In the same decade in which computers began to be used in geolinguistics projects, fieldwork began for the development of a linguistic atlas for Galician. In 1974 the Instituto da Lingua Galega (ILG), a research centre at the University of Santiago de Compostela, launched the Atlas Lingüístico Galego project with the aim of documenting the linguistic variation of the Galician language. The legacy of this project are the materials that are stored today at the ILG, and which include survey notebooks, recordings and a good number of written material with ethnographic and linguistic information concerning 167 locations in the Galician linguistic domain. All the information collected by the researchers who participated in the survey is the primary source for much of the knowledge we possess today regarding rural Galician over the last quarter of a century. The project began during a period in which the study of linguistic variation was already in crisis because of the revolution in methodology and principles that began with the development of Variationist Sociolinguistics. Despite all the objections towards onedimensional linguistic atlases, the materials provided by such projects continue to be a fundamental source for the knowledge of linguistic variation today.

ALGa is the first linguistic atlas dedicated to the exclusive study of the Galician linguistic domain through the application of geolinguistic research methods. Galician geolinguistic varieties have been the subject of study in the Atlas Lingüístico de la Península Ibérica (ALPI), an atlas project focussing on peninsular Romances varieties which began in the first third of the twentieth century under the tutelage of Ramón Menéndez Pidal and the supervision of Tomas Navarro Tomas (García Mouton 2012). The ALPI was envisaged as continuing in the wake of the first projects of European geolinguistics and in particular the Atlas Linguistique de la France (ALF), by Gilliéron, and the Sprach- und Sachatlas Italiens und der Südschweiz (AIS), by Jaberg and Jud. For the atlas of Peninsular Romance varieties, information was collected in 53 Galician locations between 1934 and 1935 (Sousa 2008). Although some Galician points were also included in subsequent geolinguistic projects (Alvar 1974), it was not until seventy years after the start of the ALPI surveys that fieldwork began for the preparation of an atlas which would was going to take up the study of spatial variation across the territory regarded as being Galician-speaking. The area investigated for the ALGa comprises localities in Galicia and in the administrative territories of neighbouring Asturias, León and Zamora. The number of points is considered far superior in the ALPI for the same territory, therefore the results provide much more detail regarding Galician linguistic variation.

According to the researchers on the project supervised by Constantino García and Antón Santamarina, a linguistic atlas of Galician varieties was undertaken for various purposes: i) to study diatopic variation at different levels of analysis (phonetics, morphology, syntax, vocabulary); ii) to develop a dictionary of spoken Galician; iii) to bear witness to the influence of social factors on variation (age, class, occupation, academic training, etc.); and iv) to identify the direction of language changes (Garcia et al. 1977: 9–10).

Despite these ambitious claims and that the traditional methods of dialectology had already found themselves affected by the crisis initiated with the first sociolinguistic studies, the ALGa is a linguistic atlas conceived according to the principles of what Britain calls the “pre-sociolinguistic era” (Britain 2014). The variety documented in the project is that pertaining to speakers of rural areas, adults, with stable residence in the survey location and male. In addition, the selection of places surveyed excluded the most populated cities and towns. The information obtained would serve to draw a picture of the spatial variation in Galician linguistic territory, but not to be the starting point for research that took into account other variables associated with social factors (gender, age, occupation, etc.), as the researchers initially claimed. In the introduction to the project’s final two published volumes, the researchers are much more cautious and honest when pointing out the project objectives and the output that can be extracted from the analysis of the materials compiled: “the ALGa seeks to take a close look at the study of the characteristics of modern Galician, and also to offer a vision of reality that allows comparison with other Romance and even non-Romance domains” (ALGa 2015: 11). Furthermore, the information contained in the ALGa notebooks would contribute to our knowledge about the spoken Galician that would be the basis of the standard variety that became widespread from the 1980s. The materials corresponding to the sections of the questionnaire dedicated to grammatical variables were essential for understanding the vitality, distribution, spread and use of the forms proposed for the standard variety (Santamarina 1988; Ramallo & Rei-Doval 2015).

Therefore, the ALGa was developed as a small domain atlas that sought to deepen linguistic knowledge regarding rural Galician in the mid-1970s. The questionnaire designed for this task was modified for the context that would be studied, both in aspects directly related to the structure and characteristics of Galician and linked to Galician cultural reality. The main innovations regarding the foundational works of dialect studies have occurred in the use of the tape recorder for recording oral witnesses and the use of a camera to record images of objects of ethnographic interest.

