ГЕОГРАФИЧЕСКАЯ СТАБИЛЬНОСТЬ КАТЕГОРИАЛЬНЫХ НОРМ ДЛЯ РУССКОГО ЯЗЫКА

Introduction

In order to study categorization it is necessary first to identify which words are used by native speakers in specific semantic categories (like “A Fish” or “A Fruit”), and to determine generation frequency of these words within categories. This variable was also named instance dominance by some researches [13], [14]. First attempts to create category norms of generation frequency were made by Cohen at al. (1957) in USA [3]. Their work was continued by Battig and Montague during the next decade [1]. Battig and Montague’s (1969) database, which contains 56 categories of English language is the most frequently cited database of generation frequency. The citation search made by Van Overschelde et al. [18] on 2002 demonstrated that it was cited over 1600 times in papers published in more than 220 different journals.

Cross-cultural and linguistic research has revealed that the content of categories varies across different cultures [21] and that patterns of phenomena and variable ratings for those categories may also vary with cultural milieu [12]. Thus using a database, that was collected from subjects of another culture, is not always acceptable. That is why similar studies were conducted in other countries as well, for example in Belgium [15], France [2], New Zealand [11], Canada [8], Israel [6], China [21], Great Britain [5], Spain [10], etc.

In order to study categorization in Russia as well it was important to create generation frequency norms for the Russian language. Some data has been published regarding 13 categories for Russian language in year 1997 [19], serving as a starting point for this line of research. Considering the ongoing changes, evolution of language content, it was important to enlarge the quantity of categories documented. Generation frequency database for 45 semantic categories was collected for Russian language later [9]. This database was collected in Moscow. Many of selected categories were the same as in the study by Battig and Montague. However, some new categories were included (for example "A Domestic Appliance", "An Organ of the Human Body").

It has been shown that some categorization phenomena depend on human experience and can vary between urban citizens and people who live in close contact with nature [12]. Thus, it is important to take into account not only cultural but also experiential factors [20], [17].

Task which is used to gather generation frequency norms can be quite sensitive not only to language and to culture aspect but to experiential factors as well [20]. It gives an impression about concept structure in population. Along with universality of concepts, it can reveal some differences between subjects, who speak the same language but live in different countries and have different environment [11], or who lives in the same environment but belongs to different cultural groups in the same country [20].

Category norms collected previously in Moscow were proved to be reliable [9]. Nevertheless, taking into account that Russia covers more than one-eighth of the Earth`s inhabited land, some differences could be suggested between distant regions. Thus, before making inferences and generalizing generation frequency norms collected in Moscow to the Russian language and the whole country, geographical stability of these results needs to be tested. Thus, it is important to test how similar generation frequency data from distant regions will be. Moscow, Irkutsk and Ekaterinburg regions were chosen for this aim.

Moscow is located in the central part of Russia. The city is playing a role of a political, economic and cultural center in Russia. Ekaterinburg is located on the borderline between Europe and Asia on the eastern side of the Ural Mountains. Irkutsk is one of the biggest cities of Eastern Siberia.

Geographical stability of psychometric data is traditionally tested through correlations between data collected in different regions.

The following suggestions can be made. Generation frequency data can be accepted as geographically stable and reliable when there are strong correlations between samples of different regions.

Method

Participants. One hundred fifty students of different universities of Moscow aged 18-26 years participated in the study as volunteers (126 females and 24 males, m=19, SD=1.59). One hundred fifty students from Ekaterinburg aged 18-24 years (126 females and 24 males, m=19, SD=1.14) and one hundred fifty students from Irkutsk aged 18-28 years (132 females and 18 males, m=19, SD=1.87) participated in this study as well. There were no significant differences in proportion of male and female participants in samples taken from three different regions (Pearson Chi-square=2.178, df=1, ps=0.132). Participants involved in this study in each region represented the same generation and the same social group. All of participants were native Russian speakers.

Procedure. The procedure used to gather the Russian category norms was similar to the procedure of Battig and Montague (1969). Participants were provided with a small notebook. The following instructions, were copied verbatim from Battig and Montague (1969), but were translated into Russian.

“The purpose of this experiment is to find out what items or objects people commonly give as belonging to various categories or classes. The procedure will be as follows: First, you will be given the name or description of a category. Then you will be given 30 sec. to write down in the notebook as many items included in that category as you can, in whatever order they happen to occur to you. For example, if you were given the category "seafood", you might respond with such items as lobster, shrimp, clam, oyster, herring, and so on. The words are to be written in the notebook, using a different page for every category. When you hear the word "Stop", you are to stop writing and go to the beginning of the next page. You will then be given the name of another category, and again you are to write the names of as many members of that category as you can think of.” The full version of the instruction can be found in the paper by Battig and Montague of 1969 [1].

The category names were read aloud by the experimenter. The participants were tested in small groups to be sure that they could work in a proper way and will not be distracted by each other. The presentation order of the categories was randomized and was different in different groups of participants. The category set for this study consisted of 45 different categories such as various natural kinds ("A Fish", "An Insect", "A Flower"), artificial kinds ("A Type of Vehicle", "An Article of Furniture", "A Musical Instrument"), names ("A Male`s First Name"), activity kinds ("A Profession", "A Sport"),  abstract kinds ("A Unit of Time", "A Unit of Distance"), etc.

Results and Discussion

The same procedure of data analysis as in previous works was used [1], [16]. No distinction was made between singular and plural or masculine and feminine versions of exemplars. Legible responses that were nonmembers were not removed from the list. For each exemplar, overall generation frequency was calculated in each region. Reliability of the generation frequency data for each of three regions was evaluated by applying the split-half method. Then the Spearman–Brown formula was applied: ρ=2r_half/(1+r_half).

Reliability data are presented in Table 1. Reliability of generation frequency data in all three regions is quite good.

In order to test geographical stability of generation frequency norms correlations between cities were calculated. All words (even these, which were named only one time) were used for this analysis. All Pearson`s correlations were significant, p<0.001. Correlations are presented in Table 1.

Table 1 - Split-half reliability coefficients of generation frequency for each region and correlation between three regions for each category