Babbling in extremely premature infants at 12 months corrected age

ABSTRACT Babbling is an important precursor to speech in infancy, and deviations from the typical babbling development can predict later difficulties in speech, language, and communication. This cross-sectional study aimed to investigate babbling and early speech in Swedish extremely premature infants. Samples of babbling were collected from 20 extremely premature infants (EPT group) at the corrected age of 12 months. Data collection was home-based and consisted of an audio-video recording of each infant playing with a parent. Presence of canonical babbling (CB), and three oral stop variables distinctive of typical babbling, and consonant inventory were assessed. The assessment was performed during a standardised observation of babbling. Data from the EPT group were compared to previously collected data of a reference group of 20 10-month-old infants without known medical diagnoses. The results showed that the EPT group had a lower proportion of infants producing CB, and that they used a significantly smaller consonant inventory compared to the reference group. Although not statistically significant, oral stops were less frequently found in the EPT group. The findings of a restricted consonant inventory and low proportion of CB in the EPT group are not surprising considering that the group has been found to be at risk of speech and language delay in toddlerhood. Still, further research is needed to explore whether babbling at 12 months can predict speech and language skills at an older age in extremely premature infants.


Introduction
Babbling is an important precursor to speech in infants, and deviation from the typical babbling development can be predictive of later neurodevelopmental difficulties including speech and language impairments.Previous research has described nontypical babbling development in clinical groups well known to be at increased risk of speech and language impairments, for example autism spectrum disorder (Patten et al., 2014;Paul et al., 2011), hearing impairment (Moeller et al., 2007), neurodevelopmental disabilities (Nyman & Lohmander, 2018), cleft palate (Chapman et al., 2001;Lohmander et al., 2011), and Fragile X syndrome (Belardi et al., 2017).Extremely premature (EPT) infants represent another group at increased risk for neurodevelopmental disorders (Serenius et al., 2016).To date, there are a few studies on babbling in groups that partly include EPT infants (D'Odorico et al., 2010;Rvachew et al., 2005;Stolt et al., 2012;Törölä et al., 2012aTörölä et al., , 2012b)).The results of these studies are contradictory, and no studies comprise only EPT infants.

Extremely premature birth
According to an estimation from 2010, about 15 million infants are born prematurely per year worldwide (WHO, 2018).Of these, 5.2% are born EPT which means that they are born in gestational week 22-27 (Blencowe et al., 2012).Among Swedish children, approximately 0.3% are born EPT, and about 300 infants survive EPT birth per year (Socialstyrelsen, 2014).Today, a premature infant born as early as in gestational week 22 can be considered for active care according to national Swedish guidelines (Domellöf & Jonsson, 2018).Neonatology has developed rapidly, and compared to 15 years ago, an increasing proportion of EPT infants survive until the age of one year (Norman et al., 2019).However, the group is affected by health problems, both in infancy and later in life.During at least the first months in life, EPT infants require intensive care in neonatal units.They are at risk for several diseases, for example bronchopulmonary dysplasia, retinopathy of prematurity, necrotising enterocolitis, intraventricular haemorrhage, and respiratory distress syndrome (Behrman & Butler, 2007).
EPT infants are also at increased risk for neurodevelopmental impairments, which include intellectual disability, cerebral palsy and/or sensory (visual or hearing) loss.In the EXPRESS study of 441 Swedish-born EPT children at the age of 6.5 years, about one-third had no neurodevelopmental impairments, while one-third had mild, and one-third had moderate or severe disabilities (Serenius et al., 2016).Furthermore, previous research has shown that there is an increased incidence of speech and language delay in children born EPT during the preschool years (Månsson & Stjernqvist, 2014;Zambrana et al., 2020).The risk of autism spectrum disorder (ASD) is also highly elevated in children born EPT (Padilla et al., 2017).
When referring to premature infants, corrected age is often reported.Corrected age is the age the infant would be if he or she were born full-term.

