounds emanating from hidden sources and to multimodal spectacles. Overall, our findings fit well within the established depiction of the development of and diagnostic differences in early joint attention skills with one important caveat: children found auditory items far more difficult to execute than multimodal ones.

For RJA, TD children developed auditory skills more slowly than multimodal ones, and even at 30 months their performance did not reach the near-ceiling level achieved for multimodal RJA. This intriguing contrast suggests that words—words that TD children likely understood as object labels early in their second year—only gradually gained their deictic force to refer to displaced entities, a force that gaze shifts and points had developed earlier (Scaife & Bruner, 1975). Furthermore, object labels clearly lacked deictic force for toddlers with DD at the end of the second year when passing auditory items was so rare that, unlike multimodal RJA items, they did not outperform toddlers in the ASD group. One reasonable explanation is that severe language delays experienced by many toddlers in both groups compounded the difficulty of understanding that when a partner calls their name and then produces an object label they are meant to locate the word’s referent.

Our results related to IJA depended markedly on which IJA variable was used. Analysis of composite IJA scores indicated that nearly every TD child even at 12 months in Study 1 and most toddlers in all three diagnostic groups in Study 2 monitored their partner’s attention when they were presented with interesting spectacles. But when higher-level IJA scores were considered, both age and diagnostic differences were marked. The ability to intentionally issue a communicative bid that invited the partner to share either multimodal or auditory spectacles was rare at 12 months of age and emerged gradually over the next year and a half. Even more strikingly, although toddlers with ASD had some success monitoring their partner’s attention during IJA items, albeit at a rate lower than toddlers with DD and TD toddlers, they essentially never intentionally bid to share any of the six spectacles we presented. The contrast of lower-level and higher-level auditory IJA scores reinforces the cogent argument (e.g., Pickard & Ingersoll, 2015) that only higher-level IJA scores capture the social motivation and intentional communication skills that are the hallmark of initiating joint attention (Charman, 2003). However, the relatively high prevalence of lower-level IJA displayed by toddlers with ASD and with DD raises the intriguing possibility that low-level monitoring of a partner’s attention might signal the child’s availability for sharing the spectacle and so might serve in interventions as an entry to nurturing higher-level IJA skill acquisition.

9.1. Limitations and Future Work

In designing the JASA, we needed to balance the benefits and limitations of using a tightly honed protocol. We confirmed that TD children from 12 to 30 months and toddlers diagnosed with ASD and DD willingly participated in the playful interactions that our friendly stranger structured and that the JASA’s manualized procedures resulted in high administration fidelity and coding reliability. But, since research on auditory joint attention is still sparse, it remains unclear if the JASA is a valid reflection of how children use auditory joint attention skills in more naturalistic settings. One reassuring note is that when the children assessed here were observed reacting to sounds during play with their caregivers, our findings over time and across groups were in line with those reported here. More specifically, paralleling our IJA findings here, we also found that for TD children spontaneous bids to share a new sound with the partner increased from 12 to 30 months of age (Adamson et al., 2019a) and that although toddlers with ASD and with DD often alerted and oriented to the sound, they rarely attempted to share the sound (Adamson et al., 2021). Moreover, in line with our RJA findings, we observed that when the caregivers scaffolded joint engagement to the sounds, the occurrence of auditory joint engagement increased from 12 to 18 months (Adamson et al., 2019a) and toddlers with ASD were less likely to sustain joint engagement than those in the DD and TD groups (Adamson et al., 2021).

There are several ways that the JASA might be used in future studies to further illuminate auditory joint attention skills. Assessing older children would help clarify when auditory joint attention skills typically consolidate and whether ASD’s adverse impact on auditory joint attention skills lessens over time, as they reportedly do for multimodal RJA but not for multimodal IJA (Mundy, 2016). It also may be instructive to use the JASA to explore the relation between auditory joint attention and joint engagement during parent-child interactions, a relation that has recently been examined for multimodal joint attention skills (Adamson et al., 2019b). Furthermore, it can be used to study whether the robust relation found between variations in early multimodal RJA skills and later receptive and expressive language (Bottema-Beutel, 2016) occurs as well for auditory IJA skills. Indeed, this link may be even stronger for auditory RJA skills given that they specifically gauge attention to a partner’s verbal input.

