1
00:00:00,278 --> 00:00:02,149
[ELECTRONIC MUSIC]

2
00:00:11,266 --> 00:00:14,888
Live Text Access: Training
for real-time intralingual subtitlers.

3
00:00:14,888 --> 00:00:17,191
This is Unit 2.
Linguistic competences.

4
00:00:17,191 --> 00:00:20,528
Element 3. How to cope
with speech related challenges?

5
00:00:22,181 --> 00:00:25,950
Speech-related challenges,
exit strategies and readability standards

6
00:00:25,950 --> 00:00:28,589
Created by UAB and SSML

7
00:00:30,513 --> 00:00:33,051
The learning outcomes for this unit are:

8
00:00:33,051 --> 00:00:35,623
Identify speech-related challenges

9
00:00:35,623 --> 00:00:38,653
when producing verbatim
and sensatim subtitles,

10
00:00:38,653 --> 00:00:44,069
apply exit strategies when necessary
and comply with readability standards.

11
00:00:46,305 --> 00:00:49,555
In this unit we will be dealing
with the following points:​

12
00:00:49,555 --> 00:00:53,477
Real-time intralingual subtitling
and simultaneous interpreting,

13
00:00:53,477 --> 00:00:56,549
exit strategies,
readability standards,

14
00:00:56,549 --> 00:00:59,616
main challenges
and we will finish with a summary.

15
00:01:02,382 --> 00:01:05,988
Real-time Intralingual Subtitling
and Simultaneous Interpreting.

16
00:01:06,501 --> 00:01:08,777
Live subtitling
versus interpreting

17
00:01:08,777 --> 00:01:11,892
Interpreters undertake
various cognitively demanding tasks

18
00:01:11,892 --> 00:01:16,187
such as retrieving, retaining,
producing and monitoring information

19
00:01:16,187 --> 00:01:17,480
almost concurrently.

20
00:01:17,725 --> 00:01:19,398
Realtime intralingual subtitling

21
00:01:19,398 --> 00:01:21,720
is sometimes linked
to simultaneous interpreting

22
00:01:21,943 --> 00:01:24,375
according to Carlo Eugeni 2008.

23
00:01:24,697 --> 00:01:28,994
Real-time subtitlers perform
two major tasks simultaneously,

24
00:01:28,994 --> 00:01:33,116
comprehending the audio comments
and producing subtitles.

25
00:01:33,531 --> 00:01:36,770
Real-time intralingual subtitlers
should pay attention

26
00:01:36,770 --> 00:01:39,375
not to overload the focus of attention

27
00:01:39,375 --> 00:01:41,617
and the language processing
and retrieval

28
00:01:41,617 --> 00:01:45,884
should be completed before
the working memory fades away.

29
00:01:46,884 --> 00:01:48,853
There are mainly three big skills

30
00:01:48,853 --> 00:01:51,315
that real-time intralingual subtitlers
have to poses:

31
00:01:52,002 --> 00:01:55,865
Phonetic or ortotypographic,
synthetic and psychocognitive.

32
00:01:56,091 --> 00:01:58,551
First, phonetic or ortotypographic.

33
00:01:59,020 --> 00:02:01,213
In the case of respeakers
every single word

34
00:02:01,213 --> 00:02:03,704
has to be pronounced 
in the clearest way possible.

35
00:02:03,972 --> 00:02:07,957
taking into consideration 
non-gramatical aspects of language

36
00:02:08,135 --> 00:02:10,871
that may cause a bad recognition
by the software.

37
00:02:11,536 --> 00:02:16,294
In the case of fast keyboard professionals
such as velotypists  

38
00:02:16,294 --> 00:02:18,284
they have to master spelling.

39
00:02:19,118 --> 00:02:20,750
Second, synthetic. 

40
00:02:21,464 --> 00:02:27,576
Deaf, read subtitles at a slower rate
in comparison to normal hearers

41
00:02:27,955 --> 00:02:33,049
Realtime subtitlers have to compress
the ST syntactically and semantically  

42
00:02:33,487 --> 00:02:36,408
trying not to lose any relevant information.

43
00:02:37,309 --> 00:02:39,370
Third, psycho-cognitive 

44
00:02:39,733 --> 00:02:44,137
Realtime subtitlers need to have
SI competences

45
00:02:44,137 --> 00:02:48,144
since they have to speak
while listening to the ST.

46
00:02:48,746 --> 00:02:49,887
Working memory.

