Exploring Speech Technologies for Language Learning

The teaching of the pronunciation of any foreign language must encompass both segmental and suprasegmental aspects of speech. In computational terms, the two levels of language learning activities can be decomposed at least into phonemic aspects, which include the correct pronunciation of single phonemes and the co-articulation of phonemes into higher phonological units; as well as prosodic aspects which include  the correct position of stress at word level;  the alternation of stress and unstressed syllables in terms of compensation and vowel reduction;  the correct position of sentence accent;  the generation of the adequate rhymth from the interleaving of stress, accent, and phonological rules;  the generation of adequate intonational pattern for each utterance related to communicative functions; As appears from above, for a student to communicate intelligibly and as close as possible to native-speaker's pronunciation, prosody is very important [3]. We also assume that an incorrect prosody may hamper communication from taking place and this may be regarded a strong motivation for having the teaching of Prosody as an integral part of any language course. From our point of view it is much more important to stress the achievement of successful communication as the main objective of a second language learner rather than the overcoming of what has been termed “foreign accent”, which can be deemed as a secondary goal. In any case, the two goals are certainly not coincident even though they may be overlapping in some cases. We will discuss about these matter in the following sections. All prosodic questions related to “rhythm” will be discussed in the first section of this chapter. In [4] the author argues in favour of prosodic aids, in particular because a strong placement of word stress may impair understanding from the listener’s point of view of the word being pronounced. He also argues in favour of acquiring correct timing of phonological units to overcome the impression of “foreign accent” which may ensue from an incorrect distribution of stressed vs. unstressed stretches of linguistic units such as syllables or metric feet. Timing is not to be confused with speaking rate which need not be increased forcefully to give the impression of a good fluency: trying to increase speaking rate may result in lower intelligibility. The question of “foreign accent” is also discussed at length in (Jilka M., 1999). This work is particularly relevant as far as intonational features of a learner of a second language which we will address in the second section of this chapter. Correcting the Intonational Foreign Accent (hence IFA) is an important component of a Prosodic Module for self-learning activities, as categorical aspects of the intonation of the two languages in contact, L1 and L2 are far apart and thus neatly distinguishable. Choice of the two languages in contact is determined mainly by the fact that the distance in prosodic terms between English and Italian is maximal, according to (Ramus, F. and J. Mehler, 1999; Ramus F., et al., 1999).