Verst-Maldaun Language Assessment (VMLA) Validation Process Database

Published: 20 January 2021| Version 4 | DOI: 10.17632/zjhfk7mm7v.4
Silvia Mazzali Verst,
Isac de Castro,
Wilson Scappini Junior,
Mônica N. de Melo,
Jean de Oliveira Ramos,
Soraya S. de Almeida,
Nathalia Raquel C. Alvarez,
Andrea C. B. Sucena,
Maria Rufina Barros,
Carlo Domenico Marrone,
Linamara Rizzo Battistella,
Maria Aparecida K. Vasconcellos,
Andrea S. Alves,
Marcos Vinicius C. Maldaun


This paper drives the process of creating VMLA, a language test meant to be used during awake craniotomies. It focuses on step by step process and aims to help other developers to build their own assessment. This project was designed as a prospective study and registered in the Ethic Committee of Educational and Research Institute of Sirio Libanês Hospital. Ethics committee approval number: HSL 2018-37 / CAEE 90603318.9.0000.5461. Images were bought by and generated the following receipts: SSTK-0CA8F-1358 and SSTK-0235F-6FC2 VMLA is a neuropsychological assessment of language function, comprising object naming (ON) and semantic. Originally composed by 420 slides, validation among Brazilian native speakers left 368 figures plus fifteen other elements, like numbers, sentences and count. Validation was focused on educational level (EL), gender and age. Volunteers were tested in fourteen different states of Brazil. Cultural differences resulted in improvements to final Answer Template. EL and age were identified as factors that influenced VLMA assessment results. Highly educated volunteers performed better for both ON and semantic. People over 50 and 35 years old had better performance for ON and semantic, respectively. Further validation in unevaluated regions of Brazil, including more balanced number of males and females and more even distribution of age and EL, could confirm our statistical analysis. After validation, ON-VMLA was framed in batteries of 100 slides each, mixing images of six different complexity categories. Semantic-VMLA kept all the original seventy verbal and non-verbal combinations. The validation process resulted in increased confidence during intraoperative test application. We are now able to score and evaluate patient´s language deficits. Currently, VLMA fits its purpose of dynamical application and accuracy during language areas mapping. It is the first test targeted to Brazilians, representing much of our culture and collective imagery. Our experience may be of value to clinicians and researchers working with awake craniotomy who seek to develop their own language test. The test is available for free use at (beginning in February, 2021)


Steps to reproduce

I. CREATING OBJECTS NAMING (ON) DATABASE: 1. Figures were bought for commercial use. Chosen criteria: High quality sharp images and photos; No trademarks or famous people (due to royalties) ; Categories: food, animal, insects, daily living objects, furniture, clothes, sports, flowers, fruits, driving vehicles, geometric forms, world well known places, country flags, folklore, music instruments, nature (strand, forest, stone, rain, moon, star), jewelry, action (people cleaning windows, playing volleyball, a child studying); plus 7 sentences, 5 numbers, counts and 3 syllables or letters. 2. They were framed in 4 ON tests of 105 images each 3. Applicators were trained by the main researcher. Test rules: a. Signed written informed consent. Information about gender, age, EL, state of origin in Brazil b. Volunteer was informed that any answer would be considered wrong. That we were searching for synonyms c. Participants were asked to name with the first word that he/she remembered d. Applicator and volunteer were seating in front of each, alone the room e. We tested using an ipad or tablet. f. The tests were administered in an identical sequence g. Time schedule was 45 minutes. Some volunteers demanded over an hour, which was pointed out by the tester for further analysis. Exposure time was 10 to 20 seconds. h. Notes taken: Different names, could not remember, did not know 5. There were two phases: a. Phase 1: First Seventy volunteers: methodology proofing. Seven MD applied the tests in 7 cities belonging to four states. Volunteers were family members, friends and house employees. These tests are part of the final results b. Phase 2: Volunteers were employees at Sirio Libanês Hospital and at Lucy Montoro Rehabilitation Center in São Paulo c. all results are shown in Table 3 6. While analyzing volunteers´ answers, a correction criteria table was created (table 1) 7. For strange words, the main researcher would check on a dictionary or, if absent, by Google images, or with testers for regionalisms 9. After the compilation of responses: Nine images were inappropriate: 1.71, 1.95, 2.23, 2.101, 3.3, 3.23, 3.81, 3.105, 4.82, and some as wrong (table 2). 10. Statistical analysis classified the images in 8 different groups, according to the 3rd template (Table 3) 11. A Final Answer Template was created (Table 2) 12. All used images are shown in Figure 1 II. CREATING SEMANTIC TEST DATABASE: I. Each slide was a combination of 3 items: one above and two below. It is a picture-picture or word-picture matching task, totaling 70 combinations. II. It was applied at the end of testing. Errors rate resulted in group stratification (Table 3)


Hospital Sirio-Libanes, Universidade de Sao Paulo, Instituto de Reabilitacao Lucy Montoro


Neuro-Oncology, Neurosurgery, Neurophysiology, Brain Mapping, Glioma, Neuropsychological Testing, Intraoperative Monitoring, Cortical Stimulation Mapping, Language