47
00:02:50,445 --> 00:02:53,917
Realtime intralingual subtitling
poses a serious challenge 

48
00:02:53,917 --> 00:02:55,917
for an individual's cognitive load

49
00:02:56,802 --> 00:03:01,337
In this sense the working memory
is a limited-capacity mechanism

50
00:03:01,605 --> 00:03:03,956
that is involved in both the processing

51
00:03:03,956 --> 00:03:06,831
and the storing
of currently active information

52
00:03:07,077 --> 00:03:08,982
while tasks are being carried out.

53
00:03:10,000 --> 00:03:12,258
An example of working memory is:

54
00:03:12,792 --> 00:03:18,861
Peter gets two apples from his mother
every day, but eats only one per day.

55
00:03:19,482 --> 00:03:22,557
How many apples does he have
at the end of the third day?

56
00:03:23,511 --> 00:03:25,858
This classic example
of mental arithmetic

57
00:03:25,858 --> 00:03:30,159
is a good illustration
of what working memory is for.

58
00:03:30,382 --> 00:03:34,807
In addition to the working memory,
the work of the real-time subtitler

59
00:03:34,807 --> 00:03:39,801
involves the four main activities
formulated by Jones in 2002

60
00:03:40,000 --> 00:03:41,571
for simultaneous interpreting

61
00:03:41,571 --> 00:03:46,300
which are: listening, understanding,
analysing, and re-expressing.

62
00:03:47,594 --> 00:03:52,663
Understanding refers to ideas
that have to be rendered as a ST.

63
00:03:52,663 --> 00:03:56,527
Obviously, the real-time intralingual
subtitler cannot understand ideas

64
00:03:56,527 --> 00:04:01,869
if he/she does not know the words
the speaker is using to express them.

65
00:04:02,423 --> 00:04:05,000
In the case of analysing,

66
00:04:05,000 --> 00:04:10,109
real-time intralingual subtitlers
may proceed to analyse the speech.

67
00:04:10,718 --> 00:04:13,627
Speeches may be of different kinds,

68
00:04:13,737 --> 00:04:20,109
such as reasoned, logical, argumented, narrative, chronological sequence, descriptive,

69
00:04:20,443 --> 00:04:22,058
purely rhetorical, under others.

70
00:04:22,817 --> 00:04:26,663
Real-time intralingual subtitlers
should differentiate

71
00:04:26,663 --> 00:04:29,536
between what is important
in the speakers utterances

72
00:04:29,536 --> 00:04:30,995
and what is secondary,

73
00:04:31,196 --> 00:04:34,839
and what is essential
and what accessory.

74
00:04:35,673 --> 00:04:40,468
Real-time subtitlers need to reformulate
the wording of the original

75
00:04:40,468 --> 00:04:43,459
to maintain the right distance
from the speaker.

76
00:04:43,638 --> 00:04:45,556
Long and complicated sentences

77
00:04:45,556 --> 00:04:48,933
are to be broken down
into easier and shorter ones.

78
00:04:49,356 --> 00:04:54,338
Relative and subordinate clauses
can be shifted around

79
00:04:54,338 --> 00:04:58,832
within a sentence or passive clauses
changed into active.

80
00:04:59,367 --> 00:05:04,541
Real-time subtitlers must use
reformulation as the tool

81
00:05:04,541 --> 00:05:08,182
that will enable them to deal
with all kinds of difficulties

82
00:05:08,182 --> 00:05:12,298
while remaining as close as possible
to the utterances of the speaker.

83
00:05:12,790 --> 00:05:15,971
To this aim, some exit strategies
can be applied.

84
00:05:16,730 --> 00:05:17,812
Exit strategies

85
00:05:20,273 --> 00:05:23,840
It is important to highlight
that there are no exact rules

86
00:05:23,840 --> 00:05:27,712
as to when to condense
and reformulate, or when to omit.

87
00:05:29,882 --> 00:05:32,666
Long and complicated sentences
are to be broken down

88
00:05:32,666 --> 00:05:35,409
into a series of easier, shorter ones.

89
00:05:36,243 --> 00:05:38,148
This is the case of reformulation.

90
00:05:38,943 --> 00:05:41,332
One of the most used techniques
in terms of reformulation

91
00:05:41,332 --> 00:05:45,000
is the so-called “salami” technique proposed by Jones in 2002.

92
00:05:45,212 --> 00:05:49,031
The author refers to salami technique
as a way of slicing up a sentence

93
00:05:49,031 --> 00:05:51,309
like cutting slices of salami.

94
00:05:51,309 --> 00:05:54,710
Speakers, usually formulate long,
complicated sentences.

