The Recognition of Verb Roots & Bound Morphemes when Vowel Alternations Are at Play Alexandra Marquis and Rushen Shi Université du Québec à Montréal Introduction For language learners, establishing a vocabulary is complex because it involves learning several levels of representations, including the phonological forms, semantic information, syntactic information, morphological regularities, etc. Even for the early vocabulary items, these representations may take infants several years to be fully acquired. One of the basic problems for the young language learners is finding the forms in the continuous speech stream that correspond to linguistically relevant units such as words and morphemes. This is because most utterances found in parental speech to infants contain multiple words (e.g., Weijer, 1998), and unit boundaries are not marked by pauses in the speech signal (e.g., Cole & Jakimik, 1978). Recent studies show that infants begin to segment word- like forms from continuous speech at around six months of age, using cues such as transitional probabilities between syllables or phonemes, stress, syllabic boundaries, or phonotactic regularities (e.g., Curtin, Mintz, & Byrd, 2001; Jusczyk, Houston, & Newsome, 1999; Nazzi, Dilley, Jusczyk, Shattuck-Hufnagel, & Jusczyk 2005; Saffran, Aslin, & Newport, 1996). Little is known about infants’ processing of bound morphemes. Bound morphemes are of great linguistic importance since they happen to be either morphosyntactic feature markers or derivational affixes. However, these units only occur as part of a word and are often not syllabic (e.g., English -s). Moreover, the acoustic and phonological realizations of these grammatical morphemes are generally reduced (e.g., Shi, Morgan, & Allopenna, 1998), compromising the syllabic boundaries with the roots. In previous studies, infants aged between 16 and 19 months have been shown to be sensitive to grammatical relations of bound morphemes (e.g., English -s, -ing; German -en) with other linguistic elements (Soderstrom, White, Conwell, & Morgan, 2007; Santelmann & Jusczyk, 1998; Höhle, Schmitz, Santelmann, & Weissenborn, 2006), suggesting that they must have recognized the bound morphemes. And in a segmentation study, English-learning infants have demonstrated that they could segment the English morpheme -ing (Mintz, 2004). In that study, infants were familiarized with sentence passages comprising nonsense words ending with -ing or ending with a pseudo-morpheme –dut, and then tested with the words with the – ing and –dut removed. Results reveal that infants preferred listening to the nonsense words affixed with the morpheme –ing during familiarization. These results suggest that infants can recognize bound morphemes despite their reduced forms and weak boundaries. What is not yet known is whether infants can recognize roots and bound morphemes that are greatly altered by morphophonological processes. Bare roots are commonly affixed with various inflectional morphemes marking features such as number, person, tense, or aspect. In certain languages, affixation may alter syllabic boundaries and sound forms roots and bound morphemes. French verb conjugation exhibits rich morphological variations like allophonic alternations and resyllabification. Tense and lax vowels are in complimentary distribution in Quebec French. No tense vowels appear in closed syllables (e.g., /di/ - /dIt/). Verb stems ending with a closed syllable containing tense vowels are subject to this alternation when conjugated with a vowel or vowel-initial suffix that reorganizes the root syllable structure. For instance, the stem [kUt] becomes [ku te] when affixed with the present infinitive and past participle /e/ morpheme. In this case, the syllabic boundary between the root and the suffix is altered by resyllabification process, which also leads to an allophonic alternation between [U] and [u]. A question that arises is whether infants are able to recognize verb stems despite such complex rule-governed alternations. The awareness of suchlike alternation is part of the native language phonology. Evidence exists about infants becoming increasingly sensitive to the phonological patterns of their native language during the first year of life. Infants’ perception begins to focus more on the native language phonemes (e.g., Kuhl, Williams, Lacerda, Stevens, & Lindblom, 1992; Werker & Tees, 1984), as they begin to show sensitivity to the distributional relations between different phonetic sounds such as phonotactic patterns (e.g., Jusczyk, Cutler, & Redanz, 1993; Jusczyk, Luce, & Charles-Luce, 1994; Mattys & Jusczyk, 2001). Infants can also use phonotactic patterns for successful word segmentation (e.g., Mattys & Jusczyk, 2001). The questions of whether and how infants may be processing phonological alternations have received little attention. One study tested infants’ learning of some phonological alternations (White, Peperkamp, Kirk, & Morgan, 2008). In this study, infants were familiarized with na or rot always followed by a stop or fricative initial disyllable (e.g., na boli, rot boli, na zuma, rot suma). Some conditions of familiarization were such that disyllables starting with