Development of alternative versions of the Logical Memory subtest of the WMS-R for use in Brazil

The logical memory test of the Wechsler Memory Scale is one of the most frequently used standardized tests for assessing verbal memory and consists of two separate short stories each containing 25 idea units. Problems with practice effects arise with re-testing a patient, as these stories may be remembered from previous assessments. Therefore, alternative versions of the test stimuli should be developed to minimize learning effects when repeated testing is required for longitudinal evaluations of patients. Objective To present three alternative stories for each of the original stories frequently used in Brazil (Ana Soares and Roberto Mota) and to show their similarity in terms of content, structure and linguistic characteristics. Methods The alternative stories were developed according to the following criteria: overall structure or thematic content (presentation of the character, conflict, aggravation or complements and resolution); specific structure (sex of the character, location and occupation, details of what happened); formal structure (number of words, characters, verbs and nouns); and readability. Results The alternative stories and scoring criteria are presented in comparison to the original WMS stories (Brazilian version). Conclusion The alternative stories presented here correspond well thematically and structurally to the Brazilian versions of the original stories.


INTRODUCTION
T he Logical Memory task from the Wechsler Memory Scale (WMS) 1 is one of the most frequently used tests for assessing episodic memory. 2 This test is composed of two stories which include 25 units or ideas each, and are verbally presented by the examiner: one about a woman (presented first), and the other about a man.
Immediately after each story is presented (immediate recall), the test subject is asked to repeat what they remember as accurately as possible, without being given any clues (free recall). 1 Thirty minutes later, subjects are asked to recall the story again (delayed recall). For both immediate and delayed recall, all correct items are summed giving a maximum score of 25 for each story. The WMS is currently in its 4 th edition, 3 and since its previous version, 4 the logical memory stories have gained an additional form of scoring. The 25-item scoring system was maintained (which counts specific units of the story), and a second scoring criteria was established, considering the recall of general topics (gist). 5 For example, the original scoring system credits points for the recall of the character's exact name and surname (one point each); the additional scoring system now also grants a point if the subject recalls that the story was about a woman, even if her name is not remembered.
Because of its great value in memory assessment, the logical memory test is often used when re-testing is needed in clinical practice and longitudinal research. Changes in memory performance are often assessed after interventions such as surgeries, 6 evolution of psychiatric disorders, 7 as well as several types of therapeutic and pharmacological interventions. 8,9 These studies involve the assessment of the same subjects on two or more separate occasions.
The logical memory test is known to be susceptible to practice effects or training, i.e., the repeated use of the same story increases the possibility that it is remembered on subsequent assessments (stimulus learning). 2,10 Thus, there is an urgent need for alternative stories. However, few studies have developed alternative stimuli. This is the first step in proposing alternative versions of cognitive tasks, the second being the study of psychometric similarity between the versions in healthy participants and clinical populations. Various studies have suggested means of constructing alternative stories to those in the WMS as discussed below. In this study, the criteria from these scientific publications were used to construct the alternative versions of the task in Portuguese.
Morris, Kunka and Rossini 11 presented an alternate version for each of the WMS-R stories that had equivalent characteristics in terms of their items (units of ideas), sentences, words per sentence, syllables per word, readability and subjective attributes (such as affective tone). In many clinical and experimental settings, however, more than one alternative story for each original may be required. Sullivan 2 proposed six stories following similar criteria. 11 The author then analyzed the comparability of stories in terms of number of words, characters, sentences and readability, as well as the performance of normal subjects in terms of immediate recall of these stories after oral presentation. Though these stories are generally similar, there were some differences between them, probably because the original stories of the WMS-R test are not equivalent in the parameters studied, 2 and also differ in terms of ease of memorization. 12 Two other publications have proposed alternative stories, applying similar methods to those adopted by the studies mentioned above. 2,11 Schnabel 10 published two alternative versions of the WMS-IV stories that were similar in structure, narrative, and readability. Moreover, he tested the compatibility of the WMS stories and the newly developed stories empirically in three large populations, comprising patients with depression, head injury, and a control group of healthy individuals. He demonstrated that performance regarding the original and the newly developed stories was interchangeable for all groups, both for immediate and delayed recall. A further study shows three short stories similar to each other, but these were not equivalent to the stories of the WMS. 13 The use of the logical memory test in Brazil poses several major difficulties. To date, the test has not been published in Portuguese for any of the WMS versions. A free translation / adaptation of the pair of stories of the WMS-R (Ana Soares and Roberto Mota) was done for clinical and research purposes at one of the major neuropsychology centers in Brazil by Camargo et al., 14 but the stories in full and the scoring criteria were not published. One article carries one of the two stories in Portuguese (Ana Soares), but no scoring criteria were provided. 15 These versions have been shared privately among memory specialists for years and are probably the most common prose recall tests used by Brazilian professionals in research and clinical work. Hence, there is no publication that includes both these adapted stories and their scoring criteria, nor scientific work on standardization or systematic adaptation of this instrument for use in Brazil.
There are stories for memory testing created in Portuguese from Brazil, but these are not equivalent to the any of the versions of the WMS, since they are shorter and have a very different thematic structure. 16,17 Given that many Brazilian professionals have been using the logical memory test and may need to re-test patients, it would be useful if the alternative stories in Portuguese were constructed so as to be equivalent to those adapted by Camargo et al. 14 from the stories of the WMS-R.
The aim of this study was to develop a set of alternative logical memory stories to those frequently used in Brazil (Ana Soares and Roberto Mota) in terms of their content, structure and linguistic characteristics. The development of the material was based on parameters proposed by previous authors, 2,10,11 and also took into account the thematic structure of the original stories. The material presented here, however, does not show the psychometric equivalence of the stories, which was tested later. 18

