Morphology: an educational and clinical support tool for speech therapy

Guillaume Duboisdindien

Context: Linguists have traditionally distinguished between inflection and derivation (Allen & Badecker, 2001). Researchers generally agree that inflectional affixes predominantly have a syntactic function, while derivational morphology primarily has a lexical-semantic function. In the context of derivational morphology, neuroimaging studies have shown that, in recognition tasks, not all derived words are subjected to the same way of analyzing / decomposing derived morphemes. Indeed, different properties would affect the probability that a derived word is processed, decomposed, or memorized as a whole word. Some of these properties are related to the phonological and semantic transparency of the derived word to the base word. Phonologically and semantically transparent words are more liable to decompositional processing.

Moreover, over the last fifteen years, a constant number of studies have been collecting developmental data and considering cognitive-linguistic models of the role of derivational morphology on language (Reed, 2008; St-Pierre, 2009) both in children and – more marginally – in adults. In addition, promising clinical and pedagogical research avenues are exploring intervention models for derivational morphology disorders (primary or secondary focus – Galuschka & Schulte-Körne, 2016) or the potential of derivational morphology to support significant skills such as spelling or vocabulary (as a better predictor of academic success) or the ability to generate complex forms and support lexical-semantic mechanisms in children or adults (Goodwin & Ahn, 2010; Colé et al., 2004; Auclair-Ouellet et al., 2017). However, the current state of knowledge in this area points to the need to develop methodologically valid tools, with the goal that researchers ensure the transfer of relevant empirical data to enable clinicians and educators to develop a rehabilitation or teaching protocol based on evidence-based intervention/teaching studies.

Given such data, the database design specifically oriented to French derivational morphology was crucial. Therefore, in parallel to the actions carried out by morphologists, NLP researchers, and engineers (lots 1, 2, 3) in order to develop the DEMONEXT database, it was required to measure upstream (lot 4) of this project, i) the needs of the future users of the tool (social validity of the tool); ii) the data coming from the materials usually used by speech-language pathologists (SLPs) and teachers in order to characterize their strengths and limitations and to take them into account in the interface. iii) Finally, it was necessary to propose trajectories of activities and strategies for SLPs and teachers in order to support them in their use of the DEMONEXT database. Moreover, to contribute to disseminating critical scientific information in morphology to these professionals who are encouraged to mobilize evidence-based pedagogical or clinical actions.

Research axes of lot 4: (led by Georgette Dal & Guillaume Duboisdindien)

Axis 1. Survey French clinicians on their knowledge of morphology and their strategies for selecting a pedagogical/clinical support tool.

Strategy: Online survey (3 sessions)

Respondents: French-speaking speech-language pathologists and students (France, Belgium, Switzerland, Canada, Morocco)

Axis 2. Conduct a critical review of French clinical/educational material in derivational morphology.

Strategy: Development of a criterion-referenced evaluation tool to measure the characteristics of these clinical/educational tools

Selected tools: Based on the catalogs of specialized publishing houses, selection of 15 pedagogical/clinical support materials specifically oriented in morphology or derivational morphology.

Axis 3. Participate in disseminating scientific data in derivational morphology for clinicians/educators using the DEMONEXT database.

Strategy: Literature review and Participatory Research Program involving project researchers and clinicians/educators

Orientations: To carry out syntheses of scientific articles in French-oriented derivational morphology (clinical studies) + To carry out clinical/pedagogical Scenarios based on +/- fictitious cases favoring A. the use of the base; B. the diffusion of scientific knowledge (based on the syntheses previously prepared with volunteers); C. the adjustment and the improvement of the pedagogical and clinical strategies in order to establish a protocol of care or teaching

Results

Axis A: DEMONEXT Survey of SLPs

Since November 2020, the DEMONEXT survey has mobilized nearly 390 SLPs in French-speaking territories (France, Belgium, Switzerland, Canada, Monaco, Luxembourg, Niger). The first survey results, whose analysis is still in progress, indicate that 35% of speech therapists perform interventions targeting derivational morphology. 10% of them evaluate their level of knowledge on this topic as low to very low, and 23% of this audience have a general knowledge of derivational morphology. However, the qualitative analysis of the spontaneous response corpus on the types of activities proposed and on the terminological and theoretical knowledge of derivational morphology suggests that these gaps are 10 to 20% higher than the self-assessed gaps. The barriers experienced by the clinicians in the survey that are similar to those reported in the literature are

  1. the feeling of illegitimacy to develop a methodology of evidence extraction and linguistic variables for practice,
  2. the lack of theoretical training in the field of derivational morphology,
  3. the lack of time and knowledge of English to access scientific literature in the English language,
  4. the preferential use of online dictionaries or word lists whose variables are not controlled or contextualized,
  5. the feeling of a marked gap (real and imagined) between the clinic and research,
  6. the difficulty and lack of time to understand the use of existing databases.

Axis B: Critical analysis of SLPs and pedagogy materials

Based on a criterion-referenced assessment scanning six major quality domains and 22 subdomains (grade 0 = maximum uncertainty > grade 4 = optimal), the preliminary results of this study (available online here) indicate that the 15 French-speaking SLPs and pedagogues’ remediation materials oriented towards morphology and derivational morphology present weaknesses in their general characteristics,

  1. in the typology of the proposed morphological tasks, and
  2. in the effectiveness of the derivational lexemes targeted for remediation.

As an illustration, quality criterion 2 Ergonomics and technical qualities show that 46.67% (n=7) of the analyzed materials scored 0 in this category; 40% (n=6) scored 2 or 3, equally distributed, and only two materials scored 4. 