95
00:05:54,710 --> 00:05:59,266
The real-time intralingual subtitler
must divide up the speaker’s sentences

96
00:05:59,266 --> 00:06:05,171
into a number of short, self- contained
and then link them as appropriate.

97
00:06:06,143 --> 00:06:09,798
Real-time intralingual subtitlers
are constantly under pressure

98
00:06:09,798 --> 00:06:13,085
with the added burden of not knowing
where the speaker is going.

99
00:06:13,398 --> 00:06:18,009
The need to be able to reformulate
so as to express oneself efficiently

100
00:06:18,009 --> 00:06:19,698
exists for all languages,

101
00:06:19,698 --> 00:06:25,413
but for some languages
it is more acute than others.

102
00:06:26,298 --> 00:06:31,650
Simplification consists in a lexical
or stylistic simplification of the original.

103
00:06:31,650 --> 00:06:36,647
A real-time subtitler is sometime faced
with highly technical material.

104
00:06:36,647 --> 00:06:40,662
For such speeches, it is desirable
to simplify the speech

105
00:06:40,662 --> 00:06:43,348
through generalisation or deletion.

106
00:06:43,804 --> 00:06:48,950
Generalisation consists in 'buying time'
by producing generic utterances

107
00:06:48,950 --> 00:06:53,493
such as replacing a segment
with a general term or idea unit.

108
00:06:53,597 --> 00:06:58,131
Deletion consists in an omission
of superfluous information.

109
00:06:58,561 --> 00:07:02,257
These techniques might apply
if the speech is too technical.

110
00:07:02,898 --> 00:07:10,518
A real-time intralingual must try to save
the essentials by simplification.

111
00:07:10,780 --> 00:07:13,112
Second, the speaker may be talking
too fast

112
00:07:13,112 --> 00:07:17,519
or in some contexts different speakers
might be speaking at the same time.

113
00:07:18,063 --> 00:07:23,060
The faithful rendering of the message
would just leave the audience confused

114
00:07:23,283 --> 00:07:25,092
also due to delays.

115
00:07:26,646 --> 00:07:31,371
Addition, real-time subtitlers
maybe faced with notions,

116
00:07:31,371 --> 00:07:36,081
cultural, or institutional references
that are not known by the audience

117
00:07:36,081 --> 00:07:37,465
and should be explained.

118
00:07:37,704 --> 00:07:40,169
The problem is that this takes time.

119
00:07:40,857 --> 00:07:45,213
Real-time subtitlers can use additions
in order to clarify the message

120
00:07:45,213 --> 00:07:51,133
and explain the meaning of a term
or wording, such as an acronym,

121
00:07:51,400 --> 00:07:57,030
the first time it is mentioned,
and after refer to it an abbreviated form,

122
00:07:57,030 --> 00:07:58,746
thus saving time.

123
00:07:58,746 --> 00:08:01,880
Addition can be used also
to maintain the coherence.

124
00:08:02,526 --> 00:08:05,550
In terms of anticipation real-time
intralingual subtitlers

125
00:08:05,550 --> 00:08:09,552
should take anticipation
when they often begin a sentence

126
00:08:09,552 --> 00:08:12,786
without knowing exactly
where the sentence is going.

127
00:08:12,999 --> 00:08:16,897
The anticipation can be possible
from the context of a meeting.

128
00:08:17,508 --> 00:08:21,712
For instance, a discussion
or negotiations, delegations...

129
00:08:21,712 --> 00:08:24,448
Positions or arguments
will become known,

130
00:08:24,871 --> 00:08:30,259
return to points already made
by the same or other speakers.

131
00:08:30,896 --> 00:08:33,249
It should be stressed
that these exit strategies

132
00:08:33,249 --> 00:08:35,349
are just means, not ends.

133
00:08:35,349 --> 00:08:42,718
Exit strategies are not and end in itself,
but make real-time subtitling possible.

134
00:08:43,123 --> 00:08:48,047
They are tools that allow professionals
to master their own decalage.

135
00:08:48,539 --> 00:08:52,839
The decalage is the length of time
between the start of the speech

136
00:08:52,839 --> 00:08:56,235
and the beginning of the text
shown on the screen.

137
00:08:56,733 --> 00:09:02,527
Long decalage allows higher accuracy
but results in long delays.

138
00:09:03,432 --> 00:09:05,691
Such exit strategies
have the potential

139
00:09:05,691 --> 00:09:08,853
to reduce the cognitive load
of real-time intralingual subtitlers

140
00:09:09,157 --> 00:09:11,606
by allowing professionals
to concentrate on fluent

141
00:09:11,606 --> 00:09:14,531
and accurate production
of the target message.

