zurück
zur Hauptseite
Rheinische Friedrich Wilhelms Universität Bonn
Hauptseminar:
Electronic Discourse WS 00/01
Dozent: Prof. Dr. Schneider
Hausarbeit über das Thema
Gambits in Spoken and Chatroom Discourse
Von Barbara Lier,
1.
Introduction
Chatrooms have
become one of the most popular forms of discourse in the digital age. For many
people they have replaced the traditional social gatherings. They opened up new
ways of communication not only between persons with the same special interest
but also for general entertainment. Despite the fact that the conversation is
written, chatroom conversation shares a lot of
important characteristics with spoken discourse. The possibility of immediate action and
reaction of conversation partners, the informality in style and the possibility
of multiple interlocutors are the most important similiarities.
As this field of research is relatively new, linguists have not yet come up
with a clear classification of chatroom discourse.
Some linguists believe that chatroom discourse is
just an enhanced type of traditional written communication. Others argue that
it is a new form of spoken communication conveyed through a different medium or
constitutes a competely new type of communication.
For this reason it may be useful to compare chatroom discourse with spoken discourse in respect to
certain linguistic phenomena. In this study we focus on the use of gambits in chatroom and spoken discourse. As gambits have a
distinctive function in spoken discourse and typically do not occur in written
conversation, a comparison might help to classify the status of chatroom conversation in linguistic terms.
2.
Thesis
The preliminary thesis of this paper
is that there are less gambits to be found in chatroom discourse than in spoken discourse. There seem to
be both social and technical reasons to support this assumption. First of all
it takes much longer to type an expression than to speak it. For this reason
participants in chatroom discourse usually try to be
as brief as possible in their expressions. Therefore gambits are likely to be
left out in chatroom discourse, as they do not
contribute to the outcome of a conversation. Especially certain categories of
gambits, namely Go-ons, Cajolers, Underscorers, Starters and Asides, are
therefore not likely to be found in chatroom
discourse.
In spoken conversation, people often use gambits to avoid speech pause
because they feel uncomfortable when silence occurs. In chatroom
discourse however, pauses within one turn of one speaker are not discernable.
They cannot be avoided as interlocutors' utterances have to be typed before
they can be seen. The possibility to type statements simultaneously in a chatroom makes the use of Starters pointless because they
are only employed in spoken language to claim the floor.
In addition to that, the etiquette of face to face conversation seems to
be considered less important in chatroom discourse.
This might be due to the physical distance and the fact that most of the
participants have never met face to face. One example for this is the Go-on
which is used to show or at least pretend one's interest in someone's
statements. In face to face communication, breaking off the conversation or
simply leaving the room would be considered rude. However, showing a lack of interest in chatroom discourse does not have to lead to an embarrassing
situation. In this environment, leaving a chatroom or
ending a conversation happens quite frequently and is considered as normal.
That is why it can be assumed that gambits, which "are used as
'discourse lubricants' plugging any conversational gap that might hinder the
smooth flow of an ongoing talk" (Edmondson/House 1981: 69), do not or at
least not very often occur in chatroom discourse.
3.
Gambits – an
Overview
Gambits, unlike
utterances, are no moves in an interactional structure, because they do not
carry the conversation forward to an outcome (cf. 1981: 69). Mostly they are
ritualised, idiomatic expressions, which are used to establish, maintain and
end a conversation (Keller?).
Although expressions
like “um”, “er”, “you know”, “I mean” look meaningless at first, they have an
important function in a conversation as they are used to fill conversational
gaps. Therefore, gambits can be considered as “discourse lubricants” used by
speakers to maintain the smooth flow of an ongoing talk. Often they occur in
connection with turn-taking, when the speaker needs time to formulate his/her
thoughts and react to what the previous speaker has said (cf. 1981: 69).
Gambits are also used as hearer-supportive devices. It is common for a hearer
to say things like “really”, “indeed”, “oh!” while listening – thereby he
clearly supports the speaker by showing an interest for and understanding of
what he says, but does not contribute to the structure of the discourse (cf.
1981: 61). The non-use or inappropriate use of gambits can be regarded as
non-supportive and might lead to a breakdown of communication altogether.
Gambits can either stand
alone, as in the above example, or occur during a speaker’s turn. If several
gambits are used within one turn they can be categorized according to the order
in which they occur. “Uptakers (if they occur) always precede Clarifiers, and
Clarifiers precede Appealers” (1981: 62). The Uptaker refers back to what the
hearer has just said. While producing his/her own move he/she may then use
accompanying Clarifiers and afterwards may append an Appealer, which points
forward to the next move of the hearer.
