Ssubtiitrite lugemiskiirus (0)

See discussions, stats , and author profiles for this publication at: https://www.researchgate.net/publication/280905030
Subtitle reading speed : A new tool for its
estimation
Article  in  Babel · January 2014
DOI: 10. 1075 /babel.59.4.02mar
CITATIONS
READS
3
179
1 author:
José Luis Martí Ferriol
Universitat Jaume I
13 PUBLICATIONS   18 CITATIONS  
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Inclusión social , traducción audiovisual y comunicación audiovisual (ÍTACA) View project
Al  content fol owing this page was uploaded by José Luis Martí Ferriol on 13 October 2017 .
The user has requested enhancement of the downloaded file.
John Benjamins Publishing Company
This is a contribution from Babel 59 : 4
© 2013. All rights reserved.
This electronic file may not be altered in any way.
The author(s) of this article is/are permitted to use this PDF file to generate printed copies to 
be used by way of offprints, for their personal use only.
Permission is granted by the publishers to post this file on a closed server which is accessible 
to members ( students and staff ) only of the author’s/s’ institute, it is not permitted to post 
this PDF on the open internet .
For any other use of this material prior written permission should be obtained from the 
publishers or through the Copyright Clearance Center (for USA: www.copyright.com). 
Please contact [email protected] or consult our website : www.benjamins.com
Tables of Contents, abstracts and guidelines are available at www.benjamins.com
Subtitle reading speed
A new tool for its estimation
José Luis Martí Ferriol
Universitat Jaume I
1.  Introduction
Subtitling, together with dubbing, is the audiovisual  translation  mode most widely 
used worldwide . The number of publications devoted to assess which one of both  
is better when viewing audiovisual material is large, and have been thoroughly 
summarized for example by Chaume (2004). The conclusion , however, is that this 
discussion is somewhat useless, in the sense that both cover different needs and 
are preferred in certain countries or regions depending on a number of econom-
ical, historical and sociological reasons. Bearing this is mind, we can state that 
Spain is among the countries (most of them European) with a solid dubbing trad-
ition. However, the introduction of DVD in the audiovisual market at the end of 
the  previous  decade, together with digitalization of  television   broadcast  and the 
rise of accessibility policies in Spain have helped consolidate subtitling as an al-
most necessary tool to watch audiovisual content in our country .
 
Many scholars have devoted their research efforts to subtitling, having most of 
them started back in the 1990s . It is not intended to make here a detailed review 
of all their contributions, since some of them like Díaz Cintas (2003) and Chaume 
(2004) have carried out this task with rigor and completeness. Instead, this pa-
per is intended to focus on the viewer as the key element of subtitling reception , 
and more specifical y, on the viewer’s ability to read subtitles, which is usual y ex-
pressed by means of two different parameters: characters per second and words  
per minute .
  Karamitroglou published the article entitled “A  Proposed Set of Subtitling 
Standards in Europe ” in issue number 2 of Translation  Journal  in 1998. This art-
icle covers many aspects in relation with subtitles, their format and the profession-
al conventions at stake . With a somewhat prescriptive intention , the author pro-
poses some linguistic “ best practices ”, as far as creation of subtitles is concerned. 
In terms of reading speed, he mentions that a tested value for “ average readers” 
(i.e. aged 14–65 years , and who belong to a middle-high social class from an edu-
Babel 59 : 4 (2013), 406–420.  © Fédération des Traducteurs (fit) Revue Babel
doi 10.1075/babel.59.4.02mar  issn 0521–9744  e-issn  1569 –9668
 
Subtitle reading speed 407
cational and social standpoint) oscil ates between  150 and 180 words per minute. 
If it is assumed that a complete subtitle of two lines usual y has between 14 and 16 
words, it can then be stated that up to 2 or 3 words can be read in a second, or simi-
larly: a subtitle with those characteristics should stay on screen about 5 and a half  
seconds . The author proposes to increase this suggested time to 6 seconds, because 
“the  brain  needs ¼–½ of a second to start  processing  the subtitle it has traced”. 
This initial period, supposedly needed for reading preparation is usual y referred 
to as “ onset ”. Likewise, the author indicates that such a subtitle should not be more 
than 6 seconds on the screen, because “this would cause automatic re-reading of 
the subtitle, especial y by fast readers”. These statements have been decisive to the 
creation of the so-called “6 second rule ”, which has become a best- known teaching 
standard for subtitling over the years.
 
Mayoral (2001, 4) mentions variable reading speeds for cinema (10 frames per 
foot of film — between 14 and 15 characters per second), as used by the Ameri -
can company 20th Century Fox, while the values for television and video would 
be lower . The author also mentions the television subtitling software program used 
by the company Screen Subtitling System Ltd, which proposes reading speed fig-
ures of 9 characters per second for adults, and half to two third equivalent ones for 
children .
 
