Geographical Review of Japan Vol. 75, No. 12, 686-708, 2002

A Geographical Study of Population Characteristics in the Province of ,

ISHII Hisao Faculty of International Studies, Kyoritsu Women's University, Hachioji 193-8501, Japan

Abstract: The objective of this paper is to analyze the spatial structure of population characteristics and its shift in the province of Gipuzkoa, Spain, which has a particularly high density of Basque speakers. By using data sets from the 1991 and 1996 State Census and applying factor analysis, several principal dimensions which regulate the spatial differentiation of population characteristics in Gipuzkoa were extracted. The spatial distribution and transi tion patterns of these dimensions were examined and certain processes of shift became obvious, such as major socio-economic change in the industrial and employment structures, rapid demographic change represented by aging and fertility decline, and an advance in reversing language shift. The dimension of social origin was the most important with regard to the spatial differentiation in Gipuzkoa. Furthermore, it is unique and noteworthy that, apart from with regard to social identity, the demo-linguistic dimensions were extracted independently at the 3rd and 4th dimensions which show relative stages of reversing language shift. As compared with the rural-urban contrast in the dimension of social identity, these demo linguistic dimensions can be found typically in areas with a relatively low density of Basque speakers, including some urban areas. This means that the reversing language shift is making a steady advance there.

Key words: Gipuzkoa, population characteristics, factor analysis, social change, Basque lan guage

element of definition for both the region and Introduction the local resident, and it can be understood that the is a principal component Gipuzkoa is one of the three provinces of at the core of Basque identity formation. Euskal Autonomi Elkartea (Autonomous Com There have been many case studies which munity of the Basque Country, or ACBC).1 have defined the Basque Country as a single Within the Autonomous Community, Gipuzkoa regional unity. In spite of the existence of the is remarkable for its high concentration of Basque ethnic minority group, geographical Basque speakers compared with the Basque studies of the Basque Country in recent decades Country as a whole.2 According to the latest have tended to emphasize functional and eco research, in the entire Basque Country there are nomic analyses (Saez de Buruaga 1969; Ferrer 711,400 Basque speakers, and 568,100 of these, and Precedo 1977; Precede and Rodriguez almost 80%, live in ACBC (Viceconsejeria de 1989). This trend has been due to the fact that Politica Linguistica 1995).3 Furthermore, more the Basque Country was one of the most impor than a half of this 80% are concentrated in tant areas of economic and industrial develop Gipuzkoa (Eustat 1999). ment in Spain from the nineteenth century, The existence of the Basque language has forming and developing a functional urban sys been so important for ACBC that the area is tem within the Spanish northern coastal zone. commonly called Euskal Herri, i.e., "the territory Internal migrants moved into the Basque Coun of the Basque language." Similarly, the resi try in large numbers, as well as to Cataluna and dents of this region call themselves Eus Valencia until the middle of the 1970s. kaldun, i.e., "Basque speakers." Thus, the However, the situation changed drastically Basque language constitutes the fundamental with the economic crisis which the Basque

686 A Geographical Study on Population Characteristics 205

Country experienced from the middle of the gion has increased because of the introduction 1970s. The reduction of the industrial sector of items about the competence and use of seriously altered employment opportunities in Basque in the State Census since 1981. This has this region. Consequently, for the first time in made it possible to observe chronologically the the twentieth century, the Basque Country saw spatial distribution pattern of demo-linguistic an increase in emigrants, reducing the net mi indicators. In this environment, various studies gration rate of ACBC from 18.2% in 1965 and about the spatial pattern and the transition 4.8% in 1975 to-5.1% in 1980. The figures for process of demo-linguistic indicators have been Gipuzkoa also fell, from 12.1% and 1.1% to made (Eusko Jaurlaritza 1989; Garmendia 1994; -6 .7% in the respective years (Diputacion Eusko Jaurlaritza and Naf arroako Gobernua Foral de Gipuzkoa 2000: 63). Also, both Gipuz 1995; Gobierno Vasco 1997, 1998, 1999; Go koa and ACBC have experienced overall popu bierno Vasco et al. 1997, 2000). lation decrease since the 1980s (Diputacion The reinforcement of the administrative Foral de Gipuzkoa 2000: 41-42). This crisis has framework of ACBC inevitably raised the im caused a relative decline in the status of the portance of Basque ethnicity. Ethnicity is an Basque functional region within the national important theme of examination for ethnic ge functional system. Furthermore, a series of ography. During the second half of the last processes of deindustrialization and deurbani century, various research projects and studies zation have caused the collapse of the economic were carried out. continuity of central places in the Basque urban Ethnic separatism and ethnicity in the politi system (Precedo and Rodriguez 1989: 28). cal field are important subjects of study for Simultaneous with this reconstruction of the ethnic geography. However, language shift and functional system, a process of reinforcement of the assimilation or integration process of ethno regional divisions based on administrative linguistic minority groups can also become boundaries has occurred. This was promoted main themes. Studies of demo-linguistic as by the approval of the autonomy of the regional pects of population studies have examined the states in Article 2 of the new Spanish Constitu acculturation or assimilation process by utiliz tion in 1978, by the recognition and rehabilita ing specific variables. Some previous studies tion of regional languages in Article 3, and by have indicated certain correlations between mi the consequent establishment of regional gration and the demo-linguistic dimension. One autonomy. Through this process, there has is a study of the correlation between human been a growing tendency to analyze the Basque flow and the socio-cultural characteristics of regional system as a cultural region, based on the French population in Ontario by Hecht the ethno-cultural and demo-linguistic aspects (1986). Another is a study of the correlation of the population. between migration and the multilingual compe In ACBC, the Autonomous State Constitution tence of the English and French population of was established in 1979. The Autonomous Quebec by Langlois and Castonguay (1993). Council approved State Law No. 10, the "Law Wessels (1994) studied the correlation between for the normalization of Basque," in 1982. Sub the Catalan population and geographical acces sequently, public education came to be ac sibility (urban-suburban traffic and informa cepted as an important element in the promo tion). On the other hand, Aitchison and Carter tion of Basque normalization. In addition, the (1999) emphasized the importance of institu normalization movement advanced in various tional support for Welsh language shift in fields such as education, administration and the Wales.4 mass media. This tendency has encouraged Language serves as an important issue for again the tendency to analyze the Basque re regionalism movements and the Basque case is gional system not as a functional region but as no exception. Therefore, for the Autonomous a cultural one. States at present, understanding the spatial pat The number of ethno-cultural studies focus tern of language use and its transition process ing on the demo-linguistic situation in the re has become one of the most important issues in

687 206 ISHII H.

Figure 1. Overview of the province of Gipuzkoa. Source: EUSTAT (1999). determining future policy. The spatial struc -Immigrant." ture of population characteristics in the exist All studies, including those from ethnic geog ing administrative units is an indispensable raphy, tend to emphasize social change in the "contact zone" (in other words subject of study. Research concerning popula , in the area tion characteristics has been developed in re where both Basque and Spanish are used along cent years focusing on factorial ecology. The side each other) and to treat lightly the transi existence of universal dimensions, such as so tion process in the ethno-cultural core area. In cial rank, life cycle, and origin, has been con the Basque Country, two provinces, Vizkaia firmed by applying the method of factor analy and Alaba, contain both "contact zones" and sis throughout different regions in Spain. For areas with the Spanish host culture dominant. example, Leonardo and Lavia (1990) applied If the whole of ACBC was an object of spatial factor analysis to and Vitoria-Gasteiz, analysis, social change in the "contact zones" and confirmed the existence of three major di within Vizkaia and Alaba would draw more mensions of residential characteristic. These attention than change in the ethno-cultural three dimensions were social rank, social origin, core region of Basque culture. and youth-oldness. Lavia (1995) also applied The objective of this paper is to explicate the factor analysis to the entire ACBC, and four spatial structure of population characteristics main factors were extracted. These were: rural and its shift in the ethno-cultural core region of industrial continuity, social rank, social origin, the Basque Country, focusing on the province and tertiary centrality. Lavia (1995: 363) indi of Gipuzkoa (see Figure 1), which is distin cated that the dimension of social origin is guished by the relative high profile of the auto sufficiently coherent with the characteristics of chthonous characteristics of local residents the ecological units in a gradation "Native compared with other provinces in ACBC. A

688 A Geographical Study on Population Characteristics 207

Figure 2. Areal units employed for analysis.

factorial analysis will be applied in order to pality, which is the minimal statistical division clarify certain dimensions which regulate the in the state census.6 An analysis of multiple formation and the shift of spatial structure in time phases is required in order to explicate the Gipuzkoa. transitional process of spatial structure. Tak ing into consideration the concordance of unit Method of Study and Analysis area in each time phase, it was decided to em ploy the data sets of two phases of time, 1991 In order to analyze the spatial structure of and 1996. Adjusting the boundaries for 1996 to Gipuzkoan population characteristics, data sets coincide with those for the units of 1991, 87 from the demographic census released by the units as shown in Figure 2 were prepared for ACBC Statistics Institute (Euskal Estatistika both time phases.7 Erakundea; Eustat) are utilized. In ACBC, the In these two time phases, variables which State Census has been conducted every five represent comprehensive aspects of population years since 1981.5 characteristics in Gipuzkoa and which have The statistical unit area employed for this common definition through both annual sec study corresponds to the udalerria, or munici tions are examined from a wide range of demo

689 208 ISHII H.