Generally, gambits are more often used by the speaker when he/she is imposing something on the hearer - as in a Request or a Suggest - or when he/she is disagreeing with him/her (Contra/Counter to what the hearer just said). (cf. 1981: 69)
It is important to note
that gambits are a distinctive feature of spoken discourse and occur less
frequently in written language. This notion is important for the study of chatroom discourse in comparison to spoken discourse, as it
can be argued that chatroom discourse shares
characteristics of both written and spoken discourse.
There are different descriptive models for gambits, one by Keller and one by Edmondson/House. In contrast to Edmondson/House, Keller distinguishes between four different types of gambits: Opener, Links, Responders and Closers. This study will be based on the 1981 model by Edmondson/House and another slightly different older model by House (1980), because they offer a more detailed and precise classification of gambit categories.
3.1 Uptakers
The first main category of gambits introduced by Edmondson/House are Uptakers. An Uptaker serves as an acknowledgement of the preceding utterance made by the interlocutor and prefaces a speaker’s move. The first subcategory would be the neutral Receipt, which basically conveys the message “I have heard what you said.” The Receipt is most often realized by the token “yes” and ”yeah” with falling intonation. Other tokens are: “hmm”, “uhum”, “aha”, “uh” or “I see”, “right”, “okay”.
Example:
A: and when you have
said that you will do something, you should stand by it
B: yes on the other hand I didn’t know about this other thing
till this morning
There are also tokens
that re-represent the original message, as in the example below. To House they
constitute their own category of gambits: the Re-Represent (House 1980:
101):
A: How much maths did
you do in your course
B: How much maths well we had two years of calculus
(1981: 71)
Receipts are used as a ritual affirmation of
the hearer’s social standing, because they are often followed by a Contra to
what has been said before. This might be called the “yes but”-phenomenon.
(1981: 72)
In contrast to the Receipt there is also the
non-neutral, emotive Exclaim.
Using an Exclaim, a speaker
shows his emotional state about what has been said before (1981: 72).
Edmondson/House distinguish between four sets of
tokens used as Exclaims:
Subset A includes tokens
like “oh”, “ah”, “really”, “indeed” that express interest, surprise or
disbelief, as in the example below:
A: John’s coming
tomorrow after all
B: Yeah
I thought he didn’t want to
Categories B and C
include tokens that are used when the previous utterance contains information
that is unwelcome or welcome to the speaker.
Typical tokens for category B are
“oh dear”, “oh lord”, “god help us”, “what a nuisance” and for category C
“super”, “great”, “terrific”, “good”.
Examples:
A: The library’s closed
this afternoon actually
B: Oh dear I’ll have to go tomorrow
then
A: So I managed to do it
on time after all
B: Good
I bet that was a load off your mind
(1981: 72)
Tokens in set D reflect
resignation or anger, for example “damn”, “Jesus Christ”, “not again”. It can
be assumed that due to the anonymity in chatrooms
even more explicit expressions might be used there.
Example:
A: John’s coming
tomorrow after all
B: Oh bloody hell couldn’t you put him
off the man bores me stiff
(1981: 73)
The third subcategory of Uptakers are the Go-ons. They are realized by the same tokens as Receipts and Exclaims but differ from them in their placing. They occur alone during an extended turn-at-talk by the interlocutor, but do not constitute an interruption. (1981: 73) With a Go-on, a speaker does not claim “the floor”. Go-ons usually have a rising intonation. (1980: 102)
Example:
A: So I went up to him
and simply introduced myself
B: Really
A: Yes I thought it
can’t do any harm and anyway in fact he was very pleasant
B: Good God
A: Yes he was really
and in fact he’s invited me out
to lunch for next week
B: Well I’ll be
damned
A: Incredible isn’t it
It will be interesting
to see whether there are any go-ons in chatroom-conversations,
because people only post complete utterances there. It is not the case that
within one turn, other participants of a conversation can simultaneously make
remarks. Therefore it is neither possible nor necessary for the hearer to use
go-ons to support the speaker while he/she is in a “flow” of talking, like in
the above example. Also, it is not possible in a chatroom
to produce an utterance without claiming the floor, as it is the case with
go-ons. Whenever a participant’s comment appears on the screen in a chatroom, he/she starts a new turn.