As far as the usage of reading speeds expressed in WPM and how adequate it 
may be in different countries, Mayoral affirms:
No se dan normas de traducción audiovisual para diferentes países (las velocidades 
de lectura se establecen habitualmente —p. ej. Ivarsson, 1992, y Screen Subtitling Sys-
tems Ltd, 1998— en palabras por minuto (WPM) aún cuando la extensión media de 
las palabras puede ser muy diferente para distintas lenguas) y el concepto de « local -
ización» por ahora abarca tan sólo elementos lingüísticos y culturales, pero no los 
que aquí nos ocupan, que están directamente relacionados con el tipo de destinatario 
(MAYORAL 2001: 14)
  The author refers to the lack of audiovisual translation norms in different 
countries, since reading speeds “as  done  by Ivarsson (1992) and Screen Subtitling 
Systems Ltd. (1998), are usual y established in WPM, while the word length can 
vary significantly for different languages , and the concept of ‘localization’ only 
copes with linguistic and cultural elements for the time being, and is thus not dir-
ectly related with the type of viewer”. These comments are considered especial y 
appropriate in order to answer some of the research questions in mind, like the re-
lationship between CPS and WPM and their more than likely variation depending 
on the language the reading speed calculations are made for.
 
Díaz Cintas (2008: 96–7) states that the famous “six second rule” can be trans-
lated in terms of CPS to values of 12, and proposes the equivalent interval in WPM 
(between 140 and 150). The scholar also  points out  figures usual y higher for DVD, 
© 2013. Fédération des Traducteurs (fit) Revue Babel
All rights reserved
408  José Luis Martí Ferriol
of about 180 WPM or 17 CPS. Additional y, the author mentions the supposed  
equivalence between both parameters, which has been proposed with the inten-
tion to normalize reading speed figures, although for English language only. This 
equivalence is based on a general assumption, according to which an English word 
has on the average five characters.
 
Toda (2009) refers again to the “10 characters per foot of film, i.e.: 15 charac-
ters per second”. As far as Spain is concerned, the author points out that reading 
speed values up to 15 CPS are common  practice , both for cinema and television. 
For other European countries, figures of 10 CPS for Belgium (Flanders) and 16 for 
Finland are mentioned in his presentation.
  As a conclusion to this initial overview of the scarce literature on reading 
speed, Romero Fresco ’s contribution (2009) must be highlighted. Although his 
article deals with reading speed considerations related to subtitles produced with 
“respeaking” techniques, it mentions the most rigorous and documented experi-
ment carried out so far regarding subtitling reading speed: the one by D’Ydewalle 
et al. (1987). The “eye tracking ” technology mentioned is being extendedly used 
nowadays in a great deal of experiments aimed at the research of audiovisual text 
reception, as well as the relevance of the information provided in this type of texts.
Using eye-tracking technology, he tested three different presentation times for subti-
tles : two lines of 32 characters in 4 seconds (approximately 192 wpm), 6 seconds (130 
wpm) and 8 seconds (96 wpm) respectively. The object of this study was to ascertain 
if the six-second rule (a full two-line subtitle displayed on screen for 6 seconds and 
shorter subtitles scheduled proportional y), accepted as common practice in most sub-
titling countries, could be validated by empirical research on reading speed. His re-
sults leave little room for doubt , the six-second rule being identified as setting the ap-
propriate reading speed for the participants. This rule has  later  on been supported by 
other scholars such as Díaz Cintas (2003), who applies it to longer lines than the ones 
referred to by D’Ydewalle (72 characters instead of 64), thus setting the recommended 
speed at 144 wpm (12 cps) (ROMERO FRESCO 2009: 114).
 
As mentioned before, the “6 second rule” seems to be the most accepted one 
among viewers . If we equate this reception best practice with the reading speed 
parameters we intend to use, it all comes down to figures of 12 CPS and 144 WPM.
2.   Commercial subtitling software programs and reading speed figures 
provided by them
The  process  of subtitling involves two  fundamental  stages: “ spotting ” (or text seg-
mentation including time assignment), and subtitle creation in itself. By spotting 
it is usual y understood the definition of the cue-in and cue-out times for a given  
© 2013. Fédération des Traducteurs (fit) Revue Babel
All rights reserved
 
Subtitle reading speed 409
subtitle. This means that it is necessary to  specify  the moment when the  dialogue  
in the original source text starts, so that the corresponding subtitle appears on the 
screen. Likewise, the subtitle must disappear from the screen when the corres-
ponding fragment of the source text dialogue covered by the subtitle finishes. The 
difference  between the cue-out and the cue-in times sets the subtitle duration. It 
must be taken into account that, if the spotting is carried out correctly, it allows the 
viewer for the reading of the subtitle without problems or difficulties. As a general 
rule, subtitle duration on screen spans between 2 and 6 seconds.
  The main constraint in the subtitle creation phase in itself, which also ac-
counts for its implicit difficulty, lies on the fact that general y only two lines (of 
about 35 characters each) are available to translate that particular piece of dialog . 
As a consequence, translation for subtitling is intrinsical y characterized by the 
need to synthesize (or condense) the information included in the source language.
 
