Those of you who read this article, and who have worked with me as an English editor, will probably know that I always aim to create subs that are displayed at the approximate rate of 20 characters per second (cps).
I arrived at this value for English subs after some research earlier last year. Fortunately, many streaming sites and public broadcasters make their subtitling rates available online. My investigation found the following rates.
A while ago, I discovered this academic article that was published in 2018 on the speed of subs.
National Libray Of Medicine
This document investigates English, Polish and Spanish subtitles at the three display rates of 12 cps (slow-paced), 17 cps (medium-paced) and 20 cps (fast-pasted). The unit “cps” stands for “characters per second”. The study’s writers are…
Agnieszka Szarkowska, PhD
Assistant professor at the Institute of Applied Linguistics at the University of Warsaw.
Olivia Gerber-MorĂłn, Ph.D.
Multilingual User Researcher, accessibility user testing, qualitative/quantitative & behavioural/attitudinal research
I found this to be very informative reading. While there is much statistical information in the article, this can be ignored as the results tend to speak for themselves.
This is all really informative. Thank you for sharing!
What I find really hard is the fact that we’re subtitling in a language that is different from the actual drama audio. Cdramas, especially, seem to pack a lot of meaning in a short sentence, making it nearly impossible to shorten. And it’s hell when they speak fast or overlap each other. I spent a terrible time with New Horizon, because they had so many meetings, and always seemed to be saying loads and loads in a few seconds.
I struggle all the time, wondering which is important - an accurate translation or shortened, easy readability. Most of the time, I tend to prioritize accurate translation. Just like the info you shared, I’ve noticed that I myself can keep up with long subtitles pretty well as I get used to reading them. I, however, can’t say this for sure, because I understand enough Korean to fill in the gaps when a subtitle passes without me finishing it. Because I spend most of my time with subtitles, either segmenting or editing them, I always feel like my length estimations may be inaccurate. Just last week, I was watching a drama with my brother, with the TCs on. I easily read both, while my brother just couldn’t understand how I could read both. He only focused on the subs.
I’m glad that people seem to be able to keep up with subs. I hate shortening sentences sometimes, because I know I wouldn’t like it if I was learning the language. If I had to choose my one fatal flaw as an English Editor, it would be concise sentences.
I’m the same. I love watching Netflix because I never have to stop the playback, rewind and then stop at the sub just to read it. While I don’t mind doing this a handful of times, if I do it too often, it removes me from the show and I lose the feeling of being part of what I’m watching. Once that happens, I usually stop viewing because I no longer have any interest in continuing.
I find that Netflix’s rate lets me have enough screen time to watch what is happening as well as to understand what is being said.
I totally get what you mean. I think that this is one of the reasons that Netflix leaves out honorifics because they take up precious space. For example, while “ge” is only a tiny Chinese word and doesn’t take much time to say, “brother” in English consumes quite a bit of subtitle space.
You’re right. However, I believe our eyes simply skip the entire word once we get used to the “look” of it.
Also, I guess I just feel weird when I hear people being addressed by their first names with no honorifics. It’s jarring and breaks my immersion more than when I have to rewind. I hate hearing “ahjumma” and seeing “Miss Geum Ok.” I was watching Lawless Lawyer the other day, and the ex-scammer says “Yes, I’ve seen Lady Justice.” It’s actually really funny because he calls her “that ahjumma,” not “Lady Justice.”
in interlingual subtitling, it is part and parcel of the process to condense and reduce the text–the underlying assumption is that because viewers can integrate information from subtitles with that coming from images and soundtrack, verbatim translation is not necessary [3]. Another reason for text reduction stems from reading speed requirements, which are meant to allow viewers to comfortably follow both the subtitles and on-screen action [3, 4]. In contrast, text reduction is not welcome in intralingual subtitling, mainly produced for the deaf and hard of hearing (SDH), who demand to have a verbatim and “uncensored” version of the dialogue to be on a par with hearing people, as well as to avoid mismatches between subtitles and speakers’ lip movements [5–8].
I use intralingual subtitles for languages that I’m learning, languages that I am not secure enough in to trust on listening alone, and in cases there is too much noise in my environment. And especially in the first case, it’s a pain if what I’m hearing is completely different from what I’m reading. I’ve noticed the differences are minimal if you watch with subtitles in the language of origin, but if you watch it dubbed in Language A with subtitles, also in Language A, the difference is so big that it becomes confusing. That said, I’m not a big fan of dubs, anyway.
As for the speed of subtitles, no matter how much research they do, it will always depend on the individual. What one person can easily read is a struggle for someone else.
I agree that Netflix subtitles are usually easier to keep up with than, say, a historical C-drama on Viki, but Netflix has the tendency of now and then suddenly place a subtitle on top of the screen instead of where I expect it and that can also cause me to discover it too late.
A word can be considered as having been processed if at least one fixation hit the word.
• Short words to the left of a fixation (within the perceptual span of three characters) can be considered processed.
• Full words within the perceptual span of eight characters to the right of a fixation can be considered processed even if they were not fixated.
• Short un-fixated words (two or three characters) between two longer fixated words can be considered as processed on the basis of predictive reading, even if they are strictly speaking not inside the perceptual span
The underlying research paper was previously introduced in community discussions here How quickly can a viewer read subtitles?