142
00:09:15,000 --> 00:09:16,234
Readability standards.

143
00:09:18,004 --> 00:09:22,955
Readability standards for subtitles
can differ depending on the country.

144
00:09:23,133 --> 00:09:28,511
A common factor is that high-quality
subtitles must be readable,

145
00:09:28,622 --> 00:09:30,225
otherwise, they aren’t accessible.

146
00:09:30,738 --> 00:09:34,528
The goal of subtitles should be
to match the intent

147
00:09:34,528 --> 00:09:40,677
and tone of the audio so that D/deaf
and Hard of Hearing individuals

148
00:09:40,677 --> 00:09:42,873
can comprehend the content.

149
00:09:43,252 --> 00:09:44,459
Punctuation.

150
00:09:44,459 --> 00:09:49,032
Real-time subtitling requires
also traditional subtitling skills

151
00:09:49,032 --> 00:09:54,271
like punctuation, which applies to both
verbatim and sensatim subtitles,

152
00:09:54,271 --> 00:09:58,256
and segmentation in the case
of sensatim subtitles.

153
00:09:59,327 --> 00:10:01,455
A major problem when dealing
with readability standards

154
00:10:01,455 --> 00:10:02,891
is the reading speed.

155
00:10:03,248 --> 00:10:06,999
The following table based on a study
carried out by Arnáiz Uzquiza,

156
00:10:06,999 --> 00:10:10,800
presents the speed-related information
in different countries

157
00:10:10,800 --> 00:10:15,143
based on number of characters,
lines and speed in words per minute

158
00:10:15,143 --> 00:10:19,406
as included in the SDH guidelines
of different countries.

159
00:10:19,406 --> 00:10:24,071
In the vase of UK, the official standards
drawn by Ofcom’s

160
00:10:24,071 --> 00:10:30,636
is that subtitling speed
should normally not exceed 180 wpm,

161
00:10:30,815 --> 00:10:33,729
which are 15 characters per second.

162
00:10:33,995 --> 00:10:38,038
Therefore, more of the times
exit strategies are applied.

163
00:10:38,945 --> 00:10:44,167
Simple syntactic structures are both
shorter and easier to understand

164
00:10:44,167 --> 00:10:46,564
than complex syntactic structures.

165
00:10:46,564 --> 00:10:48,296
So, make it simple!

166
00:10:48,970 --> 00:10:49,774
Main challenges.

167
00:10:51,903 --> 00:10:54,712
Main challenges 
are the following:

168
00:10:55,000 --> 00:10:57,026
Decalage has to be kept short

169
00:10:57,282 --> 00:11:00,634
Rigid time constraints
due to synchronisation,

170
00:11:01,076 --> 00:11:03,214
Keep up with fast dialogues 

171
00:11:03,552 --> 00:11:06,205
Terminology management,
technical terms, 

172
00:11:06,205 --> 00:11:08,290
proper names and foreign words,

173
00:11:09,009 --> 00:11:09,930
Summary.

174
00:11:11,359 --> 00:11:13,872
As a summary we would like
to stress the following points:

175
00:11:14,898 --> 00:11:19,478
Real-time subtitling
is closely linked to interpreting.

176
00:11:20,303 --> 00:11:23,302
Working memory
is a limited-capacity mechanism.

177
00:11:24,169 --> 00:11:29,443
No clear rules can be given
as to when to condense

178
00:11:29,443 --> 00:11:31,572
and reformulate, or when to omit.

179
00:11:32,330 --> 00:11:36,897
Reformulation depends
on the speech rate of the source text.

180
00:11:38,604 --> 00:11:42,275
Simple syntactic structures
are shorter and easier to understand

181
00:11:42,275 --> 00:11:44,671
than complex syntactic structures.

182
00:11:45,474 --> 00:11:48,107
High-quality subtitles must be readable,

183
00:11:48,107 --> 00:11:50,146
otherwise, they are not accessible.

184
00:11:50,949 --> 00:11:54,442
Decalage has to be kept
as short as possible

185
00:11:54,732 --> 00:11:55,586
Exercises.

186
00:11:57,773 --> 00:12:00,746
The exercises for this video lecture
are in the trainer’s guide

187
00:12:00,746 --> 00:12:02,183
and the PowerPoint file.

188
00:12:03,368 --> 00:12:04,831
[ELECTRONIC MUSIC]