Systematically modifying the JASA could help address several interesting questions about how sounds regulate early interactions. How auditory RJA bids convey meaning might be explored systematically by altering their form using experimental designs that have been profitably used to reveal that points have more deictic force than gaze shifts (Butterworth, 1997). In particular, it would be interesting to compare the effectiveness of cues that contain only an object label with those that add explicit deictic words (“look, a dog”) or directives (“find the dog”) to determine if such modifications ease the difficulties of the JASA’s current auditory RJA items. Similarly, the spectacles in the auditory IJA items could be systematically varied to discern how aspects of a sound’s hidden source affect intentional communication about sounds. One intriguing possibility is that TD toddlers and toddlers with DD might issue invitations to share sounds earlier and more often when they seem to emanate from concrete familiar entities, such as a dog or a car, than from more amorphous ones such as the knock on a door or electronic buzzer that we used here. In contrast, given the pervasive adverse impact of ASD on children’s social motivation and communicative intentions, we expect that such modifications would do little to increase their higher-level auditory IJA.

In summary, the current studies expand the assessment of joint attention skills to include items that highlight sounds in addition to the traditional items where sights predominate. Doing so helps to begin to address concerns (Akhtar & Gernsbacher, 2008) that focusing primarily on visible aspects of joint attention may overlook other important aspects of joint attention, especially in young children with ASD who display significant joint attention difficulties on assessments that do not isolate auditory cues (Bottema-Beutel, 2016). Our findings provide a foundation for further studies of auditory joint attention by demonstrating that typically developing children can use sounds to direct their own attention and to guide others’ attention although auditory joint attention skills develop more slowly than multimodal ones. Further, they reinforce conclusions that the joint attention skills, regardless of the modality of bids, are severely compromised in toddlers who screen positive for ASD risk, particularly those who are subsequently diagnosed with ASD.

Highlights.

  • Auditory joint attention skills were systematically assessed in two studies to document how toddlers respond to verbal bids and initiate bids to share sounds.

  • Like multimodal joint attention skills, auditory ones typically emerged from 12 to 30 months with the caveat that auditory skills do so far more slowly.

  • Toddlers with autism spectrum disorder (ASD) had markedly weaker auditory joint attention skills than typically developing toddlers and even toddlers with non-ASD developmental disorders.

  • Adding auditory items to traditional joint attention assessments expands our view of important variations in social-cognitive skills that may affect language acquisition.

Acknowledgments

This research was supported by a grant from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (R01HD035612). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Child Health and Human Development or the National Institutes of Health. Diana L. Robins is co-owner of M-CHAT, LLC, which receives royalties from companies that incorporate the M-CHAT(-R) into commercial products. No royalties were collected in relation to the current study. Dr. Robins also serves on the advisory board of Quadrant Bioscience, Inc.

