|Year : 2007 | Volume
| Issue : 2 | Page : 92-99
Community-based naming agreement, familiarity, image agreement and visual complexity ratings among adult Indians
Annamma George, PS Mathuranath
Department of Neurology, Cognition and Behavioral Neurology Section, SCTIMST, Trivandrum, Kerala, India
P S Mathuranath
Department of Neurology, SCTIMST, Trivandrum - 695 011, Kerala
Source of Support: None, Conflict of Interest: None
| Abstract|| |
The validity of Snodgrass and Vanderwart pictures and their norms derived on a western population on naming, familiarity, imageability and visual-complexity, is not established on a population with cultural background different from the west. We developed, therefore, a set of culturally appropriate pictures for and derived norms on Indians. Line-drawings of 103 concepts (67 from Snodgrass and Vanderwart, 36 new) from 10 semantic-categories were normed on 200 community-based older subjects. Only 31% of the Snodgrass and Vanderwart items showed a concept-agreement on the Indians comparable to western norms. Naming, familiarity and image-agreement mutually correlated but not with visual-complexity. Low-education and rural-residence tended to reduced concept-agreement. The output of this study will be of use in national and cross-national studies.
Keywords: Culture, Indian, norms, semantic memory, Snodgrass, Vanderwart pictures
|How to cite this article:|
George A, Mathuranath P S. Community-based naming agreement, familiarity, image agreement and visual complexity ratings among adult Indians. Ann Indian Acad Neurol 2007;10:92-9
|How to cite this URL:|
George A, Mathuranath P S. Community-based naming agreement, familiarity, image agreement and visual complexity ratings among adult Indians. Ann Indian Acad Neurol [serial online] 2007 [cited 2019 Oct 20];10:92-9. Available from: http://www.annalsofian.org/text.asp?2007/10/2/92/33216
| Introduction|| |
Line-drawings of pictures are used as visual stimuli for studying cognition, particularly language and semantic memory, in healthy subjects and in patients suffering from various disorders affecting cognition.,,
It is important to recognize, however, that independent of the cognitive abilities of the brain, a variety of factors related to the concept itself (such as its imageability or its familiarity in the population being tested) or its drawing (such as its visual complexity) are all known to influence the subject's ability to tell the name of the concept., Other than the Snodgrass and Vanderwart pictures and the Boston Naming Test well-evaluated widely used visual stimulus material are scarce. Even with these, the validity of the stimulus-related attributes (naming, familiarity, visual complexity and image agreement) in a 'non-western' population is not established as they were developed for a western population. This poses a major problem in their use in other populations, because, as explained above linguistic materials vary on many dimensions and it can be difficult to establish how representative a particular sample of word is of a theoretical construct in a given population. A recent study has shown that only 22% of the Snodgrass and Vanderwart pictures show equivalent relevance in the Americans and Chinese.
An important dimension having a profound influence on linguistic material is culture. Broadly defined, it includes the population's food, clothing, tools used, life style, religious practices (all of which determine the familiarity with the named concept) and language (which determines the frequency of the named concept (i.e., word-frequency)). Thus, while some concepts may have a global relevance others have relevance only in specific cultures. For example, jackfruit, which is very common in South-East Asia [Figure - 1], Panel A, is rather uncommon in Europe or North America. Similarly, asparagus, which is indigenous to North America, is virtually unknown to the vast majority of South Asians. Even amongst those with a global relevance, some may be visualized in different ways across different cultures (e.g., 'blouse' in India ([Figure - 1], Panel B) looks different from that in Snodgrass and Vanderwart pictures).
For many of the test materials used in India, lack of validation and norms has been an enduring problem. Recognizing the limitations of results obtained by the use of such tests, researchers have been working towards developing norms and validating test materials. This study is one such effort. The aim of this study was to develop line drawings for a set of culturally relevant concepts from different semantic categories and evaluate their stimulus-related attributes in the adults in the Indian subcontinent. In order to facilitate the use of such material in cross-national studies, we selected many concepts from the widely used and normed Snodgrass and Vanderwart pictures. We also added new concepts to generate additional linguistic materials that are culturally relevant to the population in the Indian subcontinent.