Go-ons
are part of the “art of good listening”. According to Edmondson/House, they
support the conversation. When they are totally absent, it might be that the
speaker stops his talk altogether.
The
use of a Go-on can be interpreted by the preceding speaker as a request for
Confirmation or Repetition. Then the Go-on initiates a Corrective or Repair
sequence and is no longer a gambit, because it now moves the conversation forward. (1981: 63). House
introduces another category of Uptakers: “Checks”, which are precisely these
requests for confirmation, repeat or clarification. Typical tokens are “Do you
really think so?”, “Could you repeat that, please?”, “I didn’t get that”
Another category introduced by House is the “Agree”. It shows that the speaker accepts the preceding communicative act as right. The token is: “yes, you’re right” (1980: 102)
3.2 Clarifiers
The second main category of gambits are the Clarifiers. Unlike the Uptakers, they do not refer to the previous utterance, but to the utterance that is uttered now. One of the two subcategories is the Cajoler. A Cajoler serves to establish or even restore harmony between the conversation partners. Its function is to make a communicative act more palatable to the hearer, it appeals for agreement or cooperation. Cajolers also have a “fumbling function” for the speaker because he suspects that what he/she is about to say might not be welcome to the hearer, he can “downplay” the impact of what he is saying (1981: 75). Edmondson/House have shown that Cajolers are the most frequently used gambit type in English (there are considerably less Cajolers used in German). “I mean” and “you know” are the most common tokens (1981: 75); the high frequency of these gambits clearly demonstrates their function to fill conversational gaps.
Example:
A: What did you think of
my lecture
B: Well erm it was you know very interesting of course but to tell the truth I thought erm you know
really you might have paid more attention to the audience
Other tokens include
“you see”, “to be quite honest”, “actually”, “in fact”, “really”. They may
occur either in pre-, mid- or post-position with respect to the move they
qualify, as in the examples below (1981: 75).
Well you know John I am rather busy at the
moment
Yeah er look John I am you know rather busy at the moment
Well I am rather busy you know John
Several Cajolers can occur in one sentence,
as the example below shows
(1981: 75):
Well you see John in fact I am rather busy at the moment you know I’d love to really
but erm quite honestly I really can’t at the moment you see.
The second subcategory is the Underscorer. In contrast to the hearer-oriented “Cajoler” it is more message-oriented. It serves to direct the hearer’s attention to a particular point the speaker is making. For that reason, Underscorers are often used in Argumentative discourse. There are a lot of different tokens for Underscorers. Some are placed in pre-position to the utterance underscored, like
“Look I’ll tell you what…“
“Well the thing is…”
“What
I’m really getting at is…”
“The basic problem is this…”
Or in post-position:
“this is
the point”
“that’s the
problem”
“this is
what I’m getting at”
Some examples for the use of Underscorers:
A: He’s a fairy he’s a
fairy all he can do is kick the ball up the field
B: Well look now come off it he’s I mean how come he’ss
captain of the team
A: Well alright then listen we’ll go down to the pub and
talk to him tonight
B: Okay but the thing is you know he may not be
there tonight
(1981: 76f)
3.3 Appealers
The third major category of gambits is the Appealer. An Appealer is used to elicit a reaction to what has just been said by the speaker. The hearer is expected to react with an Uptaker. Usually the speaker seeks to elicit agreement to what has been said, therefore Appealers are quite similar to a request for confirmation. (1981: 77)
Appealers are commonly realized by
tag-questions, such as “isn’t it”,
tags on imperatives like “will you” and also by tokens like “okay”,
“(all)right”, “remember”, “don’t you agree”, or “does he/she/they”.
Non-linguistic signals like “eh”, “uh”, “mhm” also
serve as Appealers.
Examples:
A: John you couldn’t
lend me five pounds could you
B: Erm well I’ve only
got about three pounds on me actually
A: Hm
even that would help
B: Okay well I’ll see
what I can spare is that allright
A: Great
A: Lovely day isn’t it
B: Yes it is
Except for the
tag-questions, all tokens are used with rising intonation. Tag questions with
falling intonation seek to elicit agreement, as in the example below:
A: Nice day isn’t it
B: Hmm lovely
Rising intonation may
indicate uncertainty, especially when the utterance that is referred to seems
to be in contrast with what the previous speaker has said:
A: Let’s take a taxi
B: Cost the earth
wouldn’t it
(1981: 78)
As it is with all
nuances of meaning conveyed by intonation, it remains to be seen how gambits
such as the Appealer are
employed in a chat-room discourse, as there is obviously no way to submit any
intonation of words in a written discourse.
