Research about SentenceShaper as a communication aid

When we ask whether SentenceShaper is a useful communication aid, the most basic question is: “Can people say things with the program that they could not otherwise express?”  Even if the answer is “yes”,  of course, we will also want to find out whether this makes a difference in their lives.

We will use the term “Aided effects” to refer to the difference between a person’s speech on the computer (with SentenceShaper) and their spontaneous, unaided speech.

Note that we are not talking about “BEFORE versus AFTER” here, but rather “WITH versus WITHOUT”.  Asking if SentenceShaper has aided effects is exactly like asking whether someone can walk better with a cane than without it. The cane is an effective support if it helps the person to walk while they are using it.

This page gives you a very brief overview of research about the program’s aided effects. For more details, click here to go to a page where you can download our Research Report,  or here to see a list of publications about SentenceShaper.

Does SentenceShaper help people create speech?

In the first, and most controlled, examination of this question (Linebarger, Schwartz, Romania, Kohn, Stephens, 2000), six people with agrammatic aphasia learned how to create spoken narratives with SentenceShaper, and then practiced using the program for approximately 15 hours. After this, they retold wordless videos  (which they had never seen before) with and without SentenceShaper. The aided narratives (those produced with the program) were longer and more grammatically structured than the unaided versions (those produced spontaneously, without use of SentenceShaper) for five of the six participants, despite the fact that the version of the system used in this study provided no word-finding help at all, just memory support. For two of these participants, the contrast between aided and unaided narratives was quite striking.

For example, participant DD was describing a scene in which a boy enters a fish store while the owner is hitting a fish with a mallet. The boy reaches for the fish, and his hand is hit by the mallet.

When DD described this scene without SentenceShaper, she said: “’Ooh! A fish! Ah, water’” and….uh mmm and attendant, “’here’,” and bumped his head. “’Oh boy, oh my hand, my hand, my hand.’”

In contrast, her description of the scene created on SentenceShaper was “The boy and the fishmonger is taking the fish. The boy hit his hand.”

Robust aided effects were also reported in Linebarger, McCall, & Berndt (2002a), Linebarger, Schwartz, Kantner, & McCall (2002b), Linebarger, Romania, Fink, Bartlett, & Schwartz (2008), and Albright and Purves (2008).

The first video on the “Hear it for Yourself!” Videos page contains an audio excerpt from the unaided-versus-aided comparisons in Linebarger et al (2002b). Like the comparisons in Linebarger et al (2000), these are experimentally pristine — the human examiner did not provide any help. However, the program did provide a small set of word cues on the side buttons, containing very general verbs and prepositions (no vocabulary specific to the picture), whereas SentenceShaper had provided no word-finding cues in the 2000 study.

What SentenceShaper does best: Facebook travelogues, not fast food orders!

It  can take a long time to create a message with SentenceShaper. In fact, “turning off the clock” is one of the program’s great strengths: the user is not fighting to hold onto words before they decay from memory.

Therefore, we believe that the program is most likely to be effective for messages that can be anticipated in advance (e.g., information that needs to be conveyed to store clerks, doctors, lawyers; speeches, toasts, advice; email and web postings); it is less likely to be helpful in highly interactive conversations on topics which the user cannot anticipate in advance, and it is not really needed for short, high-frequency phrases.

To put it more concretely: SentenceShaper can be very useful — perhaps uniquely so among communication aids — if you want to create a wedding toast for your son, or post a spoken travelogue about your vacation on your Facebook page. But if you just want to order coffee at Starbucks, all you really need is a set of prestored phrases or sentences; you don’t need a working memory prosthesis.

Does SentenceShaper help people self-cue “live” speech?

For those people who do create better speech with the program than without it…what next? That is, how can they use the messages that they create? This issue is explored more in the section about using SentenceShaper as a communication aid (click here), so we will just note briefly that there are two main things you can d0 with a recording of your speech: you can play it back directly to the listener, or you can practice it thoroughly ahead of time and then use it as a kind of teleprompter to speak your message “live”.  That is, you would listen to the recording through an earphone, then try to say it aloud.

A study at Moss Rehabilitation Research Institute explored this second way of using messages created on SentenceShaper. We developed a software program allowing people with aphasia to download onto a portable handheld computer spoken utterances that they had created on SentenceShaper. This handheld computer was then used to deliver their recorded messages in a set of fictitious real life situations (for example, returning an iron to the store and explaining why). Some encouraging preliminary data from this study have been reported (Bartlett, Schwartz, Fink, Lowery, Linebarger, & Schwartz, 2007; Linebarger, Romania, Fink, Bartlett, & Schwartz, 2008; True, Bartlett,Fink, Linebarger, & Schwartz, 2010).

Incidentally, the current version of SentenceShaper also allows you to play your SentenceShaper messages on a handheld device. See the section on “Sharing your Speech”  for details.

Side note…. is there an aided effect for speech recognition?

A pilot study (Dahl, Linebarger, & Berndt, 2008) suggests that SentenceShaper may allow some people with aphasia to create utterances that are more intelligible to speech recognizers than the same individuals’ spontaneous, unaided speech. Because reading and spelling impairments are widespread in aphasia, speech recognition (along with text-to-speech software that plays text aloud) may eventually make it possible for people with aphasia to interact with others in text-based environments such as the Internet. Also, improved speech recognition could open the door to treatment interventions that utilize natural language understanding software.

Summary

Research studies have demonstrated that some people can produce much better speech when they are using SentenceShaper than they can produce on their own, even if the program is modified to provide just memory support, with no word-finding assistance. This is an exciting finding in itself, because it shows that people may retain far  more linguistic ability than we hear in their day-to day-speech.

As with most communication aids, people vary in how much support they will need to achieve an aided effect and to use their recorded speech in real life. More research is needed in order to identify the best way to use this recorded speech in real life, and to assess its impact on quality of life.

This impact need not be purely logistical — regardless of whether a communication aid is helpful in accomplishing activities of daily life, it may play a crucial role in a person’s social and emotional well-being.  The need for a way to demonstrate one’s preserved intelligence and personality cannot be underestimated. Many people assume that the fragmented and halting speech of those with aphasia means that they are intellectually disabled and perhaps even child-like. For some individuals, SentenceShaper provides a unique opportunity to reveal their preserved intelligence, humor, and compassion. This is illustrated rather strikingly in the “Hear it for Yourself” Videos and the “Stories” sections of this website.