8. Software Framework

The basic software for the research is PAM, a LISP-based interactive graphics environment. PAM stands for PAttern Manipulation, and is an extension of LISP from computing with symbolic expressions to computing with text-graphic forms, [Lakin80a,80c,83a]. The PAM graphics language/system provides tree structured graphic objects together with functions for manipulating them. vmacs, the graphics editor in the PAM environment [Lakin84a,86a,86b], is the means for arranging PAM's objects into visual language phrases. Visual objects created in vmacs can be computed with by PAM functions -- including both the visual objects representing the grammars and the elements in the visual communication objects to be parsed. The vmacs/PAM graphics system is implemented in ZetaLISP on a Symbolics 36xx.

9. Related Work

Use of grammars to analyze formal visual languages was investigated some time ago. Shi-Kuo Chang parsed 2-D mathematical expressions using a ``picture-processing grammar'' [Chang71]; however the grammar itself was in a nonvisual, tabular form and the rules were encoded manually. King Sun Fu extended 1-D string grammar notation with 2-D concatenation operators and graphic objects as the terminal symbols [Fu71]; a graphical grammar to analyze stylized sketches of houses was described, but apparently never implemented. Alan Mackworth has done interpretation of maps sketched freehand on a graphical data tablet; primary local cues are interpreted with help of a grammar-like cue catalog [Mackworth83].

To contrast the present work with these earlier researches, a considerable advantage is claimed for a system supporting grammar notations which are both visual and machine-readable. That is, the linguist can directly input a perspicuous notation using vmacs. Further advantage lies in the fact that performers also use vmacs to generate the images to be parsed.