Table 1. The input variables

a Definition of these categories and items depend on the classification established by the Direction of Studies and Information of the Vice-council of Language Policy of the Department of Culture, Basque Government. The definitions are given as below. Typology of language mobility is established on the basis of crossing of the variables of Mother tongue and Linguistic competence. The linguistic evolution of the individuals is put under analysis. 35. Original Basque speakers: whose mother tongue is Basque, and who understand and speak it "well." 36. Original bilinguals: whose dual mother tongues are Basque and Spanish, and who understand and speak them "well ." 37. People with partial loss of Basque: whose mother tongue is Basque, or Basque and Spanish and who speak Basque "with difficulty," or who do not speak, understand, or read it "well." 38. People with total loss of Basque: whose mother tongue is Basque, or Basque and Spanish, and who neither speak Basque nor understand or read it "well." 39. New Basque speakers: whose mother tongue is Spanish or some other language apart from Basque, and who understand and speak Basque "well." 40. New partial Basque speakers: whose mother tongue is Spanish, or some other language apart from Basque and who speak Basque "with difficulty," or who do not speak, understand, or read it "well." 41. Original Spanish speakers: whose mother tongue is Spanish, or some other language apart form Basque, and who neither speak Basque nor understand or read it "well." Typology of language used in the home is established on the basis crossing of the reduced typology of "Mother tongue" and the variable of "Language of use in the home." The level of linguistic "activity" of the individuals is put under analysis. 42. Original Basque speakers with active Basque use: whose mother tongue is Basque and who understand and speak Basque "well," and whose language use in the home is Basque.

690 A Geographical Study on Population Characteristics 209

42. Original Basque speakers with active Basque use: whose mother tongue is Basque and who understand and speak Basque "well," and whose language use in the home is Basque. 43. Original Basque speakers with active use of both languages: whose mother tongue is Basque and who understand and speak it "well" and whose language use in the home is Basque as well as Spanish. 44. Original Basque speakers with passive Basque use: whose mother tongue is Basque and who understand and speak Basque "well" but whose language use in the home is Spanish. 45. Original bilinguals with active of both languages: whose mother tongues are Basque and Spanish, who understand and speak both languages "well," and whose language of use in the home is Basque. 46. People with partial loss of Basque with active of both languages: whose mother tongue is either Basque, or Basque and Spanish, who speak Basque "with difficulty" or who do not speak, understand, or read the language "well" and whose language use in the home is Basque as well as Spanish. 47. Partial loss of Basque with passive Basque use: whose mother tongue is Basque, or Basque and Spanish, and who speak Basque "with difficulty" or who do not speak, understand, or read it "well" and whose language use in the home is Spanish. 48. New Basque speakers with active of both languages: whose mother tongue is Spanish, or some other language apart from Basque, who understand and speak Basque "well" and whose language use in the home is Basque as well as Spanish. 49. New Basque speakers with passive Basque use: whose mother tongue is Spanish, or some other language apart from Basque, who understand and speak Basque "well" and whose language use in the home is Spanish. 50. New partial Basque speakers with active of both languages: whose mother tongue is Spanish, or some other language apart from Basque, who speak Basque "with difficulty," or who do not speak, understand, or read it "well" and whose language use in the home is Basque as well as Spanish. 51. New partial Basque speakers with passive Basque use: whose mother tongue is Spanish, or some other language apart from Basque, who speak Basque "with difficulty," or who do not speak, understand, or read it "well" and whose language of use in the home is Spanish.

graphic categories. In common with demo data matrix. linguistic variables in almost all earlier studies, Some extra procedures are required in order "percentage of Basque speakers" has been to compare the two distinct phases of time. In adopted as the single variable which represents this analysis, the method employed by Taylor the demo-linguistic situation of the population. and Parks (1975) and Wakabayashi (1987) is However, in the state census, three major lin adopted, in which two time phases are collected guistic indicators exist. These are defined by into one time-space dimension and factor analy Eustat as, "linguistic competence," "mother sis is performed on the data matrix which con tongue," and "language spoken at home." It sists of time-space units and variables. In this should be noted that in the 2nd Sociolinguistic case, it is necessary to maintain the similarity of Atlas (Gobierno Vasco 1997, 1998, 1999), newly factor structures in the two different time calculated demoliguistic indicators, which phases.8 show language shift between the three indica A factor analysis is performed on the data tors of the State Census are presented. These matrix of 174 time-space units (2 time phases•~ data sets let us verify how the demo-linguistic 87 spatial units)•~51 variables. Through a characteristics acquired at infancy subse series of procedures, a matrix of factor loadings quently change. Therefore, in this study, demo of principal factors with eigenvalues of 1.0 or linguistic variables from the 2nd Sociolinguis more are extracted. Furthermore, in order to tic Atlas are employed as well as those of the make interpretation of factors easy, a direct State Census. valimax rotation is carried out on the matrix. In accordance with the procedure mentioned In this way, each factor is interpreted based on above, the 51 variables shown in Table 1 were the matrix of rotated factor loadings, and corre prepared for analysis. Through the above pro lation with the factor structures which ap cedures, a geographical data matrix of 2-time peared in earlier studies can be verified. At this

phases•~87 statistical units•~51 variables was time, there are few previous studies which util created. To achieve the objective of this study, ize factor analysis and examine population spatial structure based on population attributes characteristics within an entire province, and was revealed by applying factor analysis to the therefore vast results of factorial ecology stud

691 210 ISHII H.

ies on urban areas within Spain and other re tended households. Considering the qualitative gions have to be applied for interpretation. aspect, this dimension is very similar to that of "social origin Next, a factor score matrix is calculated, and ," but some variables with high mapping factor scores according to two time loading such as partial loss of Basque and total sections, spatial pattern and annual transforma loss of Basque represent the population which tion process, are clarified. Moreover, a cluster is losing Basque competence in spite of the fact analysis is performed on the factor score ma that they are native Basque speakers or original trix, and regional grouping based on population bilinguals. Moreover, the variables of original characteristics is attempted. Basque speakers with passive Basque use and those of partial loss of Basque with passive

Population Characteristics and Basque use suggest the existence of a popula

Change in Gipuzkoa tion which desires to identify with the domi nant Spanish area though use of the . In the same way, the variable of new Factors and factor loadings Basque speakers with passive Basque use re

