The relationship between child age and mean length of utterance measured in morphemes (MLU) was studied in a sample of 123 middle- to upper-middle-class midwestern children, aged 17 to 59 months, conversing with mothers in free play. A significant correlation was found between age and MLU: r = .88. Age accounted for 78% of the variance when MLU was regressed on age; and MLU accounted for 77% of the variance when age was regressed on MLU. Significant nonlinear components accounted only for an additional 1% and 3% (respectively) of the variance. Ranges within one standard deviation (SD) were estimated for predicted MLUs and derived for predicted ages on the basis of linear regression. MLU increased at an average rate of 1.2 morphemes per year; predicted variability in predicted MLU increased with age.
These findings can play a useful role in three activities when the children studied are from populations similar to the one sampled here, and MLUs are obtained in the same way. They can aid in (1) finding children whose development of productive syntax requires further diagnostic evaluation; (2) finding children at a particular linguistic stage quickly; and (3) predicting the age most likely to be associated with a given MLU. The substitution of local MLU norms, sampled from the population and ages of clinical interest, can further sharpen these evaluations.Submitted on October 8, 1978