VOICE-TO-TEXT RESEARCH PAPER

 

rm3.png
rm3.png

 

This research paper looked at the use of the voice to text function of phones in the context of age, against manual text entry. There were three hypotheses that were formed: one, that manual text entry times are slower than the entry times for voice-to-text, two, that manual text entry would be faster in younger users than for older users, and three, that text entry using abbreviations (like lol, omg, ttyl) will speed up manual text entry, but will not show a similar benefit for voice-to-text entry.


Experiment One: Qualitative Interview


There were three sub-experiments outlined in this paper. The first experiment involved administering a qualitative interview to participants. This interview asks about the participants’ age, if they used voice-to-text (V2T), how they discovered V2T, which devices the participants possess which has the capability for V2T, if there are situations where V2T is helpful or not appropriate, and questions about the environmental factors while they use V2T (if applicable). The qualitative interview revealed that users have concerns for privacy and believe that V2T was inaccurate at transcribing their speech.


Experiment Two: Qualitative Survey


The second experiment employed a qualitative survey to participants, which asked about items such as their age, their native language (including a self-assessment of their English ability), and their beliefs about cell phone usage. This survey gets into more detail about the participants’ usage of V2T, including a self-assessment of their manual typing and V2T speeds. This survey revealed information about the texting styles of participants who are younger verses participants who are older. Older participants reported their manual texting speeds as slow, while also describing their typing style as using a single index finger for manual text input. This part of the study also confirmed that a majority of users who use V2T do so through their cell phone.


Experiment Three: Experimental Study


The third experiment was the main method of exploring our research questions more quantifiably (compared to the more qualitative interview and survey portion), which focused on age as an important construct involved in the speed of both manual and V2T text entry. The 37 participants of this study were divided into two experimental groups: participants over 31 years of age were classified as “old”, while participants under the age of 31 were assigned to the experimental group “young”. Participants were instructed to input six different text messages, where three of these sentences invite the participant to use abbreviations, while the other three do not. The experimental study was comprised of 12 trials: six of the trials involved the participants manually inputting the provided text messages into their phones, while the last six trials involved the participants using voice-to-text entry on their phones to copy the messages. The participants were instructed that using abbreviations in their text messages are permitted, as long as this results in the same reading and punctuation of the original message. The participants were timed using a stopwatch and their resulting text messages were scored for errors.​


Results and Discussion

A 2x2x2 MANOVA was performed for this study, with the within-subjects factors being the text input method/texting mode(V2T vs. manual) and the message type abbrevationality, (invites abbreviations vs. does not invite abbreviations), against the between-subjects factor of age group (young vs. old participants).


The statistical results of the experimental study confirmed all three initial hypotheses. The main effects for both texting mode and abbrevability were found to be statistically significant, moreover, the interaction between text mode and age category was also found to have clear, statistical significance. The results of the first MANOVA confirms our first hypothesis that voice-to-text entry is faster than manual text entry.


Additionally, the interaction effect between age and texting time and mode also confirmed our hypothesis that younger people enter text more quickly than older people. Voice to text texting speed also differed, with mean text entry times being lower for younger participants, compared to old participants, however the amount of variability between the groups was remarkably smaller between them, compared to the manual text entry condition.

The statistical results also showed lower mean texting times for typing messages using abbreviations through manual text entry, with no statistically significant difference in texting time for voice-to-text entry while using abbreviations.