Sentactics
A Computer-based Language Treatment for Individuals with Agrammatic Aphasia
By Cynthia Thompson, Audrey Holland and Ron Cole

Project Team:
Cynthia Thompson, Ron Cole, Wayne Ward, Janet Choy, Sarel van Vuuren, Nattawut Ngampatipatpong, and Jariya Tuantranont

Project Overview:

Sentactics is a computer-based clinical treatment that is designed to the improve sentence comprehension and production abilities of individuals with Broca's aphasia and agrammatism. Individuals with agrammatism, often resulting from lesions in Broca's area in the brain, have syntactic deficits. Sentactics is a computer implementation of a clinical treatment developed to address these deficits by Professor Cynthia Thompson, Director of the Aphasia and Neurolinguistic Research Laboratory, and her colleagues at Northwestern University. The treatment approach, the "Treatment of Underlying Forms" or TUF focuses on training patients to comprehend and produce complex, non-canonical sentence structures like "It was the woman who the man saved from drowning." TUF operates on the premise, substantiated by over two decades of research, that training underlying abstract properties of language in sentences with more complex syntactic structures will result in effective generalization to untrained structures that share similar linguistic properties, particularly those of lesser complexity. Interested readers can learn more about CATE, the Complexity Account of Treatment Complexity, in an article that provides the theoretical rationale for TUF. A more recent article by Thompson and Shapiro (2005) provides an excellent summary of the theory and research that have led to TUF and its implementation in the Sentactics program. According to Thompson and Shapiro:

"Our treatment investigations use a single subject experimental design in order to allow us to directly examine generalization as it emerges during treatment, while experimental control is maintained. From this work we have learned the following: (a) treatment improves production (and comprehension) of the sentence types entered into treatment, (b) generalization to untrained sentences occurs to those that are linguistically similar to those trained, (c) generalization is enhanced when the direction of treatment is from more to less complex structures, (d) treatment results in substantial changes in spontaneous discourse in most patients, (e) treatment appears to affect processing of trained sentences in real time, and (f) treatment gains can be mapped onto the brain using functional magnetic resonance imaging (fMRI)."

Development of Sentactics resulted from a collaboration between Cynthia Thompson at Northwestern University, Audrey Holland, Professor Emeritus at the University of Arizona and Ron Cole at CSLR. Janet Choy (Northwestern) and Nattawut Ngampatipatpong and Jariya Tuantranont (CSLR) helped to develop the program. The program is supported by Grant 5R21DC7377-2 from NIDCD, the National Institute of Deafness and other Communication Disorders (Cynthia Thompson, Principal Investigator). The project is now in its second year, with clinical trials now underway.

Sentactics involves "metalinguistic" methods to train aphasic participants to comprehend and produce complex sentences such as object relatives, object clefts and object wh-questions. The active form of target sentences is used to train participants to (a) comprehend and produce the verbs and the NPs representing the arguments of the verb in each sentence, (b) move the proper sentence constituents to form target sentence structures, (c) produce the surface form of the target sentence, and (d) comprehend and produce the verbs and verb arguments in their noncanonical position.

The figures below illustrate the format of the computer-based treatment within the Sentactics program.

Figure 1

Before training begins, computerized probes are presented to assess comprehension and production of both trained and untrained sentence types, using semantically reversible picture pairs as shown in Figure 1. An automated "clinician" (Sabrina) presents spoken (synthesized speech) instructions for participants to follow. In the comprehension probes, two pictures, a target picture stimulus and its foil are presented for each probe item and the participant is required to point to the picture depicting the target sentence.

For the production probes, a sentence production priming paradigm is used. Again two pictures are presented. Sabrina produces a sentence for one picture and the participant is asked to produce a sentence for the other picture using the syntax modeled. In these pre-training probes, no feedback is given to the participants.

Sentactics training starts once the probes are completed.
The training trial begins by presenting two pictures, a target picture stimulus and its foil. A comprehension probe and a production probe are given for the two pictures. This time, feedback is given for participant responses, for both correct and incorrect responses.

After feedback is given for the comprehension and production probes, the foil picture is removed, and the simple canonical sentences representing the active form of the target picture are presented on "sentence constituent cards" (see Figure 2, for example). For instance, "Pete saw the mother" and "The child painted the mother" are the simple canonical sentences representing the active form of the picture for "Pete saw the mother who the child painted". Participants are then asked to read each simple canonical sentence and to identify the sentence constituents of each sentence, such as the verb and the verb arguments (subject and object of the verb).

Figure 2

Sentactics then trains aphasic patients on how to build a complex sentence like "Pete saw the mother who the child painted" from the simple ones. To do this, Sabrina first demonstrates steps for deriving the complex noncanonical form from the simple active forms by moving the sentence constituents around. Participants then read aloud the derived structure. In addition, the participants are asked to identify the verb and verb arguments in their noncanonical position.
Finally, the sentence constituent cards are rearranged in their original canonical order and participants are instructed to form the complex sentence. Assistance is provided as required.

To end each training item, the foil picture is re-presented and the comprehension and production probes which started each training item are repeated. Feedback is again given for both comprehension and production probes.

Figure 3

Throughout training, computerized comprehension and production probes are administered to assess progress. Sentactics automatically records responses to these probes as shown in Figure 3, and generates graphs of performance patterns as shown in Figure 4. In addition, Sentactics documents patients' usage of the program by recording the session length and number of items trained in each session.

Figure 4