TheoriesScenario 1: Email | Scenario 2: Web Browsing | Claim Analysis
Scenarios
Tom is an experienced computer user interested in new technology. His latest investment is the voice recognition software DD (Dragon Dictate). Applying this tool Tom hopes that he can increase his productivity and make his work more enjoyable.
Tom gained his first impressions on the software’s capabilities during the tutorial session. These were very promising. The only point in the tutorial which made him feel uneasy is the following statement:
"When you’re dictating, always correct every recognition error that DragonDictate makes. If you don’t fix recognition errors, DragonDictate’s accuracy gets worse!"
Nevertheless highly motivated he looked forward to a real use trial. Because Tom spends a lot of his time with email communication he decided to try Dragon Dictate first with Eudora.
Tom’s expectations were high and he thought, "Using the voice recognition software the email application can be launched: Eudora will open and display the incoming mail. Using verbal commands I can:
I won’t have to use the keyboard or mouse any more. Email communication will be more convenient and efficient."
Tom’s approach to his goal is described in the subsections titled "Subtask 1", "Subtask2", "Subtask 3".
Subtask 1: Open the email application Eudora and read the new email
Tom tries to start Eudora using the [Bring Up] command he learned from the tutorial session but the program doesn’t recognize the command [Bring Up Eudora]. Instead it misinterprets the command as [Bring Up Tutorial] and therefore launches the tutorial again. After several minutes of fault finding without fixing the problem, he consults the manual. The manual tells him to place a link to Eudora in the Dragon Dictate Program Group (Claim 1). Even after doing so Dragon Dictate still misinterprets this command. Using his intuition Tom changes the name of the link from "Eudora" to "mail" and finally it works – [Bring Up Mail] starts Eudora. The program indicates that a new message has arrived. He issues the voice command [Enter] and the message, highlighted with the selection bar, opens (Claim 2). Tom uses the [Move Down] and [Stop] commands to scroll through the message. Despite his initial problem he is pleased with his first success and experiences a feeling of satisfaction.
Subtask 2: Open the editor window and dictate the reply
Tom tries to move the mouse cursor to the reply icon using voice commands. He realizes that this approach is inconvenient and time consuming, so he decides to use the menu commands instead (Claim 3). Tom searches through the menus of Eudora for the appropriate commands. In the Message Menu he finds and issues the command "reply".
After entering the dictate mode he commences to dictate the letter and immediately encounters severe recognition problems. To change the words misinterpreted by the voice software he applies the command [Scratch That] several times hoping that DD will learn from his corrections (Claim 4). The performance does not improve so Tom consults the manual and finds that the software only learns if he uses the wordlist or the [Oops] command. Some of the words Dragon Dictate is unable to understand even after a number of retries. In these cases Tom uses the [Spell Mode] to enter the letters of the misinterpreted word. He is disappointed that he has to work with the ICA (international communications alphabet) commands to spell the word instead of just the English alphabet (Claim 5).
Before sending the message Tom wants to change the first line. From the tutorial session he remembers a command called [Bottom of Document] and concludes that there should also be one with the name [Top of Document]. Tom is pleased to find his assumption confirmed.
Subtask 3: Send the reply and store the new email in one of the archive folders
For sending the reply Tom uses the same technique – he directly activates the menu message item ‘send immediately’ via voice. Now he wants to catalogue the email under a new folder. He makes a new folder using menu items and tries to name the folder. DragonDictate doesn’t recognize the word so he starts using ICA but still has problems remembering it (Claim 5). He transfers the email to the new folder using commands from the menu. Tom decides that the name of the folder is not appropriate so he tries to change it. Not willing to use the mouse he looks for a way to set the focus on the folder's name. After a period of time he finds the menu command to move the focus to the window displaying all mail folders (Claim 2). When he issues the right mouse button command in order to open the folder’s context menu the mouse cursor is still in the other window. Therefore he doesn’t get the action he wants. Instead Tom has to recall the keyboard shortcut for editing, ‘F2’, to rename the folder.
Finally Tom successfully finished the task of reading an email and sending a reply. Although he had to deal with a lot of problems and spend almost a whole afternoon for his first trial he feels relief rather than frustration. Nevertheless his euphoria is now weighted by experience. After using the software in a work situation he understands that it isn’t as easy to use as he thought it would be. Tom realizes that the software can do what he wants if he puts more time into training. More voice profile training and personal learning effort is required to achieve satisfying results (Claim 6).
Last week a friend told Tom about a software update for DragonDictate that was available over the Internet. As a user of this software, Tom wants to upgrading his system and decides to search for the update. He sits down at his computer and positions his headset and microphone comfortably on his head. With a few mouse clicks, Tom starts the DragonDictate program and waits for the voice bar to appear at the top of the screen. Soon he sees the familiar bar indicating the microphone is on and the software is loaded. To start the Internet browser, Tom gives the command [Bring up Internet Explorer]. To his surprise, the voice bar shows only a series of question marks indicating that the command was not recognized. Tom repeats the command, but again he sees the question marks. Twice more he speaks the command but with no success. Finally, Tom realizes that he forgot to change the user profile (Claim 7). His wife was the last one to use DragonDictate, so the software defaulted to her profile. With a few more mouse clicks, Tom activates his profile and starts the web browser.
