Introduction: How to Build a Talking Robot - Computational Linguistics and Talking Robots - page 10

Parallel to the work on complexity, the NEWCAT'86 parser was followed
up by the CoL'89 parser, also implemented in LISP. 22 Its purpose was to pro-
vide the earlier parser with a semantic interpretation. Because the far-reaching
ramifications of an agent-oriented approach had not yet become clear, the most
promising goal seemed to show that a strictly time-linear, strictly surface com-
positional syntax can produce traditional semantic hierarchies. 23
The CoL'89 parser provides syntactic-semantic analyses for 421 different
grammatical constructions of English. The derivation of semantic hierarchies,
based on a time-linear, surface compositional LA-grammar, was implemented
using the FrameKit software (Carbonell and Joseph 1986). This worked well
for producing the semantic trees for all constructions of the CoL'89 fragment.
Our next goal was to model (i) language interpretation as an embedding of
language content into the context of use, and (ii) language production as an
extraction of content from the context. 24 For these more demanding tasks, the
CoL'89 approach, based on trees implemented as frames, 25

was not viable.

1.6 Comparison with Other Systems

To enable the DBS robot to freely communicate in natural language, we mod-
eled it after the most evident properties of human language communication. 26
These have been reconstructed in the agent-oriented approach of the S LIM
theory of language 27 (FoCL'99, NLC'06). S LIM provides the theoretical foun-
dation for the innovations and adaptations of the DBS software machine.
The crucial issues distinguishing DBS from other current systems of natural
language analysis may be summarized as follows:

22 Thanks to Jaime Carbonell and the CMT/LTI at Carnegie Mellon University (1986-1989), who made
programming the semantically interpreted CoL parser possible.
The 1983-1986 stay at Stanford University and the subsequent 1986-1988 stay at Carnegie Mellon
University were supported by a five-year DFG Heisenberg grant. The 1988-1989 stay at CMU was
supported by a Research Scientist position at the LCL (Dana Scott and David Evans). Thanks also to
Brian MacWhinney, who supported a fruitful three-month stay at the CMU Psychology Department
in the fall of 1989.
23 Cf. FoCL'99, Sect. 21.4, for a review.
24 See FoCL'99, Sect. 22.2, for a summary.
25 In FrameKit, a desired value is retrieved by specifying the frame name, e.g., Fido , and the slot, e.g.,
has_brothers . The system returns the value(s) of the slot, e.g., Zach Eddie .
In DBS, a desired “next” proplet is activated by specifying its address in terms of the core and
the prn value. These are provided by the “current” proplet. For example, navigating from the current
proplet Fido to the successor proplet Zach would be based on the [has-brothers: Zach] feature and
the prn valuespecifiedinthe Fido proplet. The activation of the Zach proplet provides addresses for
further successor proplets.
26 A guiding principle is form follows function (Cuvier 1817/2009).