A principal axis factor analysis was per flects the existence of a population which de formed on the data matrix of 174 time-space sires to belong to the dominant Spanish area units•~51 variables. Consequently, 10 dimen through use of Spanish, despite having ac sions with eigenvalues above 1.0 were ex quired a certain level of Basque competence. tracted. Furthermore, on applying a rotation Therefore, this dimension can be defined as according to the varimax criterion, the factor a reflection of "social identity," representing loadings shown in Table 2 were obtained. a sense of social belonging and social origin These account for 75.0% of the total variance. simultaneously. Furthermore, considering the scree plot of ei Factor 2, which explains 9.2% of total vari genvalues of these factors, there is a certain ance, is remarkably small compared with Fac information gap between factor 6 and factor 7. tor 1. The variables with high loading are Therefore, it can be interpreted that the factor limited within the categories of occupation, em structure is defined by six primary and four ployment status, and industrial structure. Only secondary dimensions, and the latter are of mi one variable, unskilled workers, has a loading nor importance with regard to the total vari of over .500, in contrast with the negative high ance. Considering these characteristics of the loadings of primary industry, farmers and fish structure, the objects of examination of this ermen, self-employed, and compound family. study can be limited to 6 primary factors. When expanding the range of discussion to the Factor 1 explains 21.7% of the total variance, loadings of .400, negative scores can be found and is dominant compared with other factors. for the variables of stem family, children per As is apparent from the composition of factor household and domestic workers. In contrast, loadings, this factor presents a bipolar struc there are positive loadings for the variables of ture. High loadings (.500 and over) in absolute nuclear family, employees, and manufacturing value are restricted to the variables of house workers. This dimension, therefore, shows the hold structure, place of birth, typology of lan existence of employed workers engaged in the

guage mobility, and typology of language used manufacturing sector in the positive extreme in the home, presenting at the positive extreme and self-employed laborers engaged in the pri a population which has the exogenous charac mary industry sector in the negative extreme. teristics of nuclear family membership, persis It reflects a dimension of regional difference of tent use of Spanish, and immigration from out employment between the agricultural sector

side of ACBC. This is in contrast with the and the manufacturing sector. Consequently negative extreme, which is characterized by the this dimension can be defined as that of "agri

presence of autochthonous residents repre cultural worker-manufacturing worker." sented by Basque natives, people who use Factor 3 accounts for 8.7% of the total vari Basque daily, and who are members of ex ance. The variables which contribute to this

692 A Geographical Study on Population Characteristics 211 Table 2. Factor loadings for spatial differentiation of population characteristics in Gipuzkoa693 212 ISHII H. dimension are limited to those of linguistic industry sector, and in the negative extreme by status. The two categories considered are ty the existence of blue-collar workers. The first pology of language mobility and typology of factor extracted by Leonardo and Lavia (1990) language use in the home. The variables all in the cases of Bilbao and Vitoria-Gasteiz, and have positive high loading and therefore indi the second factor extracted by Lavia (1995), cate that the subjects achieve Basque compe were denominated as "social rank," and it was tence even when their mother tongue is a lan claimed that they represented the dimension of guage other than Basque or when they havee socio-economic status. another language and Basque as mother Verification of factor structure tongues. Simultaneously, these variables in clude Basque speakers whose mother tongue is By comparing the factor structure described Basque and who have adequate Basque compe above with those of previous studies, the char tence. Synthetically, they show a population, acteristics of the factor structure extracted in which does not use Basque actively in the this analysis can be understood. home. For this reason, this dimension can be Factor 1 extracted in this analysis corre interpreted as showing the "difference between sponds to the dimension of social identity, the acquisition of Basque competence and its which is similar to the dimension of "social use," which represents the population capable origin" mentioned in former factorial ecology of passive Basque use in spite of having re studies focusing on urban areas (Diaz 1989; cently acquired Basque capability or native-like Leonardo and Lavia 1990). At question is Basque competence. whether the dimension of social origin should Factor 4 explains 7.4% of the total variance, be extracted independently or whether it and variables which contribute to this factor should be included in other dimensions. Ac are also restricted to typology of language mo cording to Leonardo and Lavia (1990: 105), the bility and language use in the home, as with dimension of social origin "reflects the impor Factor 3. The variables with positive high load tance of the native/immigrant characteristic of ing all represent the population which uses the population, and of the set of socio-cultural both Basque and Spanish actively in the home variables assigned to this basic dimension." domain, in spite of having distinct attributes of Therefore, the dimension of "social origin" can Basque competence and mother tongue. There be defined as a combined form of "ethnic status" fore, this dimension can be interpreted as repre and "migration status" which has been ex senting those subjects with "active use of both tracted in factorial ecology studies in North languages." America (Berry 1971; Rees 1971). Factor 5 accounts for 6.8% of the total vari In factor analyses for some regions within the ance. In the negative extreme, this dimension Basque Country, social origin has been ex illustrates the existence of an elderly popula tracted as an important dimension (Leonardo tion whose members have lost a spouse and and Lavia 1990; Lavia 1995). In a factorial become single householders. In the positive analysis of ACBC by Lavia (1995), the dimen extreme, it represents a youthful population sion of social origin was extracted as the third whose members belong to nuclear family factor. In a comparative analysis of Bilbao and households. Therefore, this dimension can be Vitoria-Gasteiz by Leonard and Lavia (1990), interpreted as that of "youth-old age stage." In social origin was extracted as the second di earlier studies, similar dimensions represented mension in Bilbao. In a city such as Bilbao, by family status have been extracted as princi large numbers of immigrants are concentrated pal factors. seeking employment opportunities. The impor Factor 6 accounts for 6.1% of the total vari tance of variables regarding birthplace inside ance. This dimension can be interpreted as that and outside the region and related cultural and of "occupational status," and is expressed in the social origin indicators are reflected in this positive extreme by the presence of high status dimension (Leonardo and Lavia 1990:105). white-collar workers belonging to the tertiary Arrival year, or immigration stage, has been

694 A Geographical Study on Population Characteristics 213

identified as an important component of the the immigrant population from outside the social origin dimension in various cases (Leo state correlates with the small-scale household nardo and Lavia 1990; Lavia 1995). Occasion structure based on a nuclear unit of wife and ally, however, a dimension of immigration husband, while in contrast the non-immigrant stage is extracted independently, such as in the population correlates with the traditional cases of Madrid (Castelles 1986), and Malaga Basque extended household structure, typically (Ocana 1985). represented by multiple family nuclei and with In other cases, variables related to "origin" a traditional residential style in basseri, or contribute to the dimension of socio-economic easerio. status. For example, in the analysis of Vitoria Factor 2 represents the dimension of diffe - Gasteiz excluding the variables of immigration rence of employment between the agricultural stage by Leonardo and Lavia (1990), social rank sector and the manufacturing sector. This di was extracted as the first dimension. This di mension is similar to the first dimension of mension showed high loadings for some vari rural-industrial socio-demographic continuity ables related to "origin" besides those of socio extracted by Lavia (1995). In the case of Lavia economic status. Such a dimension, which joins (1995), in addition to the variables of certain together the dimensions socio-economic status socio-economic aspects such as occupation, em and social origin, has also been reported in the ployment status and industry, the contribution cases of Sant Adria de Besos (Sola-Morales of the variables of age group and household 1970) and of (Ferras 1977), where structure is observed. Furthermore, in the case immigrants have been concentrated mainly for of Lavia (1995), the dimension of social rank economic reasons. was extracted independently as a second di The dimension of "origin" does not necessar mension. Also, with regard to the first dimen ily correlate only with socio-economic status. sion of rural-industrial socio-demographic con The phenomenon in which variables related to tinuity, two types of unskilled blue-collar work origin correlated with multiple dimensions in a ers were considered in the bipolar structure. complex mode was reported in the factor analy These were employed unskilled workers in the sis of Alcala de Heneres by Diaz (1988). Diaz manufacturing sector and self-employed farm notes a dimension which includes the multiple ers in the primary industry sector. These two dimensions of life-cycle, social origin and immi are generally located in the same rank in axes, gration stage and which was extracted as a first and this dimension was not included in occupa factor. A dimension of socio-economic status tional rank (Lavia 1995: 323). In the same way, was also noted independently as a second fac in this analysis, the dimension represented by tor. Both of these factors accounted for almost Factor 2 is employment status, using criteria the same percentage of total variance. Diaz from within the primary industry sector and interpreted the first factor as a dimension of the secondary industry sector, which were ex "immigration ," and considered that the first fac tracted independently from the sixth dimen tor illustrates an urban development stage sion of social rank. When analyzing on the which represents the residential differentiation provincial scale, the spatial difference based on between traditional areas, where non-immi employment status in the primary and secon grant and old age groups dominate, and new dary industry sectors becomes more important areas, where immigrants and the younger age than occupational rank. Therefore, this second groups are concentrated. dimension can be interpreted as having ex Considering these results of previous studies tracted employment status within primary and with regard to household structure, it can be secondary industries, compared with the sixth stated that Factor 1 of this analysis reflects the dimension, which is concerned with occupa variety of dominant household types of non tional rank. immigrants and immigrants, placing stem fami Factor 3 and Factor 4 represent demo lies in the negative extreme and married couple linguistic dimensions because they are limited families in the positive extreme. That is to say, to the contribution of variables related to two

