20060701 knight linguistic coding
Post on 06-Apr-2018
216 Views
Preview:
TRANSCRIPT
-
8/3/2019 20060701 Knight Linguistic Coding
1/17
The Linguistic Coding of verbal and
non-verbal backchannels:A Preliminary Approach
-
8/3/2019 20060701 Knight Linguistic Coding
2/17
Initial Linguistic Codes- Recap:
Continuers- Maintaining the flow of discourse
Convergence Tokens- Marking agreement and disagreement
Engaged Response Token- High level of engagement, with theparticipant responding on an affective level to the interlocutor
Information Receipt Token- Marking points of the conversation
where adequate information has been received
-
8/3/2019 20060701 Knight Linguistic Coding
3/17
A key concern of this project is to explore how, in conversation,verbal and visual realisations of backchannels interact within andacross such categories (and beyond!).
However in its present form, such a coding scheme is onlyproficient for the use of audio data, spoken linguistic
backchannels forms, and provides no utility in its definitions todefine and encode non-verbal backchannels seen in the imagedata and for integrating this information with the verbal data.
Therefore we need to develop a more all-encompassing codingscheme which can be used for both verbal and non-verbalforms.
Limitations & Future Requirements:
-
8/3/2019 20060701 Knight Linguistic Coding
4/17
PRAAT is a computer program that enables you toanalyse, synthesize, and manipulate speech- so can beused to explore the phonetic patterns of backchannels.
Developed by researchers at the Institute of PhoneticSciences based at the University of Amsterdam
It is free to download online and available for general use
Note on technical restrictions in using our data in PRAAT
Linguistic & Methodological Approach:
-
8/3/2019 20060701 Knight Linguistic Coding
5/17
Methodological Approach:
Aim: to explore any potential relationships between the actualexistence of non-verbal (i.e. head nods) and verbalbackchannels, as well as the duration, pitch and intensity of
these.
We have focused specifically upon our training data
We have explored whether there are (in) consistencies:
Across all instances- with nods, without nod Between those occurring with nods and those occurring
without head nods Within the groups of yeah and mmm (the most frequent
backchannels) specifically
-
8/3/2019 20060701 Knight Linguistic Coding
6/17
Sample size = 100 verbal backchannels, from 275 total 50 co-occur with head nods (43 female, 7 male) 50 occur without head nods (41 female, 9 male)
84 are spoken by the female supervisor
16 are spoken by the male supervisee
We are aware that we need to take into account the fact that the results willobviously vary according to who actually speaks as the phoneticcharacteristics of the student, male will be different to the supervisor, female
Sample information:
-
8/3/2019 20060701 Knight Linguistic Coding
7/17
Sample Data- Backchannel Length (secs):
= with nod = no nod
0
0.5
1
1.5
2
2.5
3
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49
Backchannel Number
Len
gth
(Seconds)
-
8/3/2019 20060701 Knight Linguistic Coding
8/17
Sample Data Results- Length in detail (secs):
Male Data: Female Data:
0
0.5
1
1.5
2
2.5
3
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43
Backchannel Number
L
en
g
th
(seco
n
d
s)
= with nod = no nod
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1 2 3 4 5 6 7 8 9
Backchannel Number
Length
(seconds)
-
8/3/2019 20060701 Knight Linguistic Coding
9/17
Sample Data Results- Pitch (Hz) & Intensity (dB)
0
50
100
150
200
250
300
350
400
1 3 5 7 9 11 13 15 17 19 2 1 23 25 27 2 9 31 33 35 3 7 39 41 43 4 5 47 49
Backchannel Number
Pitch
(dB)
= with nod = no nod
0
10
20
30
40
50
60
70
80
1 3 5 7 9 11 1 3 15 1 7 19 2 1 23 2 5 27 2 9 31 3 3 35 3 7 39 4 1 43 4 5 47 4 9
Backchannel Number
Intensity
(Hz)
-
8/3/2019 20060701 Knight Linguistic Coding
10/17
Sample Data Results- Yeah in detail:
- Focus on FUNCTION, linking back to initial linguisticcodes
- 31 * yeah in the sample, 16 with nods, 15 with no nods
- 6 male (3 + nods, 3 + no nods), 25 female (13, 12)
- 101 in total, making the second most frequentbackchannel (mmm = 102 occurrences)
- Similar investigations have been conducted across eachof the backchannel forms
- Extensions and future plans
-
8/3/2019 20060701 Knight Linguistic Coding
11/17
Sample Data Results- Yeah Focus:
= with nod
= no nod
0
10
20
30
40
50
60
70
80
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Backchannel Number
Intensity(dB)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Backchannel Number
L
en
g
th
(S
eco
n
d
s)
0
50
100
150
200
250
300
350
400
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Backchannel Number
Pi
tch
(Hz)
-
8/3/2019 20060701 Knight Linguistic Coding
12/17
Sample Data Results- Yeah Function Results
= Continuer
= Convergence
Token0
10
20
30
40
50
60
70
80
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Backchannel Number
Intensity
(d
B)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Backchannel Number
L
eng
th
(secs)
0
50
100
150
200
250
300
350
400
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Backchannel Number
Pitch
(Hz)
-
8/3/2019 20060701 Knight Linguistic Coding
13/17
All Sample Data- Function Focus: Length (sec)
= Continuer = Convergence Token
= Engaged Response = Information Receipt
0
0.5
1
1.5
2
2.5
3
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61
Backchannel Number
Length
(Secs)
-
8/3/2019 20060701 Knight Linguistic Coding
14/17
All Sample Data- Function Focus: Pitch (Hz)
= Continuer = Convergence Token
= Engaged Response = Information Receipt
0
50
100
150
200
250
300
350
400
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61
Backchannel Number
Pitch
(Hz)
-
8/3/2019 20060701 Knight Linguistic Coding
15/17
= Continuer = Convergence Token
= Engaged Response = Information Receipt
0
10
20
30
40
50
60
70
80
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61
Backchannel Number
Intensity(dB)
All Sample Data- Function Focus: Intensity (dB)
-
8/3/2019 20060701 Knight Linguistic Coding
16/17
Future Developments:
- To look in more detail at:
- The actual timings of the verbal and non-verbal
backchannels- Where they occur in discourse
- Between what words and lengths of pauses they
occur
This will allow us to examine whether there is a link
between, for example, the time when the interlocutormakes and statement and when a response is made,and the value/ function of the response.
-
8/3/2019 20060701 Knight Linguistic Coding
17/17
Although for the basic PRAAT explorations we have used thesupervision data but we also have collected multiple forms ofother data too for future investigation
Future Developments- Data:
top related