Babbling and early speech
During the first year of life, infants undergo four consecutive stages of vocal development (Oller et al., 1999).During the first two stages, the phonation stage and the primitive articulation stage, the infant begins to control phonation and articulation.At the third stage, the expansion stage, the infant starts to produce both vowels and consonants, yet lacking the rapid transition that is characterising the fourth, and important, canonical stage.The canonical stage is a major milestone in early speech development and is characterised by utterances consisting of at least one vowel and one consonant connected by a rapid transition (Oller et al., 1999).The early syllables used in canonical babbling (CB) work as building blocks for children to produce their first words (McCune & Vihman, 2001;McGillion et al., 2017;Stoel-Gammon, 2011).CB appears at the age of 6-9 months (Nathani et al., 2009;Oller et al., 1999;Stoel-Gammon, 2011) and is normally well established at 10 months (Lang et al., 2019;Oller et al., 1999).The onset of CB seems to be stable across languages (Lang et al., 2019;Nathani et al., 2009), and it is not dependent on socioeconomic status (Eilers et al., 1993).If not established at 10 months, the absence of CB might predict a later delay in expressive language (Oller et al., 1999).It can also be an early sign of a late detected neurodevelopmental disorder such as ASD, Rett syndrome or Fragile X syndrome (Lang et al., 2019).

Consonant inventory
Another central aspect of early speech development is phonemic inventory.Previous research on phonemic inventory in infants has mainly focused on consonants, and in particular the so-called true consonants (Morgan & Wren, 2018).Whereas the concept of CB includes syllables with glides (for example/jaja/), neither glides, essentially vocalic, nor glottals, are considered to be true consonants (McCune & Vihman, 2001).These syllables are still classified as CB but appear at an earlier age and can be considered less mature (Nathani et al., 2009;Stoel-Gammon, 1989).The first true consonants appearing in babbling are often anterior stops, and the dental/alveolar stops (d/t) are very common (Morgan & Wren, 2018).These consonants can therefore be considered as typical of CB.In other groups at risk for non-typical speech development, such as infants with cleft palate, a significant relationship between dental/alveolar stops at 18 months and consonant proficiency at 3 years has been found (Klintö et al., 2014;Lohmander & Persson, 2008).Furthermore, usage of the bilabial stops (b/p) has been found to be of special importance for the child to be able to make the transition from babbling to referential use of words (McCune & Vihman, 2001).Proportion of CB syllables in the infant's babbling at 12 months is correlated to the age when the first words appear, and consonant use at the same age is related to phonological skills at 3 years of age (Stoel-Gammon, 2011).
A restricted consonant inventory in CB has been found for groups of infants at risk for, or having, neurodevelopmental disorders that are more common among EPT infants, for example ASD and cerebral palsy.For infants at high risk for a later diagnosis of ASD, Paul et al. (2011) concluded that compared to infants with lower risk of ASD, high-risk infants produced a significantly lower number of different consonants in the first year of life.For children with cerebral palsy, number of different consonants produced by the child at age 2 years predicted functional communication and speech at age 5 years (Pennington et al., 2020).

