Phonetic invariance and phonological stability: lithuanian pitch accents

We argue that phonetically invariant realizations of phonological categories imply the synchronic and diachronic imperviousness of such categories to phonological rules and sound laws. We claim that phonetic invariance is the foundation of phonological stability. The category we discuss in this contribution is the pitchaccent. We provide a parametric phonetic description of this phonological category. By means of a parametrization technique we apply this description to the contrastive pitch-accents of Lithuanian. The statistic differences between acute and circumflex pitch-accents derived by the parametrization provide a basis for the discussion of synchronic and diachronic behavior of the phonetically nonbalanced phonological contrasts. 1. PITCH-ACCENT PARAMETERS A pitch-accent is a characteristic F0 shape realized over a stress bearing unit. Languages in which word stress is characterized by specific F0 contours are called pitch-accent languages. In languages (e.g. Swedish) where a pitch-accent is realized over at most two syllabic nuclei we speak of the syllabic pitch-accent [2]. In languages (e.g. Japanese) where pitch-accent is realized over two units of phonological weight, we speak of moraic pitch-accent languages [10]. In all languages with characteristic intonations (possibly all human languages) pitch-accents are realized over the sentence stress bearing units [13]. Irrespective of the accent bearing unit we use a pitch-accent description as defined by the following set of phonetic parameters [15]: • p: form of movement (-1 ≤ p ≤ 1) This is the most important of pitch-accent parameters as it defines the pitch-accent contour. Typical forms of pitch-accent movement are rise (LH, p=1), fall (HL; p=-1) and rise-fall (LHL; p=0) • d: alignment of accent (in fraction of the stress bearing unit) Alignment is decisive for the perception of pitch-accent. Typically pitch-accents are aligned with phonological heads of stress bearing units (i.e. syllabic nuclei or head morae). These phonological heads are aligned with phonetic steady states. If, however, the alignment does not correspond with the phonetic steady state but with a transition, pitch-accent is perceived not as a movement (LH; HL) but as a level pitch (L; H) [11]. • s: steepness of movement slope This parameter defines the span of movement over the accent bearing unit(s) and is particularly important for the perception of accent in cases where accent is spread over two syllables or two morae • l: accent base (in Hz) • h: amplitude of movement (in Hz) These two parameters define the thresholds of pitch perception and they are particularly critical in the perception of consecutive pitch-accents [8]. The parametric model is represented by a polynomial function. When the parameter values p=-0.5; d=-0.5; s=0.8; h=30 Hz and l=100 Hz are provided, the function defines a pitch-accent like the one in figure 1. Figure 1: Prototypical pitch-accent representation generated by the parametric model. −1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1 100 105 110 115 120 125 130