Supplemental Material for Generic References to Gender Predict Essentialism and Stereotyping Even When They Express Counter-Stereotypic Ideas

In the main text, we include demographic information for the 192 families that completed the initial language session, summarize the study design, and briefly describe our coding protocol.Here, we provide further details on our transcription and speech coding procedures.Although transcription and speech coding was conducted on the full set of (parent and child) utterances, the analyses in this manuscript focus on parent language only.Transcription procedure.All transcription and speech coding was conducted using the Datavyu video coding software (2014).The transcription strategy was adapted from the rigorous procedure used by the PLAY project (Adolph et al., 2019).The initial phase of transcription coding involved marking the timed onset and offset of each utterance, and identifying who the speaker was (child, parent, other).A valid utterance was defined as any unit of speech separated by grammatical closure, intonation, or prolonged pause (the statement "Look at this girl!She's painting her nails," for example, would be coded as two distinct utterances).Only communicative utterances made by real humans were transcribed; any voices originating from a person not present on camera, voice assistants, or instances where non-communicative sounds were made (e.g., unintelligible speech, whistling, sighs), were excluded from transcription.Speech coding procedure.After transcription was complete, a team of trained research assistants conducted speech coding in a series of iterative passes.All coders were trained with the first author using sample videos; all valid session videos (n = 192,