695 214 ISHII H.

demo-linguistic categories: typology of lan increased rapidly to 46.53% in 1996/97.9 Con guage mobility and typology of language use in sequently, the phenomenon of a concentration the home. Demo-linguistic variables have been of new Basque speakers within the younger used as indicators of social origin or human generation of 20 years and under was observed mobility until now. In this analysis, however, (Gobierno Vasco 1997: 60). This section of the several variables related to the demo-linguistic younger generation has acquired competence situation contribute to the first dimension of in Basque as a second language at school. As social origin. The variables which contribute to for their parents' generation, it is common to Factor 1 indicate a tendency toward devotion use Spanish at home. Consequently, in spite of to either the Basque or the Spanish cultural the increased acquisition of Basque compe domain. These variables could serve as an indi tence, the status of the Basque language has cator not only of the dimension of social origin come to be reduced in the home (Gobierno but also of membership of autochthonous popu Vasco 1999: 53). lation groups and immigrant groups. Further Not only age structure, but also Basque popu more, the variables which contribute to the lation density in the home10 could have an third and the fourth dimensions illustrate trans effect on domestic language use (Gobierno movement between the specific indicators of Vasco 1997: 60). In Gipuzkoa, regional differ the Basque and Spanish languages, and repre ences in Basque population density are notice sent the dimension of diversity in population able, and there are differences between the characteristics based on language. areas with a high concentration of home units One possible reason behind the figures ob containing Basque speakers and those with a tained in the extraction of Factor 3 is a certain low density. To such a spatial demo-linguistic growth tendency in recent years of the popula difference, the recent demo-linguistic element tion which is classified within the category of of the increase in the number of new Basque "new Basque speakers ," who have acquired speakers has been added, and consequently a Basque competence recently. According to the complicated linguistic environment has been 2nd Sociolinguistic Atlas, "new Basque speak generated. ers" are those who have sufficient competence Factor 4 indicates that language use in the to understand and hold conversations in home is influenced principally, at first, by Basque, although their mother tongue is not Basque speaker density in the language net Basque. "Partial new Basque speakers" are work, and secondarily, by relative linguistic those who understand and speak Basque to a competence (Gobierno Vasco 1997: 40). It is certain extent and whose mother tongue is not also mentioned in this study that the frequency Basque. In Gipuzkoa, new Basque speakers of Basque use in the home declines drastically numbered 3.6% of the population of 5 years or as Basque population density drops, compared over in 1986, and increased rapidly to 9.3% in to the cases of Basque use between friends and 1996 (Gobierno Vasco 1999: 82). within the workplace. This dimension reflects The principal reason for such an increase is the recent increase in the number of Basque the advance in the normalization of Basque in speakers, and the subsequent rise in communi the education sector. The Basque language was cation network density. introduced into public education after the es Factor 5 and Factor 6 in this analysis corre tablishment of regional autonomy. At the be spond to two universal dimensions, namely ginning stage, when three models of language family status and socio-economic status, which oriented education were introduced in the have been commonly extracted in earlier facto 1982/83 academic year, 23.41% of the students rial ecological studies for a variety of urban in compulsory education in Gipuzkoa learned areas in Spain or in ACBC. Furthermore, a according to "Model D," which utilizes Basque tendency for family status to be particularly as a language of material description and in significant in its contribution to variables re struction, and accompanies Spanish language lated to age structure corresponds to the find instruction. The students registered in Model D ings of former studies about regions in ACBC or

696 A Geographical Study on Population Characteristics 215

Figure 3. Spatial distribution of factor scores Figure 4. Spatial distribution of factor scores (1991). (1996). elsewhere in the nation. However, the extrac Spatial differentiation of population charac tion of the socio-economic dimension as Factor teristics 6 is at variance with the tendency of socio economic position to appear as an important It becomes possible to examine the spatial and principal dimension in former studies of distribution patterns and transition processes ACBC. In previous studies of various regions in of population characteristics in Gipuzkoa by Spain, nevertheless, it has been pointed out that mapping the factor scores for the six factors socio-economic status is not necessarily ex according to two annual phases as shown in tracted as the first dimension (Diaz 1989). Figure 3 and Figure 4. In the case study of Bilbao and Vitoria In the case of Factor 1, high score units ap Gasteiz by Leonardo and Lavia (1990), the pear frequently for both years in urban areas youth and old age factor with a high loading of and the surrounding zone. In Donostia-San Se age structure hierarchy variables is extracted bastidn and the surrounding zone, high scores in both cities. In Lavia (1995), variables repre of over 1.0 a appear. Donostia-San Sebastian senting age structure (below 10 years, over 60 developed as a center for tourism and related years) and household status (persons per house tertiary industry. At the same time it has been hold, female employment, housewives) contrib the administrative center of the province. ute to one dimension, accompanying the contri There has been a concentration of immigration, bution of an extensive range of variables such from both inside and outside the province, of as industrial structure and employment status. people in search of employment opportunities Also, a dimension classed as "rural-industrial in the tourism service sector and related indus sociodemographic continuity" was extracted. tries. This could be a principal cause of the Throughout these factorial analyses, the contri appearance of high scores in the provincial butions of variables which represent roles capital and the surrounding commuter zone. within the household apart from age structure, High score units are also found in inland household status and housing are reported. industrial zones, such as the Valley which

697 216 ISHII H.

extends from Zumarraga to Legazpi, and also in throughout the central axis of this area from , Arrasate, and other places where in north to south, improvement and double dustrial polygons are located. There is a certain tracking work on National Road 1 is under way. degree of concentration of immigrants from This principal and important national route outside ACBC. High scores are observed also in connects this area to the north (Donostia-San regional centers, such as , the border city, Sebastian and the French border), and to the , the second largest city in Gipuzkoa, and south (the state capital, Vitoria-Gasteiz and the their surrounding areas. In contrast, some local national capital, Madrid). In these areas, undu urban centers such as Tolosa and show lating physical features have been an obstacle low scores below 0ƒÐ. Formerly, Tolosa was the and have delayed road network improvements. of Gipuzkoa, and the history of its Improvements of the main road network and development as an industrial center of the area the accompanying activation of interchange since the medieval era is reflected in the sense with external regions influence the loss of tradi of identity of the population and the low score. tional attributes of origin and identity in this In the case of Azpeitia, which has prospered area. Consequently, such an environment since the medieval era as a commercial center of raises factor scores in the area. the Urola-Kostaldea region, its low score re A clear spatial distribution pattern cannot be flects the attributes of the autochthonous popu observed in the case of Factor 2, which repre lation which has lived there. sents the dimension of employment in primary Relatively low score units below -1.0ƒÐ are and secondary industry. This may be because concentrated in mountainous rural areas ex secondary industry employment in Gipuzkoa is tending from to . In addition, decentralized and exists throughout the whole many low score units below -0.5ƒÐ can be region (Ferrer and Precedo 1977: 56). There found in the region of Urola-Kostaldea, where fore, high score and low score units are inter mountains and highlands extend close to the mingled. shore of the Cantabircan sea. This area, extend High scores do not appear generally in urban ing from the regions of Tolosaldea and Goierri areas where employment in the secondary in to Urola-Kostaldea, is called the "cradle of the dustry sector is concentrated, although high Gipuzkoan dialect of the Basque language," 11 scores are rather more common in urban pe where the physical feature of undulating moun ripheral zones. This could be because employ tains has restricted population mobility within ment clusters of workers who live in the sur the sub-regions since early times, and a unique rounding commuter belts are reflected in the dialect has been conserved until today.12 high scores of the urban periphery rather than When comparing the factor scores of the two in urban centers, where a variety of employ annual phases, a remarkably noticeable rising ment clusters exist. High scores appear in the score is observed in areas where low scores mountainous of eastern Tolosal were dominant in 1991. Such areas are the dea in both 1991 and 1996. In eastern Tolosal

peripheral zone of Donostia-San Sebastian, dea, and show scores over Arrasate, a newly industrialized urban center in 1.0ƒÐ, and in the northern part, shows the western part of the province, Tolosaldea over 0.5ƒÐ. In all of these mountainous rural