Until now, the ALGa has been the basic source for Galician dialectology research. Analyses of Galician linguistic variation and proposals for dialectal division have so far been based on the materials collected as part of the ALGa project and therefore reflect rural Galician spoken in Galicia in the 1970s. These materials have been the basis for most studies of Galician dialectology conducted over the past three decades, whether these have been the interpretation of simple answer maps like those of Álvarez Blanco (1983), González González (1984) and Dubert & Sousa (2002), or those based on aggregate data analysis (the most classic examples being Santamarina 1982 and Fernández Rei 1985; 1990), and those applying more sophisticated quantitative procedures, such as Álvarez et al. (2006), Dubert (2011) and Sousa (2006). In recent years, ALGa has also served as the basis for conducting analytical studies of linguistic change in real time, such as those carried out by González et al. (2002), Louredo Rodríguez (2014), Rodríguez Lorenzo (2012) and Sousa (2010; 2012).

For both the undertaking of some of that research especially that developed in the 1980s and 90s, and the editing of the first volumes published by ALGa researchers, data was collected individually by obtaining access to books used in the original field research or printed maps from the volumes that were being published. Data collection for this procedure is very laborious as it requires the investment of considerable time and extreme care in handling the original documents. The impossibility to access data in a simpler manner and the lack of files in digital format restricted the ability to perform a quantitative analysis of ALGa for decades and even hindered the edition of the first volumes.

The first ALGa volume, published in 1990, was produced in an almost handmade way, passing information manually from notebooks to maps and pages of notes (ALGa 1990; Sousa 2004). The second volume was prepared using a graphic design program for the cartographic representation of the results, thus improving the appearance of the maps and lessening the correction work. During the drafting of the third volume (ALGa 1999), the project team had the opportunity to meet other groups working on linguistic atlases and began to reflect on the need to computerize the development process of the volumes. On the one hand, it was necessary to automate the process of map making in order to make it manageable, but an urgent task was also to seek a new medium to all project materials in order to ensure their conservation and enable a better use of the dialectal information.

The first attempt to carry out the task of computerizing the ALGa materials took place in the late 1990s. At that time, several linguistic atlas computerization projects had already been launched and therefore there were models available for dialectal database design and for setting up a computer-aided mapping system (Bauer & Goebl 2000; Kretzschmar 2001; Aurrekoetxea 2008; Olariu & Olariu 2014; Kumagai 2016). First it was necessary to transfer the material from ALGa in a format that would guarantee conservation and enable usage through different types of applications (geographic information systems, web services, networking applications quantitative analysis, etc.). The materials collected during dialectal fieldwork are usually heterogeneous in nature and require a different approach. The documentation registered for ALGa consists of three types of materials: field notebooks, sound files and written documentation.

3.1 Field notebooks

The field notebooks are the paper questionnaires in which the answers of the respondents in each of the locations researched were written down. For each of the 167 points that form the ALGa network, there is a 93-page book with 2712 questions and the phonetic transcription of the corresponding answers (Fig. 2). The questionnaire is divided into sections according to the type of linguistic information sought: phonetics (1–148), morphology (149–386), syntax (387–526) and lexicon (527–2712). The answers were transcribed using a phonetic alphabet originally developed for peninsular varieties by Tomas Navarro Tomas (1966) and used in most Iberian geolinguistic developed during the twentieth century. On the even pages there is complementary information regarding the answers noted on odd pages: explanations, sayings, indications as to the use and the validity of the answers, drawings etc.

Fig. 1

Map of locations surveyed for the ALGa.

Fig. 2

Page 68 of the notebook for A Gudiña (O-21).

The first task to be performed in order to ensure the conservation of the notebooks and ease of reference was scanning. A high quality PDF file was created for each of the field notebooks. The files were identified by the code assigned to each of the locations surveyed (a letter identifying province followed by a number).

3.3 Visual data

In addition to the field notebooks and sound recordings, and to accompany the questions of an ethnographical nature, the ALGa project also has an important collection of drawings and photos. The drawings are contained in field notebooks and were made by the same researchers (Fig. 3). A collection of photographs contains a total of almost a thousand images, most of them from farm implements and objects associated with traditional professions. This graphic material is scanned and each file is identified by reference to a location and the corresponding question on the questionnaire.

Fig. 3

Drawing linked to question 1003 angarela ‘stretcher’, Ribadulla notebook.

