Maternal speech to infants of 13 months in two different cultural contexts (Padua, Northern Italy and Ruoti, Southern Italy), is examined to verify how cultural variation can influence the contents of that speech. This study seeks to analyze and compare the language directed to infants in terms of two prominent classes of speech - affect salient and information salient. 50 mothers and their firstborn children were individually videotaped in naturalistic free-play interaction at home (30 from Padua and 20 from Ruoti; 25 males and 25 females). Maternal speech was transcribed and coded in terms of the primary function of each utterance. Amounts of affect-salient and information salient maternal speech were analyzed in terms of frequency and were also transformed into proportions. Analyses of variance (ANOVA by sex and site) and analyses of covariance (ANCOVA) with SES, mother age and educational level as covariates, were conducted.The results showed similarity in amount of affect-salient contents of the speech in the two samples. The mothers seem to share feelings and contribute to emotional exchanges via their affect-salient speech to babies. Some differences in the two sample appear in the contents of information-salient language.