# Segmentation ## Where to find files: * **Main folder**: Manuscript with a detailed desciption of the theoretical background, hypotheses, methods, and results of the study (`segmentation_manuscript.pdf`) with appendices. * **Hypotheses**: Theoretical and operational hypotheses. * **Stimuli**: Audio and visual files as they were presented in the familiarisation and test phase. A detailed description of how we edited the files can be found in the manuscript. * **Data**: Raw data, processed data, data aggregated by participant, data, and group summary statistics. * **Results**: Outcome of the *t*-test and equivalence testing performed. * **Code**: Repository hosted in GitHub that (if everything goes well, should be completely reproducible). It contains: * **R scripts** used to import and process data, and to perform statistical analyses * **RMarkdown** files used to generate he manuscript and appendices. * **Communications**: Poster presented at the Workshop on Infant Language Development (WILD) 2019 in Potsdam (Germany). ## Summary of the study This experiment was the first of an experimental series aimed at investigating how **Catalan-Spanish bilinguals** infants encode new word forms, and how they deal with phonological overlap between translation equivalents. We tested whether Jusczyk and Aslins (1995)'s **Head-turn Preference Procedure (HPP)** was a suitable paradigm to our aims. Due to the lack of evidence of natural speech segmentation in our sample, we concluded that the use of this paradigm did not suit the aims of this study. This study ws aimed at investigating whether **monolingual and bilingual 8 months-old infants** processed phonemic vowel exchanges differently when embedded in similar word forms. In the **familiarisation phase**, infants listened for 2 minutes to several sentences in their dominant language (either Catalan or Spanish). The sentences embedded a made up word (*gon*, *mus*, *for*, or *pul*). Half of the infants listened to sentences embedding *gon* and *mus*, whereas the other half listened to sentences embedding *for* and *pul*. At **test**, all infants listened to the four words in isolation. Two of them were familiar, two of them were novel. At the beginning of each test trial, a rotating pinwheel was displayed in the central screen to grab the infant’s attention. Once the infant fixated the pinwheel, the pinwheel disappeared from the central screen and appeared randomly in either the right or the left screen. Once the infant fixated at the side pinwheel, one of the four words was played repeatedly with 500 ms interval between repetitions until the infant looked away for more than 2000 ms At this point, the pinwheel disappeared and the word stopped being played, finishing the trial. In each trial, the experimenter coded on-line how long the infant looked to the pinwheel while the word was being played. We compared infants’ looking time in trials in which a familiar word was played and in trials in which an novel word was played. Differences in looking time towards familiar vs. novel words would suggest that infants recognised familiar words. This experiment was part of **Mireia Marimon**'s and **Gonzalo García-Castro**'s MSc theses at Pompeu Fabra University. The first was supervised by Prof. Nuria Sebastian-Galles. The second was co-supervised by Dr. Santolin and Prof. Nuria Sebastian-Galles.