Vocal babbling in a wild parrot shows life history and endocrine affinities with human infants

Abstract

Prelinguistic babbling is a critical phase in infant language development and is best understood in temperate songbirds where it occurs primarily in males at reproductive maturity and is modulated by sex steroids. Parrots of both sexes are icons of tropical vocal plasticity, but vocal babbling is unreported in this group and whether the endocrine system is involved is unknown. Here we show that vocal babbling is widespread in a wild parrot population in Venezuela, ensues in both sexes during the nestling stage, occurs amidst a captive audience of mixed-aged siblings, and is modulated by corticosteroids. Spectrographic analysis and machine learning found phoneme diversity and combinatorial capacity increased precipitously for the first week, thereafter, crystalizing into a smaller repertoire, consistent with the selective attrition model of language development. Corticosterone-treated nestlings differed from unmanipulated birds and sham controls in several acoustic properties and crystallized a larger repertoire post-treatment. Our findings indicate babbling occurs during an early life-history stage in which corticosteroids help catalyse the transition from a universal learning programme to one finely tuned for the prevailing ecological environment, a potentially convergent scenario in human prelinguistic development.

Keywords: Forpus passerinus; corticosterone; cubic clustering criterion; parrot; vocal babbling; vocal learning.

Figure 1.

(a–f) Spectrographic examples of portions of bouts of babbling from each treatment group and period. Roman numerals identify analogues of functional call types: i = nestling contact call, ii = adult contact call, iii = adult warble, iv = alarm. (Online version in colour.)

Figure 2.

Effects of CORT on babbling and selective attrition. (a–c), k-means clustering of babbling repertoires during CORT treatment based on eight spectrographic measurements. (d–f) During the post-treatment period, untreated (control) and sham control groups (oil) showed a reduced number of clusters (optimized k = 17,13, respectively) (d,e), while the CORT-treated group (f) had high diversity (optimized k = 25). Axes are principal components 1 and 2. (g–i) Repertoire sizes of individuals by treatment and treatment period. CORT-treated birds universally increased their repertoire during post-treatment, while controls tended towards decline. Unique symbols and colours differentiate individuals. (Online version in colour.)

Figure 3.

(a) Mean repertoire size per day (±95% confidence interval (CI)) as function of mean total number of elements of babbling produced each day for each treatment group. (b) Mean repertoire size (±95% CI) as a function of age and treatment group. All groups increased repertoires during the treatment period. During the post-treatment period, controls and unmanipulated individuals' repertoires declined, while CORT-treated individuals increased repertoires. k is the number of clusters from k-means cluster analysis. (Online version in colour.)