The authors thank Taylor Bradley, Claire Cusack, Anita Hasni, Lauren Schmuck, and Sarah Vogt for their help with this study. We also gratefully acknowledge the parents and children who participated in this research.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Adamson LB, & Bakeman R (1991). The development of shared attention during infancy. In Vasta R (Ed.), Annals of Child Development, Vol. 8 (pp. 1–41). London, England: Kingsley. [Google Scholar]
  2. Adamson LB, Bakeman R, Suma K, & Robins DL (2019a). Sharing sounds: The development of auditory joint engagement during early parent-child interaction. Developmental Psychology, 55, 2491–2504. 10.1037/dev0000822 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Adamson LB, Bakeman R, Suma K, & Robins DL (2019b). An expanded view of joint attention: Skill, engagement, and language in typical development and autism. Child Development, 90, e1–e18. 10.1111/cdev.12973 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Adamson LB, Bakeman R, Suma K, & Robins DL (2021). Autism adversely affects auditory joint engagement during parent-toddler interactions. Autism Research, 14, 301–314. http://doi.org/10w.1002/aur.2355 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Adamson LB, & Dimitrova N (2014). Joint attention and language development. In Brooks P & Kempe V (Eds.), Encyclopedia of language development, (pp. 299–304). Los Angeles, CA: Sage. [Google Scholar]
  6. Adamson LB, Kellerman A, Suma K, Vogt S, Hasni A & Bakeman R (2016). Joint attention skills assessment: Initiating and responding to sounds and sights (JASA; Technical Report 21). Georgia State University; website https://bakeman.gsucreate.org/DevLabTechReport21.pdf [Google Scholar]
  7. Akhtar N, & Gernsbacher MA (2008). On privileging the role of gaze in infant social cognition. Child Development Perspectives, 2, 59–65. 10.1111/j.1750-8606.2008.00044.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bakeman R (2005). Recommended effect size statistics for repeated measures designs. Behavior Research Methods, 37, 379–384. 10.3758/bf03192707 [DOI] [PubMed] [Google Scholar]
  9. Bakeman R (2018). KappaAcc: Deciding Whether Kappa is Big Enough by Estimating Observer Accuracy (Technical Report 28). Retrieved from Georgia State University; website: https://bakeman.gsucreate.org/DevLabTechReport28.pdf [Google Scholar]
  10. Beers AN, McBoyle M, Kakande E, Dar Santos RC, & Kozak FK (2014). Autism and peripheral hearing loss: A systematic review. International Journal of Pediatric Otorhinolaryngology, 78, 96–101. 10.1016/j.ijporl.2013.10.063 [DOI] [PubMed] [Google Scholar]
  11. Beuker KT, Rommelse NNJ, Donders R, Buitelaar JK (2013). Development of early communication skills in the first two years of life. Infant Behavior & Development, 36, 71–83. 10.1016/j.infbeh.2012.11.001 [DOI] [PubMed] [Google Scholar]
  12. Bottema-Beutel K (2016). Associations between joint attention and language in autism spectrum disorder and typical development: A systematic review and meta-regression analysis. Autism Research, 9, 1021–1035. 10.1002/aur.1624 [DOI] [PubMed] [Google Scholar]
  13. Bruner J (1983). Child’s talk: Learning to use language. New York: W. W. Norton. [Google Scholar]
  14. Butterworth G (1997). Origins of joint visual attention in infancy. Monographs of the Society for Research in Child Development, 63, No. 4, 144–166. [DOI] [PubMed] [Google Scholar]
  15. Charman T (2003). Why is joint attention a pivotal skill in autism? The Royal Society, 358, 315–324. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Cicchetti DV (1994). Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment, 6, 284–290. 10.1037/1040-3590.6.4.284. [DOI] [Google Scholar]
  17. Clifton RK (1992). The development of spatial hearing in human infants. In Werner LA & Rubel EW (Eds.), Developmental Psychoacoustics (pp. 135–157). Washington, DC: American Psychological Association. 10.1037/10119-005 [DOI] [Google Scholar]
  18. Cohen J (1998). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum Associates. [Google Scholar]
  19. Fenson L, Marchman VA, Thal DJ, Dale PS, Reznick JS, & Bates E (2007). MacArthur–Bates Communicative Development Inventories: User’s guide and technical manual (2nd ed.). Baltimore, MD: Brookes Publishing. [Google Scholar]
  20. Ganea PA, & Saylor MM (2013). Representational constraints on language development: Thinking and learning about absent things. Child Development Perspectives, 7, 227–231. 10.1111/cdep.12045 [DOI] [Google Scholar]
  21. Gardner W (1995). On the reliability of sequential data: Measurement, meaning, and correction. In Gottman JM (Ed.), The analysis of change (pp. 339–359). Hillsdale, NJ: Erlbaum. [Google Scholar]