Quality criterion 3 Target population indicates that 60% (n=9) of the materials are at grade 0. In addition, 13% (n=2) share grades 1 and 2. and 27% (n=4) of the materials are classified as grade 4.

Quality criterion 5, which concerns the theoretical validity of the material, indicates a majority of grade 0 with a score of 73% (n=11), 13% (n=2) of the materials achieved grade 2; one material achieved grade 3, and only one material achieved grade 4.

In addition, the instructions for morphological activities (adapted from the Taxonomy of Berthiaume et al., 2010) are primarily oriented toward derivation tasks (Task A) and structural analysis tasks (Task B). However, 13 of the 15 supports revealed that most of the proposed tasks fit into more than one subtask category or were not specific enough to be formally labeled by the study researchers. In addition, it is not explicitly stated whether these tasks are to be completed in oral or written form, or both, except in the two materials that score very well in their overall characteristics.

Finally, when studying in more detail the lexemes frequently used in these remediation tools, the analyzed occurrences do not stand up to structural and diachronic examination. Indeed, the morphological pattern inX frequently used in these materials is not productive in French, except with deverbal bases in -able. Thus, the relevance of the most frequent lexemes prefixed with in- SLPs materials is questionable. As found in our study, lexemes such as INHUMAIN or INJUST are not suitable primary stimuli (Task A). In particular, the specific analysis of the derived occurrences challenged the logic of the two-remediation media that had the best overall characteristics. The task’s context (task A or B) seems to be at odds with the target lexemes proposed in most materials. At least, this raises the question of whether it is indeed morphological mechanisms that are engaged in this type of task associated with this type of schema or whether it does not cognitively activate the child’s memory skills or lexical stock (keeping in mind the biases related to inequalities of access to the lexicon, see French study by Rassel et al., 2020).

These data indicate that a critical analysis of materials for educational and clinical use is relevant to enabling SLPs and teachers to make informed, evidence-based choices in their practice and their materials. Additional studies should be considered in this direction, and the development of research at the interface of linguistic expertise and clinical needs.

Axis C: Clinical scenarios and awareness of the use of the DEMONEXT database and the EBP approach

Several actions to raise awareness of the scientific and clinical data in the field of derivational morphology and the DEMONEXT database have been initiated or are still ongoing.

Dissemination of study summaries in French:

For the team of interdisciplinary researchers of the DEMONEXT project, the commitment to a program of dissemination of scientific data to the public of clinicians but also pedagogues, families, and patients/students is an exciting opportunity. However, respecting all knowledge and recognizing the diversity and equality of this knowledge, whether it is that of the researcher or that of the clinical and pedagogical actors, is a sine qua non condition for a sustainable commitment within this research program, the Agence Nationale de la Recherche.

In this will of diffusion and transmission, we propose a corpus of about twenty syntheses of scientific articles in French with clinical or pedagogical orientation in free access and downloadable on the site DEMONEXT. An article synthesis aims to contribute to the dissemination of evidence in French for French-speaking clinicians and educators within the framework of the DEMONEXT Participatory Research Program. We remind you that this document does not replace an expert opinion. Finally, podcasts of conferences intended for speech-language pathologists training at the University of Lille (T.M. Tran) are also available online.

Dissemination of clinical scenarios in French:

Three clinical scenarios are currently being finalized (release June 2022) to allow clinicians and educators to contextualize a care/teaching approach based on evidence on the one hand and accurate patient/student data on the other. Here, a sample tutorial is already available (Cattini & Duboisdindien, 2019). These scenarios have the advantage of presenting step-by-step methods (supporting video vignettes) to

  1. present the clinical decision based on the patient’s preferences/complaints/needs;
  2. present the use of the DEMONEXT database and the use of other French databases in order to control the measured variables and the intervention targets according to the care objective (frequency, developmental age, imageability, etc. );
  3. To model progressive instructions and activities, from the introductory session to a new affix to the restitutive session, including an intermediate session illustration.

A clinical study using the DEMONEXT database is also underway for a young adult with a learning disability in written language (Duboisdindien, Cattini & Dal). Examples of mini-lessons using forms derived from the database will also be presented for a clinical case to help pedagogues in the constitution of pedagogical supports in class.

To go further:

In general, the DEMONEXT database has the potential to accompany pedagogues and clinicians on several reflective levels to secure their professional practice. First of all, by offering a tool dedicated explicitly to queries of French-derived words developed by morphologists, NLP specialists, psycholinguists, and clinicians. The social and theoretical validity of the database is guaranteed by a set of scientific and clinical strategies that consider the limitations that derivational morphology represents for these users and the means of getting around them (information leaflets, critical theoretical and technical knowledge, pedagogical capsules, pedagogical files). Secondly, by proposing realistic and contextualized scenarios based on clinical/pedagogical data rather than activity/gamebooks or simple lists of words from the database that would be detached from their context of use. These scenarios are based on the same framework in order to guarantee the replication of these actions and thus reinforce the learning of users who would not be at ease with the use of the database, the constitution of word lists to measure the performance of the learner/patient, the appetite for concrete illustrations of what can be undertaken with DEMONEXT. The tool’s limits are also transparently addressed by proposing video illustrations to combine the words from DEMONEXT with variables from the databases offering additional selection criteria that were not in the objectives of DEMONEXT (frequency variable, imageability, etc., for example). Finally, by its direct action with users by promoting a Participatory Research Program or seminars dedicated to the interest of derivational morphology in educational and clinical actions. Other interesting strategies could support these modalities to reach the most significant number of people, particularly by developing training in the database and e-learning workshop sequences.