logo Enthusiasms No 29
A collection of shorter pieces on subjects of
interest, outrage or enthusiasm ...


The Folk Song and Broadside Indexes

Compiled by Steve Roud, Southwood, Maresfield Court, High Street, Maresfield, East Sussex TN22 2EH.
Tel: 01825 766751   Email: sroud@btinternet.com  Price: £50 ($100 overseas) for the Original subscription; this includes all the current data plus periodic updates for at least one year.

Introduction:

Readers of the better examples of British CD booklets may have noticed that, for the last few years, 'Roud Numbers' have been appearing alongside, or in place of for ordinary songs, the familiar Child Numbers allocated to classic ballads.  Having just received the first 2002 update of the Indexes, I though that a piece about Steve Roud's Folk Song and Broadside Indexes would make interesting reading for those of you unfamiliar with them.

Steve Roud, formerly Honorary Librarian of the Folklore Society, has been compiling data on Broadside Ballads and Folk Songs for a number of years, and has, more recently, been making the results of his work available to both Institutions and the public as a pair of computer databases.  Currently they contain between them almost 235,000 records.  The data is supplied in comma-delimited text form, ready to be imported into your own database program, once you have set it up ready to receive and display the data - not a particularly difficult task.  This year, Steve is also supplying the Indexes already set up for Microsoft Access '97 and Filemaker Pro (v4), and you may find that you can Import these popular formats directly into your own Database program.

Each record contains information about the ocurrence of a song - in a book, broadside, manuscript, on a sound recording, etc. - and there are 20 fields to each record.  A complete list of these fields is given below, but the more important ones are Title, First line, Performer, Collector, Place, Date, Type, Roud number, Other number.  Rather obviously, few, if any, records will contain data in all 20 fields.

The Two Indexes:

The Folk Song Index and the Broadside Index have a separate identity but can be used to full effect in conjunction with each other.  They share the same structure (in particular, the same fields), to facilitate the swapping of data.

The Broadside Index started out as a simple database of songs which had been printed on pre-1900 broadsides, chapbooks, and popular songsters.  The Index functions as a major finding-aid in its own right, but was also designed to be of particular assistance to the folk song researcher.  In its latter role, it soon became clear that a wider range of genres needed to be listed and, although broadsides and chapbooks still constitute the bulk of the Index, the following types of song are increasingly included: music hall, parlour ballads, blackface minstrel, community songbooks, ballad operas, radio singers' folio albums, hillbilly and race records, and so on.  There is no official cut-off date for inclusion, but in practice there is very little material from after the Second World War.

The Folk Song Index comprises English-language traditional songs wherever in the world they are found.  Unpublished material such as field-recordings and manuscripts is included as well as published books, articles, sound-recordings and so on.  No hard-and-fast definition of 'traditional' is at present possible, and the term is used loosely here to denote all songs 'collected' as folk songs from non-professional singers, with the exception of performers who constitute part of the 'folk song revival'.  For example, singers such as the following, normally called 'folk singers', are excluded: Pete Seeger, Joan Baez, Bob Dylan, Martin Carthy, Ewan MacColl, A L Lloyd, the Dubliners, the Spinners, and so on.

Once a song has been included on its own merits as a song with at least one 'traditional' version, earlier appearances of the song are included to aid historical and comparative research.  These items are copied from the Broadside Index, and do not in themselves denote traditional performances.

Usage:

The way I use the Indexes is probably reflective of a fairly general approach, and may be of help in visualising how they might be of use to you.  The first thing is to find at least one example of the song you're looking for - thus, a search in the TITLE1 field of the Folk Song Index for, say, 'Barley Straw'.  If this doesn't produce a result, try 'Barley Straw, The'

My database program (Dataease) is ageing and unsophisticated, but it does allow me to switch instantly from 'Record View' to 'Table View', so I now switch to Table and have a look at how many examples of the song I've found, and to look at their Roud numbers.  I do this because, as here, there is sometimes more than one song using the same title.  A quick look at the FIRST LINE and SUBJECTS fields convinces me that Roud No 118 is the song I'm after, so I then do a second search, removing 'Barley Straw' from the TITLE1 field and substituting '118' in the ROUD NO field.