Babbling and early speech in preterm infants
The first well-known study on the onset of CB in premature infants was published nearly 30 years ago (Eilers et al., 1993).The infants were born in gestational week 30-36, which means that there was no EPT infant included in that study.This first study concluded that onset of CB was not delayed in premature infants compared to full-term infants.
To date, there is a limited number of studies on babbling and early speech in premature infants that include EPT infants (D'Odorico et al., 2010;Rvachew et al., 2005;Stolt et al., 2012;Törölä et al., 2012aTörölä et al., , 2012b)).To the best of our knowledge, there is no study that includes only EPT infants.The four studies mentioned above had inclusion criteria not only defined by prematurity, but also by maximum birth weight, which varied from 1000 g for Törölä et al. (2012aTörölä et al. ( , 2012b) ) to 2000 g for D 'Odorico et al. (2010).Regarding the degree of prematurity, Törölä et al. (2012aTörölä et al. ( , 2012b) ) included extremely and very preterm infants, while the other three (Rvachew et al. (2005), Stolt et al. (2012), andD'Odorico et al. (2010)) included extremely, very, and moderately preterm infants.Furthermore, the methods for assessment used by these authors varied and included phonetically transcribed audio-video recordings (D'Odorico et al., 2010;Törölä et al., 2012aTörölä et al., , 2012b)), phonetic and acoustic analyses of audio recordings (Rvachew et al., 2005), and parental questionnaires (Stolt et al., 2012).All authors compared the preterm infants with full-term infants, and corrected age was used for the preterm infants when comparing them to full-term infants.In addition to varying methods for assessment, the outcome variables varied, and that complicates the comparison between these studies.For CB, neither Törölä et al. (2012aTörölä et al. ( , 2012b) ) nor Stolt et al. (2012) found any difference regarding onset of CB between premature and full-term infants.For the premature infants in the study by Rvachew et al. (2005), there was a delay in the onset of CB for preterm infants, but only in a subgroup of infants with a secondary diagnosis of bronchopulmonary dysplasia.For the group of infants investigated by D'Odorico et al. ( 2010), the proportion of canonical syllables was lower at 12 months in the premature group compared to full-term infants.Concerning consonant inventory, the preterm infants described by D'Odorico et al. ( 2010) used fewer different consonants than the full-term infants at both 12 and 18 months.This finding was, however, not seen in the study by Rvachew et al. (2005), and in Stolt et al. (2012) and Törölä et al. (2012aTörölä et al. ( , 2012b)), data on consonant inventory were not reported.Regarding the potential of babbling to predict speech and language development in premature infants, Stolt et al. (2012) concluded that for the medically fragile premature infants, slow vocalisation development in infancy could predict language skills at 24 months.For the infants studied by D' Odorico et al. (2010), phonological ability at 12 months predicted vocabulary at 18 months, and consonant inventory at 12 months predicted vocabulary size at 24 months.For the infants described by Rvachew et al. (2005), however, canonical syllable ratios in infancy could not predict vocabulary at 18 months.Thus, the findings seem contradictory regarding both onset of CB and consonant inventory, whereas there is some support of a relationship between size of consonant inventory and expressive vocabulary.Together, considering the limited amount of previous research and contradictory results, the need for additional knowledge on babbling and early speech production in EPT infants is clear.

Aim
The aim of this study was to investigate the presence of CB and consonant production in EPT infants at 12 months corrected age, and to compare these results with data from a reference group of 10-month-old typically developing infants.The following research questions were posed: • How does presence of CB in the group of EPT infants compare with a reference group of typically developing infants?• Are there any differences between the group of EPT infants and the reference group regarding presence of oral stops, oral stops with anterior, and specifically dental/ alveolar placement, and if so, which?• Are there any differences between the group of EPT infants and the reference group in consonant inventory measured as number of different true consonants, and if so, which?

Design
A cross-sectional design was employed to investigate babbling and early speech in EPT infants.The EPT infants in this study, hereafter referred to as the EPT group, were all included in the main study Stockholm Preterm Interaction based Intervention Program (SPIBI) (Baraldi et al., 2020).They were born in 2018 or 2019, and at that time, all families who were about to be discharged with their EPT infant from any of the neonatal wards in Stockholm County were invited to the SPIBI study.All infants in the EPT group had been admitted to neonatal intensive care units, where they had been cared for in incubators, receiving at least respiratory support and tube-feeding.Exclusion criteria for entering the SPIBI study were if none of the parents could communicate in Swedish or English, or if the infant needed long-term inpatient hospital care.Inclusion criterion in the current study was that at least one parent should be Swedish speaking.Exclusion criteria were, in addition to those in the SPIBI study, if the child had a tracheostomy and therefore lacked the ability to phonate.The data collection period was from November 2019 to August 2020.

