Computing Perspective in Dialogue - Computational Linguistics and Talking Robots - page 198

10.4.1 N ONLANGUAGE CONTENT IN THE INTERROGATIVE MOOD

What did you write? STAR − 0
In analogy to 10.1.1-10.1.4, the content and its anchor may be represented
here as the following set of proplets:

10.4.2 A NCHORED STAR-0 CONTENT OF WH INTERROGATIVE

⎡
⎣

⎤
⎦

⎡
⎣

⎤
⎦

verb: write
cat: interrog
sem: past
arg: toi what
prn: 625

S: Paris
T: 1930-07-04
A: Simone de B.
R: J.-P. Sartre
prn: 625

⎡
⎣

⎤
⎦

⎡
⎣

⎤
⎦

noun: toi
fnc: write
prn: 625

noun: what
fnc: write
prn: 625

As in 10.1.4, the connection between the STAR and the content is established
by a common prn value, here 625 . The indexical toi points at the R value of
the STAR-0.
Given that there is no significant time difference between the formation of
the content and its use for language production, there is no need to derive a
separate speaker perspective on the content (in contrast to the transition from
10.1.1 to 10.2.1). Instead, Simone proceeds to realize the surface in 10.4.1
by using (i) 10.4.2 as a STAR-1 content and (ii) the following DBS graph
analysis:

10.4.3 Q UESTIONER AS SPEAKER :DBS GRAPH ANALYSIS OF 10.4.1

(i) semantic relations graph (SRG) (ii) numbered arcs graph (NAG)

write write

3

1 2 4

toi what toi what

(ii) signature

V

(iv) surface realization

1 2 3 4

N N

What did you write_?

As shown in Sect. 7.4, a NAG and a proplet representation of a content (here
10.4.2) are sufficient to quasi-automatically derive an LA-speak grammar
which realizes a corresponding surface.
Now it is J.-P.'s turn to interpret the incoming surface (presented in English):