“…Note that, in that artificial language, TPs between syllables were higher within word boundaries (TP = 1.0) than across word boundaries (TP = 0.33), thus making the extraction of TPs a reliable cue for word segmentation. Since then, many other works have provided support for the involvement of SL mechanisms in other levels of language acquisition, such as word-referent associations (e.g., Saffran and Estes, 2006 ; Estes et al, 2007 ; Hay et al, 2011 ; Breen et al, 2019 ), grammatical categorization (e.g., Mintz, 2003 ), the establishment of long-distance dependencies in different grammatical structures (e.g., Gómez, 2002 ; Newport and Aslin, 2004 ; Gómez and Maye, 2005 ; Thompson and Newport, 2007 ; Kidd, 2012 ; Hsu et al, 2014 ), and literacy skills (e.g., Arciuli and Simpson, 2012 ; Spencer et al, 2015 ; Sawi and Rueckl, 2019 ; Lages et al, 2022 ). Statistical learning is, thus, assumed as a powerful mechanism that enables children to detect the regularities embedded in the spoken (and written) language even without awareness or intention to do so, and to use that “knowledge” to make predictions about “what comes next,” which not only facilitates language processing but also creates the conditions for children to scale up to the extraction of other (higher) levels of regularities that mastering a language requires.…”