This gives me 134 instances of the song, and I can now start looking for the information I want - which I generally do by sorting the records on a particular field.  A sort on TITLE1 shows me that The Jolly Beggar is the most common title; that Harry Cox's Barley Straw is quite unusual; and throws up Davie Faa with its interesting Gypsy connections.  A sort on the PLACE field confirms my supposition that it's a Scottish song (some 65 instances), but also shows me that it was very well-known in England (with 45).  This field can be rather misleading, in that you can easily assume that a song was particularly popular in a certain area of the country, when the truth is merely that it (like Somerset) was heavily trawled by a number of assiduous collectors.  Similarly, a concentration of instances from Norfolk often turn out to be all from Walter Pardon!

Sorting on TYPE reveals that although the majority of sourses are textual (books, manuscripts, journals, etc), there are in fact some 44 sound recordings ranging from the BBC Harry Cox of 1947 through to Doc Rowe's recording of Lizzie Higgins in 1988.  So I know that the song has remained popular among singers until quite recently, and can guess that some older Travellers may still be singing it today.  I also notice that very few broadside entries are present, so I switch to the Broadside Index to check.  Here it's very easy to see that a number of the 'book' entries in the other Index are in fact broadsides - but published in books.

Searching on the DATE field is rather disappointing, principally because most of the earliest sightings of a song are undated.  All you generally learn is what you already knew - that most songs were collected in the first half of the 20th century and that most sound recordings were made in the 1960s and '70s.  In this example, though, a PREVIOUS SOURCE field reveals that The Gaberlunzie Man appeared in Ramsay's Tea-Table Miscellany of 1750.

The SOURCE field is probably the most useful, and I find that - should I want to compare some versions from sources in my possesion - I can find examples in Purslow's Foggy Dew, MacColl & Seeger's Travellers' Songs, Sam Henry's Songs of the People, and no fewer than eight CDs and LPs, including our own Smith Family's Band of Gold (MTCD307).

I hope this brief overview has given some indication of just how useful the Indexes can be.  Information of the sort I've outlined would have taken hours to assemble from books and record sleeves, and even then, would have only been gleaned from the comparitively few sources in one's own collection.  Here, more or less everything is accessible.

However, Steve can only include the information he knows about, and any user of the Indexes is sure to come upon a gap in the record sooner or later.  When this happens - instead of grumbling about it - contact him and give him the information ... then it will be available to all users in the next update!

A subscription to both Indexes costs £50 - money well-spent, in my opinion.  Contact Steve at the address at the head of this piece.  For those who don't want, or can't afford, to buy them, copies are available at several institutions including The Vaughan Williams Memorial Library, London; Taisce Ceoil Dúchais Èireann, Dublin; and the School of Scottish Studies, Edinburgh.  What follows is a selection of extracts from the Indexes' User Manual which I hope will answer most of the questions you may have at this point.

Rod Stradling - 5.1.02

Roud Numbers:

The numbering system was devised to assist in identifying versions of songs, even when they are hidden under variant titles.  All versions of a particular song are given the same number.  Thus, by identifying and searching on the relevant Roud number, the user can find all the indexed versions.

The identification of 'different versions of the same song' is intellectually problematic, but necessary in practical indexing terms and will be adequate for the majority of the Index's users.  With important exceptions, the Anglo-American folksong corpus is remarkably stable and thus largely amenable to this treatment, but the identification of similarity between 'versions' is not an exact science and is necessarily subjective.  It is certainly the case that further refinement in numbering will be necessary, but it is equally certain that this will only be possible once the rough groundwork has been done.  Any future sub-division of Roud Numbers will be by means of decimal numbers rather than letters.

