| Summary: | The elicited speech samples (divided into word-list and quantity material) consists of about 50 words chosen to represent all (known) Swedish vowels, consonants, and certain consonant combinations as well as quantity. The vowels occur in a phonetically balanced context, mainly a dental/alveolar one. Mono- or disyllabic words are repeated about thre times,and quantity words about five times. For some dialects, additional words were elicited in order to test dialect-specific pronunciation distinctions. Whenever possible, the words were not elicited by reading or repeating but by memory activation, e.g. by filling in the missing word of a sentence (e.g. "When you receive a gift it is polite to say ___.") or opposites (e.g. "The opposite of 'here' is ___."), etc. Simple and obvious contexts and target words were used as far as possible in order to avoid unwanted side-effects, such as speakers who believe they are participating in an knowledge test. The recordings were made in the consultant's home environment. Date of recording given is guaranteed to be correct only for year. Abstract Phonetic and phonological data on today’s Swedish dialects are scarce. The aim of this project is to record and analyze comparable speech samples from about 110 dialect areas. A central goal is to establish phonetic and phonological typologies on the basis of segmental and prosodic characteristics. The project which is scheduled to be carried out during 1998-2003 is a collaboration between the phonetics departments in Lund, Stockholm and Umeå. Method A selection of dialects was chosen in such a way as to enable phonetic and phonological analyses of a broad spectrum of segment inventories (vowels and consonants) and prosodic systems (intonation, word accent and quantity). Regionally conditioned voice qualities (phonation types) were also considered. Speech samples were collected from two groups, young and elderly. From each dialect area 3 male and 3 female speakers in an age interval from 55 to 75 were documented. The younger group consists of male and female speakers in the age interval from 20 to 30 years. The elderly informants were recruited from a rural population rooted in a traditional dialect. The younger informants were preferred to be of a similar geographical and educational background. In general, however, the younger generation had a longer formal education than the elderly generation. The contact was made via organizations which study local geography, history and folklore. The recordings took place in the informants’ homes. The recordings contained both spontaneously produced and elicited speech materials. The spontaneous material consisted of interviews with one informant at a time. The elicited material consisted of words, phrases and sentences created with the purpose of documenting dialectal variation in segmental and prosodic features. The elicited speech samples (divided into word-list and quantity material) consists of about 50 words chosen to represent all (known) Swedish vowels, consonants, and certain consonant combinations as well as quantity. The vowels occur in a phonetically balanced context, mainly a dental/alveolar one. Mono- or disyllabic words are repeated about thre times,and quantity words about five times. For some dialects, additional words were elicited in order to test dialect-specific pronunciation distinctions. Whenever possible, the words were not elicited by reading or repeating but by memory activation, e.g. by filling in the missing word of a sentence (e.g. "When you receive a gift it is polite to say ___.") or opposites (e.g. "The opposite of 'here' is ___."), etc. Simple and obvious contexts and target words were used as far as possible in order to avoid unwanted side-effects, such as speakers who believe they are participating in an knowledge test. The recordings were made in the consultant's home environment. Swedish was spoken throughout the session Birthdate given is guaranteed to be correct only for year. The .ord files contain manual segmentation of the target words of the elicited material. Inter-transcriber consistency (e.g. segmentation of words with initial plosives) is not very high because of a relatively large number of transcribers. The .seg files contain manual segmentation and phonetic transcription of the target phonemes in the elicited material. To increase inter-transcriber consistency for the vowels, two sets of reference vowels (one female set, one male set) were used for comparison. The .ant files contain comments related to the other written resources (orthographic and phonetic transcriptions), such as if the speaker was laughing or coughing, and whether there was any disturbing background noise.
|
|---|