METHODS
Three alternative stories for each of the stories in Portuguese (Ana Soares and Roberto Mota) 14 were proposed by the authors following the criteria below, based on previous studies. 2,10,11 1. General structure of the plot: each story involved the presentation of character, a conflict, an aggravation / complements to the main plot, and a resolution. The alternative stories to that of Ana Soares involved female characters whereas those of the Roberto Mota stories had male characters.
2. Specific elements: all stories presented the characters with a first name, surname, occupation and place of origin; the stories that were alternatives for the female story had numerals and the male ones had closure statements with a specific word that can be considered slightly out of the context that can be scored even if recalled alone (equivalent to the word 'shark' in the Roberto Mota story).
3. Grammatical structure and details: all stories were equivalent regarding the grammatical classes of words, number of words, letters, verbs and nouns per idea item and in total; 4. Units of general topics (gist): units that summarize the essential ideas of the excerpts of the story were considered, allowing little specific recollection to be granted a point on the secondary scoring system as mentioned in the introduction section. The units of general themes of the female stories were guided directly by those proposed by the WMS-III, which maintains the story equivalent to that of Ana Soares, with a total of 7 themes. The male stories had these units freely defined by the present authors, since the story that led to the Roberto Mota version was not maintained in the most recent versions of the test in its original English version; this secondary scoring has a total of 8 units for the stories of the male characters. 5. Readability: reading complexity was calculated with the Flesch Reading Ease Scores adapted for local use in Portuguese. Scores between 75 and 100 denote very easy to read, from 50 to 75 easy, 25 to 50 difficult and below 25 very difficult to read. 19 Following the creation of the alternative stories, they were presented to a group of experts from the Psychology Department of the Universidade Federal de Sao Paulo, including memory researchers from different areas of Brazil (various sessions including more than 15 experts in each). This group of experts proposed changes that were necessary to make the stories easy to understand and equivalent to each other. The authors then determined the scoring criteria, which were subsequently evaluated by the group of memory experts. The final selection of acceptable responses was defined by consensus among the authors. The Appendix shows the scoring criteria for each story.

RESULTS
The original stories in Portuguese 14 , and their organization according to thematic units, can be seen in Table  1. The new alternative stories are presented in Table 2, which also shows the equivalence to the original stories according to various parameters such as: thematic content, structure (sex of the character, their origin and profession, what happened and the resolution of the problem), grammatical details, as well as readability of stories, all of which were scored as easy to read according to the Flesch Readability Ease Scores (between 50 and 75). 19 The Appendix shows the scoring criteria with the acceptable answers, both for the main score, which requires greater accuracy in recall, as well as for the secondary scoring, which evaluates the ability to remember general themes (gist). Examples of incorrect responses are also presented.

DISCUSSION
The logical memory test is one of the most used tasks for assessing episodic memory in both clinical and research settings. 2 In many cases, reassessment of patients is needed. [6][7][8][9] The repeated use of this kind of test leads to learning effects due to many factors, 10 including recall of the story content when the same story is told more than once. The use of different stories that have similar characteristics is one of the methods used to improve the validity of repeated assessments. 10 The present study presents three alternative stories for each of the freely translated stories from the WMS-R, as these are frequently used in Brazil despite the lack of an official translation and of validation in the Brazilian population. The creation of the alternative stories was based on criteria proposed by authors who developed English versions of the task (narrative structure, emotional content, readability, number of words, etc.). 2,10,11 The comparability of stories in terms of the plot, structure and the linguistic features were presented here, and detailed information on their psychometric equivalence can be found in the study of Martins et al., 18 which shows the performance of a non-clinical sample on the immediate and delayed recall of the alternative stories.
Besides the alternative stories that are similar to the Portuguese version of the WMS-R, the material presented here also includes the additional scoring criteria (gist or general themes) introduced in later versions of the WMS. 3,4 This additional score has important clinical and experimental value, since it helps differentiate levels of impairment and allows a more detailed investigation of memory not only as an all-or-nothing phenomenon, but rather as the storage of stimuli at different levels of specificity. 20   No. of verbs 11 11 11 9 No. of nouns 18 19 20 20 test. For research purposes, it is recommended that matching control groups be used. However, until appropriate validation studies and norms for the WMS logical memory test are published in Brazil, the alternative stories and detailed correction criteria presented here may be used. These items may also represent a first step toward the standardization and production of norms of this task for the Brazilian population. Lastly, it must be born in mind that even with the use of alternative versions, practice effects can occur as a result of decreases in test anxiety due to familiarity with the test setting, testing procedure, and/or experimenter or clinician, as well as due to procedural learning and development of test strategies. 10 The impact of these factors warrant attention in future studies employing the use of more than one story in the same individual, and may differ depending on the cognitive status of the patient.