In order to enable the employment of dialect data from different perspectives (searches, index compilation, variation analysis, results visualization, diachronic data analysis, etc.), the first step was to organize and structure the information so that each linguistically registered form could be characterized with respect to all the parameters considered in the original investigation. Despite attempts made in recent years to establish a catalogue of classes and properties for data from geolinguistic projects, we still do not have a standardized ontology that has been put into practice and evaluated. The proposal submitted for Atlante Sintattico d’Italia (ASIt) by Bucci is one of the most detailed and complete, but has still not even been implemented for the project for which it was designed (Bucci et al. 2014). This shortcoming and the heterogeneity of these same geolinguistic projects have been used to justify the difficulty of reusing data and the limited presence of this type of projects in open repositories of linguistic data (LLOD).

Using the aforementioned proposal for classifications of geolinguistic materials (Buccio et al. 2014) as a guide, it is possible to suggest a basic organization of information in three main areas: i) geographic location data; ii) data on the informer subjects; and iii) linguistic data documentation. In a way, this distinction was already the foundation of the ALGa database’s first structure (Sousa 2004).

Class i) corresponds to the table named Puntos, and class iii) consists of the contents of the Tipos, Respostas and Cuesitonario (Fig. 4). The content of the current database is already organized on the basis of a distribution of information in these three areas: towns, locations and documentation (Fig. 5).

Fig. 4

Basic outline: structure of the first version of the database.

Fig. 5

Schematic structure of the current ALGa database.

4.4 Using the database

The ALGa project’s current database and the mapping display system are designed with commercial software. The database uses Microsoft SQL Server 2000® and its access interfaces were designed using the Access Data Project (ADP). ADP is a file format for Microsoft Access that contains information about the project, and enables database objects to be connected and consultation and editing forms to be designed. Fig. 5 shows the components and the fundamental relationship of the tables and objects that constitute the database. The index used to relate the different tables is the code assigned to each of the locations surveyed.

Access to the database is realized by means of an interface (ADP) that allows the three basic tasks that an operator can perform to be selected: editing of materials, querying and mapping.

The appearance of the editing application is presented in Fig. 6. This interface is used to enter the information on the basis of notebooks and to make corrections regarding the information transcribed. With the selection of a question from the questionnaire, the operator has access to all information related to the selected answer and the image of the questionnaire page on which the information is found. At the top of the window, data appears identifying the edited question (question number, the text of the question on the questionnaire and the identifying name of the question). In the central part is the page image from the original questionnaire where the question is found. This image always pertains to the question and location selected. At the bottom of the window appears the list of survey points (the left box), where the location on which to work is selected. In the central square, larger in size, fields are located that collect the different ways of transcribing the answers: an alphabetic code that identifies the answers (type), phonetic transcription (IPA Answer) transcription in conventional orthography (Orto Answer), transcript to be included in the printed map (Legend) and notes made by researchers. To the left of the window are two boxes that provide information about forms registered as answers: number of phonetic and orthographic variants, and variants of these relationships. For the question used as an example (Fig. 6) 28 phonetic variants were collected and 8 in conventional orthography.

Fig. 6

Editing application; question 1882 espertar ‘to awake’, point O-21 (A Gudiña).

The information contained in the database can also be accessed through a search interface. Through consultations, information can be obtained on all answers and additional information recorded in notebooks. It is possible to perform a search on the basis of the question number, locality code, a given semantic field, a string of sounds or graphemes, etc. Lists of variants associated with the same question indicating the frequency with which they appear in the database and calculations on the occurrence of certain linguistic forms (sounds, strings of sounds, morphemes, words and complex constructions) can also be obtained in a straightforward manner. The connections between the different tables that constitute the data base enables the relationship shared by the points where form or forms sought were recorded may be established in all cases. Fig. 7 presents as an example the results table for the form espertar ‘to awake’: locations where it was registered (Locality code), questionnaire questions where the answer appears (Question number) and associated transcripts (IPA, Orto and Legend).

Fig. 7

Table of search results for variant espertar ‘to awake’.

The design of the current ALGa database was used as a model to develop the web application project for the preparation and publishing of the ALPI materials. This application enables online editing work to be performed and in the future allow consultations to be made regarding the materials (García Mouton 2016).

The design and features of the ALGa database enable developed cartographic visualizations of the results to be produced in a simple manner. The database is connected with a GIS programme that immediately creates thematic maps. The connection between the data base and GIS enables, through the selection of a question number from the program outline, a map and legend with relevant symbols to be generated immediately (Fig. 8; the selection square with all questionnaire questions appears at the top left of the image).

Fig. 8

Screenshot of GIS program employed for cartography.

