Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Cooperative Dialogues While Playing

Adventure
David Albert
July 23, 2020

Abstract
To collect examples of the use of hedging phrases in informal con-
versation, three pairs of subjects were (separately) asked to cooperate
in playing the computer game Adventure. While one typed commands
into the computer, the two engaged in conversation about their op-
tions and best strategy. The conversation was recorded on audio tape,
and their computer session was saved in a disk file. Subsequently, the
audio tape was transcribed and merged with the computer record of
the game, to produce a combined transcript showing what was typed
along with a simultaneous running commentary by the participants.
This report contains the complete transcripts and a discussion of the
collection and transcription methodology.

1 Intent and Methodology


To collect examples of the natural use of hedging phrases in dialogue, I asked
three pairs of subjects to participate in a study. Each pair was asked to sit
at a computer terminal and play the computer game Adventure [Advent75]
for approximately forty-five minutes. One member of the pair was asked
to type, but both were encouraged to discuss the game and the options
available to them. Both their spoken conversation and their interaction with
the computer were recorded, the former on audio tape and the latter, by use
of the script program (which produces a complete transcript of a computer
session), into a disk file.

1
2 TRANSCRIPTION 2

Adventure was chosen as the subject domain for several reasons. First,
and most importantly, it is a game which provides the player with a closed,
well-defined world in which certain commands always produce the same re-
sults, and about which questions can be asked that have definite answers.
Since the purpose of this study was to find out how people expressed uncer-
tainty in themselves or inquired about uncertainty in others, it was important
to provide a domain in which it is possible to be certain as well as uncertain
of a fact, and to recognize whether someone else is correct in their statements
of fact. Secondly, Adventure is a text-based game, making possible the sub-
sequent task of merging the players’ conversations with their actions in the
game, since it was possible to save a complete transcript of everything typed
and all of the computer’s responses. Third, Adventure is well-known among
students in computer science, and provided a domain in which I could easily
find both experts and novices. I call an “expert” one who is familiar with the
game and knows some of the correct strategies for solving the game’s puzzles.
A “novice” is one who has never played the game and is not familiar with
the domain.
Dialogues 1 and 3 are between two experts, while dialogue 2 is between
a novice and an expert. For dialogue 2, I asked the novice to assume the
role of typist; for the other two cases, I allowed the participants to choose
between themselves.

2 Transcription
After each pair had played for forty-five minutes, I transcribed the audio
tape recording of their conversation, and interpolated the typescript of their
computer session with their spoken comments. It was usually, but not always,
possible to tell where the oral conversation fit in the written typescript.
Frequently one or both participants would speak what they were typing, or
would read a message on the screen. In a few cases, especially where a long
string of similar commands were typed during a time when little was said, I
made my best guess as to how long the commands took to type in or as to
where a comment seems most appropriate.
The complete transcripts of Dialogues 1, 2, and 3 are reprinted as an
Appendix to this document. In each, the computer typescript is indented
eight spaces, while all the spoken dialogue is printed flush with the left-hand
2 TRANSCRIPTION 3

margin. In Dialogue 1, the participants are S and T, with S typing. In


Dialogue 2, A, a novice, is typing while C, an expert, is helping. Dialogue
3 is between L and M, with M typing. (The letters used as designators do
not in any way identify the actual participants.)
In the diverse literature using speech transcriptions, no single standard
has emerged for notating the accents, intonation patterns, pauses, interrup-
tions, and other features that are so common to the spoken word and es-
sentially unknown to written language. Nor is there complete agreement on
what these features are. Rather, each transcriber lists the features important
to her particular problem, and chooses symbols for these features which are
suited to the available typography. Deborah Tannen [Tannen84] lists a page
of “transcription conventions” for pauses, stress, pitch, intonation, volume,
and interruptions; and Peter Bull [Bull89] and Gail Jefferson [Jefferson89]
analyze and notate in detail the nuances of conversational interruptions and
silences.
While retaining general information about pauses and interruptions, I
have chosen to ignore many of the subtle distinctions of accent, pitch, and
intonation in my transcription since they are not directly relevant to my
research. Furthermore, my choice of symbols reflects the fact that I was
restricted typographically to the ASCII character set. Thus, for example,
where Tannen [Tannen84] represents interruptions and simultaneous talking
with vertical brackets and a vertical superimposition of the two texts, I use
the in-line symbols described below.

