Neonatal oral behaviors: Their coding and organization

Abstract

A comprehensive system for coding neonatal oral behaviors was developed as part of a larger investigation of neonatal imitation. A review of the literature indicated the need to interpret imitation effects with caution, since most studies report reliability coefficients either in the form of percent agreement or the Pearson r, without correcting for chance agreements. In addition, most authors report coefficients that are based on sum frequencies of individual categories between two raters over a large interval of time, as opposed to specific point-by-point agreements. These coefficients might very well be overestimates of the true reliability.;Data for this investigation was obtained from videotapes of 83 newborns (mean age = 40.4 hours, SD = 21.9). Subjects were presented with at least two of the following three facial gestures in a randomized order: Tongue Protrusion, Mouth Opening, or Passive face, for a period of 160 seconds each. Three pairs of raters coded the infants' oral behaviors using an 11-category system, with the main ones being full, partial, and slight tongue protrusion and mouth opening gestures. Reliability, calculated on a point- by-point basis, using the kappa coefficient to correct for chance agreement, averaged.63 across three pairs of raters for 61 subjects.;Reliability coefficients were also calculated at 80-second intervals for 27 of the 61 subjects for whom reliability was calculated on a point-by-point basis. Comparison of the two methods typically used in previous research demonstrated that the interval method consistently produced higher kappa coefficients than did the point-by-point method. Therefore, the data used to draw conclusions in neonatal imitation studies may contain more error variance than is indicated by the reported reliabilities.;The relationship between the various oral behaviors was investigated through factor analyses conducted for each modeling condition, and across all conditions. Results yielded a consistent factor structure, with tongue protrusion and mouth opening gestures separating into two distinct factors. Although there was no evidence in the factorial structure that imitation actually occurred, these findings suggest that the organization of neonatal oral behaviors makes it possible to activate these gestures differentially.