Neural Bases of Proactive and Predictive Processing of Meaningful Subword Units in Speech Comprehension

To comprehend speech, human brains identify meaningful units, like words, in the speech stream. But whereas the English ‘She believed him.’ has three words, the Arabic equivalent ‘sạddaqathu’ forms one word with three meaningful subword units, called morphemes: a verb stem (‘sạddaqa’), a subject suffix (‘-t-’), and a direct object pronoun (‘-hu’). It remains unclear whether and how speech comprehension involves morpheme processing, above and beyond other language units. Here, we propose and test hierarchically nested encoding models of speech comprehension: a NAÏVE model with word-, syllable-, and sound-level information; a BOTTOM-UP model with additional morpheme boundary information; and PREDICTIVE models that process morphemes before these boundaries. We recorded MEG data as 27 participants (16 female) listened to Arabic sentences like ‘sạddaqathu.’ A temporal response function analysis revealed that in temporal and left inferior frontal regions, PREDICTIVE models outperform the BOTTOM-UP model, which outperforms the NAÏVE model. Moreover, verb stems were either length-AMBIGUOUS (e.g., ‘sạddaqa’ is initially mistakable for the shorter stem ‘sạdda’, meaning ‘blocked’) or length-unambiguous (e.g., ‘qayyama’, meaning ‘evaluated’, cannot be mistaken for a shorter stem) but shared a uniqueness point, beyond which stem identity is disambiguated. Evoked analyses revealed differences between conditions before the uniqueness point, suggesting that, rather than await disambiguation, the brain employs PROACTIVE predictive strategies, processing accumulated input as soon as any possible stem is identifiable, even if not uniquely. These findings highlight the role of morphemes in speech and the importance of including morpheme-level information in neural and computational models of speech comprehension.