The use and reliability of Tait video analysis inassessing preverbal language skills in profoundlydeaf and normally hearing children under12 months of age

M.E. Tait a, T.P. Nikolopoulos b,*, P. Wells a, A. White a

aThe Ear Foundation, Nottingham, United KingdombDepartment of Otorhinolaryngology, Athens University, 116 George Papandreou Street,Nea Philadelphia, Athens 143-42, Greece

Received 21 February 2007; received in revised form 8 May 2007; accepted 14 May 2007

International Journal of Pediatric Otorhinolaryngology (2007) 71, 1377—1382

www.elsevier.com/locate/ijporl

KEYWORDSHearing impairment;Communication;Child;Preverbal;Hearing aid;Cochlear implant;Observation;Interaction;Deaf;Speech;Outcomes

Summary

Background: Assessment measures in evaluating preverbal skills and their progress invery young deaf children are lacking. However, their importance is highlighted by therecent trend of implanting children under 1 year of age. Tait video analysis is atechnique for assessing preverbal communication behaviours in very young childrenwith hearing impairment and has been found to be strongly related to speechdiscrimination and intelligibility outcomes post-implantation.Aim: To assess feasibility and inter-user reliability of Tait video analysis in assessingpreverbal communication skills in children under 1 year of age.Material and methods: Ten children (five profoundly deaf and five normally hearing)under 1 year of age were assessed by Tait video analysis. Three observers analysed thesamples independently, according to the established protocol.Results: There was complete agreement on 305 judgements and 8 discrepanciesbetween observers over all the measures. Four of the discrepancies occurred in thesamples of deaf children and four in the normally hearing samples. Statistical analysisrevealed that the correlation coefficients between the different observers wereextremely high ranging from 0.94 to 1 (perfect agreement). All of them were foundto be statistically significant ( p < 0.01).Conclusion: The very high rate of inter-observer reliability suggests that the videorecordings of children under 12 months can be scored consistently, and Tait video

* Corresponding author.E-mail address: Thomas.Nikolopoulos@Nottingham.ac.uk (T.P. Nikolopoulos).

0165-5876/$ — see front matter # 2007 Elsevier Ireland Ltd. All rights reserved.doi:10.1016/j.ijporl.2007.05.011

1378 M.E. Tait et al.

analysis is therefore a valid method of monitoring the development of vocal andauditory preverbal skills in very young deaf children, either following cochlearimplantation or using acoustic hearing aids.# 2007 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

Cochlear implantation is now a widely acceptedmethod of management in profoundly deaf children[1—8]. This has brought together surgeons, teachers,and speech/language pathologists in order that thesechildren may reach their potentials regarding thedevelopment of spoken language. However, thedecreasing age at implantation, especially in youngchildren under 1 year following very early diagnosisby newborn hearing screening programs, has broughtclinicians new challenges. It is important to evaluatetheprogress of these children, particularly in thefirstyear after implantation, but of course standard testsand procedures are not possible with children of thisage. Their progress can be monitored through inter-viewswith parents and carers [9,10], and these inter-views, though necessarily subjective, with peoplewho are with the children for all or most of theirwaking hours are very helpful and reveal detail onaspects of communication behaviour which are oftennot observed in the clinic.

Another observational method is to use videorecordings of children in the home environmentinteracting with someone they know well. This hasan advantage over live observation in that the recordof the interaction is permanent and can be viewedrepeatedly. Although children under 12 months havelittle, if any, understanding and use of spoken lan-guage, their development of communication can beassessed by examining their preverbal communica-tion skills. These preverbal behaviours are the nat-ural precursors of language development in allchildren, whether deaf or normally hearing. Prever-bal skills include appropriate eye contact, conversa-tional-style turn-taking, autonomy and auditoryawareness of the sound of speech. They constitutethe normal pattern of language development whichbegins in early infancy.