Participants
During the data collection period, 25 infants met the inclusion criteria for the current study.
Out of these infants, five were excluded due to unsatisfying quality of data, such as too short recordings, unexpected events that interrupted the data collection, and poor sound quality that made babbling analysis impossible.Thus, 20 infants were included in the study and their demographic characteristics are presented in Table 1.Information on additional diagnoses that existed among the infants was obtained from the Swedish Neonatal Quality Register.
Reference group Lohmander et al. (2017) collected audio-video recordings from 30 typically developing infants, who were recruited from paediatric health centres in the Stockholm County.Out of these 30 infants, 20 were selected to form a reference group for this study.In the EPT group, there was a higher proportion of boys, and the 20 typically developing infants were selected to match the EPT group based on sex.The mean age of these 20 infants was 10 months and 4 days (9 months 21 days − 10 months 21 days).They had at least one Swedish-speaking parent and no known medical diagnoses.The audio-video recordings of the infants in the reference group had been collected at local paediatric health centres by two final-year speech and language pathology students.The infants were offered a fixed set of age-appropriate toys, and the parents were asked to play with their infant as they normally do at home.The mean length of the recordings was 37 minutes (range 30-45), and the mean volubility was 4.7 utterances per minute (range 2.4-7.3).The procedure is further described by Lohmander et al. (2017).

Ethical considerations
All parents provided written consent prior to entering the main study, approved by the Regional Ethical Review Board in Stockholm (Dnr.2017/1596-31) and an amendment included the babbling study (Dnr.2019/05169).

Data collection
The data collection was carried out in collaboration with a scheduled follow-up visit at 12 months of corrected age in the main study (Baraldi et al., 2020).At this point, parent and child were audio-video recorded together while playing at home.The recordings were made with a video camera (Panasonic HC-V380 or Panasonic HC-V750).The latter was equipped with an external microphone (Røde NTG4+).A portable audio recorder (TASCAM DR-22WL) was used simultaneously with the video camera in most cases.The purpose of the separate audio recording was for back up.The recordings were collected by one or two persons involved in the project.Before collecting data from the EPT group, two infants were recorded to practice the handling and use of technical equipment.The parent was told to play with the infant as they normally do.Most of the infants played with age-appropriate toys during the recording, their own, or the same fixed set of toys as for the reference group.However, since the infants were at home, some of them also chose to play with other things they were familiar with such as kitchen utensils or books in shelves.The recordings were finished when the infant had presented at least 75 utterances, if the parent wanted to end the play, or if the infant had become hungry or tired.The mean length of the audio-video-recordings was 29 minutes (range 10-40).The mean volubility (number of utterances per minute) was 2.78 (range 0.15-7.3).While no question was posed as part of the procedure, the parents usually confirmed spontaneously that the babbling was representative for their infant.This is in accordance with the opinion that parents typically recognise their child's babbling very well (Oller et al., 1998).
The visiting researchers avoided communication with the infants during the recordings to promote interaction between parent and infant and tried to stay out of sight of the infant as much as possible.Due to the outbreak of the COVID-19 pandemic, the visitors left the families' homes during the play in some cases to minimise the risk of infection spreading.
One infant was recorded twice; the quality of the first recording was not fully sufficient for the analysis since the infant had a feeding bottle in her mouth for most part of the recording.One video recording was lost due to technical problems, and therefore one babbling observation was carried out based upon the audio recording.

