The purpose of this dissertation is to investigate how Korean speakers'' production and perception of the English speech sounds in word boundaries differ from English speakers''.
In order to figure out the difference, speech production and perception experiments were carried out. The results were something like this: The difference between Korean speakers'' and English speakers'' production of a sequence of two English words in the word boundaries is that the duration of Korean speakers'' production of consonants, the VOT, and vowels are relatively longer than that of English speakers''. This can be accounted for by the fact that the Korean speakers participating in the experiment tend to realize the English speech sounds in the word boundaries in a very similar way to the production pattern they acquired when they learned L1(i.e. Korean). It indicates that their L1 affects the production of their English(i.e. L2) speech sounds. This can be accounted for in terms of the phonetic category assimilation(Flege, Schirru, & MacKay, 2003), which refers to a phenomenon in which L2 learners need not create a new phonetic category for L2 when L1 and L2 both possess speech sounds that are close, but are not physically identical, to one another in phonetic space.
In the experiment, the English speakers linked two consecutive words together and produced them as if they were one single cluster. Consequently, phonological processes such as resyllabification, aspiration, and deletion were found in their production. In contrast, liaison did not take place in the Korean speakers'' production. Rather they ended up producing two consecutive words individually, hence distinguishing the word boundaries. Due to this, glottalization, pause, and unreleased stops were commonly found in the Korean speaker''s word boundary production.
This dissertation is also mainly concerned with exploring how the perception of two consecutive words by native Korean-speaking listeners differs from that of two consecutive words by native English-speaking listeners. Unlike words or phrases represented in a written text, there is no clear-cut distinction of word boundaries whatsoever in continuous speech. Listeners have to use various kinds of information, including lexical/semantic cues, syntactic/morphological cues, phonotactic cues, prosodic/rhythmic cues, and allophonic acoustic-phonetic cues, which can function as cues that may be involved in the process of identification of word boundaries(Cutler & Norris, 1988; Gow & Gorden, 1995; Nakatani & Dukes, 1997; McQueen, 1998; Sanders & Neville, 2000; Altenberg, 2005; Ito & Strange, 2009).
Identifying word boundaries is a necessary process for the easy understanding and perception of continuous streams of speech sounds. Specifically, Korean listeners find it difficult to identify English word boundaries in continuous speech unless they are provided with reliable information or clues as to the marking of word boundaries. Despite the fact that the Korean listeners used the same perception cues as the English listeners, the perception rate of the two consecutive words in word boundaries by the Korean listeners were lower than that of the two consecutive words in word boundaries by the English listeners. The main reason for this is that the perception pattern of the former is different from that of the latter. To be more precise, the difference in speech production between the Korean speakers and the English speakers may lead to that in speech perception between those two groups of the speakers. The results of the experiment shows that speech perception is based on speech production, which suggests that the former is correlated to the latter.