Version française

Grindmill songs of rural Maharashtra
A cultural-sociological database

In the context of projects Grindmill songs of Maharashtra

conducted by the Centre for Cooperative Research in Social Sciences (CCRSS, Pune), musical forms of rural Maharashtra have been being documented and analysed with the help of a computerised database.

The database

This relational database is a set of six tables containing texts, pictures and sounds. Relations and numbers of records of each table are shown below.

SONGS (109997)





TUNES (90)

This database was implemented under FileMaker™, ultimately versions 5-6.
It is now available in several export formats, see:
All data are accessible on a recently implemented thematic server:


is a set of villages and hamlets (see diagram).

Each location is defined by its name (following the census), taluka, district and state. Valley information may optionally be included.

Each location is assigned a “corpus” (Mawal, Western Maharashtra...) used for the classification of grindmill songs.

Several fields are available in this file to store textual information about the village (in Marathi and English) and pictures.

A portal connected to the PERFORMERS file displays the pictures of performers belonging to each location.


is a set of persons who supplied information contained in the database (see diagram).

Each person is identified with a full name and assigned a location, gender and cast. Biodata in Marathi and English have been entered for a few performers. A portrait photograph is stored when available.


is a set of texts of grindmill songs (see diagram).

Each song is assigned a “song number” indicating its order in the performance.

Each song is assigned a performer (the main informant during its recorded performance). Location, corpus, cast, photographs, etc., are looked up from the LOCATIONS and PERFORMERS files.

Each song is assigned a classification code characterising the main, and optionally secondary, semantic content of its text.

Each song may be assigned a recording index, if it has been tape-recorded, in which case tune information may be looked up from the TUNES file and displayed.


is a set of semantic units used in grindmill songs (see diagram). This set has been elaborated on the basis of a large corpus of songs collected in area covering the “Mawal corpus”.

Each unit is assigned two codes. The first one is the paragraph number in the classification scheme designed prior to the database setup. The second one is an equivalent alphanumeric encoding compatible with search procedures in the FileMaker database system.


is a set of DAT (Digital Audio Tape) cassette recording sections (see diagram).

Each section is identified by its cassette number and program number (index) on the cassette.

Each fragment may be assigned a tune number, in which case the sound examples stored in the corresponding tune will be displayed. To avoid ambiguity, program numbers on the DAT recordings determine sections with a unique tune.

Each fragment is assigned a location. Related information is looked up from the LOCATIONS file.

Consistency checks in the SONGS database verify that the location of the recording and the location of the performer of each musical piece are identical.


is a set of melodic forms (see diagram).

Each tune is characterised by a few sound examples: original grindmill songs, other types of songs in various musical genres or performance contexts, interpretations by classical and film music performers, etc.

A simplified musical score of the tune is stored in text format. It was created by a MIDI keyboard using the sargam notation system of Hindustani music.

Each tune has a “comment” field in which statements are entered to facilitate the classification, notably observed similarities with other tunes.

A classification of tunes has been initiated in 1997 by Véronique Bacci, a student in musicology at the University of Aix-en-Provence (France).

Linguistic analysis

A table named LEXICON contains words in Marathi along with their spelling variants and derived forms. Each form may be automatically searched in the corpus of grindmill songs (the SONGS file, see supra) or in a corpus of popular myths compiled in a separate project.

The FileMaker™ script automatically selects songs or myths containing occurrences of the target word. It returns the total number of occurrences and (in the case of grindmill songs) the set of semantic units characterising the contexts in which this word is used, and the set of talukas in which songs containing it have been collected. This is very useful for quantitative statements about the semantic and geographical domains covered by each word.