Abstract

The UK-Communicative Development Inventory (CDI) is an adaptation of the MacArthur-Bates CDI
questionnaires and has been newly developed, standardised and normed for British children between
8 – 18 months. This parent-report instrument assesses children’s communication and
language. Research in other languages has shown that CDI instruments assessing infant language show
good stability with language up to the preschool years on a group level. However,
the prediction of language delays/disorders for individual children was unsatisfactory.

This research examined the predictive validity of the UK-CDI for the first time. The aim was to
establish if the UK-CDI subscales (Gestures, Phrases Understood, Production and Comprehension) were
associated with later language scores up to 36 months, if continuity of language depended on UK-CDI
ability group (low, low-average, average-high, high), if other factors needed to be
considered to predict later language, and if language delay at 24 or 36 months could be predicted
for individual children using the UK-CDI at 12 or 18 months.

Families were from the East Midlands (UK) and took part at four time points (N = 82). Parents
completed the UK-CDI and Family Questionnaire at 12 and 18 months, the Lincoln Toddler CDI at 24
months and the 3-year parent report language measure and the British Ages and Stages
communication subscale at 36 months. At 18 and 24 months, children also participated
in standardised tests assessing language (Preschool Language Scales) and, cognitive and motor
ability (Bayley Scales of Infant and Toddler Development).

Results showed that UK-CDI scores at 12 and 18 months were correlated with language scores up to 36
months, albeit language was more stable from 18 months onwards. The associations were usually
strongest with the same category at the closest follow-up testing. In addition,
stronger correlations were found between UK-CDI scores and other parent-report measures compared to
in-person assessments. When investigating the stability of language separately for ability group, language was most stable for high ability children. Furthermore, high ability children remained significantly better than low ability children on most language measures up to 36 months. However, low ability children improved over time in terms of vocabulary but continued to show slow
grammar development. In addition, early language (UK-CDI scores) was the best predictor of future
language. The other factors (i.e., prematurity, gender, SES, sibling status, ear infections, sleep,
family risk of dyslexia or speech or language problems, cognitive and motor skills) influenced
later language but their contributions were not consistent and depended on the outcome measure used
and the time tested. Therefore, only UK-CDI scores at 12 and 18 months were used to predict
language delay at 24 and 36 months using receiver operating characteristic curves. To achieve
clinically useful levels of classification accuracy, the UK-CDI cut-off scores had to be higher
than the norm-referenced 25th percentile. Depending on the criteria used for delay,
Production and Gestures at 18 months predicted delay at 24 months and Production at 18
months predicted delay at 36 months. Implications of using high cut-off scores were
discussed.