  22. Gomes H, Molholm S Chrisodoulou C, Ritter W, & Cowan N (2000). The development of auditory attention in children. Frontiers in Bioscience, 5, d108–120. doi: 10.2741/Gomes Gravel, J., [DOI] [PubMed] [Google Scholar]
  23. Lord C, Rutter M, DiLavore PC, Risi S, Gotham K, & Bishop S (2012). Autism diagnostic observation schedule (2nd ed.). Torrance, CA: Western Psychological Services. [Google Scholar]
  24. Melis AP, Call J, & Tomasello M (2010). 36-month-olds conceal visual and auditory information from others. Developmental Science, 13, 479–489. doi: 10.1111/j.1467-7687.2009.00892.x [DOI] [PubMed] [Google Scholar]
  25. Moll H, Carpenter M, & Tomasello M (2014). Two- and 3-year-olds know what others have and have not heard. Journal of Cognition and Development, 15(1), 12–21. doi: 10.1080/15248372.2012.710865 [DOI] [Google Scholar]
  26. Morales M, Mundy P, Delgado CEF, Yale M, Messinger D, Neal R, & Schwartz HK (2000). Responding to joint attention across the 6-through 24-month age period and early language acquisition. Journal of Applied Developmental Psychology, 21, 283–298. 10.1016/S0193-3973(99)00040-4 [DOI] [Google Scholar]
  27. Mullen EM (1995). Mullen Scales of Early Learning. Circle Pines, MN: American Guidance Services. [Google Scholar]
  28. Mundy PC (2016). Autism and joint attention: Development, neuroscience, and clinical fundamentals. New York: Guilford Press. [Google Scholar]
  29. Mundy P, Block J, Delgado C, Pomares Y, Van Hecke AV, & Parlade MV (2007). Individual differences and the development of joint attention in infancy. Child Development, 78, 938–954. 10.1111/j.1467-8624.2007.01042. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Mundy P, Delgado C, Block J, Venezia M, Hogan A, & Seibert J (2003). Early Social Communication Scales (ESCS). Coral Gables, FL: University of Miami. [Google Scholar]
  31. Murza KA, Schwartz JB, Hahs-Vaughn, & Nye C (2016). Joint attention interventions for children with autism spectrum disorder: A systematic review and meta-analysis. International Journal of Language and Communication Disorders, 51, 236–251. 10.1111/1460-6984.12212 [DOI] [PubMed] [Google Scholar]
  32. Nelson K (1996). Language in cognitive development: Emergence of the mediated mind. Cambridge: Cambridge University Press. [Google Scholar]
  33. Olejnik S, & Algina J (2003). Generalized eta and omega squared statistics: Measures of effect size for some common research designs. Psychological Methods, 8, 434–447. 10.1037/1082-989X.8.4.434 [DOI] [PubMed] [Google Scholar]
  34. Picard KE, & Ingersoll BR (2015). Brief report: High and low level initiations of joint attention, and response to joint attention: Differential relationships with language and imitation. Journal of Autism and Developmental Disorders, 45, 262–268. 10.1007/s10803-014-2193-8 [DOI] [PubMed] [Google Scholar]
  35. Robins DL, Casagrande K, Barton M, Chen CA, Dumont-Mathieu T, & Fein D (2014). Validation of the Modified Checklist For Autism in Toddlers, revised with follow-up (M-CHAT–R/F). Pediatrics, 133(1), 37–45. 10.1542/peds.2013-1813 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Robins DL, Fein D, & Barton M (2009). The Modified-Checklist for Autism in Toddlers-Revised with Follow-up (M-CHAT-R/F). Self-published. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Saylor MM (2004). Twelve- and 16-month-old infants recognize properties of mentioned absent objects. Developmental Science, 7, 599–611. 10.1111/j.1467-7687.2004.00383.x [DOI] [PubMed] [Google Scholar]
  38. Scaife M, & Bruner JS (1975). The capacity for joint visual attention in the infant. Nature, 253, 265–266. 10.1038/253265a0 [DOI] [PubMed] [Google Scholar]
  39. Tukey (1977). Exploratory data analysis. Reading, MA: Addison-Wesley. [Google Scholar]
  40. Watson LR, Poston V, & Baranek GT (2003). The attention-following and initiating joint attention protocol (JA protocol). Unpublished manuscript.
  41. Werner H, & Kaplan B (1963). Symbol formation. New York: Wiley. [Google Scholar]
  42. Wetherby A, & Prizant B (2002). Communication and symbolic behavior scales developmental profile—first normed edition. Baltimore: Paul H. Brookes. [Google Scholar]
  43. Widaman KF (2006). Missing data: What to do with or without them. In McCartney K, Burchinal MR, & Bub KL (Eds.), Best practices in quantitative methods for developmentalists (pp. 64–97). Monographs of the Society for Research in Child Development, 71(3, Serial No. 285). [DOI] [PubMed] [Google Scholar]
  44. Williamson RA, Brooks R, & Meltzoff AN (2015). The sound of social cognition: toddlers’ understanding of how sound influences others. Journal of Cognition and Development, 16, 252–260. doi: 10.1080/15248372.2013.824884 [DOI] [Google Scholar]

RESOURCES