For the production of maps for the ALGa volumes that are currently being edited, the ArcMap tool is used from the ArcGIS Desktop package, created by ESRI®. A first map is created automatically by selecting a question from the questionnaire and via the same application site. On this map, symbols are arbitrarily assigned (one colour for each different variant; Fig. 9). The editor must choose the shape and colour of the symbols that allow a better understanding of the geographical distribution of the variants associated with the variable that is represented (phonetic, morphological, lexical, syntactic, etc.; Fig. 10 shows an example of a map for a lexical variable that has been edited and ready for publication). The link between the database and the GIS allows more complex displays to be obtained: data representation from different sources, frequency analyses of segments, combination of diverse types of data (linguistic, economic, geographical, historical etc.).

Fig. 9

Map generated automatically with the results of question 1011 hórreo ‘raised granary’.

Fig. 10

Map edited for publication with the results of question 1445 teixugo, porco teixo ‘badger’.

The application for using the database and the mapping software are employed solely for internal use within the ALGa research team for the development andediting of hard copy volumes. Some of geolinguistic projects implemented in recent decades used cartographic display applications designed specifically or for use with web applications that allow simple format representations.

The use of GIS software, as used for the ALGa project, offers several advantages with respect to any of these other options. The GIS programs use standard formats which guarantee interoperability of the information processed and the continuous renewal of applications. Moreover, this type of software allows maps to be exported in formats that facilitate the work of editing the final results.

In addition, currently there are free code applications that have the utilities required in order to allow a not very experienced user to produce linguistic displays without difficulty and for free.

In 1974, the first fieldwork began for the collection of the materials that would allow the first rigorous and reliable description of modern Galician’s geolinguistic varieties to be drawn up. The work of documenting, analysing and editing these valuable materials began at the same time as computer technologies began to be applied to dialectal studies. These innovations are now being used for the edition of the 12 volumes that will comprise the complete work.

The development of the ALGa database seeks to preserve the original materials from the research and advances in the employment of this valuable information. The project co-ordinators continue to work in order to enlarge and complete the database with additions that perfection and help disseminate the results: the inclusion of visual fields for drawings and photos, the introduction of new fields with information from the analysis of the answers (morphological segmentation, detailed semantic characterization, the linking of related answers, etc.), incorporating information from other sources (etymology, synonyms, bibliographical references, etc.), etc.

Amongst the tasks that are intended to be undertake in the future is also the updating of the design and database components. This work is guided by two fundamental purposes: i) the use of free code applications, both in the database (MySQL, MariaDB, PostgreSQL) and in the mapping application (QGIS and OpenLayers); and ii) the open availability of all information related to the project. For the latter objective we are designing a web application that allows the consultation of information, its cartographic representation and also free data access for the user who is interested in studying Galician geolinguistic variation. In the academic field, there are numerous geolinguistic projects that allow information to be consulted and to download data in formats compatible with several analytical applications. In this area, a remarkable project, both for its design and ease of access to all information, is the World Atlas of Language Structures (Dryer & Haspelmath 2013). The ALGa project must participate in the current trend in the field of academic research that promotes the development of infrastructures that ensure the conservation of data, permit free access to information and make resources available that enable the straightforward employment of materials (Sousa 2016).

Linguistic atlases, and geolinguistic projects in general, are projects that are financed almost exclusively with public funds in all the stages of their development. Therefore, it is the obligation of researchers to make all the information collected in these projects available to the scientific community. Furthermore, all those of us who are involved in geolinguistic research recognize that the results of the linguistic atlases development projects rarely achieve the dissemination deserved. At best the volumes of atlases published in hard copy eventually end up being difficult to access in a few specialized university libraries. Indeed, linguistic atlases published in hard copy are works that are difficult to handle and in which is not always easy to find information.

Geolinguistic studies should avail of all means made available by technology today (Kretzschmar 2013). Disseminating knowledge regarding variation in languages and facilitating access to information so that it can be explored from various perspectives can only redound to the benefit of the projects in which we are involved.

This work was funded in part with the help of the Ministry of Economy and Competitiveness of the Spanish Government (project FFI2015-65208-P), the Galician Government and the European Union (GRC2013/40). I would like to express my gratitude to César Osorio for his crucial assistance in the design and development of the database and the automatic map preparation system. The Atlas Lingüístico Galego was developed at the Instituto da Lingua Galega at the Universidade de Santiago de Compostela thanks to the continuing contribution of the Galician Government (Dirección Xeral de Política Lingüística da Consellería de Cultura, Educación e Ordenación Universitaria) and the Fundación Barrié.