I have used the following notational marks in transcribing the oral dis-
course. A comma (,) indicates an ordinary conversational pause with rising
intonation (suggesting that there is more to come). A double comma indi-
cates a noticeable conversational pause of approximately one half second. A
pair of angle brackets (<>) indicates a longer pause of approximately one
second; multiple pairs are used to indicate pauses of several seconds. Where
a voice is low enough to be inaudible on the tape, I indicate the problem
with [mumble]. A question mark in brackets ([?]) indicates that, although
the voice was loud enough to be heard, I was not certain of the words. The
symbol [ha] indicates a short laugh, and [sigh] an audible sigh. The nota-
tion [reading] indicates that the speaker appears to be reading, sotto voce,
the words on the computer terminal screen.
Where one person speaks throughout, I begin the phrase or passage with
2 TRANSCRIPTION 4

that person’s identifier and a colon (e.g., S:). If the person finishes normally,
I end with a period. Three dots at the end of a phrase or in the middle of a
passage indicates that the speaker’s voice has trailed away; that is, that the
final syllable has been drawn out. In contrast, a sharply-clipped break, in
which the final syllable is cut short, is indicated by a hyphen.
Where someone is interrupted but continues to speak, I enclose the inter-
ruption in square brackets and begin the interruption with the identifier of
the interrupter. For example,

S: How are [T: fine] you?

indicates that S asked “How are you?” and that T squeezed in the word
“fine” approximately between the “are” and the “you” of S’s question.
Where the two speak together, I enclose the words they spoke simultane-
ously in brackets, indicate who spoke what with an identifier and colon, and
separate the words of the two with a double slash (//). Thus, in

S: I guess, um ah, [S:do you want to type // T: we’re at the


end of the] [T: road].

S begins with “I guess, um ah,” at which point he and T speak together for
a few words. T reaches the phrase “end of the” while S is saying “type,”
and then finishes his sentence with the word “road” while S is silent.
I use the asterisk (*) at the beginning and end of a word or syllable to in-
dicate special emphasis.1 I also use the exclamation point and question mark
to indicate the presence of the intonations and emphasis normally associated
with those marks.
1
This asterisk convention comes from the USENET computer research bulletin board,
where participants, restricted to the ASCII character set, cannot use underlining or italics
for emphasis. Interestingly, a standard has developed, despite the absence of any explicit
discussion on the issue, in which three levels of stress are differentiated: asterisks denote
the first level, all-capitals the second, and the two combined denote the highest level of
stress.
The only changes I made to the typed record were to indicate typing
errors that were corrected with the backspace key by replacing the control
code actually stored by the script program with the notation < BS >.
A small number of other unusual situations are described, in brackets.
For instance, at one point the telephone rang, and I note that event with
[phone rings]. I make one or two comments towards the beginning and end
of some of the transcripts; my identifier is D.

3 Conclusion
The complete, combined transcripts, reprinted in the Appendix, show three
pairs of people working cooperatively towards solving the problems presented
in the domain. The dialogues are a rich source of the features normally found
in discourse, and lend themselves to a variety of analyses. Although I have
investigated primarily the use of hedging phrases (which come up frequently
throughout all three dialogues), examples from this corpus have also been
used in an examination of issues in cooperative planning, and in a study of
referring phrases. I expect that other researchers will find additional uses
for the transcripts. They may be obtained as computer files upon request to
albert@endor.harvard.edu.
References
[Advent75] The version used is the 350-point version running in C, written
by William Crowther and Don Woods and standardized around
1975.

[Bull89] Psychological approaches to transcription, Peter Bull, Chapter


7 of [Roger89].

[Jefferson89] Preliminary notes on a possible metric which provides for a


’standard maximum’ silence of approximately one second in
conversation, Gail Jefferson, Chapter 8 of [Roger89].

[Roger89] Conversation: An Interdisciplinary Perspective, Derek Roger


and Peter Bull (eds.), Intercommunication Series, Vol. 3, Mul-
tilingual Matters LTD, Philadelphia PA (1989).

[Tannen84] Conversational Style: Analyzing Talk Among Friends, Deborah


Tannen, Ablex Publishing Corporation, Norwood NJ (1984).

[Taylor87] Analyzing Conversation, Talbot J. Taylor and Deborah


Cameron, Language and Communication Series, Vol. 9, Perga-
mon Press (1987).
A First Dialogue
Both participants in this dialogue, S and T, had played Adventure exten-
sively before this session. S, who was typing, was an undergraduate student
concentrating in Classics. T was a graduate student in Computer Science.
The participants did not know each other before this session.
B Second Dialogue
In this session, A, at the keyboard, had never played Adventure before, while
C had extensive experience with the game. Both were undergraduate stu-
dents in Computer Science. At the time of this study, C was a teaching
assistant in a course that A was taking, and the two knew each other fairly
well.
C Third Dialogue
Both L and M, the two participants in Dialogue 3, had played the game
before this study, but neither had played for several years. M, an under-
graduate student in Computer Science, did the typing during this session. L
was a graduate student in Computer Science, and the two participants had
met on occasion but did not know each other well.

You might also like