An innovative approach to language software for aphasia
This page is part of the SentenceShaper Manual. As it comes near the end of the manual, some things may not be obvious if you haven't yet read the manual. Right-click if you would like to download the SentenceShaper Manual, then choose "Save target as..." (The report is in PDF format.)
Once the basic mechanics of the system have been acquired, subjects may still need additional supervised practice in order to achieve what we will term an aided effect, i.e., the superiority of utterances created with SentenceShaper over utterances created spontaneously. (In contrast, treatment effects arise when subjects’ spontaneous speech improves after a period of SentenceShaper use.) The suggestions below derive from our limited experience in treatment studies involving mild to moderately impaired non-fluent subjects. Different approaches may be required for more severely impaired or fluent aphasic subjects.
Replaying his/her recorded speech may possibly serve to develop a subject’s monitoring skills.
If the user has very poor metalinguistic ability, however, and fails to recognize even gross errors, intensive use of the system might conceivably be detrimental, since it would expose him to ill-formed utterances on a regular basis. Since aphasic individuals who have made treatment gains have often shown increases in the proportion of their sentences that are well-formed, however, we do not believe that exposure to imperfect speech samples is a major concern.
We suggest that users be encouraged to click the Off Button as soon as they "run out of words" and, more generally, that they record fairly small chunks of speech in each shape. This has several advantages:
If a user does record something that is very important to them or to the caregiver, but it does have long pauses, it is possible to edit the pauses out with sound editing software, which can be purchased inexpensively on the internet. It’s beyond the scope of this manual to go further into this topic, but we wanted to make you aware of the possibility.
As indicated elsewhere in this manual, and in publications about SentenceShaper/CS (especially Linebarger & Schwartz, 2005), we strongly suggest that you encourage subjects to use the verb and preposition Side Buttons, for two distinct purposes:
"Underneath the tree is a manger. Above the manger is a star. In front of the manger Mary and Joseph were kneeling. Behind the tree is a plug. From the light to the plug lights are sparkling."
It may be that word associations, which have been shown to play an important role in normal sentence production, are not fully utilized by aphasic patients because they have trouble retaining in memory the words that they have just produced. Therefore, these words decay from memory before they can trigger useful word associations to move the process of sentence construction forward.
We hypothesize that SentenceShaper, which helps to refresh the user's spoken words in memory, may allow the user to draw on these word associations during sentence production. That is, SentenceShaper enhances the effect of word associations because it allows replay of material already produced, allowing the user to build on associations to these words. In addition, SentenceShaper may help the user to draw on word assocations by means of the "light" verbs and prepositions on the Side Buttons. These words, because of their high frequency and semantic flexibility, are especially effective in eliciting lexical associations appropriate to a specific context. For example, in the Christmas tree example above, the various preposition words proved startlingly effective in triggering words for items which that patient typically associated with her Christmas tree.
Note that aphasic users often find the use of the Side Buttons unnatural and disruptive, especially in the early stages of system use. Nonetheless, it may be worth persevering, through constant reminders and by assigning Side Button exercises for homework. In these exercises, the subject is provided with a list of the Side Button words (or a screen shot of the SentenceShaper Assembly Area), and is asked to create one sentence incorporating each word.
SentenceShaper is designed to allow creation of longer and more complex structures than may be possible in spontaneous speech. However, users may need to be pushed toward the creation of more ambitious structures in the beginning, as they may be in the habit of relying on nouns and other fragments to convey their messages.
Both lexical and morphological "perfectionism" should be discouraged. We rarely correct users’ morphological errors, as it may distract from the deeper process of grammatical encoding. More important still, long searches for the "perfect" word are very disruptive to sentence production, and we encourage circumlocution and lexical flexibility. The generality of the Side Button verbs allows them to replace more precise but unavailable words. For example, the Side Button verb "go" can replace many verbs of motion ("run," "walk", "crawl", etc.). Even if a Side Button word cannot be found to replace the elusive word, it may suggest an alternate way to express the speaker’s idea. Such flexibility should be strongly encouraged.
In order to encourage independent use, we suggest that you provide only minimal assistance after the early training sessions, and that this assistance be focused upon development of effective techniques. We suggest that you almost never provide the actual word that the subject is searching for, but rather that you encourage the user to exploit the system to help him deal with the block.
The user should be encouraged to click the Off Button as soon as s/he is finished speaking, to avoid recording silence.
We encourage most (but not all) users to record any word that comes to mind, even if it cannot be incorporated into the sentence at that time. The Work Area can be used to store elements that may be useful later in the sentence.
It is best to leave a few empty slots in both the Sentence Assembly Area and the Narrative Assembly Area, so that elements can added and be re-ordered.
For a longer narrative, it may be useful to create a "draft" of the final version by recording a few words for each sub-part of the final narrative and moving these chunks to the Narrative Assembly Area. Then the user can go back and flesh out each chunk into a full sentence. For example, a first draft of the Cinderella story might include "stepmother" in the first purple bean, "fairy godmother" in the second, "party" in the third, "shoe" in the fourth, etc.
Here we note a few tasks that have been used in previous studies with SentenceShaper and its earlier prototype (CS). The goal in these studies was to bring about treatment effects – that is, gains in subjects’ unaided, spontaneous speech – not simply to develop effective use of the system as a communication aid.
Studies to date suggest that narrative production – retelling of short videos, movies, or television programs -- may be particularly effective in bringing about changes in subjects’ spontaneous speech. Less demanding tasks such as single picture description and formulaic emails ("Hi, how are you? It is raining today" etc.) appear to be less effective than narrative, which requires the speaker to work off a complex internal representation rather than, for example, naming a picture designed to elicit a single sentence.
For subjects who are too severely impaired to retell an entire movie or television show, "embellished picture description" (creating an imaginary "sequel" or "prequel" to a picture or photograph) or the description of multi-picture sequences or wordless picture books may introduce a degree of message-level complexity while reducing the memory load imposed by retelling materials without any visual stimulus.
Subjects who can only produce a word or two at a time during their first attempt at producing a sentence may benefit from going back and re-recording separate words (each stored in its own shape) into a single sequence (stored with one shape). This may require frequent replay of the individual shapes. This may impact not only linguistic fluency but also morale, as many subjects find it encouraging to hear themselves speaking in longer, prosodically correct utterances.
As noted above, we strongly encourage exercises targeting the Side Buttons, such as providing subjects with a screen shot of the Assembly Area or a list of Side Button words and asking them to create a sentence with each word, checking off the word on the list or screen shot each time.