The process of subtitle development can be undertaken in a manual fashion , o 
by means of commercial subtitling programs. A series of those are available on the 
market: some of them are free software, others can be downloaded from specific  
sites on the Internet as a “demo” version for a limited period of time, and others 
require the  purchase  of a license, which can be  acquired  by paying in some  cases  
significant prices.
  Here follows a list of the main commercial subtitling programs and the sites 
from where they can downloaded or purchased:
 
– Subtitle Workshop (http:/ subtitle-workshop.softonic.com/)
 
– Subtitul@m (http:/ subtitul-m.software.informer.com/)
 
– Substation Alpha (http:/ www.videohelp.com/ tools /Sub_Station_Alpha)
 
– WinCAPS (http:/ wincaps-int.software.informer.com/)
 
– Fab (http:/ www.fab-online.com/eng/subtitling/broadcast/subtmpg.htm)
 
–  Spot Software (http:/ www.spotsoftware.nl/)
 
–  Swift (http:/ www.softelgroup.com/Our- Products /vtx)
  All of them are of great help in the process of subtitle generation, and some 
of them generate a warning during their use, if the number of specified characters 
has been exceeded, for example. Spotting is much easier with these programs as 
wel , since it is usual y necessary just to press one key so that the cue-in and cue-
out times are recorded. In some cases, these programs also provide the user with 
reading speed parameters (CPS and/or WPM), an aspect of utmost personal inter-
est. It is curious, though, that these values only seldom coincide, when compared 
for different programs.
  The above -mentioned statement is il ustrated in Table 1. It depicts values for 
reading speeds expressed in CPS and WPM, which have been calculated by some 
(3) of the seven programs listed above for a sample set of 8 subtitles. Among these 
© 2013. Fédération des Traducteurs (fit) Revue Babel
All rights reserved
410  José Luis Martí Ferriol
ot
ps
ps
ps
ps
ps
ps
ps
ps
Sp
11 c
18 c
14 c
18 c
11 c
14 c
13 c
10 c
m
m
m
m
m
m
m
m
n 2010.
ift
ps / 101 wp
ps / 203 wp
ps / 107 wp
ps / 149 wp
ps / 105 wp
Sw
9 c
15 c
13 c
14 c
9 c
121 wp
182 wp
119 wp
pain) i
m
m
m
m
m
m
m
m
astelló, S
e I (C
aps
s
ps / 174 wp
ps / 160 wp
ps / 168 wp
ps / 121 wp
ps / 124 wp
aum
m
inC
ps / 106 wp
ps / 115 wp
ps / 108 wp
ra
W
8 c
14 c
13 c
14 c
9 c
10 c
10 c
9 c
rog
niversitat J
g p
f 
ber o
ters
he U
rac
t t
ubtitlin
um
N
cha
35 / 37
25
21
al s
12 / 13
16
15 / 18
12 / 6
10
ci
erchán a
er
m
y co
ation
erezo M
ur
D
2:09
4:03
1:03
2:00
2:07
1:12
1:16
1:00
eatriz C
ulated b
y B
alc
ut
p b
s c
er
et u
et
m
as s
ara
Time in/o
00:00:45:12 / 
00:00:47:21
00:00:50:10 / 
00:00:54:13
00:01:04:09 / 
00:01:05:12
00:01:29:15 / 
00:01:31:15
00:01:51:13 / 
00:01:53:20
00:02:39:04 / 
00:02:40:16
00:02:42:02 / 
00:02:43:18
00:02:43:23 / 
00:02:44:23
p 
hich w
peed p
ed u
ne w
g s
how
ted.
ger o
ig
eadin
ho s
rres
f a b
he r
e a
f t
ews w
er
ss.
art o
he J
ht?
lip w
ty, mi
nly a p
aid t
ret
parison o
…
s o
p?
t s
reen s
ot a lig
ter
rong ?
ther
y p
e.
om
ews …
et
s fa
er
ou g
se.
able i
 C
ntac
he g
ur
hat’s u
is t
ad n
y co
at’s b
hat’s w
t’s hi
xcuse m
es?
ave y
Subtitle
ith t
– W
– B
M
w
Th
– W
– I
You’re v
– E
– Y
H
Of co
te: Th
Table 1.
1
2
3
4
5
6
7
8
No
© 2013. Fédération des Traducteurs (fit) Revue Babel
All rights reserved
 
Subtitle reading speed 411
three software tools “WinCaps” and “Swift” are included, the ones which provide 
figures for both reading speed parameters. Values for CPS generated by “Spot” 
are also shown. The table additional y includes the subtitle text and the in and 
out times for each of the studied subtitles. This means the readers have all the in-
formation which is needed available, in case they wish to make the calculations 
 themselves.
 
A  simple and quick analysis of the data allows us to check that the exceptions  
are the very few instances where the values proposed by the three programs co-
incide. It seems as though the simple process of  counting  characters and dividing 
the resulting number by the subtitle duration must be somehow implemented in a 
different way in these programs, given the results shown. As far as word counting, 
it is more than likely that “WinCaps” and “Swift” are programmed to make some 
sort of initial assumption (as the one mentioned before that every set of five char-
acters makes  up a word), although this presumed assumption must clearly not be 
the same in both applications, since differences in results of up to 15% (see subtitle 
number 2, the one with a longer duration and the biggest number of characters of 
the set) have been recorded.
 
