±«Óãtv

Word At A Time Subtitles

Author:

Abstract

We present a prototype for a novel form of subtitle presentation for mobile using the Rapid Serial Visual Presentation (RSVP) technique common to a range of speed reading applications. We discuss it's suitability for subtitling as well as presentation techniques to adapt it for subtitling purposes. We outline further areas of research that would need to be undertaken before RSVP could be considered a candidate for subtitling.

Introduction

Our audience research indicates around 90% percent of subtitle users use subtitles for the purpose of aiding comprehension of content. These subtitle users typically use a combination of visual and audio cues, including lip reading in order to comprehend the content. We decided to do some exploratory work into the use of rapid serial visual presentation as a technique for subtitling which would allow the subtitle user to spend more time viewing the content on screen which we hope will both increase immersion in the content and give the viewer more time to pick up on other visual cues to aid their comprehension.

Rapid Serial Visual Presentation

RSVP has made headlines recently [1] with commercial applications [2] claiming to help people speed read at between 250 to 1000 words per minute. The claim is that you spend 80% of there time moving between words in what is known as a "saccade" and 20% of the time processing the content. By presenting the words in the same location in a rapid serial fashion the hope is that the time spent on saccades between words is removed and you are in theory able to read much faster. There is also some work done to align the sequence of words on their "optimal recognition point", which is the point within a word that our eyes are supposed to seek in order to comprehend them.

RSVP has also been met with criticism. Critics claim that comprehension is negatively impacted [3] by the style of presentation, and that the ability for the eye to move back through content in traditionally presented text is important, especially when reading about concepts or words the reader may be unfamiliar with. 

Figure 1.  Example of our subtitle presentation style

RSVP For Subtitling

A common complaint for subtitle users is that time spent reading subtitles often distracts them from the content being presented on the screen and while they are reading on-screen actions can be missed. The ±«Óãtv's on-line subtitling editorial guidelines specify a maximum speed of 140 words per minute [3] (with 180 being permitted in exceptional circumstances). If the claims of speed reading applications are accurate and reading rates can be pushed to 500 words per minute and upwards it presents an enticing area of research for subtitling.

An important part of reading with RSVP style is that the eye must always be looking at the presentation area otherwise words may be missed which could cause lack of comprehension. This leads to difficulties when using RSVP for subtitling as the eye naturally follows the action on screen that could cause the user to miss critical words. To help to overcome this in our prototype we've chosen to place the subtitles covering the majority of the screen [Figure 1]. This will hopefully obscure the areas of interest in such a way as to direct the reader’s attention to the subtitles, but for a significantly shorter amount of time than they would of spent reading traditional subtitles.

We developed this work on mobile because the variation in screen real estate between devices often means subtitles are presented smaller than ideal, so our choice to present over the majority of the screen ensures a suitably sized font for all devices.

Figure 2: (left) Subtitles as viewed on mobile, (right) A sequence of successive subtitles. 

Observations

Our final prototype presented the words at 600 words per minute and begun displaying the subtitles as the characters began talking. We chose to subtitle the “Inside No 9” episode “The Harrowing”, a 30-minute comedy drama. 

While watching the episode on a mobile it was noted that unfamiliar character names and words were difficult to read. This might suggest that non-fictional and educational content may suffer if subtitled this way.

The style of presentation worked well for drawing attention to the subtitles and for many of the subtitles the ability to see the actors mouth after they had been presented felt like it helped  with comprehension. Instances where it was hard to see an actors mouth felt more difficult to read.

Sometimes the pacing of certain words felt off, this is probably due to each word being presented for the same length of time, this could possibly be remedied by giving a longer presentation time to longer words.

Further Work

RSVP is one of many possible techniques for subtitling ranging from traditional presented subtitles, to subtitles which become an immersive part of the content. The work we've undertaken so far looking into RSVP for use on mobile is a small area of the many possible novel subtitle presentation techniques that could be explored.

Undertaking this work has helped to broaden our view of the range of potential subtitling techniques as well as raising a number of questions.

To take the work further we would need to perform user testing to validate the effectiveness of this subtitling technique and confirm or deny our observations. We would explore the comprehension of the subtitles on users with a range of reasons for using subtitles and compare this against traditionally presented subtitles. 

We would also look at the effectiveness of the technique specifically for the purpose of subtitling on mobile, comparing it against various other strategies for mobile subtitling.

Further research could also be conducted to validate the claims and criticisms of RSVP reading applications such as:

• Eye tracking to determine the lack of saccade in reading an RSVP style.
• Eye tracking to validate that words have an "optimal recognition point".
• Evaluating the level of comprehension of content read using RSVP with a focus on the style of content being presented.
• Determining the style of content best suited to reading with RSVP.
• There are also a number of variations within this subtitling style that could be explored to refine the experience further. Examples of variations could include 
• Words per minute of the subtitles.
• Presentation style (full screen, partial screen, bottom of screen).
• Alignment of the optimal recognition point of words.
• Timing of words.

References

1. ±«Óãtv News, Could Spritz boost your reading speed?.
2. Spritz, The Science. 
3. Ars Technica, Speed reading apps may kill comprehension.
4. ±«Óãtv, Inside No 9 - The Harrowing.