Phonetic Realization And Phonological Analysis Of Focus In Standard Chinese

The subject matter of this dissertation is the relation between intonational structure, prosodic structure, and focus in Standard Chinese (hereinafter SC). It is an investigation of prosodic effects of various kinds and different numbers of focus in SC, on the one hand, and of the accent patterns conveyed by the focus, on the other. The research examines three aspects: (i) the acoustic manifestations of the entities bearing focus, that is, wh-elicited focus, syntax-marked focus (lian-marked or shi-marked focus), and interaction of these two kinds of focuses; (ii) the phonological representation of the pattern and hierarchical levels of accents induced by the focus in discussion; and (iii) a theoretical explanation of the corresponding relation between focus and accent in SC. To accomplish this aim, this dissertation employs (a version of) the theory of Intonation, as developed by Pierrehumbert and colleagues, Ladd, Grice, Gussenhoven, and others, (a version of) the theory of the Prosodic Hierarchy, as developed in work by Selkirk, Nespor, and Ladd and (a version of) the theory of Optimality Theory, as developed by Gussenhoven and Yip, among others. The chosen approach here is simultaneously theoretically and empirically based, much along the lines of laboratory phonology research, in which experimentally collected speech data is used to investigate questions about the abstract categories of phonological structure.The samples sentences employed in this dissertation are confined to declarative sentences with one or more (single, double or multiple) focuses. The inclusion of focus in the domain of research is motivated by the fact that the focus affects phrasal prominence, prosodic phrasing, and other aspects in intonation in various languages. It is expected that the present examination of focus may provide important evidence for the prosodic organization of SC.Fundamental frequency (F₀), together with duration, is an acoustic measure that is adopted to signal the contrast of the accent patterns. The underlying tonal target H or L is employed in dealing with the phonetic realization. It is mainly concerned with the following aspects: (i) The single focus condition, no matter what kind of the focus is, i.e., wh-operators elicited information focus, Han-marked focus or shi-marked focus, the accent is correlated with the focus. Taking the F₀ into consideration, the syntax-marked focus observes resemble phonetic nature with the wh-elicited information induced focus, specifically, in the under-focus domain, the F₀ ranges of the focused words are expanded as the H and L tones of both focused syllable are raised. The accent conveyed by the syntax-marked focus also exerts compressive effect on the following constituents and this effect can extend to the every end of the sample sentence. The tonal combinations of the focused constituents contribute greatly to the specific manner of the accent manifestation, that is, the accent is realized by modifying the original tonal targets H or L. As for the durational adjustment, the focus can trigger significant effect on the durational lengthening; (ii) In the syntax unmarked sentence, in double focus condition, what is crucially involved in the acoustic reflection of focus is the hierarchical level of accents. In the syntax unmarked sentence, concerning the F₀ and durational pattern, the observations are threefold: (a) different levels of focus can lead to different magnitude of pitch register raising and durational lengthening in one sample sentence, i.e., wh-elicited theme focus and rheme focus, between these two, the primary role is due to the rheme focus; (b) double rheme focus exhibits similar acoustic manifestations, namely, each of the focused unit is characterized by F₀ lifting and durational lengthening; (c) consistent no lowering of the F₀ on the right constituent reveals the fact that the accents consist status difference; (d) the accent of multiple rheme focus displays distinct properties in comparison with the single and double focus condition that the distribution of the accent exhibits no corresponding relation with the focus. Only the rightmost focused constituent serves as the anchor to realize accent in the surface form; (e) The hierarchical level differences of accents are adequately accounted for by the nuclear accent and pre-nuclear accent distinction in SC. The nuclear accent bears the characteristics of obligatoriness and uniqueness, the pre-nuclear tone is of optional and secondary; (f) focus is also assigned a crucial role in determining the phrasing at the intermediate phrase boundary in SC; (iii) In regard with the syntax-marked sentence, acoustic evidences for the double focus are provided based on the relation between the syntax-marked focus and the wh-elicited rheme focus: (a) they can add together to induced nuclear accent correlated with larger F₀ excursion and durational lengthening, and more obvious F₀ drop of the postfocus constituents; (b) they may co-exist with each other on different components, when the syntax-marked focus locates in the proceeding position of the rheme focus, these two focuses can be remarked by pre-nuclear accent and nuclear accent. These two accents also observe phonetically degree-based difference, in that the latter is marked by more obvious phonetic features than the latter. Crucially, when the syntax-marked focus is preceded by the rheme focus, it would be deaccented by the compressive effect of the rheme focus in the foremost position.Based on the empirical investigations as well as the existing literature on the phonological study of intonation, this dissertation further proposes the phonological representation of the linguistically significant non-lexical configurations in SC. The grammar states that the intonation contour consists of four types of categorical phonological events, i.e., nuclear accent, pre-nuclear accent, onset and boundary tone. Among these four entities, the nuclear accent and the boundary tone bears obligatory nature, whereas, the appearance of the pre-nuclear accent and the onset are optional. The intonation pattern is constructed by the phonological events in linear sequence, the events are represented by the primitive tonal targets H(High) and L(Low). The phonetic realization of any given H or L tone depends on a variety of factors (status and numbers of focus, position in utterance, etc) that are essentially important to the identity of those tonal targets. Overall trends of contours (e.g., local range expansion or gradual compression of overall range) mostly reflect the operation of the localized linguistic factors. The essential properties of the nuclear accent and pre-nuclear accent distinguish from each other, former one bears unmarked and potentially unique feature, and the latter is restricted by the focus condition, it only appears in the double focus utterance. Thereafter, it contains status difference between these two accents, they observe primary and secondary status distinction. Evidences of 'focus without accent' and 'accent without focus' in SC further demonstrate that sentence-level accent is not only a matter of where accent is located, but involves a specification of relative 'strong or weak' relation in the prosodic structure, which in turn determines the distribution of the accent in the surface form. Under single focus condition, the metrical relation is 's-w' which determines the accenting and deaccenting phenomenon in the surface form. Under double or multiple focus environment, the metrical structure is captured as 'w-s' which determines the distribution of the nuclear accent in the rightmost position elicited by the focus.Within the framework of Optimality Theory, a ranked the generation of the accent in the surface form is conducted in three steps: (i) the analysis of accent patterns generation, i.e. H*, L*, LH* or H*L; (ii) the association of the accent; (iii) the location of the prosodic boundary.