Analyses
The analysis of the recordings was made by a standardised observation of babbling and by calculation of canonical babbling ratio (CBR UTTER) ).Before assessing the data, the observer practiced the observation method jointly with an experienced user of the method during three sessions.During these sessions, two to four recordings of different infants were analysed.Differences in assessments that occurred were discussed with the aim of calibrating assessments.
The standardised observation of babbling (Lieberman & Lohmander, 2014;Lohmander et al., 2017) was carried out by one observer who listened to each recording and made an assessment on presence or absence of CB, oral stops (including (b/p), (d/t) and (g/k)), anterior oral stops (including (b/p) and (d/t/)), and dental/alveolar oral stops (including (d/t)).In the standardised observation of babbling, these variables are binary and judged as present or not present.The observer then listened to the recordings once again to assess the consonant inventory.While listening, the observer marked consonants produced by the infant in a chart showing the Swedish consonant phonemes.Only consonants that occurred at least twice and in canonical syllables were counted.Then the total number of different true consonants produced by each infant was counted.Since four infants did not produce any consonants at all, this part was only carried out for the 16 infants who produced at least one consonant.
Furthermore, during an additional observation of the recordings, the observer counted every speech-like utterance and every canonical utterance to calculate the infants' canonical babbling ratios, CBR UTTER , which is the proportion of utterances with at least one canonical syllable.The observer counted every utterance and those which were also canonical by making tally marks on a paper.If CBR UTTER ≥15%, the infant was judged to be in the canonical babbling stage (Nyman & Lohmander, 2018).To be judged as an utterance, the vocalisation had to be separated by a breath or a pause in the speech long enough for a potential breath.Screams, laughter, burps, whispers, and other vegetative sounds were not judged as utterances (Nathani & Oller, 2001).An utterance was judged as canonical if it included at least one consonant and one vowel connected with a rapid and smooth transition (Oller et al., 1998).One infant had a noisy breathing, which made it impossible to distinguish breath sounds from speech-like utterances.Therefore, this assessment was carried out only for 19 of 20 infants.
For the reference group, data on CBR UTTER , consonant variables, and number of different true consonants were available from Lohmander et al. (2017).However, presence of CB was parent reported in that study.The assessment on presence or absence of CB based on observation of babbling was therefore made separately for the current study by the same observer as for the EPT group.

Interobserver reliability
Four recordings (20%) were selected randomly to be assessed by a second observer (last author).The percentage of agreement on presence/absence of the four variables CB, oral stops, anterior oral stops, and dental/alveolar oral stops was calculated.For CBR UTTER percentage of agreement on presence (i.e.≥15%) or absence (i.e.<15%) was calculated.The two observers completely agreed on presence or absence for all variables including the calculation of CBR UTTER to be ≥15% or not.For the consonant inventory assessment, the difference between the two observers was at most one consonant.

Intraobserver reliability
A second assessment of four recordings (20%) was done by the first observer.Three of these recordings were chosen because they were perceived as more difficult to assess, while the fourth was selected randomly.The repeated assessment by the main observer showed 100% agreement for all dichotomous variables including the calculation CBR UTTER ≥15%.

Statistical analyses
Descriptive statistics were calculated for all variables.For the dichotomous outcome variables, Fisher's exact test (2-sided) was used to analyse differences between the groups.For the consonant inventory variable, number of different true consonants, Mann-Whitney U test for independent samples was used.For all statistical calculations, IBM SPSS version 27 was used.

Presence of CB
For the EPT group, 14 of 20 infants (70%) produced CB according to the dichotomous assessment by observation, while all 20 infants in the reference group produced CB according to the same method (Figure 1).The difference between the groups was significant (Fisher's exact test p = 0.02).
For CBR UTTER , descriptive statistics are shown in Table 2.For the dichotomous variable of CBR UTTER ≥15%, there was a higher proportion of infants with CBR UTTER ≥15% in the reference group, however the difference was not significant (Fisher's exact test p = 0.155).Distribution of CBR UTTER in the two groups is shown in Figure 1.

Presence of oral stop variables
Oral stops, and with anterior placement were used by 16 infants in the EPT group (80%), while all infants in the reference group used oral stops, and with anterior placement.Oral stops with dental/alveolar placement were present in 15 infants in the EPT group (75%), compared to 19 (95%) of the infants in the reference group (see Figure 2).No significant differences between the groups were found for the three oral stop variables: presence of oral stops (Fisher's exact test p = 0.11), anterior oral stops (Fisher's exact test p = 0.11), and dental/alveolar oral stops (Fisher's exact test p = 0.18).

