Syllable structure and gemination in Maltese

Little is known on the phonetics and the phonotactic constraints of Maltese. This dissertation sheds light on aspects of syllable structure and geminates in Maltese in order to contribute to understanding how the sound system of the language is structured. The study begins by describing the possible syllable structures in Maltese, carefully defining the onsets, nuclei and codas attested in its syllables. Furthermore, the syllabification processes employed in Maltese are discussed. \nThe dissertation then moves on to its primary focus: geminates in Maltese. In relation to syllable structure, geminates in word-medial position are considered to be ambisyllabic, however, the syllable affiliation of word-initial and word-final geminates is under scrutiny. In addition to word-medial geminates, Maltese also has word-initial and word-final geminates. Previous descriptions of Maltese word-initial geminates (e.g, Azzopardi 1981) have claimed that such geminates are preceded by the epenthetic vowel [ɪ]. Based on a series of production studies, I provide acoustic evidence to examine the articulation of word-initial geminates, and show that this epenthesis occurs almost always when the preceding word ends in a consonant. However, when the preceding word ends in a vowel, there are a number of strategies which speakers employ. Subsequently, in a perception experiment, I show that native speakers of Maltese are insensitive to true word-initial geminates (#ss); results indicate that native speaking Maltese listeners could not discriminate between true word-initial geminates (#ss) and word-initial singletons (#s). However, they were able to discriminate between word-initial geminates that were preceded by the epenthetic vowel (#ɪss) and word-initial singletons (#s). Therefore, I argue that this vowel is part of the phonological representation of word-initial geminates, and I discuss implication of these results for lexical access. \nIn addition, I compare word-initial and word-medial geminates and word-final and word-medial geminates. As expected, the most robust correlate is constriction duration as geminates are always longer than singletons. Other correlates, such as voice onset time, does not serve as a correlate to gemination in Maltese. However, the duration of the vowel before word-medial geminates is shorter than the vowel before word-medial singletons and this can serve as a correlate to gemination in production. \nFinally, I address the consequences that these results have for phonological representation. Following the current literature on gemination, I propose a moraic representation for geminates in Maltese, regardless of their position in the word.