All studies concerned with investigating linguistic forms as they occur in everyday usage must rely on data collecting in the field through interviews with informants. This is to ensure that what is collected will be an acceptable sample, which will meet with the requirements of the study being undertaken. Certain factors are common to this data collecting process whether the study be lexical, descriptive or comparative.
Compiling a dialect glossary or dictionary means exploiting as fully as possible the totality of expression in everyday use amongst the people of an area. A dialect dictionary devoting itself to the whole language is an immense project. One aspect of the lexicographical projects undertaken in Finland, for example, had as its aim 'a dialect dictionary exhaustively expounding the vocabulary of Finnish dialects, dialect forms and distribution' .14
It was calculated that for such a dictionary a specific number of 'complete' parish vocabularies would be needed. These 'complete' vocabularies would include:
The common vocabulary of the community in its everyday living which in itself will vary considerably according to the predominant occupations of its inhabitants.
The vocabulary particularly associated with those predominant occupations as well as other crafts and skills.
The vocabulary related to the whole range of aspects which form human experience: bird, plant and animal names, natural features, folk lore, hobbies and pastimes, etc.
The particular modes of expression of a community in its telling idioms and phrases which may often be peculiar to the region.
Such collecting should reflect regional differences in the vocabulary - the local forms of the 'same word' and the local words for the 'same thing' - and will include morphological variations.
To describe the sound patterning - the phonological encoding - and the composition of words and the manner in which they are ordered to form complete sentences the morphology and syntax - of a variety of speech within one defined area, requires a copious sample of speech-data if it is to be adequate enough to reflect the range of patternings in that variety within the levels mentioned.
The kind of questions to be answered may be illustrated in an example from the phonology of a variety of Welsh. It shows the importance of sound distribution in the patterning. In this variety the distinctive diphthong [æə] occurs. Native speakers will use it in monosyllabic forms:
/tæəd/ father ., /tæən/ fire
/fæa/ broad beans., /ɬæə Ɵ/ milk
In monosyllabic structures in surrounding areas this sound does not occur. It is replaced there and in most other areas of Wales by the sound [ɑː].
The process of collecting will reveal a certain ordering, a pattern, and pose questions for the investigator. He will want to know if [ɑː] occurs at all in the south-east Wales variety and whether [æə] occurs in structures other than monosyllabics. He will find that the sound [ɑː] occurs also in the variety but in other positions, namely in the first syllable of a disyllabic structure:
[pɑˑtäɬ] pan, [kɑˈtu] to keep
[nɑˈpöd] to recognise
In this position the pattern corresponds to what is common in other areas. The distribution of a sound in a system can be a factor which distinguishes varieties of speech. This is the case here. Apart from the fact that the presence of [æə] in the inventory of sound distinguishes the variety, the occurrence of [ɑː] does not display equivalence of distribution with other varieties and thereby also differentiates this variety.
What is collected must identify all the speech sounds - vowels and consonants - used, where they occur in structures, their positional variants, their possible combinations in consonant clusters or diphthongs, the range of structures, syllabic stress, junction variation and all such features which make up the phonological encoding. Likewise how word elements combine and are composed into word-forms must emerge from the collected data.
To ensure that he has collected the necessary data the researcher must resort to directed investigation of the speech usage, formulating questionnaires whose purpose is to reveal specific linguistic structures, and to follow up initial results with further questionnaires designed to throw light on particular problems. What goes into a questionnaire will be based on the investigator's experience of the language and especially on what is known regarding specific varieties of it. But the arrangement of a questionnaire must be thought out and related to the way language is normally used in the community, if the samples are to approach what could be deemed as a natural response. In order to do this the investigator must be conversant with the pattern of livelihood within the various communities of "the area and must centre his enquiries on topics related to his informants' everyday experience.
A comparative study provides a conspectus of the linguistic usage at a certain period in time. At the planning stage for such a survey two aspects relating to its range have to be decided. The first of these has to do with determining the territorial extent of the field investigations. This should not prove difficult. The survey may cover an agreed area of the country with the remaining parts being apportioned to be investigated by other centres, or it may be the intention to cover the whole country from the working base. In a case of this kind, as happened in Ireland, practical co-operation between interested centres in providing financial resources and trained staff can ensure that one centre is able to plan and undertake such a survey for the whole country.15
The second aspect relates to the scope of what is to be collected in each location chosen for investigation. This depends on the kinds of variation in usage to be surveyed. Basic to all these types obviously is the regional variation, but within this divergence the differences related to social dimensions and age groups are included in the wider ranging surveys. Criticism was often made of the earlier surveys in that they confined themselves to the speech of the older generation and to one social group, frequently a rural informant. 16 In The Linguistic Atlas of the United States it was maintained, however, that 'all population centres of any importance are regularly included, and, in principle, all social levels are represented' .17
After deciding the scope of the survey, the collecting of data as in the case of the other studies must meet with the criterion of adequacy of evidence for what is proposed. Sufficient data must be obtained from each place of investigation to provide information for the ensuing analysis. In order to evaluate the kinds of phonological encoding likely to occur regionally this concept of the adequacy of evidence for each set of data collected was formulated thus:
'Care was taken to provide sufficient material for a rather full description, both phonemic and phonic, of the pronunciation of each informant, and hence for determining the regional and social distribution of the phonic variations of all the phonemes of American English, and for establishing differences in phonemic structures. '18
In the later stages of analyzing the data 'the essential soundness of the expectation' was proved. 19
As we are again in search of natural speech-forms the data must be collected from the language of common interaction within the community. For this reason:
'Regional and local expressions are most common in the vocabulary of the intimate everyday life of the home and the farm -not only among the simple folk and the middle class but also among the cultured ... Food, clothing, shelter, health, the day's work, play, mating, social gatherings, the land, the farm buildings, implements, the farm stocks and crops, the weather, the fauna, and the flora -these are the intimate concern of the common folk in the countryside, and for these things expressions are handed down in the family and the neighbourhood that schooling and reading and a familiarity with regional or national usage do not blot out. 20