(mainly in Tolosa), and Goierri. In the case of settlements, paper mill factories are commonly the surrounding zone of Donostia-San Sebas located, and the employment of local residents tian, the rise in the score is caused by the in is reflected in the high scores. creasing influence of the central place and the On the other hand, low scores appear in , subsequent enforcement of endogenous charac , and within Urola-Kostakdea. teristics. In Arrasate, an improvement of the Those municipalities specialize in forestry and highway network in recent years connecting cattle raising. There are also low scores in the area to the state capital, Vitoria-Gasteiz, had Getaria and , which are fishery the effect of raising the score. In the case of the bases. The rising scores from 1991 to 1996 in area extending from Goierri to Tolosaldea, Tolosaldea and Goierri are particularly remark

698 A Geographical Study on Population Characteristics 217

able. In this area, a series of shifting processes the mass media, administration, and education. within the industrial structure from primary to These phenomena act to reduce the differences secondary industry, and within the employ between Basque competence and Basque use in ment structure from self-managed to employed, the home. Furthermore, it is possible to inter are under way. pret that the results show a qualitative trans The spatial distribution of the factor scores of formation of peoples' negative perceptions of the third dimension in 1991 is characterized by the Basque language, which had been consid the high scores in large urban centers such as ered a "low language" formerly. Recent change Donostia-San Sebastian and Eibar and their in the linguistic situation was reflected at the

peripheral zone. This spatial pattern reflects time of the State Census, and the data value of the existence of a population, which acquired Basque use rose. Basque competence with the advancing process With regard to Factor 4, low scores appear in of the normalization of Basque within schools urban centers such as Donostia-San Sebastian, in large urban centers and in suburbs where Eibar, Irun, and its periphery. Furthermore, exogenous characteristics are dominant. there are low scores also in mountainous mu High scores are also found in the inland ur nicipalities within the regions of Tolosaldea, ban center Tolosa and its surroundings. Al Goierri and Urola-Kostaldea. Since low scores though it is an area in which autochthonous reflect a low incidence of bilingualism, the low attributes are dominant, the high scores reflect scores in urban centers reveal Spanish monolin a process of Basque speaker revival accompa gual use, and in mountainous regions, Basque nied by the popularization of Basque education monolingual use. With regard to urban centers, since the 1980s in an urban center where the high scores are found in some traditional cities, inclination to the Spanish realm had formally such as Tolosa and Onati. This could be be been dominant because of active interchange cause the existence of a population with exoge with the external zone. Simultaneously, high nous attributes appears in bilingual use in spite scores are observed in the rural municipalities of the fact that it is situated within the Basque within the regions of Deba Beherea and Deba influenced zone. Garaia. This is because these regions have a In the time series, there was a noteworthy relatively low density of Basque speakers, and rising score in the western part of the province, the effect of Basque education has been notice especially in the regions of Deba Beherea and able compared with the high density zone. Deba Garaia. A rising score is in fact observed By contrast, low scores are found in the also almost all over the province except for in mountainous rural areas within the regions of the mountain municipalities within the regions Tolosaldea, Goierri, and Urola-Kostaldea. In of Tolosaldea and Goierri. these regions, active Basque use is observed The wide variety of demo-linguistic environ within an environment with a high density of ments cannot be explained simply in terms of Basque speakers, and the existence of actual human mobility. The advance of the Basque Basque speakers is reflected in the low score. normalization process (particularly in the field In a time series, the tendency toward score of public education) since 1982 has produced a decline at a wider, provincial level is remark steady increase in the number of Basque speak able. High scores of over 1.0ƒÐ can be observed ers in the younger generation. The percentage only in a few urban centers such as Donostia of Basque speakers of two years and over in - San Sebastian, Hondarribia, Eibar and Tolosa. Gipuzkoa increased from 21.5% in 1981, to Two causal points should be considered. First 24.6% in 1986, 26.3% in 1991, and finally to ly, originally, Gipuzkoa had already arrived at a 30.9% in 1996 (Eustat 1999:162). According to maximum level of Basque identification be this data, the recovery of Basque language de cause of the existence of a high density of velopment can be interpreted optimistically. Basque speakers. Second, the population with Nevertheless, the rate of Basque use for conver active Basque use in the home has increased in sation in the home domain is declining slightly; response to ongoing Basque normalization in from 13.8% in 1991 to 13.6% in 1996 (Eustat

699 218 ISHII H.

1999: 171). These data represent the relation whole province in 1996. This decline could be ship between Basque and Spanish within caused by the transformation of family struc present Basque society. In Gipuzkoa, where ture which Gipuzkoa has experienced in recent Basque speakers are mostly concentrated years. The recent transformation of household among the three provinces of ACBC, the rate of structure is characterized by a decrease in the Basque speakers with active Basque use in the number of nuclear family households with chil home has declined as far as 23.6% for the 25-29 dren, stem family households, and compound age group, compared with 32.6% for the 5-9 family households, in contrast to an increase in age group or 34.4% of the 2-4 age group. These the number of nuclear family households with data reflect the ongoing phenomenon of the out children, single parent households, and sin relinquishing of Basque use in some age groups. gle householders (Eustat 1999: 132). It is obvi This occurs when Basque users leave the public ous that, behind this structural change, there is education system and become independent of a series of demographic transitions represented family. In spite of having used Basque in the by fertility decline, aging and an increasing home domain in childhood, for these people divorce rate. The provincial scale of this Spanish has become the high language of the change exceeds the national level and also the workplace and Basque the low language. whole EU level.13 Especially, in the mountain Although linguistic competence increases ous rural area within Tolosaldea and Goierri, with education, everyday language use is infl low scores of -1.0ƒÐ and below appear because uenced by the status of languages in society. depopulation and aging in this area accelerate Improvements in the status of Basque and the the transition process of age structure composi establishment of the Basque-based educational tion. model have achieved steady success. This is The spatial distribution of the scores for also true within the administrative scene. Fur Factor 6 is characterized by high scores in thermore, the Basque government is promoting Donostia-San Sebastian and its suburbs. High a plan for the normalization of Basque use in scores are also found in some traditional com the workplace environment in order to improve mercial centers such as Azpeitia and Tolosa, the low status of the language. The binary and their surroundings. In any case, high linguistic situation which currently exists is scores commonly appear in central places and known as diglossia, and the validation of their surroundings, where the population con Basque status through public education has yet tinues to increase as a result of the occupational to appear. characteristics of the upper socio-economic The principal characteristic of the distribu class who specialize in tertiary industry and tion of the Factor 5 score is in time series trans reside in the inner-cities and their suburbs. In formation. In 1991, low scores of -1.0ƒÐ below coastal areas, luxury seashore resorts, such as are found only in inland mountainous units, Hondarribia and , show high scores, re and scores of -1.0-0.5ƒÐ appear both in flecting the occupational status of the upper mountainous units and also in large urban cen class who have residences or second houses in ters such as Donostia-San Sebastian and Eibar. these resort areas. Low scores are found in the Conversely, high scores are found notably in units with small populations in the mountain

peripheral zones of Donostia-San Sebastian, and ous areas within Deba Garaia, Tolosaldea, and this could be a result of the emigration from the Goierri. The rural characteristic of these areas urban area of the younger generation which is reflected in the low scores. Although a simi desires housing space which is both economical lar pattern of spatial differentiation is main and large. High scores are observed also in the tained throughout the time series, an overall inland municipalities within the industrial increase is observed in the province as a whole.

polygons and their surroundings. This is a This is a result of the shift of industrial struc reflection of the presence of a younger genera ture to tertiary industry which is ongoing tion who find employment in those areas. throughout Gipuzkoa. Factor scores decline remarkably in the

700 A Geographical Study on Population Characteristics 219

Figure 5. Typology of areal units in Gipuzkoa (1991, 1996).

(-.697) and extremely low for Factor 2 Regional Typology Based on (-2.505), as shown in Table 3. The cluster is Population Characteristics characterized by identification with the auto chthonous society, and contains self-managed employees in the primary industry sector, a Cluster analysis sedentary population, and Basque speakers. Utilizing the results of the factor analysis, This cluster is heavily represented in the integration and categorization of each dimen authentic rural municipalities such as Aia, or in sion was attempted in order to clarify regional fishing industry villages such as Getaria, within typology. A cluster analysis was adopted as the a high concentration area of Basque speakers method to categorize regional groups. Spe extending from Tolosaldea, Goierri, to Urola cifically, by using a factor score matrix of 174 - Kostaldea. If the spatial distribution of Cluster time-space units (2 time phases•~87 spatial A in 1991 is compared with that of 1996, the units)•~6 dimensions, the Ward's Method clus number of units belonging to this regional ter analysis was performed, in which proximity group decreased from 8 to 5 as a result of shift is calculated and categorized and statistical of industrial activity from primary to other units were combined together one after another sectors. hierarchically depending on proximity. At the Cluster B is distinguished by low mean scores stage where there was a sharp change in infor in Factor 1(-.632) and Factor 3 (-.596), and by mation loss, a dendrogram was cut. Conse a high score in Factor 2 (.947). Therefore, this is

quently, nine regional groups were identified as a group of units with strong identification with shown in Figure 5. the Basque autochthonous realm, and a sparse Cluster A has low mean scores for Factor 1 population with Basque recognition in spite of a

701 220 ISHII H.