The method of Tait video analysis is a well estab-lished method for the assessment of preverbal lan-guage skills in children with acoustic hearing aidsand cochlear implants [11]. Measures from the ana-lysis, taken before and 12 months after implanta-tion, have been found to be strongly related tospeech discrimination and intelligibility outcomes3, 4 and 5 years post-implantation [12—15]. Interestin the video analysis technique has led to manytraining courses being provided. Its widespread

use in other countries besides the UK has given riseto questions as to the reliability and repeatability ofthe method. The complexity of detail and the jud-gement required on the part of the analysts couldlead to considerable variation in scoring, whichwould limit the usefulness of the technique [16].A study in 2001 therefore examined the reliability ofthe method across different observers, whoachieved a high degree of consistency on theirassessments of children between the ages of 2and 5 years [14]. However, more and more childrenare now being given implants under the age of 2years, and increasingly under 12 months of age.

The aim of the present study was to assess fea-sibility and inter-user reliability in assessing prever-bal communication skills in children under 1 year ofage.

2. Material and methods

Ten 2 min samples were taken from longer recordingsmade in the home. These were made in good lightingand reasonably quiet conditions. The camerarecorded the child almost full-face, with a profileview of the adult if they were sitting opposite thechild (sitting alongside presented no difficulties ofobservation). In this way, any signs, gestures or facialexpressions could be clearly seen. Children werefilmed in interaction with their mothers, activitiesbeing chosen that were interesting and that wouldpromote interaction. A picture book was alwaysincluded. All the children were under 12 months ofage, ranging from 9 to 11months. Five of the childrenrecorded were profoundly deaf, using acoustic hear-ing aids and awaiting cochlear implantation, and fivewere normally hearing, in the same age range. It wasthought important to include normally hearing chil-dren for two reasons: first, themeasures are based onnormal preverbal development; and second, themeasures include observation of auditory processing,which was less likely to be observed in profoundlydeaf children before implantation, but on which itwas necessary to check reliability across observers.Characteristics of the children are listed in Tables 1and 2. Regarding the five deaf children, all wereprofoundly deaf, as measured by auditory brainstemresponses. All children, deaf and hearing, werewithin thenormal rangeofdevelopment, forexample

Preverbal skills 1379

Table 1 Characteristics of children on the 10 videosamples

Samplenumber

Deaf or normalhearing

Age of child atinitial recording(months)

1 Deaf 112 Deaf 103 Deaf 114 Normal hearing 115 Normal hearing 106 Deaf 97 Deaf 98 Normal hearing 99 Normal hearing 10

10 Normal hearing 11

Table 3 Scoring sheet (observer A, case no. 1)

Vocal turns 5 of 8Vocal autonomy 1 of 8Non-looking vocal turns 3 of 8Gestural turns 2 of 8Gestural autonomy 0 of 8No response 1 of 8

sitting up unsupported by the time they wereassessed. The measures of preverbal behaviours[15] are as follows.

2.1. Turn-taking

Initially, ‘turns’ are identified. These are instanceswhere the child has an opportunity to communicate.Opportunities occur when the adult has left a pause,but also included are instances where the childinterrupts the adult’s communication. The turnsare further classified as vocal, where the child hasused voice to communicate, with or without theaddition of sign, gesture or facial expression; or asgestural, where sign, gesture or facial expressionare used, without vocalisation. Eye contact made bythe child with their mother is classified as gesturalcommunication. Vocal and gestural turn-taking areeach calculated as a percentage of the total numberof turns. Instances where the child does not take theopportunity to communicate are classified as noresponse.

2.2. Autonomy

Children’s use of autonomy in their turns is assessedby counting the number of turns in which theycommunicate something which cannot be directlypredicted from the adult’s preceding turn. As withturn-taking, autonomy can be either vocal or ges-tural (silent). For example, a child may push away

Table 2 Characteristics of the five deaf children in the stu

Case number Age at diagnosis Age at hearing aid

1 2 months 3 months2 3 months 5 months3 3 months 6 months6 2 months 3 months7 1 day 1.5 months

something that is offered and point to a preferreditem. This would be classified as gestural autonomyif done silently, or as vocal autonomy if vocalisationwas also used. Both vocal and gestural autonomy areexpressed as percentages of the total number ofturns.

2.3. Auditory awareness

Auditory awareness of the adult’s speech is mea-sured by the non-looking vocal turn (NLVT). A NLVToccurs when a child vocalises communicatively intheir turn after not being in eye contact with theadult during the adult’s previous turn, no visualclue, such as a sign or a pointing finger, having beengiven, and the child’s vocalisation not occurring as avocal clash with the adult’s. Again, the number ofNLVTs is expressed as a percentage of the totalnumber of turns.