Data regarding reading speed parameters like the one shown in the previous 
table turned out to be the  clear motivation to design and program an alternative 
tool, one as simple as possible which would just make the calculations the easiest 
way with no assumptions whatsoever.
3.  New tool to estimate reading speed parameters
This section introduces the tool which has been designed and programmed to es-
timate reading speeds, as well as the dialogue boxes the user will encounter when 
using the macro and also the final table the program generates to present the re-
sults. According to the functional specification defined at the beginning of the 
project (a project aimed original y at the provision of a tool which would pro-
vide quantitative results in a teaching environment), it was decided to program 
this new application as a macro for MS Word using source code written in Visual  
Basic . This means that the above mentioned “tool” which makes the reading speed 
calculations is just a macro programmed in Visual Basic for Word. The subtitles 
to be analyzed (including both the times and the set of characters) must be cop-
ied with a specified format into the Word file containing the macro, which must 
be subsequently executed.
  This new application makes no assumption of any kind: it simply calculates 
the subtitle duration by deducting the cue-in time from the cue-out time, and it 
then counts the characters for any given subtitle. Based on these two intermediate  
© 2013. Fédération des Traducteurs (fit) Revue Babel
All rights reserved
412  José Luis Martí Ferriol
results, the code divides the latter by the former , thus obtaining a value expressed 
in CPS for each of the subtitles. As far as words, the piece of logic has been pro-
grammed to consider a “word” as any set of alphanumeric characters preceded 
and followed by a blank character . In the case of punctuation signs, if they are 
placed next to the word with no blank spaces in between (like the usual subtitling 
practice of putting hyphens at the beginning of a line to indicate that dialogue be-
tween more than one speakers will be included in the same subtitle), the logic con-
siders them as  part  of the word they precede or follow (a  similar  logic will be ap-
plied to comas following words).
 
It is now intended to il ustrate the process of using this macro with one prac-
tical example: the one with the set of 8 subtitles compiled in the table above. Those 
subtitles must be in the first place copied into a Word file which contains the mac-
ro. The format to be used is il ustrated below , and it is crucial that this specific for-
mat be followed, if the user wants the macro to run without errors and to produce 
good values. For each subtitle, the cue-in time (hh:mm:ss:cc) appears first, then a 
blank  space  follows, and then comes the cue-out time in the same format. Then, 
after another blank space, the subtitle text is next. If the subtitle is a two- liner , the 
separation of both lines is indicated by the  slash sign (/), preceded and followed 
by a blank space. All this information regarding formats is presented in a dialog 
box prompted by the macro when initial y run, although it will be probably more 
clearly understood by the reader as shown below for the example proposed. Thus, 
the set of 8 subtitles to be analyzed by the macro should look like the following 
when copied to the Word file which contains the reading speed application: 
00:00:45:12 00:00:47:21 - What’s up? / - Bad news…
00:00:50:10 00:00:54:13 My contact said the Jews who showed / up with 
the green slip were arrested.
00:01:04:09 00:01:05:12 That’s better…
00:01:29:15 00:01:31:15 - What’s wrong? / - It’s his father .
00:01:51:13 00:01:53:20 You’re very pretty, miss .
00:02:39:04 00:02:40:16 - Excuse me. / - Yes?
00:02:42:02 00:02:43:18 Have you got a light ?
00:02:43:23 00:02:44:23 Of course .
 
Once the subtitles have been copied into the file with the appropriate format, 
the macro is to be run. Having done this, a dialog box containing the most import -
ant information regarding this tool is prompted to the user (see Figure 1).
 
Apart from the text shown (which insists on the importance of sticking to the 
pre-specified formats and reminds the user that security checks for macros must 
be lowered or deactivated to be able to run this), the box also includes a button  
© 2013. Fédération des Traducteurs (fit) Revue Babel
All rights reserved
 
