Structured treatment can improve predicate argument structure impairment

Abstract

Background: Complementation information is a type of lexical-syntactic information that determines which syntactic environments a verb can be inserted into. It includes information about the verb’s predicate argument structure (PAS), the thematic role of each argument, and the verb’s subcategorisation frames. Complementation information might be impaired in aphasia. Various procedures have been used in treatment studies aimed at improving the production of verbs or sentences (including the verb with its arguments); these have usually yielded an improvement in treated verbs, without generalisation to untreated verbs.

Aims: In this study, we evaluated a structured treatment procedure, aimed at improving impaired PAS information, and examined whether it improved the production of verbs with their arguments in sentences. The research questions were these: (a) Will treatment improve the production of verbs with their arguments in sentences? (b) Will improvement generalise to untreated verbs? (c) Will improvement be maintained after the end of the treatment? and (d) Will improvement generalise to connected speech (in a task of telling a story in response to a series of pictures)?

Methods & Procedures: Two chronic aphasic patients with impairment in PAS information participated in the study. Their complementation information, and specifically their PAS information, was assessed prior to treatment, immediately after treatment, and in follow-ups 6 weeks and 6 months after treatment. The treatment consisted of instruction and practice on the number of arguments different verbs select, and taught the participants a strategy they could use. The practice was organised hierarchically, with regard to the number of arguments a given verb requires and the amount of cueing given.

Outcomes & Results: Following treatment, a significant improvement was found in the participants’ ability to produce sentences with the correct number of arguments. This improvement was generalised to untreated verbs and to connected speech and maintained for (at least) 6 months after treatment. However, neither participant showed improvement in other language skills, or even in other types of complementation information (i.e., subcategorisation frames).

Conclusions: The findings suggest that structured treatment focusing on PAS information can improve the use of this specific type of information, manifested in the production of sentences including the verb with its arguments. Theoretically, the findings support the view that complementation information is represented separately from other types of language information, and they suggest that different types of complementation information might be represented separately.

Keywords:

• treatment
• complementation information
• predicate argument structure
• aphasia
• Hebrew

Acknowledgements

We thank Naama Friedmann for her constructive advices throughout the work on this research. Many thanks to Aviah Gvion, Dror Dotan, Noga Balaban, Ronit Szterman and Maya Yachini for fruitful discussions of this research and helpful comments on the article. Thanks also to Maya Menahemi for her help with participants recruitment.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

^1. Because the control group in Biran and Friedmann (2012) study performed at ceiling level on the tests, with small or no variance, it was impossible to use t-tests and consequently to apply dissociation analyses that are based on Crawford and Howell’s (1998) t-test (Crawford & Garthwaite, 2007). Therefore, Biran and Friedmann’s analyses were based on the decision that for individuals with brain damage, a score lower than 85% indicates impaired knowledge (along the lines advocated by Willmes, 1990.) (See Biran & Friedmann, 2012 for elaboration.)

^2. We included in the treated items also verbs that yielded correct responses in the pretreatment assessment to make it possible for the untreated verbs to improve. (As if all correct responses were in the untreated group, it might have been impossible to reach a significant improvement in this group, given the relatively small set of items. Biedermann & Nickels, 2008; B. Biedermann, personal communication, September, 2012, June, 2014). However, there were only few items with correct responses in the pretreatment assessment.

As for frequency and length, post-hoc analyses, for each participant, revealed no significant difference in frequency or in length between lists of treated and untreated verbs. (For HY—frequency: t(18) = 0.34, p = .73; length: t(18) = 0.34, p = .19. For AB—frequency: t(32) = 1.3, p = .2; length: t(32) = 0.16, p = .87).

^3. In Hebrew, gender and number manifest themselves in different verb, adjective, and pronoun inflections, which agree with the noun they are syntactically connected to in the sentence. In this study, each verb agreed in gender and number with the subject of the sentence. Sentences included inflections for masculine and feminine (the two genders that exist in Hebrew) and three numbers (singular masculine, singular feminine, and plural). All verbs were presented in third person.

^4. This significant improvement was also found when considering the full lists of verbs that were administered to HY in each of the assessments. In the first follow-up assessment, HY produced 76% grammatically correct sentences (48/63)—significantly better than pretreatment (7/52), χ² = 31.25, p < .0001. In the second follow-up assessment, she produced 68% grammatically correct sentences (21/31)—still significantly better than pretreatment, χ² = 25.6, p < .0001, and not significantly different from her performance immediately after treatment and 6 weeks after it (p > .05).

^5. AB produced all the sentences without a subject (without the external argument, the agent). Considering only the production of the complements of the verb (the internal arguments), he produced 20/46 sentences with the correct number of complements. (A comparison of this performance with his performance after treatment was also significant.) Note that Hebrew is considered to be a partial pro-drop language, in which first and second person pronominal subjects may be omitted in past and future tense (Melnik, 2007). As mentioned earlier, all verbs in the sentence production test were given in third person.

^6. This significant improvement was also found when considering the full lists of verbs that were administered to AB in each of the assessments. In the first follow-up assessment, AB produced 88% grammatically correct sentences (43/49)—significantly better than his performance before treatment (0/46), χ² = 73.75, p < .0001. In the second follow-up assessment, he produced 61% grammatically correct sentences (41/67)—still significantly better than pretreatment, χ² = 44.18, p < .0001, but also significantly lower than immediately after treatment and 6 weeks after it (p < .01).