Table 3. Mean scores of factors in each cluster

Note: Figures in bold are .500 and over in absolute value. recent acquisition of Basque competence. lingual use in the home domain. Examples are There is a high rate of employment in the mountainous rural districts where aging is pro manufacturing sector. Although this group is ceeding quickly and where Basque speakers are found mainly in the municipalities within dominant, such as some municipalities within Tolosaldea, Goierri, and Urola-Kostaldea (simi Tolosaldea and Goierri. lar to Cluster A), it includes principally inland Cluster E is characterized by high mean units within the industrial polygon and its sur scores in Factor 1 (1.304), Factor 3 (.746) and roundings. Components of Cluster B increased Factor 4 (.553), and by low scores in Factor 4 notably from 10 in 1991 to 21 in 1996. This is (-1.099) and Factor 6 (-.703). Therefore, this a reflection of changing employment patterns can be defined as a group of regional units in the population, which have shifted from the distinguished by a population which is charac primary industry sector. terized by exogenous identity. In this cluster In Cluster C, a high mean score appears in there are large number of Spanish speakers and Factor 5 (1.025) and a relatively high one in immigrants, and Spanish monolingual use in Factor 4 (.445). In contrast, Factor 6 (-.456) the home domain is common, in spite of a ten shows a relatively low mean score. Therefore, dency toward recent acquisition of Basque com this is a group of regional units where the petence. The cluster is, moreover, distinguished younger generation is relatively prevalent, bi by the existence of a younger age group and lingualism in the home is widespread and in lower occupational status. In 1991, Cluster E which relatively low status workers exist. In contains Irun, inland industrial centers such as this category are included small-scale local ur Arrasate, Legazpi, and Renteria, and the zone ban units such as Villabona, Azcoitia and Onati, surrounding Donostia-San Sebastian. The num and also Tolosa's surrounding municipalities. ber of components decreases from 11 in 1991 to The number of components decreased consid 3 in 1996. erably from 27 districts in 1991 to 7 in 1996. Cluster F has an extremely high mean score This is obviously a reflection of decline in fer in Factor 3 (1.767), and also a high mean score tility which has been notable in the whole in Factor 4 (.790). There is a low mean score in province. Factor 5 (-.560). This indicates a regional Cluster D has low scores in Factor 1 (-.667), group in which passive Basque users are domi Factor 4 (-.886) and Factor 5 (-.844). This is a nant in spite of the existence of a population group of regional units in which a population with recent acquisition of competence in the showing remarkable identification with auto language. In contrast, bilingual language use is chthonous Basque society is dominant and in observed in the home domain because of the which aging and depopulation is underway. relatively high density of Basque speakers, and The importance of bilingual use is relatively the cluster is distinguished by the presence of low because of the dominance of Basque mono an aging population. Traditional urban centers

702 A Geographical Study on Population Characteristics 221 such as Tolosa, Eibar and their surrounding relatively low score in Factor 5 (-.438). There municipalities belong to this cluster. fore, the cluster contains the regional units dis Cluster G shows extremely high mean scores tinguished by identification with the Spanish in Factor 1(2.081) and Factor 6 (1.828), and also dominant realm. It reflects an absence of new a high mean score in Factor 3 (.902). Simultane Basque speakers with passive Basque use and ously it shows low mean scores in Factor 4 at the same time the existence of the impor (-.895) and Factor 5 (-1.181). Therefore, this tance of bilingual use in the home. An aged is a group of regional units which contains a population group also predominates in this population who identify with exogenous at cluster. A further characteristic of this regional tributes and who have high occupational status group is its complete absence in 1991, and the represented by employment in the service new appearance of 20 units in 1996. Units sector. Furthermore, it contains an aging popu which belonged to Clusters C and E in 1991 lation, new Basque speakers with passive Bas moved to this category. This shift can be inter que use and passive bilingual use in the home preted as a reflection of recent fertility decline domain as a result of the low density of Basque and aging, both of which are in rapid progress. speakers in the area. In other words Spanish Examination of regional typology formation monolinguals are dominant. This cluster has an extremely restricted composition, contain The characteristics and distribution patterns ing in 1991 only one unit, Donostia-San Sebas of these 9 clusters thus became clear, and it now tian, and in 1996 four units, including Donostia subsequently became possible to examine the San Sebastian and its surrounding municipali regional structure of Gipuzkoa based on these ties. It is indicated that, in this small number regional groups. of units, in other words in the primate city of In all clusters apart from C, F, and H, the the province and its periphery, the number of certain contribution of Factor 1, i.e.; the dimen new Basque speakers has increased steadily sion of social identity, is observed. Of these 6 although the Basque population is small. Si clusters, E, G, and I all have an extremely high multaneously there exists a dominant identific mean score for Factor 1. The set of Cluster E, G ation with the Spanish realm and dependency and I could be defined as a superior group with on the tertiary industry sector. exogenous social identity dominant. There are Cluster H has an extremely high mean score 12 units belonging to Cluster E in 1991. Of in Factor 6 (1.755), and relatively high scores in these, 3 units shifted to Cluster G and 6 units to Factor 4 (.537) and Factor 5 (.534). Therefore, Cluster I in 1996. The shift from Cluster E to this is a regional group distinguished by the Cluster G can be explained as a result of the existence of a population with a higher occupa process of industrial shift to the tertiary sector, tional achievement in tertiary sector employ and fertility decline. The shift to Cluster I can ment, and a tendency toward bilingual use in be understood by a recent advance in reversing the home domain. The population is also domi language shift. nated by the younger age group. Some coastal Cluster I did not exist in 1991, and 20 compo resort municipalities such as Hondarribia and nents appeared in 1996. Compared with Clus Zarautz, and suburbs of Donostia-San Sebastian ters E and G, the mean scores of Factor 3 and are included in this cluster. This regional group Factor 4 in Cluster I are inverted. Therefore, reflects dependency on the tertiary industry Cluster I is marked by the absence of new sector such as tourism within such resorts, the Basque speakers with passive Basque use, and existence of an economically active population by the importance of bilingual use in the home. in suburbs of the largest urban center, and a The appearance of such a regional group, show relatively high density of the Basque popula ing a dimension of the demo-linguistic environ tion network. ment in which Basque competence is newly Cluster I has high mean scores in Factor 1 acquired and bilingual use in the home is domi (1.346) and Factor 4 (.811). This is in contrast to nant, is a reflection of recent advances in the a low mean score in Factor 3 (-.764) and a Basque normalization process, or reversing lan

703 222 ISHII H.