On the 10 video samples the number of turns wasbetween 7 and 11. These turns had first to be identi-fied, then classified as vocal, gestural or no response,and then further classified as showing either vocal orgestural autonomy, or no autonomy. Finally, vocalturns had to be classified as ‘non-looking’ or as‘looking’. Therefore, themaximumnumber of judge-ments that could be made on any one video samplewas between 28 and 44, which would occur if all theturns were taken vocally. If no turn was taken at all,the number of judgements would be between 14 and22. The number of judgements on the 10 videosamples ranged from 15 to 43. The variation in thenumber of judgements is attributable to the varyingnumber of turns, and to the varying number of vocalturns taken by different children.

An example–—scoring sheet is given in Table 3.Statistical analysis used Pearson’s correlation

coefficients.When thedistributionswerenotnormal,

dy

s fitting Onset of deafness Cause of deafness

From birth Congenital3 months MeningitisFrom birth CongenitalFrom birth CongenitalFrom birth Congenital

1380 M.E. Tait et al.

Table 4 Raw data of all children studied as assessed by observer A

Case Vocal turns Vocalautonomy

Non-lookingvocal turns

Gesturalturns

Gesturalautonomy

Noresponse

1 deaf 5 out of 8 1 of 8 3 of 8 2 of 8 0 of 8 1 of 82 deaf 0 of 11 0 of 11 0 of 11 10 of 11 3 of 11 1 of 113 deaf 1 of 11 0 of 11 0 of 11 8 of 11 0 of 11 2 of 114 hearing 2 of 9 2 of 9 0 of 9 6 of 9 2 of 9 1 of 95 hearing 0 of 10 0 of 10 0 of 10 7 of 10 0 of 10 3 of 106 deaf 10 of 11 7 of 11 7 of 11 1 of 11 0 of 11 0 of 117 deaf 0 of 7 0 of 7 0 of 7 5 of 7 1 of 7 2 of 78 hearing 8 of10 1 of 10 8 of 10 2 of 10 0 of 10 0 of 109 hearing 3 of 10 1 of 10 0 of 10 3 of 10 1 of 10 4 of 1010 hearing 7 of 10 3 of 10 5 of 10 1 of 10 0 of 10 2 of 10

Spearman’s rank correlation coefficients were used.Statistical significance was accepted at the p < 0.05level.

3. Results

Statistical analysis revealed that the correlationcoefficients between the different observers wereextremely high ranging from 0.94 to 1 (perfectagreement). All of them were found to be statisti-cally significant (p < 0.01).

The agreements and discrepancies in scoring(Tables 4—6) were as follows:

Identification of turns: there was agreement on97 turns identified by all three scorers. Only obser-ver C and on one only occasion identified (on sample6) an additional turn that was not identified byobservers A and B.

Vocal turn-taking: there was full agreement on 36vocal turns, with no discrepancies.

Vocal autonomy: there was full agreement on 35vocal turns as showing, or not showing, vocal auton-omy. Observer B classified only one turn (on sample1) as showing vocal autonomy, which was not soclassified by observers A and C.

Table 5 Raw data of all children studied as assessed by ob

Case Vocal turns Vocalautonomy

Non-loovocal tu

1 deaf 5 of 8 2 of 8 3 of 82 deaf 0 of 11 0 of 11 0 of 113 deaf 1 of 11 0 of 11 0 of 114 hearing 2 of 9 2 of 9 0 of 95 hearing 0 of 10 0 of 10 0 of 106 deaf 10 of 11 7 of 11 7 of 117 deaf 0 of 7 0 of 7 0 of 78 hearing 8 of 10 1 of 10 8 of 109 hearing 3 of 10 1 of 10 0 of 1010 hearing 7 of 10 3 of 10 5 of 10

Gestural turn-taking: there was full agreementon 43 gestural turns. There were also three discre-pancies on this measure:

� O

se

kinrn

bservers A and C classified a gestural turn (onsample 4) which was classified as no response byobserver B (see also under no response).