| Materials and Methods|| |
Development of pictorial material
We drew up a list of concepts from a number of sources, which included, the Snodgrass and Vanderwart drawings, pictures from children's picture books and course books and the Malayalam dictionary. Over a period of one year and following many informal discussions with colleagues across the country, a team comprising of a neurologist, a speech therapist, a neuropsychologist, a linguist (with an expertise in Malayalam and proficiency in English) and community leaders from a senior citizen's forum short listed eleven items in each of the following ten categories and formalized their respective drawings- animals, birds, insects/reptiles, fruits, vegetables, vehicles, inanimate objects, musical instruments, tools, clothing and body parts. The short listing was dictated by common knowledge and was done on the basis of the universal familiarity of the concept across gender and different cultural identities within the Indian subcontinent in general and south India in particular. For formalization of the drawing of a concept we mainly considered the pictorial representation likely to be most familiar for the given concept as dictated by common knowledge. Thus, modifications to/replacements of pictures were done when the team considered the available representation to be unfamiliar to the population. We used 67 concepts from Snodgrass and Vanderwart without (n = 40) or with (n = 27) modification/replacement of the pictures. We added 36 new concepts and obtained the line drawings for these and the modified items of Snodgrass and Vanderwart pictures drawn by professional artists. All the pictures/drawings were hand drawn or traced, then scanned into the computer at a resolution of 300 dpi or more and optimized for their sizes. Each drawing was printed on cards of size 11 x 15 cm large enough to be visible to our elderly study subjects.
Evaluating the stimulus-related properties of the object-naming material
Subjects: Naming and familiarity was evaluated in community-based randomly selected 200 cognitively unimpaired elderly subjects (Group-A), who were native speakers of Malayalam living in the district of Trivandrum, Kerala, in southern India. Of the 200 subjects (in Group A), eight were excluded as they failed on the screening questionnaire for visual/physical problems. Thus 192 subjects consisting of 102 males (53.7%) responded to naming agreement and familiarity. Mean (SD) age and education were 61.06 years (13.69) and 7.64 years (6.37) respectively. There were 51 subjects in 21-54 years age group, 56 in 55-64 years, 49 in 65-74 years and 36 in 75 and above. Furthermore, 87 subjects had primary education or less (0-4 years of formal education), 20 had middle school level of education (5-8 years), 34 had high school/college education (9-12 years) and 51 had university education (≥ 12 years). One hundred and two subjects (53%) were urbanites, 56% were Hindus, 24.6% Christians and 19.4% Muslims.
Image agreement and visual-complexity was evaluated in another group of subjects (Group B) after the data on naming and familiarity was collected. Forty cognitively unimpaired individuals (spouses or caregivers of patients attending a Neurology outpatients department in a tertiary referral center) who had a formal education of 10 or more years and were native speakers of Malayalam constituted Group-B (mean age = 52.05 ± 9.90 years, education = 14.23 ± 3.39 years, Males = 16). We limited the image agreement and visual complexity evaluation to subjects having ≥ 10 years of formal education as we realized that subjects with lower levels of education in our population found it rather difficult to comprehend these notions or make an independent and graded judgment on these attributes. Participants in both groups were screened using a brief questionnaire for a) any significant cognitive complaints or overt neurological diseases that could affect cognition (which included, memory related complaints, dementia, psychiatric illness, epilepsy, stroke, head injury and alcoholism) and b) presence of any physical impairments like visual or hearing handicap.
Language : Interviews were conducted in Malayalam, a Dravidian language spoken by the people of Kerala. Across Kerala, Malayalam has 12 major dialects. Like many other oriental languages, it allows multiple synonyms for many words, especially nouns.
Test environment : Group - A subjects were administered the test in a well-lit room of their homes to ensure a comfortable and familiar environment. Group-B subjects were tested in the quiet and well-lit neuropsychology testing rooms in the hospital.
Procedure : Demographic and cultural details (religion, area of residence (rural/urban) and dialect of Malayalam (used colloquially)) were recorded. A random sequence was generated to prevent bias towards any particular category and the stimulus cards were presented accordingly. All subjects were presented the stimulus cards in this same sequence. Group A subjects were asked to first rate the familiarity and then tell the name. Group B subjects first rated image agreement and then rated visual complexity. All the interviews were conducted by trained psychologists in Malayalam and recorded on audiotapes.
Familiarity: Familiarity was defined as the degree to which one came in contact with or thought about the item on the card. Participants (Group A) were asked to rate the familiarity on a 5-point analog scale from 0 to 4 (0 representing unfamiliar and 4 representing very familiar).
Naming: Participants (Group A) were asked to identify each item by telling the first name that came to their mind on seeing the drawing of the item on the stimulus card. All instructions were provided in Malayalam and as a part of the initial instruction, all the participants were asked to generate Malayalam names. Many individuals in this part of the world are bilinguals with good fluency in Malayalam as well as English and many words in English are often used in colloquial Malayalam since some of them do not have a Malayalam equivalent. There were ten concepts (Harmonium, Penguin, Tractor, Spanner, Guitar, Scooter, Jeep, Telephone, Belt) in our list, which did not have an explicit Malayalam name.