3.4 Other types
There are also gambits
that do not fit into the three main categories, as they are mostly used to fill
gaps and don’t refer backwards or forward or modify a move internally, the
Starter and the Aside. The Starter usually appears at the beginning of a speakers turn-at-talk, but it is not to be confused with an
Uptaker. It basically just constitutes a claim to the “floor”, indicates that
somebody is about to say something. The only tokens are “well” and “now”, as
for example in:
“Well I
think myself…”
“Oh well if you say that
of course…”
“Now you know it’s not
easy…” (1981: 79)
To clarify the
sequential order at the beginning of a turn-at-talk, it is important to note
that a Starter appears after an Uptaker, but before a pre-posed Clarifier, like
in:
I see well I mean are you sure (1981: 79)
Starters can also appear
in mid-utterance, as Re-Starters, for example when the speaker wants to correct
himself:
“One of us ought well probably I ought to tell him the truth”
Starters are likely to
precede “non-cooperative” moves, such as Contras or Counters, in contrast to
Satisfies where they usually don’t occur. (1981: 80)
Examples:
A: Can you let me have a
cigarette
B: Well
erm I’m down to my last one actually
There is also the Aside, which occurs rarely but is quite useful to fill conversational gaps. They can occur anywhere in a turn-at-talk and they appear, as if the speaker was talking to himself. But they also have the function to show that the speaker is conversationally still present, although he is “engaged” with something else at the moment. (1981: 82)
Example:
Yes now when was that I know I’ve got it written down here somewhere
won’t be a minute let me see now ah yes it was in October last year. (1981: 65)
They look like
self-prompts and often stimulate the hearer to provide the prompt in order to
assist the fumbling speaker. Tokens include:
“Erm oh let me see now”
“Yes what’s it called erm”
“Oh it’s erm it’s on the
tip of my tongue”
4.
Research material:
For the purpose of this
study we chose a corpus of about 17,000 words, taken from various sources.
For the
face to face conversation we decided to use the well established Santa Barbara Corpus of Spoken American English. Since most of the chatrooms have several participants we chose the
conversations with the titles "Lambada" and
"Raging Bureaucracy” because they included more than two
participants. We took two chats, one chatroom is called "Parents R Us" from US-AOL,
and another from the IRC (Internet Relay Chat) with the title “Chat Office ". The first one
was logged on
In
order to achieve accurate results, it is important to compare only
conversations taking place between speakers of the same native tongue. The
participants of the conversations taken from the Santa
Barbara Corpus are all native speakers of American English. However, it is a
problem to determine the identity of participants in an internet-chatroom. People do not reveal their identity in chatrooms, they are only known by “nicknames”. Even when
people are willing to give information about their background, there is no way
to verify its accuracy. In order to minimize the probability that there are
utterances of non-native speakers in the research material, we chose one chatroom from US-AOL and also looked for indications that
the participants in the chat were Americans. Examples for these indications
would be the fact that people talk about their salaries in US-Dollars in “Chat Office”. Also, the people in “Chat Office” frequently state that they are at
work at the moment and the chat took place during normal business hours in the
There
are basically two different kinds of chats: Chatrooms
for participants who want to exchange experiences or thoughts about a specific
topic and chatrooms where people talk just for
"fun". We logged one of each category. The "Parents R Us"
chat is visited by young parents who talk/type about problems and experiences
with their children, which is why we thought that it might be comparable to the
"Raging Bureaucracy" conversation. The other one
includes such important topics such as "shlongs
in oatmeal", “piercing” or “masturbating at work”. There may be more
important topics as the ones mentioned before but those kind
of topics occur very often in chats with an open topic.
5.
Results of the
Comparison Between Chatlogs
and Spoken Discourse
Comparing the types of
gambits and their frequency of occurrence in chatroom
discourse and spoken discourse in the material chosen to evaluate, the
following results were found:
5.1 Uptakers:
Receipts:
In the “Parents R Us” chat there
were 4 Receipts, 12 in “Chat Office”, as compared to 36 in the “Lambada”-conversation and 17 in “Raging Bureaucracy”. This
makes a total of 16 Receipts in all chats as opposed to 53 in all spoken
discourse.