Subtitle reading speed 413
NOTA INICIAL:
•  Este  fichero contiene una macro programada en VBA (“Visual Basic for Applications”) para
   MS  Word,  que  calcula  la  legibilidad  de  subtítulos  introducidos  por  el  usuario  de  forma
   automática. Es  decir , que se trata de subtítulos exportados desde un programa convencional
   de subtitulación, o “ripeados” desde un DVD. Los subtítulos exportados deben contener un
   tiempo de entrada del subtítulo, un tiempo de salida y el texto del propio subtítulo, en una o
   dos líneas. 
•  Como  ejemplo del formato, véase a continuac ión unos subtítulos en alemán de la película
   “Lolita” (S. Kubrick, 1962):
00:03:29:36 00:03:31:40  Nein , ich bin Spartakus.
00:03:31:76 00:03:33:56  Oder  wol en Sie für  mich  / die Sklaven befreien?
00:03:33:56 00:03:35:28 Sind Sie  Quilty ?
•  La  macro  genera valores  en  CPS  (caracteres  por  minuto)  y  WPM  (palabras  por  minuto,
   “words per minute”, en inglés), para cada uno de los subtítulos exportados, y sólo funcionará
   sin errores para subtítulos exportados en este mismo formato. La presente versión contiene
   lógica que tiene en cuenta correcciones para cambio de minuto y de hora en los tiempos de
   entrada y salida.
•  Para  ejecutar  la  macro,  denominada  “Automatización”,  puede  ser  necesario  habilitar  las
    macros  en  Word  o  reducir  el   nivel   de  seguridad  de  las  macros (pasándolo  a  “baja”), 
   respectivamente, dependiendo de si la versión de MS Word es 2007 (.docx) o 2003 (.doc).
•  Este  fichero  que  incluye  la  macro  debe  tener  ya  introducidos  los  subtítulos  exportados.
   Debe  introducirlos  primero  antes  de  ejecutar  la  macro. Los  principales  resultados  de  los
   cálculos (tanto en CPS como en WPM) se presentan para cada subtítulo en una única tabla
   de Word. Dicha tabla se puede pasar directamente a  Excel , y realizar cálculos (estadísticos).
• Dado que estamos ante un prototipo, creado únicamente con un fin docente e investigador,
   el potencial usuario deberá ser consciente de sus posibles limitaciones, así como hacer uso
   de  esta  aplicación con las lógicas reservas.
Figure 1.  Dialog box prompted to the user when executing the macro
(“OK”) that the user must press so that the tool can proceed. Having this been 
done, a new dialog box appears on the screen (some sort of second check), which 
informs the user that the reading speed calculations are going to be performed 
for the proposed set of 8 subtitles (if we follow with the example). Final y, another 
confirmation push -button has to be acknowledged (“OK”) to let the macro go 
ahead, make the CPS and WPM calculations and generate an overview table with 
the results. For the particular example used in this  paper , the overview table is set 
out in Table 2.
 
The table is placed by the macro into a new Word document; this means that 
it can now be stored with a dedicated name. In any case, the table contents (and 
more specifical y, its last two columns which include the reading speed para-
meters) will probably be exported to a spreadsheet (Excel) or a statistical analysis 
program (Minitab or SPSS ) for its additional processing and quantitative study.
© 2013. Fédération des Traducteurs (fit) Revue Babel
All rights reserved
414  José Luis Martí Ferriol
Table 2.  Reading speed calculations as calculated and presented in a table by the new tool
Subtitle 
number Subtitle text and time
Duration Characters Words CPS WPM
1
00: 00: 45: 12 00: 00: 47: 21 - - - 
2,09
23
6
11
172
What’s up? / - Bad news…
2
00: 00: 50: 10 00: 00: 54: 13 My 
4,03
72
14
18
208
contact said the Jews who 
showed / up with the green slip 
were arrested.
3
00: 01: 04: 09 00: 01: 05: 12 That’s 
1,03
16
2
16
117
better…
4
00: 01: 29: 15 00: 01: 31: 15 - - - 
2
33
7
16
210
What’s wrong? / - It’s his father.
5
00: 01: 51: 13 00: 01: 53: 20 You’re  2,07
25
4
12
116
very pretty, miss.
6
00: 02: 39: 04 00: 02: 40: 16 - - - 
1,12
18
5
16
268
Excuse me. / - Yes?
7
00: 02: 42: 02 00: 02: 43: 18 Have 
1,16
21
5
18
259
you got a light?
8
00: 02: 43: 23 00: 02: 44: 23 Of 
1
10
2
10
120
course.
4.  Proposed results: summary, comparison and discussion
In order to carry out an easier comparison between the reading speed values sup-
plied by commercial subtitling programs and the ones proposed by the new ap-
plication, the following table has been generated by compiling all the necessary 
information. Besides , since all the preliminary data needed to make the calcula-
tions is available to the reader (the subtitle times and their texts), it is possible to 
make this calculations manual y: a simple exercise of obtaining durations for sub-
titles, counting characters and words and performing straightforward mathemat-
ical operations. The reader is encouraged to proceed: the results they will come 
up with will be the same as the ones obtained by the macro. As a consequence, it 
can be stated that the new application is just an automation tool which (once thor-
oughly tested) performs the same calculations that can be done by hand, but more 
quickly and without errors.
  In Table 3 (overleaf) the values shown in bold are the ones obtained by the 
new macro.
© 2013. Fédération des Traducteurs (fit) Revue Babel
All rights reserved
 