In the case of these concepts, the English equivalent was taken as the Malayalam name, since one of our end points was also to determine concept agreement. However, if there was a Malayalam equivalent for a concept whose name the subjects generated in English, they were asked to give the Malayalam name. Only the Malayalam name was taken for estimating naming agreement. If they failed to generate a name, they were asked to respond with a 'do not know'. Responses were rated as correct if, as determined by the linguist, it was the correct terminology for that item in any of the known dialects of Malayalam.
Image agreement: Participants (Group B) were first asked to rate (on a 3-point scale: 0 for 'dissimilar', 1 for 'somewhat similar' and 2 'very similar') how similar the drawing of an item on the stimulus card was to their mental visualization of the item. Prior to presenting the stimulus card, the experimenter said aloud the most common name of the picture and participants were given several seconds to visualize the item in their minds. Then the drawing was presented.
Visual complexity: Participants (Group B) were asked to rate the complexity of each drawing, with complexity defined as the amount of detail or intricacy of lines in the drawing and not the concept it represented, on a 3-point rating scale from 0 to 2 (0 for 'very simple', 1 for 'simple' and 2 for 'complex'). To facilitate the subject to comprehend this definition, an example with visual complexity ratings, shown in [Figure - 2], was provided to all subjects at the beginning of the test.
Group - A participants' responses to line drawings were analyzed both for familiarity and naming. To determine the familiarity rating, the scores were averaged for each item. Two measures of naming agreement were used: the percentage of subjects who responded with a correct response and H statistics. Correct response included all synonyms and was a measure of the concept agreement. H statistics was computed for each item according to the formula
i =1 where k is the number of unique correct names given for a picture and pi is the proportion of the sample providing each unique name. H = 0 when there is perfect agreement among all participants and increases as agreement decreases. Thus H is sensitive to how widely distributed responses are over all of the unique names that are provided for an item. For estimating the image agreement and the visual complexity, we calculated the average score on each of these parameters for each of the items. On naming, the proportions of 'do not know'/incorrect responses across the different categories in each of the demographic and cultural denominations were also estimated. To study the effect of the demographic and cultural denominations on the ability to generate correct names, number of items on which there was failure/incorrect generation of name was regressed on the education, age, gender, religion and area of residence subgroups. We also correlated the naming, familiarity, image agreement and visual complexity ratings of our cohort. On regression and correlation statistics, p values ≤ 0.05 were considered significant and ≤ 0.1 were considered indicative of a trend. Finally, on each of the common concepts from Snodgrass and Vanderwart between our study and that of Yoon et al , we used Chi square test to compare the proportions on concept agreement of our cohort with that of the older Americans and Chinese in their cohort.
We also correlated concept agreement, naming agreement and familiarity ratings of our cohort with that of the older Americans and Chinese in theirs.
| Results|| |
On naming, there were a total of 19673 responses of which 3% (n = 591) were 'do not know' and 27%
(n = 5324) were incorrect. The 'do not know' and incorrect responses were largely distributed in subjects with education ≤ 4 years (65%), but comparably distributed across the different age categories (between 20% and 30%), gender (56% vs. 44%) and rural-urban (49% vs. 51%) and almost proportionately across the different religions (51%, Hindus vs. 31.5% Christians, vs. 17.5% Muslims). Forward regression of the 'do not know' and incorrect response on the age, education, gender, rural-urban and religion categories showed that only education (a= -8.99; CI: -10.59 to -7.39; p <0.001) and rural-urban (a= 0.16; CI:0.07 to.25; p=0.001) had significant effects. For each of the 103 items the percentage of subjects giving the correct names (concept agreement) is shown in [Table - 1]. More than 50% of subjects on 86, ≥ 66% of subjects on 63 and ≥ 90% of subjects on 40 items gave the correct names. [Table - 1] also shows the average familiarity rating on each of the 103 items.
[Table - 1] shows the image agreement and visual complexity ratings for each of the 103 items. On image agreement, 86 of the 103 items received a rating of 'very similar' by 66% or more subjects. On the remaining 17 items, 14 received a rating of 'dissimilar' by 15% of subjects or less and only 3 items, Spade, Cymbal and Tractor, received a rating of 'dissimilar' by 16 to 25 % of subjects. On visual complexity 17% (n=17) of the items were rated as complex, 43% (n=45) as simple and 40% (n=41) as very simple.