Exclaims:
27 Exclaims were found in the
“Parents R Us”-chat, 37 in “Chat Office”. There were 31 Exclaims in “Lambada” and 24 in “Raging Bureaucracy”, making a total of
64 Exclaims in all chats and 55 in all spoken discourse.
Go-ons:
There were 5 Go-ons in the “Parents
R Us”-chat, 10 in “Chat Office” as compared to 44 in “Lambada”
and 24 in “Raging Bureaucracy”, resulting in a total of 15 Go-ons in all chats
and 68 in all spoken discourse.
5.2 Clarifiers
Cajolers:
No Cajolers were found in the
“Parents R Us”-chat, and one was found in “Chat Office”. In contrast to that,
there are 36 Cajolers in “Lambada” and 58 in “Raging
Bureaucracy”. Therefore, one Cajoler occurs in all chats, while there is a
total of 94 Cajolers in all spoken discourse.
Underscorers:
No Underscorers occur in “Parents R
Us”, 2 were found in “Chat Office”. There are 5 Underscorers in “Lambada” and 9 in “Raging Bureaucracy”, making up a total
of 2 Underscorers in all chats and 14 in all spoken discourse.
5.3 Appealers
There are 5 Appealers in “Parents R
Us”, 3 were found in the “Chat Office”, and 7 and 11 in “Lambada”
and “Raging Bureaucracy”, respectively. This makes up a total of 8 Appealers in
all chats and 18 in all spoken discourse.
5.4 Other categories
Starters:
There are 14 Starters in “Parents R
Us”, and 11 in “Chat Office”, compared to 34 in “Lambada”
and 24 in “Raging Bureaucracy”, making up a total of 25 Starters for all chats,
and 58 in all spoken discourse.
Asides:
There are 2 Asides in “Parents R Us”
and no Starters at all in “Chat Office” and in both spoken conversations. This
leads to a total of 2 Asides in all chats and 0 in all spoken discourse.
During the anlysis of the research material a high frequency of “lol”- and “lmao”-expressions
(abbreviations for “laughing out loud” and “laughing my ass off”) in chatroom-discourse was observed. It appears that these “lol”s sometimes have the
function of a Go-on, as they are often uttered between two turns of one speaker
and clearly seem to take a hearer-supportive function. In some cases they also
seem to fulfill the function of an Exclaim, as they
present an emotional reference to the previous utterance and often occur at the
beginning of a turn. Therefore, it seemed appropriate to include these “lol”s into this study. In
order to present a comprehensive picture and due to the fact that “lol” could be meant to represent actual laughter,
all occurances of laughter in spoken discourse are
also listed here.
There are 65 “lol/lmao”s in “Parents R Us”, and 95 in “Chat Office”,
while there are 127 “laughter”s in “Lambada” and 96 in “Raging Bureaucracy”. This makes up a
total of 160 “lol/lmao”s in all chats and 223 “laughter”s in all spoken discourse.
An integral part of chatroom-discourse is the use of so-called “emoticons”, a
sequence of symbols meant to represent various kinds of facial expressions or
emotions, such as the well-known “smiley”: :-).
It is, however, difficult to find an equivalent for these “emoticons” in
spoken discourse. Smiling or other facial expressions or voice intonation could
be considered to have a similar function. Nevertheless, these “emoticons” are a
deliberate expression of somebody’s feelings while facial expressions like
smiling are often made unconciously. During the
research of several chat logs, it became obvious that the usage of emoticons
differs greatly as can be seen in the two chats presented in this paper.
There are only two
“emoticons” used in “Parents R Us”, while there are 70 in “Chat Office”.
There are many reasons
that speak for an inclusion of laughters, “lol/lmao”s and emoticons in this comparison between chat
and spoken discourse, as mentioned above. Nevertheless they seem to be too
vague to justify a quantitive comparison of those
features. The question whether a laughter that might be spontaneously uttered
in spoken discourse has the same quality as a deliberatly
typed “lmo” in a chatroom
would require further research. The same applies to emoticons compared to
“real” smiles, which would additionally require a video recording of the
conversations to include facial expressions in this analysis. In addition to
that, our study is based on Edmondson/House’s classification of gambits which
does include neither laughters nor smiles as gambits.
Therefore, those two groups are not included in our comparing figures 1, 2.1
and 2.2 of this paper.
6.
Conclusions
The total of gambits in
spoken discourse compared to the total of gambits found in chatroom
discourse seems to confirm our preliminary thesis, that there significantly
less gambits to be found in chatroom discourse. In
Chart 1, it can be seen that the total of gambits in spoken conversation is
more than two times greater than the total of gambits found in the chats.