guage shift, mainly in the home domain. The particularly high density of Basque speakers. units belonging to Cluster I are found in the The results can be summarized as follows. industrial areas in the west and south of the At first, in order to resolve the pattern of province, such as Deba Beherea, Deba Garaia residential differentiation based on the popula and Goierri. In Deba Beherea and Deba Garaia, tion characteristics in Gipuzkoa, State Census the density of Basque speakers has been rela data for 1991 and 1996 were used, and geo tively low, and the same can be said of some graphical matrixes of 87 units•~51 variables municipalities specializing in manufacturing were prepared for each time phase. Simultane activities in Goierri. In the same way, units of ously, in order to resolve simultaneously the Cluster I are also found in the southern suburbs spatial transition pattern form 1991 to 1996, of Donostia-San Sebastian. The shift of this the geographical matrixes of each time phase area to Cluster I reflects the recent advance in were combined, transforming to a single data reversing language shift in this area, which is matrix composed of rows of 174 time-space strongly influenced by identification with the units (2 time phases•~87 units) and columns of exogenous realm. Although in these areas, the 51 variables. density of Basque speakers is relatively low for Secondly, as a result of applying a factor Gipuzkoa, the rapid shift to Cluster I is caused analysis to the geographical matrix, 10 factors by the existence of an adequate Basque speaker were extracted. These accounted for 75.0% of network in the home domain to allow a certain the total variance. The principal 6 factors rep degree of reversing language shift from domi resent dimensions of, sequentially from Factor nant Spanish use to active bilingual use. 1, social identity, agricultural-manufacturing Regional groups in areas where industrial ac workers, difference between acquisition of tivity is dominant reflect the regional charac Basque competence and its use, active bilingual teristics of each in spite of being use, youth-old age cycle, and occupational correlated in a complex way with several other status. dimensions such as life cycle, social origin, and Factor 1 represents the dimension of social occupational status. With regard to the kind of identity, which shows the ethno-cultural diver industrial activity, Cluster A specializes in the gence of the Gipuzkoan population. Factor 5 primary industry sector, Clusters B, C, and E and Factor 6 correspond respectively to the specialize in the secondary industry sector, and dimensions of family status and socio-economic Clusters G and H are strong in the tertiary status extracted commonly in earlier factor industry sector. analyses. The contribution of demo-linguistic Moreover, the transformation of employment variables to the dimension of social origin has structure can be understood by examining the been discussed in previous studies. However, in shift process between typologies from 1991 to this paper, variables representing demo 1996. A shift from Cluster C and E dominance, linguistic characteristics also contributed inde

with high secondary industry sector employ pendently to Factor 3 and Factor 4, reflecting ment, to Cluster G and H dominance, with high respectively the demo and socio-linguistic tertiary industry sector employment is espe situations in Gipuzkoa. cially notable. In particular, Irun, and Next, by mapping factor scores, examination Renteria, which shift from Cluster E to Cluster and comparison of the principal dimensions of H, are satellite cities of Donostia-San Sebastian, 1991 and 1996 were attempted, in order to ver and are influenced strongly by the industrial ify the spatial patterns and the time-series shift structure of the province. of each dimension. Finally, in order to clarify regional character Conclusion istics, clustering of regional typology was at tempted in accordance with the similarity of This paper has aimed to analyze the spatial factor scores. A cluster analysis was performed structure of population characteristics and its on six principal factors. Consequently, nine shift in the province of Gipuzkoa, which has a regional types based on the population charac-

704 A Geographical Study on Population Characteristics 223

teristics of Gipuzkoa were drawn out. Moreover, the two dimensions of language Throughout this analysis, certain shift whose presence and characteristics have be processes became obvious, such as socio come evident in this analysis cannot be ex economic changes to industrial and employ plained simply in terms of a bipolar structure. ment structure, demographic change repre These two dimensions, showing a level of pas sented by aging and fertility decline, and a sive domestic use of Basque in spite of the certain degree of advance in reversing lan acquisition of Basque competence, and another guage shift. level of active use of both languages in the As mentioned above, the dimension of social home, correlate with important stages of re origin was the most important for spatial differ versing language shift between two extreme entiation based on the population attribution in poles. Gipuzkoa. Furthermore, it is unique and note worthy that the demo-linguistic dimensions Acknowledgments were extracted independently as the 3rd and 4th dimensions. However, the presence of This research was partially supported by the Minis demo-linguistic dimensions is considered to be try of Education, Culture, Science, Sports and Tech nology, Grant-in-Aid for Encouragement of Young natural rather than exceptional. Scientists, 12780065, 2001. I would like to express In Gipuzkoa, as in ACBC and in my sincere gratitude to the professors of the Institute general, an enthusiastic regionalism movement of Geoscience of the University of Tuskuba, espe can be observed and a revival of regional lan cially to Professor Nobuo Takahashi, for their worth while advice. I am also grateful to the staffs of Eustat guage is taking place. Formerly, in the Basque and the Department of Culture of the Basque Govern country, many nationalist leaders had insisted ment, who offered me precious data sets and informa on the special nature of the racial status of the tion. Basque ethnic group, and a movement empha (Received 8 August 2001) sizing the pureness of racial blood was preva (Accepted 3 April 2002) lent. However, currently in ACBC, including Gipuzkoa, the State Government is the primary Notes force in Basque language normalization and promotion politics. The principal political goal 1. In this paper, names used for the regional units within ACBC reflect the official denomination is to give the Basque language higher status in of the municipalities by Eustat on 3 May 1996. various aspects of the demo-linguistic environ 2. The territory of the Basque Country in the ment. broad sense is situated partly in but Within the dimension of social identity, this mainly in Spain, extending across the interna study has identified a situation of cultural plu tional border. In France, the Basque Country is ralism which supports the typical dualistic situated in the Departement du - atlantiques and traditionally divided into three environment of a bipolar structure of Basque provinces, Lapurdi, Behe-Nafarro and Zuberoa. autochthonism and exogenous non-autochtho In Spain, the Basque Country consists of two nism. However, the contributions of birthplace autonomous communities, Nafarroako Doru and linguistic competence do not mean an ex Kommunitatea (Charter Community of Na act correlation between migration status and varre) and ACBC. The latter is subdivided into three provinces, Araba, Bizkaia and Gipuzkoa. demo-linguistic status. Indeed, in this analysis, 3. In contrast, in France only 59,500 Basque recent migration does not contribute to either speakers exist. In there are 70,500. In the dimension of origin, or to the demo the , although the num linguistic dimensions; therefore, an interpreta ber of Basque speakers is small, the rate tion which does not depend on migration is reaches 41.20% of the total population of this required. The importance of an institutional area. In Navarre, the rate of Basque speakers is only 14.13% of the State population. framework in ACBC, such as the support and 4. There are also many sociolinguistic studies on the promotion of the Basque language in public the assimilation process of ethnic minorities. education and in public administrative organi Bills et al. (1995) analyzed the correlation be - zations, must be recognized. tween the distribution of the population whose

705 224 ISHII H.

mother tongue is Spanish and the distance As of May 1 1996, there were 250 municipali from the border in the southwest zone of the ties in ACBC and 88 municipalities in Gipuz United States, and referred to the phenomenon koa. There is a sub-unit of municipality, the of assimilation to the dominant group in pro auzoa, or barrio; however, it does not cover all portion to the distance from the booboboboborder.Stol municipalities but partly some urban areas. -zenberg (1990) indicated that the employment 7. Modification of statistical boundary of munici status of the Hispanic population in the United palities within the 5 years only happened in States correlates not with their geographical , which was subdivided into distribution but with competence in English, - Gaztelu and Gaztelu. and referred to the phenomenon of occupa 8. Applying the same procedure, a factor analysis tional assimilation depending on competence was applied to each geographical matrix of 87 in English. Similarly, Hidalgo (1995) indicated units•~51 variables of 1991 and 1996, and 11 the existence of the same process of assimila factors were extracted in both years. At this tion to dominant society, apart from in some time, coincidence coefficients of each factor traditional communities where the reversing loading of 1991 and 1996 were examined. language shift, i.e., a shift to Spanish, can be From factor 1 to factor 9, although there is observed, and pointed out the importance of some exchange of order, high values of coinci the geographical and historical environment. - dence coefficients of 0.750 and over are ob 5. Actually, in Spain, seven Autonomous States served. Therefore, the factor structures of both (, Galicia, Canarias, Catalonia, Madrid, years are sufficiently equivalent mutually, and Navarre and ACBC) conduct an original State it can be said that almost the same factor struc Census. In ACBC, since 1981, Zentsu (the Cen - tures were extracted in both years. Conse sus) has been conducted in the first year of each - quently, it is possible to apply a factor analysis decade, and Errolda (the Population Register) to the data matrix composed of 174 time-space has been conducted in the 6th year of each units (2 time-phases•~87 units)•~51 variables. decade. Each Autonomous State conducted 9. Besides Model D, there are Model A and Model Census research in parallel with the National B. In Model A, Spanish is utilized as a language Census till 1991, under the collaboration of the of material description and instruction accom National Statistics Institute (Instituto Nacional - panying Basque language instruction. In de Estadistica: I.N.E.). However, in 1996, it Model B, both Basque and Spanish are used for became difficult to get the collaboration of education activity. In Gipuzkoa, in the 1982/ I.N.E., because I.N.E. limited the items of the 83 academic year, 61.23% of students learned Population Register which are considered nec in Model A. However, the students registered essary for the elaboration of electoral roll by in Model A decreased drastically to 28.93% in the revision of the Census Law in 1990. Under 1996/97. At the first stage of introducing these this situation, seven Autonomous States con models in public education, there was another ducted in 1996 their own Population Register type, Model X, which neither used nor taught research on the same scale as the previous Cen Basque. However, actually, this model has sus or Population Register. Therefore, al practically disappeared. In 1996/97 year, only though the character of the 1996 Population 0.63% of total students were registered in Register differs from the previous researches, Model X. there is no difference in the statistical contents 10. In ACBC, when the density of Basque use in the because exactly the same statistical method home is 80% or over, the rate of Basque speak was employed. In this study, in order to avoid ers overall reaches 69.7% in spite of the fact confusion, the State Census and the Population that there are 12.3% Spanish speakers. By Register conducted every five years from 1981 contrast, if the density of Basque use is 25% or to 1996 by Eustat are both referred to as "State below, the rate of Basque speakers overall is Census." merely 0.0% and that of Spanish speakers 6. In the State Census, a minimal statistical unit is reach 98.6% (Gobierno Vasco 1999: 51). the udalerria, which corresponds to the most 11. The are classified into 8 dia basic administrative unit. Apart from the mu lects (Intxausti 1992). nicipality, Eustat releases statistical figures of 12. The percentage of the population whose larger regional scales. The largest regional mother tongue is Basque reaches 64.5% in scale is the lurralde historiko, or province. In Tolosaldea, 72.4% in Urola-Kostaldea, and ACBC, there are three provinces, Alava, 47.4% in Goierri, compared with 20.3% in Vizcaya and Gipuzkoa. The next is the eskual ACBC as a whole, and 38.6% in the province of dea, or region, which composes each province. Gipuzkoa (Eustat 1999: 246). There are 20 regions in ACBC, 7 of them in 13. In the same period, at the national level, the Gipuzkoa. The next level is the municipality. rate for the age group of 20 years or below