Consonant inventory measured as number of different true consonants
The EPT group had a significantly smaller consonant inventory (median = 2 (0-4)) than the reference group (median = 5 (3-7)); Mann-Whitney U = 11.5, p = <0.001,see Figure 3.The most frequently produced consonants in the EPT group were bilabial oral stops (p/b) and dental/alveolar oral stops (d/t), which were used by more than 50% of the subjects.For the reference group, (b/p), (d/t), (g/k), and the nasal consonants (m and n) were used by more than 50% of the subjects.55) 56 ( 28) 31% ( 13

Discussion
The present study was designed to investigate presence of CB, oral stop variables significant for typical babbling, and consonant inventory at 12 months corrected age in a group of EPT infants in Sweden.Compared to a reference group of infants two months younger, the EPT group had lower results on all outcome variables.However, there was a large variation in the EPT group, and only two group differences were statistically significant: presence of CB and number of different true consonants.
The first question in this study was related to the presence of CB in EPT infants.Despite the age difference between the groups, there was a lower proportion of infants in the EPT group who produced CB, statistically significant when based on a simple observation of presence or absence of CB.When presence of CB instead was based on calculation of CBR UTTER , the group difference was not statistically significant, presumably because the variation of CBR UTTER was much larger in the EPT group than in the reference group.Furthermore, it is notable that the differences between the two methods were only seen in the reference group where three children differed when assessed with the two methods.In the study group, the assessment result was identical using both methods.The finding of a lower proportion of infants producing CB in premature infants supports the studies by D' Odorico et al. (2010), and by Rvachew et al. (2005) but for the latter study only in a subgroup of less healthy infants with a secondary diagnosis of bronchopulmonary dysplasia.However, this finding is contrary to the Finnish studies by Stolt et al. (2012) and Törölä et al. (2012b) who did not find a lower proportion of CB in premature infants.
Turning to the second question concerning presence of the variables on oral stops: presence of oral stops, anterior oral stops, and dental/alveolar stops described by previous research to be distinctive of typical CB, there were no statistically significant differences between the EPT group and the reference group.However, there was a lower proportion of infants producing them in the EPT group, where 75% of the infants produced all three oral stop variables, even though only 70% produced CB.This result is explained by the fact that two EPT infants produced oral stops but had a low proportion of syllables.Since 20% of the infants in the EPT group did not produce any consonants at all, it is a consequence that a subgroup of them did not produce any of the three reported variables on oral stops.
Regarding the third question on consonant inventory, the current results showed that the EPT group had a significantly smaller consonant inventory at the age of 12 months compared to a reference group of full-term infants at the age of 10 months.This finding supports the results of D 'Odorico et al. (2010) but not the results of Rvachew et al. (2005).What stands out was also that there were four infants in the EPT group who did not use any consonants at all.In the reference group of infants two months younger, all infants used at least three different true consonants.Molemans et al. (2012) showed that smaller sample sizes can result in a lower CBR, and consequently a lower proportion of infants being judged as being in the canonical babbling stage.Although we aimed to collect a data sample of at least 75 utterances per child, this goal was not reached for half of the 20 EPT infants.There are several possible explanations.Some of the EPT infants had significant motor delays, which made the data collection procedure fatiguing.In some cases, unexpected events interrupted the play and we had to finish the recording earlier than what would be optimal, and consequently the data samples became more restricted.These circumstances might have led to underestimation of the CBR and consonant inventory in the group of EPT infants.Yet, there were two EPT infants who produced less than ten utterances although the recordings were long, about 35 minutes.Late talkers are known to be less talkative than typically developing children (Molemans et al., 2012), which means that it can be more difficult to collect a sample of utterances of optimal size.The fact that children at risk for speech and language delays and disorders produce less vocalisations has been described for example for Fragile X syndrome (Belardi et al., 2017) and for autism spectrum disorder (Patten et al., 2014).The relatively low number of utterances in the EPT group can reflect on this group being at risk for speech and language delays.