However, there are great
differences in the respective gambit categories. Although most gambit types
occur less frequently in chatroom discourse (as can
be seen in Figure 1), only the number of Cajolers, which was the most
frequently used gambit type in spoken discourse, clearly met our expectations
as it occurred only once in the chatrooms. This might
be due to the fact, that a speaker’s turn in chatroom discourse has to be typed and is only visible to
the other participants, when it is deliberatly sent
to the chatroom. The extremely low number of Cajolers
in chatroom discourse supports our assumption that
the function of Cajolers, which are used to fill gaps within one turn, are
unnecessary in this discourse environment. Also, speakers seem to feel less comitted to make a potential unwelcome utterance more
palatable by using Cajolers. This might be due to the often anonymous
environment in a chatroom.
The other categories -
with the exception of Exclaims - followed the same pattern, although not to the
same degree. The percentage of Underscorers in chatroom
discourse compared to spoken discourse is slightly higher (2 vs. 14), and there
are about a quarter to one half as many Receipts (16 vs. 53), Go-ons (15 vs.
68), Starters (25 vs. 58) and Appealers (8 vs. 18) in chatroom
discourse as compared to spoken discourse.
It is, however, quite
astonishing that 2 Asides were found in the “Parents R Us” chat, while no
Asides were found in the spoken discourse. Asides are generally one of the
rarest types of gambits and usually occur in a longer turn-at-talk of a
speaker. Asides also don’t seem to be a distinct feature of chatroom-conversation,
as there were only 2 Asides found in “Parents R Us” and none at all in “Chat
Office”. Therefore, it can be assumed that the occurrence of Asides in the
“Parents R Us” chat is well within the margin of error, respectively due to
coincidence.
It
has already been mentioned that the number of Exclaims in chatroom
discourse exceeds the the number in spoken discourse
(64 vs. 55). A closer look at these numbers seems necessary: There are 27
Exclaims in “Parents R Us”, a number right between the 31 in “Lambada” and the 24 in “Raging Bureaucracy”. However, there
are 37 Exclaims in “Chat Office” which accounts for the high total of gambits
in all chatroom conversations. Looking at the content
of “Chat Office” it becomes obvious that the discussion that takes place is
very emotional. The aim of the participants is not to exchange information but
to clearly demonstrate their attitude towards different topis
and also to draw attention to themselves. This could
partly explain the high frequency of Exclaims in this particular chat.
Summarizing, the
preliminary thesis has only partly been confirmed. In spite of the initial
assumption that there were almost no gambits to be found in chatroom
discourse, it has turned out that the frequency of gambits is quite
significant. One possible explanation for this surprising fact could be that
participants in a chatroom conversation envision themselves
to be in a face to face conversation. That is why they are tempted to use
gambits, although they are usually superfluous in chatroom
conversations. People who regularly take part in chats might think that they
are actually talking and not typing and therefore change their linguistic behavior accordingly. This might also be due to the higher
degree of informality in chatrooms, as opposed to
other written forms of communictation. It has, for
example, already
been determined that chatroom-conversation is similar
to telephone conversation in several respects. It might be for this reason that
people start to type almost exactly the way they would talk. This could be the
explanation why there are so much more gambits in chats than we initially
assumed. The results of the study seem to further support the thesis that chatroom discourse is positioned between spoken and written
discourse. Although less gambits were found in chatroom than in spoken discourse, the substantial number
of gambits pushes the position of chatroom discourse
a little closer towards spoken discourse.
Bibliography:
Edmondson, W./House, J. 1981: Let’s Talk and Talk about it. A Pedagogic Interactional Grammar of English. München u.a.: Urban & Schwarzenberg.
House, J. 1980: Gambits in deutschen und englischen Alltagsdialogen. In Kühlwein, W./Raasch, A. (eds.): Sprache und Verstehen. Kongressberichte der 10. Jahrestagung der Gesellschaft für Angewandte Linguistik, Mainz 1979. Vol. 2. Tübingen: Narr, 101-107.
Keller, E. 1979: Gambits:
Conversational strategy signals. Journal
of Pragmatics 3, 219-238. (Repr.
in Coulmas (ed.) 1981, 93-113).
Santa Barbara Corpus of Spoken American English - Part
I. University of California Santa Barabara Department
of Linguistics, 2000