Over the years I have seen viki viewer complaints about “too long” subtitles significantly decline. There used to be complaints that Viki’s English subbing on Korean drama was unprofessional word by word translation but that has rarely been the case. Rather than subtitled dialogue being a mere gist of what is said I have advised newbies to include the meaning of every word they hear in the subtitle. I have made a conscious effort to make a best effort replication of what the native speaker’s experience in watching a Kdrama is so that if a news broadcast gives background information then our sub should contain the information even if the resulting subtitle exceeds the old school rule of thumb of 45 characters per line. Under traditional guidelines a maximum length subtitle contained two lines of 45 characters each on screen for six seconds. Dozens of studies have shown average adult reading speeds can comfortably accommodate far more characters per second than the old school rule of thumb. I have also considered that watching a subbed drama on one’s own screen gives the reader the opportunity to “ smell the roses “ and to pause to read the note explaining an idiom or historical note which gives viki drama watching a unique flavor. Research conducted dozens of years ago showed that people could understand spoken speech at twice the normal rate of speech so it is natural to assume subtitles which require faster reading can also be comprehended well
@mirjam_465 – Sorry. It was a research study I learned about in the stone ages when I was a graduate student in psychology. I don’t have the exact citation but I remember being intrigued with the results. Here is one of the many studies on the subject from that era: The technical term for speeding up speech is “speech compression.” https://academic.oup.com/joc/article-abstract/18/3/236/4560861 (originally published in 1968)
This quote is from the study “comprehension remained at approximately the same level as rate of presentation increased from 175 wpm to 325 wpm” See – speech rate was nearly double the 175 wpm but comprehension remained the same!! This means you can fast forward almost double and understand everything. Most recently focus on compressed speech has been on the decline of comprehension with age. So some people are advocating a compressed speech test to detect cognitive problems in the aged.
I have been cursed/blessed with the ability to retain what I read so I am chock full of trivia.
Now, Viki and those other streaming sites, with the fashion for Asian content, have trained people to be patient with subtitles. The more foreign films and dramas people watch, the more their ability to read and comprehend quickly is growing.
What I mean is that we have today a large group of people who can consume foreign content with subtitles much more easily - and more quickly - than previously. The speed will not be the same for someone who is watching their first foreign drama and someone who has already watched about a hundred!
At the same time people’s reading and understanding skills in general are decreased. The new generations have a lack of concentration and a lack of reading and understanding complex written texts, that’s why modern novels have very often very short and simple sentences as if it’s for kids (old authors of course still write very good, complex sentences but the new generation of authors tend to write in a very poor way without any interesting stylistically syntax/wording…).
I assume the younger viewers are more focused on pictures so they might interprete the visuals more instead of looking at the subtitles (I’m thinking about Viki’s most popular genres like RomComs and BL. These genres have rarely complex, long sentences and topics and the storyline and character’s behaviour is mostly very similar, not just in BL and RomComs but also in many historical and fantasy stories that I watched from Asia here).
This makes perfect sense to me. While I’m sure that there will be a number of views on this, I personally see it as the difference between subtitles used for educational purposes and subtitles used for entertainment.
From my perspective as a school teacher, subtitles used for educational purposes must be 100% accurate and faithful to what they explaining. From an entertainment perspective, though, subs need to provide me with enough information to understand what is happening on screen and the gist of what is being said. Because I can only speak English, what actors say in other languages completely bypasses my understanding and subs become the window that allows me to enjoy what I’m watching. I also need time to be able to absorb what’s happening onscreen.
When I use video clips for educational purposes, I am always stopping, rewinding and replaying sections because I want to learn something. When I watch something for entertainment, though, if I have to stop, rewind and replay too often, I become frustrated and my enjoyment levels plummet.
Yes. There’s no escaping that this will always be the truth in many of life’s endeavours. Statistically, though, most people tend to be around the “average” of just about everything when a large population is considered. So while there will be very fast and very slow readers, they will be not as abundant as those within a specific range of the average speed. It’s the same with teaching. I always taught at a speed that the majority of my students could cope with. I gave additional, advanced material to those who were quicker on the uptake and I worked individually with those who struggled to keep up.
Thank you for posting this link. I haven’t come across this paper before.
During my teaching years, I did a great deal of professional development on Dual Coding Theory (Allan Pavio) and Cognitive Load Theory (John Sweller). Because of this, I intend to study this paper in detail.
At first glance—and I might be wrong about this—it seems that the paper helps develop understanding about how to maximise learning through using what a person sees in addition to what that person hears at the time of seeing (i.e. Dual Coding Theory).
The paper also mentions cognitive load. Prof. John Sweller’s work in this area is outstanding. Many years ago, I discovered that if I wanted to be a good teacher, I needed to appreciate cognitive load and how much new learning material an average student could cope with during a class. When I started my journey into Sweller’s theory, what initially surprised me were the limitations that cognitive load put on learning (and, more generally, every day life).
So thank you again for this link. I’m quite excited about spending a couple of days working my way through this information.
Thank you for posting this link. You’ve opened a doorway into an area that I haven’t done any research in. I love learning about new things and I’m looking forward to seeing where this door will take.