[Table - 2] shows the summary statistics for the four stimulus-related attributes. The H statistics and visual complexity had low means and a positive skew while familiarity and image agreement had high means and negative skews. [Table - 3] shows correlation between the four stimulus-related parameters. Naming had highly significant correlation with familiarity and significant correlation with image agreement while familiarity had a highly significant correlation with image agreement. However, visual complexity did not correlate with either naming or familiarity and showed a trend towards correlation with image agreement.
Three items, potato, chisel and cherry scored below the second standard deviation (SD) on overall correct naming and familiarity shown in the summary statistics. Potato, chisel, hammer, nose, spanner, owl and bear received image agreement ratings below 2SD. Watermelon and cymbal received familiarity ratings below 2SD.
We compared our results with that reported by Yoon et al . The older Americans (age = 66.47 ± 4.24; education = 15.88 ± 2.52) and Chinese (age = 64.68 ± 3.38; education = 16.73 ± 1.35) were significantly ( P <0.001) older and had more number of years of formal education than our cohort. The correct naming (concept agreement) percentages of our cohort were not significantly different from that of the older Americans on 21 (31%) and of the older Chinese on 26 (39%) of the 67 common concepts. Comparable concept agreement across all three cultures was seen on 11 (16%) of the 67 concepts. All these items are indicated in [Table - 1]. On correlation, our cohort showed a significant correlation on familiarity, naming agreement ( H statistics) and correct naming (concept agreement) with the older Chinese, but only on correct naming with the older Americans [Table - 4].
| Discussion|| |
It is not uncommon for investigators to find subjects with different cultural background performing very differently on items in the Snodgrass and Vanderwart pictures. Two important reasons for this include differences in the familiarity with the concept and/or the mental image evoked by the word. This is more evident in the elders who have lived most parts of their active life unexposed to the influence of globalization.
The community-based cohort responding on naming agreement and familiarity largely consisted of older adults. The demographic profile of the cohort shows a heterogeneous mix of different ages, cultural backgrounds and educational abilities and reflected the demographic profile of the subcontinents' population in general and Kerala in particular. We therefore feel that the norms obtained on this cohort may be useful in the population at large. However, the results of this study need to be replicated in different parts of the subcontinent before it can be considered applicable to the subcontinent. Given the demographic heterogeneity in the population, a major concern for us at the time of selection of items was an inability of subjects from specific demographic/cultural denomination to generate the correct name for an item for want of culture-fairness of the concept. On naming agreement, our results showed that proportions of 'do not know' or incorrect responses were high when compared to other such studies and were largely determined by education levels. The demographic denominations of age, gender and religion did not influence naming and area of residence had a very small effect. This suggests that the items selected were generally fair across the different demographic/cultural denominations except perhaps education, where the failure or incorrect generation of name was mostly limited to individuals with low formal education.
Low mean and positive skew on H statistics and visual complexity show that many concepts had a high name agreement and were visually simple line drawings. In contrast the high means and negative skew on familiarity and image agreement suggest that many concepts were familiar and generally matched the mental imagery for that concept. These results are similar to that of Himmanen et al on the concepts in BNT. Results of correlation analysis suggest that there was greater naming agreement with increasing familiarity and better image agreement. Furthermore, greater familiarity was associated with better image agreement. Although unlike the findings of Himmanen et al our study failed to show any significant association of visual complexity with image agreement or with familiarity, it did show that there is a trend towards greater image agreement when the visual complexity is less. We suspect that our failure to demonstrate significant association with familiarity is probably attributable to our use of a 3-point scale for familiarity as against their 5-point scale.
Only three items potato, chisel and cherry, received scores below the normal range on two or more stimulus-related parameters. Cherry is a rare fruit in this part of the world, available seasonally and generally found in up-market areas. We suspect this could be one of the reasons for cherry having a low familiarity and naming. The line drawing of a chisel that we used was slightly different in form from the one that is commonly used by carpenters in this region. Furthermore, unlike the hammer or screwdriver, chisel is a more specialized tool in the carpentry kit that is used less commonly by lay people and consequently is not very familiar to the vast majority of our subjects. Nevertheless, many of those who failed to name it identified it as a tool. We were surprised that potato figured in this list of items with low correct naming, familiarity and image agreement scores. We find it difficult to explain this since common knowledge tells us that it is widely available, used and familiar to public. Our line drawing of the potato was very simple as suggested by the low score (below first quartile) on visual complexity. Although it had 'eyes' drawn on it, many subjects named it as a stone. It is possible, as suggested by the low image agreement scores that the drawing of the concept was perhaps oversimplified. This is further strengthened by our observation that when used as an item for the pointing task of semantic battery, (where, in an array of line drawings, the subject has to point to the picture of the concept whose name is announced by the examiner), subjects point to this item correctly.
