Understanding long before speaking

Communicative development inventories, questionnaires completed by parents and validated in many languages, measure words understood and words produced separately. The two curves never coincide for several years.

The most widely used, published in the United States in the early 1990s by Larry Fenson and colleagues, served as the model for most adaptations. Relying on the parent has a practical reason: no one-off observation captures all the words a child uses or recognises.

Around sixteen months, understood vocabulary commonly exceeds produced vocabulary several times over. The gap narrows gradually without disappearing: it persists in adults, who understand far more words than they use.

The methods rest on gaze. In the preferential looking paradigm, two images are presented and one of them named, then the time spent looking at each is measured. Gaze directed at the named image is taken as evidence of comprehension.

Those measures are indirect and disputed. A child may look at the named image without understanding the word, through learned association, and the absence of gaze does not establish the absence of comprehension.

Electrophysiological measures, which record the brain's response to an expected or unexpected word, offer another route. They raise the same problem of interpretation: a different response to two stimuli establishes discrimination, not necessarily comprehension in the sense an adult means it.

The normative data of the inventories publish wide percentiles. At eighteen months, the gap between the tenth and ninetieth percentile of produced vocabulary runs to tens if not hundreds of words.

That spread is why clinical thresholds are set very low and coupled with other criteria. An isolated production delay, without difficulty in comprehension, is not enough to characterise a disorder.

Surveys regularly find an association between social background and vocabulary measured at school entry. The association is robust, present in several countries, and its interpretation remains disputed.

An often cited American study had claimed a considerable gap in words heard by background. Later reanalyses and replications concluded that the gap was markedly smaller than the popularised figure, while confirming that a gap exists.

The inventories have been adapted in more than twenty languages, which allows trajectories to be compared. The major milestones are found everywhere, but their timing varies with the properties of the language, notably the position of verbs in child-directed speech.

Those comparisons are a reminder that a norm established in one language does not transfer as it stands. Children exposed to two languages follow yet another timetable, and total vocabulary matters more than that of a single language.