The Parallel Encoding and Target Approximation (PENTA) Model
Xu (2005)
- Communicative functions (far left) are parallel to each other without a hierarchy.
- Each function is coded by an encoding scheme (second left) which specifies how each of the articulatory parameter (middle) will be modified by the function.
- The target approximation (TA) model then generates surface F0 through syllable-synchronized sequential target approximation, using parameters jointly specified by all participating functions.
- Thus surface F0 and its features, including shape, slope, peak alignment, elbow, etc., do not directly correspond to any phonological or phonetic units. Rather, they are the end products of the TA process whose control parameters are specified by the encoding schemes, each corresponding to a specific communicative function.