Unlike the cohort of Yoon et al , our cohort was randomly selected from the community for naming and familiarity and did not exclude subjects on the basis of education. Thus, the difference in the age and the education levels between the cohorts of the two studies was only to be expected. Of the 67 common Snodgrass and Vanderwart concepts used in the two studies, correct naming proportions of the Indian was comparable to either the Americans or the Chinese on less than 40% and with both on only 16% of the concepts. Interestingly, naming agreement, concept agreement and familiarity, all showed good correlation between the Indian and the older Chinese. This seems to suggest that the acquaintance with these concepts in Snodgrass and Vanderwart is perhaps analogous for subjects from both these cultures. While the differences in age and education limits the validity of these comparisons and correlations between the two studies, it allowed us to identify some of the concepts in Snodgrass and Vanderwart pictures that could be used for cross-national studies. It also provides insights into the differences in acquaintances with concepts across cultures, emphasizes the need to establish culturally/educationally stratified norms and underscores the need for validation of the stimulus-related attributes of pictures when used in cross-national studies.
In conclusion, we have developed a set of line drawings of concepts that are largely culture-fair for the population of the Indian subcontinent. We have also generated norms on each of them on the stimulus-related attributes of naming, familiarity, image agreement and visual complexity. We believe that this material and data will be useful to investigators studying language or semantic memory in the Indian subcontinent. Finally we have identified concepts with comparable concept agreement between different cultures, which will be useful for future cross-national studies.
| References|| |
|1.||Albert MS, Heller HS, Milberg W. Changes in naming ability with age. Psychol Aging 1988;3:173-8. [PUBMED] [FULLTEXT]|
|2.||Bayles KA, Tomoeda CK. Confrontation naming impairment in dementia. Brain Lang 1983;19:98-114. [PUBMED] |
|3.||Freedman M, Oscar-Berman M. Breakdown of cross-modal function in dementia. Neuropsychiatry Neuropsychol Behav Neurol 1997;10:102-6. [PUBMED] |
|4.||Snodgrass JG, Vanderwart M. A standardized set of 260 pictures: Norms for name agreement, image agreement, familiarity and visual complexity. J Exp Psychol 1980;6:174-215. |
|5.||Bates E, D'Amico S, Jacobsen T, Szekely A, Andonova E, Devescovi A, et al . Timed picture naming in seven languages. Psychon Bull Rev 2003;10:344-80. |
|6.||Kaplan EF, Weintarub S. The Boston naming test. 2nd ed. Lea and Febiger: Philadelphia; 1983. |
|7.||Himmanen SA, Gentles K, Sailor K. Rated familiarity, visual complexity and image agreement and their relation to naming difficulty for items from the Boston naming test. J Clin Exp Neuropsychol 2003;25:1178-85. [PUBMED] |
|8.||Yoon C, Feinberg F, Luo T, Hedden T, Gutchess AH, Chen HY, et al . A cross-culturally standardized set of pictures for younger and older adults: American and Chinese norms for name agreement, concept agreement and familiarity. Behav Res Methods Instrum Comput 2004;36:639-49. [PUBMED] [FULLTEXT]|
|9.||Director of Census Operations, Kerala. District Census Handbook, Trivandrum. Village and Town Directory. Village, Panchayat and Townwise Primary Census Abstract. Tiruvananthapuram: Director of Census Operations, Kerala; 1991. |
|10.||George A, Mathuranath PS. Primary progressive aphasia: A comparative study of progressive nonfluent aphasia and semantic dementia. Neurol India 2005;53:162-6. |
[Figure - 1], [Figure - 2]
[Table - 1], [Table - 2], [Table - 3], [Table - 4]
|This article has been cited by|
||Response Letter to Drs. Vidal, Villas Boas, Valle, and Fukushima
| ||Joe Verghese,Pavagada S. Mathuranath |
| ||Journal of the American Geriatrics Society. 2013; 61(7): 1249 |
|[Pubmed] | [DOI]|
||Picture-Based Memory Impairment Screen for Dementia
| ||Joe Verghese,Mohan L. Noone,Beena Johnson,Anne F. Ambrose,Cuiling Wang,Herman Buschke,Vayyattu G. Pradeep,Kizhakkaniyakath Abdul Salam,Kunnukatil S. Shaji,Pavagada S. Mathuranath |
| ||Journal of the American Geriatrics Society. 2012; : n/a |
|[Pubmed] | [DOI]|