If the relatively small sample sizes of utterances in the EPT group have underestimated the presence of CB and consonant inventory, the age difference between the EPT group and reference group might instead have led to an underestimation of potential group differences.While the infants in the EPT group were on average 12 months (corrected age) when the babbling data was collected, the typically developing infants in the reference group were two months younger.The production of CB has been shown to be sensitive to small changes in age at this stage (Nathani et al., 2009).This means that an aged-matched reference group might have revealed larger group differences that could not be found in our material.Furthermore, the infants in the EPT group were recorded at home while the infants in the reference group were recorded in a clinical setting.Being in the safe and well-known environment at home might have been a positive circumstance for the EPT group and encouraged them to vocalise more.However, this possible advantage seems negligible for the EPT group since the volubility in this group compared to the reference group was obviously lower.Additionally, there was large variation in number of utterances in the EPT group seemingly without a relationship with length of recording.
Our aim was that the study group should be representative of the population of EPT infants.Compared to the 456 EPT infants in the EXPRESS-study (Serenius et al., 2016), the mean gestational age at birth was very similar (25.4 weeks for EXPRESS, 25.3 for the current study); likewise, the sex distribution was comparable (54% male for EXPRESS, 55% male for the current study).In contrast to the earlier studies by D 'Odorico et al. (2010) and Rvachew et al. (2005) on babbling development in premature infants, comorbidities such as neurological impairments were not exclusionary criteria.This means that the present study was more homogenous in terms of all infants being born EPT, but also more heterogenous in terms of cooccurring neurological sequels and additional diagnoses.The risks of neurodevelopmental disorders increase the earlier premature infants are born, and cooccurring diagnoses can lead to an even greater risk.One can therefore hypothesise that the EPT group in this study is at greater risk for speech and language disorders and delays than the premature infants in the previous published studies.However, our findings of restricted consonant inventory and a smaller proportion of infants producing CB, are similar to the work of D'Odorico et al. ( 2010) who described a group of premature infants who were both healthier and older in terms of gestational age at birth than the EPT infants in the present study.
The restricted consonant inventory and low proportion of infants with CB in the EPT group is not surprising, considering that they have been found to be at risk of speech and language delay in toddlerhood.This risk has for example been described in the EXPRESS study (Månsson & Stjernqvist, 2014).In that study, 399 children born EPT, and a matched control group of 366 children born full-term, were assessed with the Bayley Scales of Infant and Toddler Development (Bayley, 2006) at 2.5 years (corrected age for the EPT children).The EPT children had significantly lower language scores, and about 15% had moderate or severe delay in expressive and/or receptive language functions (test scores at < −2 SD).
Since this study was carried out at one time-point only, further follow-up of speech and language skills in the EPT group is needed to explore whether babbling at 12 months can predict speech and language at an older age in the EPT group.

Conclusion and implications
The present study showed that babbling and early speech in 12-month-old EPT infants differed from a reference group of 10-month-old typically developing infants in significant ways.While no difference in presence of oral stops typical for canonical babbling was found between the groups, the EPT group had a significantly lower proportion of infants with CB and smaller consonant inventory.These differences were found even though the EPT infants were two months older than the infants in the reference group.
To our knowledge, this is the first study focusing on babbling and early speech in a group of only EPT infants.This study provides reference data on expected babbling and early speech in EPT infants at 12 months of corrected age.

Funding
The study was supported by Region Stockholm.

Figure 1 .
Figure 1.Distribution of CBR UTTER in the EPT group (n = 19) and the reference group (n = 20).

Figure 2 .
Figure 2. Proportion of infants in the two groups, EPT group (n = 20) and a reference group (n = 20) producing the dichotomous variables canonical babbling, oral stops, anterior oral stops, and dental/ alveolar oral stops.

Figure 3 .
Figure 3. Number of different true consonants produced by the infants in EPT group (n = 20) and the reference group (n = 20).

Table 2 .
Descriptive statistics on CBR UTTER .