Subtitle reading speed 415
pm
pm
pm
pm
pm
pm
pm
pm
w tool
ps / 172 w
ps / 208 w
ps / 117 w
ps / 210 w
ps / 116 w
ps / 268 w
ps / 259 w
ps / 120 w
Ne
11 c
18 c
16 c
16 c
12 c
16 c
18 c
10 c
ot
ps
ps
ps
ps
ps
ps
ps
ps
Sp
11 c
18 c
14 c
18 c
11 c
14 c
13 c
10 c
m
m
m
m
m
n
atio
m
m
m
ps / 203 wp
ps / 107 wp
ps / 149 wp
pplic
ift
ps / 101 wp
ps / 105 wp
Sw
9 c
15 c
13 c
14 c
9 c
121 wp
182 wp
119 wp
ew a
m
m
m
m
m
m
m
m
he n
nd t
s a
aps
m
ps / 174 wp
ps / 160 wp
ps / 168 wp
ps / 121 wp
ps / 124 wp
ra
inC
ps / 106 wp
ps / 115 wp
ps / 108 wp
W
8 c
14 c
13 c
14 c
9 c
10 c
10 c
9 c
rog
f 
g p
ber o
ters
rac
ubtitlin
um
N
cha
12 / 13
25
35 / 37
72
16
16
15 / 18
33
25
25
12 / 6
18
21
21
10
10
al s
ci
er
mm
ation
ur
 09
 09
 03
 03
 03
 03
 00
 00
 07
 07
 12
 12
 16
 16
 00
 00
y co
D
2:
2:
4:
4:
1:
1:
2:
2:
2:
2:
1:
1:
1:
1:
1:
1:
ut
 12 / 
 21
 10 / 
 13
 09 / 
 12
 15 / 
 15
 13 / 
 20
 04 / 
 16
 02 / 
 18
 23 / 
 23
ulated b
 45:
 47:
 50:
 54:
 04:
 05:
 29:
 31:
 51:
 53:
 39:
 40:
 42:
 43:
 43:
 44:
alc
 00:
 00:
 00:
 00:
 01:
 01:
 01:
 01:
 01:
 01:
 02:
 02:
 02:
 02:
 02:
 02:
lues c
Time in/o
00:
00:
00:
00:
00:
00:
00:
00:
00:
00:
00:
00:
00:
00:
00:
00:
peed va
ho 
lip 
g s
ews w
reen s
ss.
eadin
f r
he J
he g
ht?
ty, mi
ary o
aid t
ith t
…
ret
p?
t s
p w
ted.
ot a lig
ter
rong?
ther
y p
e.
ews…
ntac
rres
et
s fa
er
ou g
se.
ur
 Summ
ed u
hat’s u
ad n
y co
e a
at’s b
hat’s w
t’s hi
xcuse m
es?
ave y
Subtitle
er
– W
– B
M
show
w
Th
– W
– I
You’re v
– E
– Y
H
Of co
Table 3.
1
2
3
4
5
6
7
8
© 2013. Fédération des Traducteurs (fit) Revue Babel
All rights reserved
416  José Luis Martí Ferriol
  A  preliminary analysis of the results shows some coincidence in the CPS 
values obtained by the macro and the program “Spot”, namely for subtitles num-
ber 1, 2 and 8. For subtitle number 2, the CPS value turns out to be so high because 
only 4,03 seconds have been allowed for a 72-character subtitle. A more academic 
proposal would indicate that at least 5 seconds would be desirable for this subtitle 
to be read comfortably, or that the text should have been reduced, for example by 
deleting “who showed up”.
 
Another regularity which can be observed , as far as CPS values are concerned, 
is that the ones obtained by “WinCaps” are systematical y lower than the mac-
ro ones, being this difference  around  2 or 3  units . An exception to this  concerns  
subtitles numbers 6 and 7, where those “WinCaps” values are much lower (6 and 
8 units, respectively). This more than significant divergence can be found in the 
WPM values for these same two subtitles as wel , where the commercial program 
results are less than half than the macro ones (121 vs. 268 and 124 vs. 259).
 
The above mentioned regularities can also be pointed out for the values calcu-
lated by “Swift” (in some cases, only WPM results have been recorded). Like “Win-
Caps ”, this program systematical y produces lower values, both for CPS (some 2 or 
3 units) and for WPM (except for subtitle number 8).
 
Figures for subtitles 1, 4 and 6 (i.e. the ones which include dialogue and whose  
lines start with hyphens) need some additional clarification which may explain  
higher values for WPM when the calculations are made by the macro. As it was 
explained before, the application source code considers a “word” any character se-
quence located between blank spaces. In the above mentioned instances, the sub-
titler has left blank spaces between the initial hyphen and the first letter of the first 
word of each subtitle line. This is not considered as a “best practice” in the field  
professional conventions, and has also been recorded in some of the most used 
subtitling manuals (like Díaz Cintas, 2003), since one character is spoilt, or ra-
ther left unused, for no obvious reason. Due to this, it is always recommended not 
to leave any blank space between the hyphen (if used when needed) and the first 
word in the line. The macro source code has then been designed and programmed 
according to subtitling best practices. This explains the increased number of words 
for subtitles 1, 4 and 6, which lead to higher WPM values in the case of macro cal-
culations. In any case, WPM values higher than 200 in some subtitles clearly indi-
cate their difficulty to be read, and may be an indication of incorrect hyphen usage.
  Subtitle number 7, on the other side, also shows very high WPM values, be-
ing a one-liner with 5 words and just above 1 second to read it. “Swift” is the only 
commercial program which seems to have traced this, while the real figure (as ob-
tained manual y of by means of the application) clearly indicates that it will be very 
difficult for a viewer to be able to read it properly.
© 2013. Fédération des Traducteurs (fit) Revue Babel
All rights reserved
 