� O

bserver C classified a gestural turn (on sam-ple 6) which was not classified as a turn byobservers A and B (see also under identificationof turns).

� O

bservers A and C classified a gestural turn (onsample 9) which was classified as a no response byobserver B (see also under no response).

Gestural autonomy: there was full agreement on43 gestural turns as showing, or not showing, ges-tural autonomy. None of the discrepancies listedunder gestural turn-taking was classified as showinggestural autonomy.

Non-looking vocal turns: there was full agree-ment on the 36 vocal turns as being ‘looking’ or‘non-looking’, with no discrepancies.

No response: there was full agreement that 15 ofthe turns were no response. There were three dis-crepancies:

rver B

gs

Gesturalturns

Gesturalautonomy

No response

2 of 8 0 of 8 1 of 810 of 11 3 of 11 1 of 118 of 11 0 of 11 2 of 116 of 9 2 of 9 1 of 97 of 10 0 of 10 3 of 100 of 11 0 of 11 1 of 115 of 7 1 of 7 2 of 72 of 10 0 of 10 0 of 102 of 10 1 of 10 5 of 101 of 10 0 of 10 2 of 10

Preverbal skills 1381

Table 6 Raw data of all children studied as assessed by observer C

Case Vocal turns Vocalautonomy

Non-lookingvocal turns

Gesturalturns

Gesturalautonomy

No response

1 deaf 5 of 8 1 of 8 3 of 8 2 of 8 0 of 8 1 of 82 deaf 0 of 11 0 of 11 0 of 11 10 of 11 3 of 11 1 of 113 deaf 1 of 11 0 of 11 0 of 11 8 of 11 0 of 11 2 of 114 hearing 2 of 10 2 of 10 0 of 10 7 of 10 2 of 10 1 of 105 hearing 0 of 10 0 of 10 0 of 10 7 of 10 0 of 10 3 of 106 deaf 10 of 11 7 of 11 7 of 11 1 of 11 0 of 11 0 of 117 deaf 0 of 7 0 of 7 0 of 7 5 of 7 1 of 7 2 of 78 hearing 8 of 10 1 of 10 8 of 10 2 of 10 0 of 10 0 of 109 hearing 3 of 10 1 of 10 0 of 10 3 of 10 1 of 10 4 of 1010 hearing 7 of 10 3 of 10 5 of 10 2 of 10 0 of 10 1 of 10

� O

bserver B classified one turn as no response (onsample 4), which was classified as a gestural turnby observers A and C.

� O

bserver B classified one turn as no response (onsample 9), which was classified as a gestural turnby observers A and C.

� O

bserver C classified one turn as no response (onsample 6), which was not classified as a turn byobservers A and B.

Over all, there was complete agreement on 305judgements, and there were 8 discrepanciesbetween observers over all the measures. Four ofthe discrepancies occurred in the samples of deafchildren and four in the normally hearing samples.

4. Discussion

Since Tait video analysis is one of the very fewassessment measures in evaluating preverbal skillsin very young children, high inter-observer reliabil-ity would give significant value in the method andcontribute to its wide use internationally. Theresults of the present study indeed showed a veryhigh degree of consistency across the three obser-vers using TAIT video analysis in children under theage of 12 months, both normally hearing and pro-foundly deaf. All three observers were experiencedusers of the analysis with children over the age of 18months, but observers B and C had less experienceof using the analysis with children under 12 months.In fact, all three observers found little difference inthe analysis procedure with these very young chil-dren compared with children around the age of 2years. The main difficulty, shown in the discrepan-cies in scoring gestural turn-taking, was in decidingwhen the child was communicating and when theywere simply interested in the toy, book or activity.Often the key to this decision was found in thechild’s visual regard. A child could be so engrossed

in the object of discourse that they missed oppor-tunities to communicate; however, a look towardsthe adult, for example to share enjoyment, wasalways interpreted as communication.

The high degree of agreement between theobservers supports the use of video analysis as areliable and consistent method of monitoring youngchildren’s preverbal behaviours. The authors havefound it straightforward to teach the method togroups of educators and therapists in 2-day work-shops. When groups of participants have been askedto transcribe and score the same recordings inde-pendently of the other groups they have achievedsimilar results. The same has been found in work-shops in different parts of Europe, as the procedureis not restricted to English speakers.