706 A Geographical Study on Population Characteristics 225

declined from 28.5% to 24.3%, and the rate for area. Vitoria-Gasteiz: Eusko Jaurlaritzaren Argi the age group of 65 years and over increased -talpen Zerbitzu Nagusia. from 13.2% to 15.4%, and furthermore in the Gobierno Vasco 1999. II. Soziolinguistikazko mapa: 3. EU, the figures changed from 25.5% to 23.7%, area. Vitoria-Gasteiz: Eusko Jaurlaritzaren Argi and from 13.2% to 14.4% respectively (Eustat -talpen Zerbitzu Nagusia. 1999). Gobierno Vasco, Gobierno de Navarra, and Institut Culturel Basque 1997. Encuesta sociolinguistica de Euskal Herria 1996: La continuidad del euskera II. References Vitoria-Gasteiz: Servicio Central de Publicaciones del Gobierno Vasco. Aitchison, J., and Carter, H. 1999. Cultural empower Gobierno Vasco, Gobierno de Navarra, and Institut - ment and language shift in Wales. Tijdschrift voor Culturel Basque 2000. Encuesta sociolinguistica de Economische en Sociale Geografie 90:168-183. Euskal Herria 1996: La continuidad del euskera II: Berry, B. 1971. Introduction to the logic and limita Comunidad Autonoma Vasca. Vitoria-Gasteiz: tions of comparative factorial ecology. Economic Servicio Central de Publicaciones del Gobierno Geography 47: 209-219. Vasco. Bills, G., Chavez, E., and Hudson, A. 1995. The geogra Hecht, A. 1986. Variations in the socio-cultural phy of language shift: Distance from the Mexican boundaries of Ontario's French ethnic popula border and Spanish language claiming in the tion: A geographical interpretation. Etudes Cana southwestern U.S. International Journal of Sociol - diennes 2: 115-126. ogy of Languages 114: 9-27. Hidalgo, M. 1995. Language and ethnicity in the Castells, M. 1986. La ciudad y las masas. Madrid: Ali "taboo" region: the U.S.-Mexico border. Interna anza Editorial. tional Journal of Sociology of Languages 114: 29 Diaz, M. 1988. El espacio social en la ciudad de Alcala - 45. de Henares. Madrid: Editorial de la Universidad Intaxusti, J, 1992. Euskera: La lengua de los vaseos. Complutense. Donostia-San Sebastian: Elkar. Diaz, M. 1989. Hacia un modelo de diferenciacion Langlois, A. and Castonguay, C. 1993. Mobilite geolin residencial urbana en Espana: La aportacion del guistique de la population de langue maternelle analisis de areas sociales y la ecologia factorial. francaise au Quebec et en Ontario. Canadian Jour Estudios Territoriales 31: 115-133. nal of Sociology 18: 383-404. Diputacion Foral de Guipuzcoa 2000. Estructura Lavia, C. 1995. Areas sociales en el sistema urbano economica de Gipuzkoa. Donostia-San Sebastian: vasco. Onati: IVAP. Diputacion Foral de Guipuzcoa. Leonardo, J., and Lavia, C. 1990. Hacia un modelo Eusko Jaurlaritza 1989. Soziolinguistikazko mapa: general de diferenciacion residencial: Analisis 1986. urteko erroldaren araberako euskal autonomi comparativo de Bilbao y Vitoria-Gasteiz. Ciudad y elkarteko azterketa demolinguistikoa. Vitoria Territorio 83: 97-110. - Gasteiz: Eusko Jaurlaritzaren Argitalpen Zerbitzu Ocana, M. 1985. Sobre los factores de diferenciacion Nagusia. de las areas sociales. Paralelo 37: 389-402. Eusko Jaurlaritza and Nafarroako Gobernua 1995. Precedo, A., and Rodriguez, R. 1989. El sistema ur Euskararen jarraipena. Vitoria-Gasteiz: Eusko bano del Norte de Espana en la transition postin Jaurlaritzaren Argitalpen Zerbitzu Nagusia. - dustrial. Ekonomiaz 15: 18-53. Eustat 1992. Censo de poblaeion y viviendas 1991. 6 Rees, P. 1971. Factorial ecology: An extended defini volumes. Vitoria-Gasteiz: Administration de la C. tion, survey and critique of the field. Economic A. de Euskadi. Geography 47: 220-233. Eustat 1999. Biztanleriaren eta Etxebizitzen Estatisti Saez de Buruaga, G. 1969. Ordenacion del territorio: el - karen Emaitza Nagusiak. Vitoria-Gazteiz: Euskal caso del Pals Vasco y su zona de infuencia. Madrid: Aeko Administrazioa. Guadriana Publicaciones. Ferras, R. 1977. Barcelone: Croissance d'une metropole. Sola-Morales, M. 1970. Factorializacion de caracteris : Anthropos. ticas de una area suburbana. Revista de Geograf is Ferrer, M., and Precedo, A. 1977. El sistema urbano 2: 159-186. vasco: Las ciudades de Guipuzcoa y Vizeaya. Stolzenberg, R. 1990. Ethnicity, geography, and occu Durango: Leonaldo Zugaza. pational achievement of Hispanic men in the Garmendia L., M. C. 1994. El proceso de normalization United States. American Sociological Review 55: linguistica en el Pais Vasco: Datos de una decada. 143-154. International Journal of Sociology of Languages Taylor, P., and Parkes, D. 1975. A Kantian view of the 109: 97-107. city: A factorial ecology experiment in space and Gobierno Vasco 1997. II. Soziolinguistikazko mapa: 1. time. Environment and Planning A 7: 671-688. area. Vitoria-Gasteiz: Eusko Jaurlaritzaren Argi Viceconsejeria de Politica Linguistica 1995. La conti -talpen Zerbitzu Nagusia. nuidad del euskera. Vitoria-Gasteiz: Servicio Cen Gobierno Vasco 1998. II. Soziolinguistikazko mapa: 2.

707 226 ISHII H.

tral de Publicaciones del Gobierno Vasco. teristics in Catalonia. Tijdschrift voor Econo Wakabayashi Y. 1987. Urban factorial ecology in mische en Sociale Geografie 85: 130-140. space and time within the city region of Hi White, P. 1984. The West European city: A social geo roshima. Geographical Review of Japan 60A: 431 graphy. London, New York: Longman. - 454. (JE) Wessels, C. 1994. Accessibility and language charac (JE): written in Japanese with English abstract

708