Subtitle reading speed 417
  Hopeful y this analysis and comparison of the results generated by commer-
cial subtitling software and the new application will shed some light to some of the 
inconsistencies brought to the fore . Stil , it is considered that these divergences are 
even highlighted by a practice which may be termed as “aggressive spotting” for 
the subtitles shown, with reading times which, from a personal point of view, seem  
to be sometimes too short.
5.  Potential uses of the new tool
This macro is currently being used in an intensive way to obtain reading speed fig-
ures for large series of commercial subtitles. Even thousands of them can be ana-
lyzed with a reasonable amount of time and effort, and some results for several case 
studies and language combinations will be presented in future publications.  While 
this research usage can in itself justify the tool design and programming, it is un-
derstood that, given the tool simplicity and robustness, it can also find its applica-
bility in subtitling teaching courses. This twofold utilization in different environ-
ments is elaborated below.
  As far as research, if big- size subtitle corpora are analyzed via the new tool, 
new conclusions regarding some of the research questions implicitly stated above 
can be drawn. Among them, the feasibility of choosing either one of the two read-
ing parameters (CPS vs. WPM) in a more consistent fashion, the possible correla-
tion between them to be obtained from empirical data and not based on previous 
assumptions, and especial y, their behavior across different languages, which may 
lead to a selective use of the one (probably CPS) which would be language-inde-
pendent. These topics, for the time being only introduced as open questions, may 
lead to subsequent additional research tasks aimed at confirming them empirical-
ly or experimental y with the assistance of the application tool. The tasks may be 
grouped to provide more consistent research jobs .
 
Regarding the use of the application as a teaching assistance tool, it can provide 
the lecturer with some quantitative support to the more intuitive “6 second rule” 
(updated in some universities already to a similar “5 second rule” for 2-liners of 35 
characters each). The use of the rule is widespread, although there are situations 
where we lecturers just use it as a quality parameter (mainly related to the spotting 
phase) in sentences like “it does not fit” or “it can’t be properly read”, and then pro-
pose the students to copy their subtitles into a commercial program (usual y “Sub-
title Workshop”, being it a free-downloadable software) to let them check as viewers 
if the rule has been properly applied or not. Unfortunately , this program does not 
make reading speed calculations that the students may use as an additional check.
© 2013. Fédération des Traducteurs (fit) Revue Babel
All rights reserved
418  José Luis Martí Ferriol
 
As an alternative to this  qualitative   approach , it may be suggested to students to 
export  their subtitles to a Word file which contains the macro, and to run it. By using 
the  result  table after macro execution, more specific quantitative quality assurance 
criteria can be specified, for example: that calculated readings speed oscil ate be-
tween 10 and 14 CPS (o between 120 and 170 WPM). It is also to be kept in mind that 
the macro has been designed and programmed while bearing in mind the standard 
formats generated by “Subtitle Workshop” when exported to a Word file, so that the 
necessary format adjustment previous to  running  the macro is minimized.
  To finalize this paper, which tries to foresee tentative applications in the re-
search and teaching environments of a new reading speed tool, it must be stated 
that all the effort has been made in an attempt to somehow clarify the apparent 
confusion which surrounds this concept and its expression in terms of two differ -
ent parameters.
References
Chaume, Frederic . 2004. Cine y Traducción. Madrid: Cátedra. 336 pp.
Díaz Cintas, Jorge. 2003. Teoría y práctica de la subtitulación inglés – español. Barcelona: Ariel. 
412 pp.
Díaz Cintas, Jorge. 2008. “Teaching and Learning to Subtitle in an Academic Environemt” In 
The Didactics of Audiovisual Translation, ed. by Jorge Díaz Cintas, 89–105. Amsterdam : 
John Benjamins. DOI: 10.1075/btl.77.10dia
Karamitroglou, Fotios. 1998. “A Proposed Set of Subtitling Standards in Europe”. Translation 
Journal 2. http:/ accurapid.com/journal/04stndrd.html. 12 pp.
Mayoral, Roberto. 2001. “El espectador y la traducción audiovisual”. En La traducción en los 
medios audiovisuales, ed. by Frederic Chaume, y Rosa Agost, 33–48. Castellón: Universitat 
Jaume I, Servei de Publicacions.
Romero Fresco, Pablo . 2009. “More Haste less Speed: Edited versus Verbatim Respoken Subti-
tles”. Vial 6. http:/ webs.uvigo.es/vialjournal/pdf/Vial-2009-Article6.pdf. pp. 109–133
Toda, Fernando, and González Iglesias. 2009. “Spoken Language and ICC: Managing Cultur-
al Diversity in Dubbing and Subtitling in Spain”, Boras NIC 2009. http:/ nic.hb.se/index.
php?id=12. 17 pp.
Abstract
This article presents a new tool which has been developed in order to make reading speed cal-
culations, and whose results are expressed by means of the two parameters most used in the 
field: CPS (characters per second) and WPM (words per minute). Topics such as the suitability 
of each one of the parameters versus the other, the potential correlation which may exist among 
them, as well as their behavior across different languages are open to discussion in the very lim-
ited available bibliography on the subject .
© 2013. Fédération des Traducteurs (fit) Revue Babel
All rights reserved
 