Subtask 1: Searching the Internet
Since Tom does not know the location of the DragonDictate home page, he decides to search for it. He gives the command [Open] followed by [Yahoo] and soon the search page appears on the screen. Next, Tom needs to move the mouse to the query box. After saying [Mouse Grid], the screen is divided into nine numbered sectors like a tic-tac-toe board (Claim 8). Tom chooses the third sector by saying [3]. Immediately, this sector is subdivided into nine more sectors. With one more iteration, Tom is able to move the mouse pointer to the query box. Once there, he changes the software to [Dictate Mode] and pronounces each word of his search phrase: "dragon", "dictate", "software", and "upgrade". He says [Enter] to submit the search, but he is surprised to see the word "enter" appear on the screen. Tom realizes that he has not exited the [Dictate Mode] (Claim 9). He changes modes by saying [Command Mode], erases the word "enter" with several [Backspace Key] commands, and then submits his query. When the search results appear, Tom scans the page to find the link he wants. Since the pointer is just left of the link, he says [Mouse Left] to move the pointer over the link. As the pointer crawls over the link, Tom repeatedly says [Stop], but the system does not respond. The pointer finally stops at the left side of the screen. Frustrated, Tom reaches for the mouse, moves the pointer to the link, and clicks (Claim 3).
Subtask 2: Downloading the Upgrade
The download page appears, and Tom skims the text for the download procedure. The web site requires the completion of a registration form before download. Tom says [Mouse Grid] and the familiar grid appears. While he is homing in on the target link, the phone rings. Tom gives the command [Go To Sleep] to turn off the microphone and answers the phone. When he returns and says [Wake Up], the mouse grid has disappeared (Claim 10). Using the [Mouse Grid] commands again, he reaches the link and loads the registration form. With a series of [Tab] commands, Tom reaches the top of the form. He changes to [Dictate Mode] and says "Tom" to enter his name. Between each of the fields, Tom switches between the [Command Mode] and the [Dictate Mode] to complete the form. However, when he reaches the city field, the software does not understand the name of his city. After some time, Tom resorts to entering the name manually by spelling it with the ICA rather than correcting the system (Claim 5). When the form is complete, Tom says [Mouse Grid] to move the pointer to the download link. However, when the grid appears, the link button is divided by the grid lines. Tom pauses since he is unsure which sector to name. Finally, he chooses sector [2], homes in on the link, and downloads the file.
Subtask 3: Creating a Shortcut
Tom wants to remember this site for future reference, so he decides to add the page to his favorites list. He gives the command [Add to Favorites], and a new dialog box appears (Claim 11). He chooses the command [New Folder] and another dialog appears for the folder name. Tom changes to [Dictate Mode] and says "DragonDictate," but the software does not recognize the word. Tom decides to use the ICA to spell the name and changes back to [Command Mode]. This procedure takes some time, however, since Tom cannot remember all the keywords, and some of the words are not recognized. Tom corrects the wrong words, but he has to use the [Backspace] command to remove the incorrect characters that were displayed (Claim 12). Once the folder appears, Tom says [OK] to add a link to the page in his new folder.
(1) You can start applications with DD after placing a short cut to the application into the DD folder
+ can open any application with DD
+ limits the size of the vocabulary DD has to deal with, improves
recognition
+ can choose a different name for an application (eg. Word
instead of Microsoft Word)
- have to the read manual to know that a
short cut is required in the DD folder
- have to know how and where to create a short cut
- have to do this for all applications
(2) DD facilitates navigation in the window using screen object names and keyboard key names
+ verbal sequence follows Direct
Manipulation actions of the mouse on menus
+ the active user does not need to learn new commands
+ supports immediate use of software
- access of icons is impossible
- exploratory learning is required to recognize that this feature
is available
- people who use toolbar icons may not know the menu structures
(3) Using voice commands to move the mouse pointer
+ provides intuitive and consistent
direction controls for moving the pointer
- poor accuracy, targeting problems
- mouse movement speed is low
(4) Having the software learn from every spoken word, instead of only from corrections
+ system can faster adopt to the
user’s voice
- performance gets worse if user
doesn’t correct recognition errors
- user might feel pressure to correct every error
(5) The ICA is always used for spelling
+ increase recognition accuracy of the
alphabet
+ the use of letters of the alphabet is consistent throughout the
application
+ user might be interested in learning this standard code for use
in different situations
(eg. spelling his name on the telephone)
- the user might not be interested in the
ICA at all
- the user has to learn the communications alphabet
- learning to use ICA slows performance
(6) Using an adaptive learning strategy for user profiles
+ allows the system to improve
recognition accuracy over time
– requires significant training to
achieve acceptable performance
– can be very inconvenient since users must initiate error
correction
– can contribute to poor performance if users do not correct
errors
(7) Defaulting to the last user profile
+ is convenient for single users
– can easily allow users to start with
the wrong profile
– can promote corruption of profiles by wrong users
– can frustrate users who forget to change profiles
(8) Controlling the mouse movements with a grid
+ allow precise control of the mouse
pointer
+ allows for faster control of the pointer than direction
commands
– is still much slower than the mouse
– can cause confusion when the objects of interest appears
in two sectors
(9) Changing between modes in the control languages
+ limits the vocabulary the system must
understand by context
+ allows the system to easily distinguish between words and
commands
– users easily forget to change modes
– multiple mode changes are very cumbersome
(10) Removing the mouse grid after returning from sleep mode
+ restores the system to status quo when
the users breaks from work
– may cause the user to repeat
previous work
(11) Shortcut commands such as [Add to Favorites]
+ allow rapid access to program
functionality
+ consistent since name mirrors the menu option
– may be hard to remember since they
are not displayed on the screen
(12) Failing to reverse incorrect commands in the [Command Mode]
+ simplifies the required
"intelligence" of the voice control system
(e.g. the system does not need to deduce how to
reverse a paste command)
+ is consistent with commands that cannot be reversed (i.e. Save
File)
– leaves the correction to the user
– is inconsistent with system behavior in the [Dictate Mode]