Temporal interpretation

The first step in phonetic interpretation of a prosodic structure, is to assign a temporal interpretation to each node in the structure. This is done by specifying rules for:

durations for constituents which are treated as terminals in the phonetic interpretation
overlap between adjacent nodes in the prosodic structure

Across syllables, a left-to-right model is used:

the duration of a constituent is the sum of the durations assigned to the daughter constituents, minus the overlap specified at the boundary

Within syllables, a coproduction model is used:

the duration of a constituent equals the duration assigned to the prosodic head of that constituent
weak nodes are overlaid on their strong sister nodes at the left or right edge, depending on whether the configuration is weak-strong or strong-weak, respectively

In the coproduction model the following rules apply:

if no duration is specified for a weak node, the duration is equal to the amount of overlap between the weak node and its strong sister node
if no overlap is specified between a weak and a strong node, the amount of overlap equals the duration of the weak node
a negative amount of overlap specifies the amount of non-overlap between a weak node and its strong sister node

Example:

Duration:

   Nu => 600.
   E => 250.
   F => 200.

Overlap:

   On + Ri => 200.
   A + B => -80.
   E + F => 120.

Analysis

Temporal interpretation

Note: these rules are overly simplistic: normally, rules for duration and overlap make use of feature values encoded on a constituent; also, we don't normally use the category labels A, B, C, D, E and F.

Back to Table of Contents

Arthur Dirksen / adirksen@prl.philips.nl / January 1995