Development Inventories to Additional Languages*

Philip S. Dale, Larry Fenson, and Donna Thal

November 3, 1993

These adaptations are motivated by both research and clinical needs. From a research perspective, cross-linguistic research has provided much fruitful data for formulating and evaluating theories of language development (Slobin, 1985; 1992). The extent to which the early phases of language development, in particular, remain similar or diverge in languages of differing structure provides crucial evidence for the existence and nature of basic "operating principles" of language development. Yet cross-linguistic research, like research on the acquisition of English, has been generally limited to the study of very small samples. A comparison of two or three children acquiring one language, with two or three children acquiring a different language, has only limited interpretability in the absence of information about variability among children acquiring those languages. An apparently large contrast between the two groups may simply reflect sampling fluctuation in the two populations; alternatively, a genuine difference may be obscured by sampling variation in the opposite direction. Information is needed on the nature, time course, and stability of individual differences in the acquisition of each language. This basic science understanding of variation is also essential for the diagnosis and remediation of language disorders. Investigation of individual differences necessarily requires large samples of children; parent report is ideally suited for this purpose.

This brief list of suggestions is directed to investigators preparing to develop a communicative development inventory based on parental report.

Each new inventory is necessarily an adaptation, not a translation, of the CDIs. Languages and cultures differ substantially in both the form and content of their communication systems, and there is every reason to believe that even in the earliest phases of development differences will be noticeable in gestural communication, vocabulary, and grammar. For example, Ogura et al. (1993) included bowing as an early-emerging gesture, while Jackson-Maldonado, etl al. (1993) include "tortillitas" (little tortillas), a variant of pattycake used in Mexico. In the domain of vocabulary, it is obvious that there can be major differences in clothing, food, and household items.
Grammatical features are likely to be even more different from one language to another. This raises problems for measuring vocabulary, as well as grammar. Consider, for example, the problem that Italian researchers face in constructing a list of nouns and verbs. Which inflected form of the noun or verb should they use on the word checklist? Caselli and Casadio followed the convention of using singular nouns and verbs in the infinitive form. Languages with a rich inflectional morphology system are especially difficult. The English list includes different entries for "am," "are," and "be." In a language like Italian, listing all forms separately in this way would enormously lengthen the list. In the Sentence Complexity section, assessing grammatical development, the selection of which aspects of grammar to include, and how to capture early syntax and morphology must be done on the basis of knowledge of the acquisition of each language. We believe the forced-choice, sentence pair format developed for the English-language CDI will prove valuable in most languages.
The English-language CDIs have many subparts. The core components of the CDI: Words and Gestures are a 396 word vocabulary checklist (comprehension and production), and a 63 item list of gestures. The core components of the CDI: Words and Sentences are a 680 word vocabulary checklist (production only) and a set of 37 sentence pairs to assess sentence complexity. These have the greatest validity as general measures of language development for English, and similar scales are likely to be the most important in other languages. It is not essential to adapt the other, more minor portions of the CDIs.
It is very important to keep in mind from the outset the need for multiple iterations in the development process. The CDIs represent the culmination of nearly 20 years of research, beginning with interviews. Even as questionnaires, they have evolved through more than half a dozen forms in the past decade. In each cycle of revision, previously collected data have been used to modify, add, or omit selected items in order to improve clarity, internal consistency, and validity. It is best to develop the first version or two on a small scale, concentrating on obtaining the information necessary to revise the inventory before proceeding to a larger-scale norming study.
In the process of iteration, it is most effective to begin with a more open-ended format, in which parents are invited to list additional words and gestures, and perhaps even sentences. In this way, a more inclusive list of potential items appropriate for the widest possible range of young children in the linguistic community of interest will be generated. At each step, information from a modest number of parents across the full age range can be used to modify the inventory on the basis of item frequency, clarity of questions, etc. Information from language samples is also highly valuable for identifying possible additions. Later in the development process it is important to shift to the checklist format for the collection of norms, in order to remove the variance that would otherwise be introduced by parental reporting style and recall abilities.
Items with relatively low frequency are appropriate for the instrument (along with higher frequency items), in order to provide an assessment across the full range of age and language ability. In the development of the English-language forms, only items with an overall frequency of less than 5% were dropped on the basis of frequency alone.
The developers of the English-language MacArthur Communicative Development Inventories had the advantage of being part of a sizable collaborative team, with an adequate amount of time (many years), and access to substantial samples of children. This will often not be the case in other linguistic communities. We believe the highest priority should be given to going through the revision cycle at least twice with a modest number of parents (perhaps 25-30) before attempting to obtain norms from a larger number of children. For the norming process itself, we recommend a minimum of 40 children at each age for which norms are being obtained. (It may not be appropriate or necessary at first to obtain norms at each month of age, depending on the intended use of the instrument.)
Finally, even though numerous studies have documented the validity of the CDIs for English, it is highly desirable to conduct validity studies of the newly adapted forms, comparing parental CDI information with information from structured tests and/or language samples.

* Portions of the material in this paper have been adapted from:

Fenson, L., Dale, P. S., Reznick, J. S., Bates, E., Thal, D. J., & Pethick, S. J. (1994). Variability in early communicative development. Monographs of the Society for Research in Child Development, 59 (5), 1-189.



Last Updated June 2013


copyright 2003 CDI Advisory Board