In conclusion, video recordings of children under12 months can be scored consistently, and the ana-lysis is therefore a valid method of monitoring thedevelopment of vocal and auditory preverbal skillsin very young deaf children, either followingcochlear implantation or using acoustic hearingaids. In addition, the method should also be con-sidered in monitoring the progress of young childrenwhose language is delayed or deviant.

References

[1] E.A. Beadle, D.J. McKinley, T.P. Nikolopoulos, J. Brough,G.M. O’Donoghue, S.M. Archbold, Long-term functional out-comes and academic-occupational status in implanted chil-dren after 10 to 14 years of cochlear implant use, Otol.Neurotol. 26 (6) (2005) 1152—1160.

[2] T.P. Nikolopoulos, S.M. Archbold, G.M. O’Donoghue, Doescause of deafness influence outcome after cochlear implan-tation in children? Pediatrics 118 (4) (2006) 1350—1356.

[3] H.W. Francis, J.K. Niparko, Cochlear implantation update,Pediatr. Clin. North Am. 50 (2) (2003) 341—361.

[4] T.P. Nikolopoulos, K.P. Gibbin, D. Dyar, Predicting speechperception outcomes following cochlear implantation usingNottingham children’s implant profile (NChIP), Int. J.Pediatr. Otorhinolaryngol. 68 (2) (2004) 137—141.

1382 M.E. Tait et al.

[5] J.B. Tomblin, L. Spencer, S. Flock, R. Tyler, B. Gantz, Acomparison of language achievement in children withcochlear implants and children using hearing aids, J. SpeechLang. Hear. Res. 42 (1999) 497—511.

[6] T.P. Nikolopoulos, D. Dyar, S. Archbold, G.M. O’Donoghue,Development of spoken language grammar followingcochlear implantation in prelingually deaf children, Arch.Otolaryngol. Head Neck Surg. 130 (5) (2004) 629—633.

[7] S.S. Hehar, T.P. Nikolopoulos, K.P. Gibbin, G.M. O’Donoghue,Surgery and functional outcomes in deaf children receivingcochlear implants before age 2 years, Arch. Otolaryngol.Head Neck Surg. 128 (2002) 11—14.

[8] T.P. Nikolopoulos, G.M. O’Donoghue, S.M. Archbold, Age atimplantation: its importance in pediatric cochlear implan-tation, Laryngoscope 109 (1999) 595—599.

[9] A. Lesinsky-Schiedat, A. Illg, R. Heermann, B. Bertram, T.Lenarz, Pediatric cochlear implantation in the first and inthe second year of life: a comparative study, CochlearImplants Int. 5 (2004) 146—159.

[10] A. McConkey Robbins, D.B. Koch, M.J. Osberger, S. Zimmer-man-Phillips, L. Kishon-Rabin, Effect of age at implantation

on auditory skill development in infants and toddlers, Arch.Otolaryngol. Head Neck Surg. 130 (2004) 570—574.

[11] M. Tait, Video analysis: a method of assessing changes inpreverbal and early linguistic communication followingcochlear implantation, Ear Hear. 14 (1993) 378—389.

[12] M. Tait, M.E. Lutman, The prediction value of measures ofpreverbal communicative behaviours in young deaf childrenwith cochlear implants, Ear Hear. 18 (1997) 472—478.

[13] M. Tait, M.E. Lutman, K. Robinson, Preimplant measures ofpreverbal communicative behaviour as predictors of cochlearimplant outcomes in children, Ear Hear. 21 (1) (2000) 18—24.

[14] M. Tait, T.P. Nikolopoulos, M.E. Lutman, D. Wilson, P. Wells,Video analysis of preverbal communication behaviours: useand reliability, Deaf. Educ. Int. 3 (2001) 38—43.

[15] M. Tait, M.E. Lutman, T.P. Nikolopoulos, Communicationdevelopment in young deaf children: review of the videoanalysis method, Int. J. Pediatr. Otorhinolaryngol. 61 (2001)105—112.

[16] S.M. Archbold, M.E. Lutman, T.P. Nikolopoulos, Categories ofauditory performance: inter-user reliability, Br. J. Audiol.(1998) 7—12.