Subtitle reading speed 419
  It seems that part of this confusion has spread over to commercial subtitling programs, most 
of which present values for either one or both parameters to the final user. It has been confirmed 
that different subtitling software programs calculate different reading speed values (both in CPS 
and WPM) for the same set of subtitles. Due to this, a very simple software application which 
calculates values for CPS and WPM is presented. It has been willingly designed to make no as-
sumptions, but simply to count characters and to make very straightforward time algebra with 
the subtitle in and out times. The new tool allows for empirical and quantitative research of large 
series of subtitles at a time, and it may also play an important role in the classroom environment.
Keywords:  subtitling, subtitling software, subtitle reading speed, characters per second, words 
per minute
Résumé
Cet article présente un nouvel outil qui a été élaboré pour calculer la vitesse de lecture et dont 
les résultats sont exprimés au moyen des deux paramètres les plus fréquemment utilisés dans  
le domaine : CPS (caractères par seconde) et MPM (mots par minute). Des thèmes tels que la 
pertinence de chaque paramètre par rapport à l’autre, leur corrélation potentielle, ainsi que leur 
comportement dans différentes langues sont sujets à discussion dans la bibliographie très limi-
tée, disponible sur le sujet.
  Il semble que cette confusion se soit étendue en partie aux programmes de  sous -titrage com-
merciaux, dont la plupart présentent une valeur pour l’utilisateur final en ce qui concerne l’un 
ou l’autre paramètre ou les deux. Il est confirmé que des logiciels de sous-titrage différents cal-
culent des valeurs de vitesse de lecture différentes (tant en CPS qu’en MPM) pour le même en-
semble de sous-titres. C’est pourquoi nous présentons une application informatique très simple 
pour le calcul des valeurs de CPS et de MPM. Elle a été volontairement conçue pour ne pas 
émettre d’hypothèses, mais pour simplement compter les caractères et effectuer des calculs nets 
et précis du temps d’entrée et de sortie des sous-titres. Le nouvel outil permet la recherche em-
pirique et quantitative de grandes séries de sous-titres simultanément, et il pourrait aussi jouer 
un rôle important dans l’environnement de la classe.
Mots clés : sous-titrage, logiciel de sous-titrage, vitesse de lecture des sous-titres, caractères par 
secondes, mots par minutes
About the author
José Luis Martí Ferriol obtained a Bachelor’s Degree in Industrial Chemistry from the Univer-
sity of  Valencia in 1985, and earned his PhD degree in audiovisual translation at the Universitat 
Jaume I, Castellón (Spain) in 2006. For more than 21 years, José Luis combined his profession-
al activity in several chemical process production companies with his education in Translation 
Studies. Since 2007, José Luis is a full-time researcher, and a full-time lecturer as well since 2009, 
at the Universitat Jaume I. His main research field of interest is audiovisual translation, and is 
specialized in comparing translations for dubbing and subtitling. José Luis teaches audiovisual 
and scientific translation, and has been involved in the development of software tools both to 
automate translation conventions for dubbing and to make subtitle reading speed calculations. 
© 2013. Fédération des Traducteurs (fit) Revue Babel
All rights reserved
420  José Luis Martí Ferriol
Additional y, he occasional y acts as a free-lance translator of scientific and technical assign-
ments, and has several publications (including two  books ) related to audiovisual translation re-
search.
Address: Universitat Jaume I, Av. Sos Baynat s/n, 12071 Castelló de la Plana, Spain
E-mail: [email protected]
TRANSLATIO  issn  1027 -8087
Bibliographie – Lexicographie – Histoire – Recensions
Bibliography – Lexicography – History – Reviews
Revue Trimestrielle publiée par la
Féderation Internationale des Traducteurs
International Federation of Translators
Information et commande chez Dr. R. Haeseryn
Redacteur en chef responsable
Heiveldstraat 245, b- 9040 Sint-Amandsberg (Belgique)
Tél. et télecopie: +32-9-2283971
COMMANDITÉ
par
E S O
Euro-Schulen- Organisation
© 2013. Fédération des Traducteurs (fit) Revue Babel
All rights reserved
View publication stats

Document Outline

• Subtitle reading speed: A new tool for its estimation
  • 1. Introduction
  • 2. Commercial subtitling software programs and reading speed figures provided by them
  • 3. New tool to estimate reading speed parameters
  • 4. Proposed results: summary, comparison and discussion
  • 5. Potential uses of the new tool
  • References
  • Abstract
  • Résumé
  • About the author