It must be noted that the system is not nearly so effective with material from other traditions, most notably Afro-American genres such as spirituals and blues.  These are, in general, far more textually 'fluid', and although numbers are assigned to entered songs from these traditions, the user should be aware of the limitations of this approach.

For songs to be given the same number, a reasonable degree of textual similarity has to be present, in addition to any similarity of plot.  In isolated instances, the form of the song is allowed weight, for example the 'Alphabet' songs of Lumberman, Sailor, Soldier, etc. in which the format is "A is for the…., B is for the…".

In cases where a particular rendition appears to be made of identifiable pieces of more than one numbered songs, two or more Roud Numbers can be assigned.

000 numbers: Where songs are entered on the basis of the title only (such as in a repertoire list or collection inventory), these entries are assigned the temporary number 000.  If further information comes to light, this number can be changed.  Editors who plan to provide Roud Numbers in their publications should never quote 000, but should always contact Steve for a new number to be assigned.

Changed numbers: Where mistakes have been made by the compiler, or further information becomes available, numbers assigned to particular versions may need to be changed.  Periodic updates to the Index will reflect these changes.  Where numbers are completely replaced by others, the superseded number will not be used again for another song.

The Basic Model:

It is important to note that the basic model used for material entered in the Folk Song Index is that of:
  1. A singer who knows the song
  2. A collector who notes the song and thus produces a manuscript, field-recording, etc.
  3. An editor who publishes the song in some way; as book, journal article, issued recording, etc.
This model is necessary on the theoretical level, but in practice it is often problematic.  The roles can be played by the same person, and editors do not always make it clear exactly how a song reached them, or who played what part in the sequence.  There can be a chain of people involved as 'collector', and fine distinctions sometimes have to be made on slender or ambiguous evidence: a woman who 'notes' a song from her grandmother is counted as a 'collector', but if she 'learnt' the song from her grandmother, she becomes the 'singer' herself.  Commercial recordings, in particular, do not fit this pattern very closely.  In addition, the model implies a face-to-face interaction between collector and singer, but in many cases singers have sent in manuscripts and letters to a collector, and in others it is the singer's own manuscript 'songbook' which is the source of the collected song.

The Folk Song Index focuses firmly on the songs themselves.  Singers' names (where known) are included, but no other details such as their age or occupation.  Place and date of recording are included, but no information on where or how the recordings were made, and so on.

The Multiple Entry Principle:

Newcomers to the Indexes sometimes expect one entry per 'song' – i.e. one computer record which includes all the known versions.  As there are often dozens, or even hundreds, of versions of particular songs, this approach is totally impractical, and the structure of the Indexes is thus based on a multiple-entry system, as follows.  The basic unit of entry in the database is the performance of a song.  A singer sings a song (or writes it down), and this is a unique performance.  S/he may sing the song again, but this is a different performance.  Obviously, if a performance is not noted down or recorded, it cannot be indexed beyond the level of 'X sang a song called Y on date Z' and there are indeed some entries in the Index on these lines.  Most entries, however, are to song performances which have been noted or recorded in some way, and these notations and recordings can be termed representations of the performance.  This opens up the possibility of multiple representations of a single performance.  A tape-recording, for example, can be duplicated; it can be issued on a range of LPs, CDs, and so on; it can be transcribed, and this written transcription can be duplicated in any number of journals, books, and so on.  A second transcription can be made by another person, which differs from the first.  The added complication is that each representation is subject to possible deliberate editorial change – verses omitted, words replaced, verses re-arranged.

For singers who were documented by many fieldworkers over a number of years, it is often difficult to be sure whether separate representations refer to different performances, and this problem will only be solved (and then not with complete confidence) once all the representations have been identified, located, and indexed.  A major planned development of the Indexes will provide a master catalogue, which will provide space for an overview of each song, but this is still under development.

Until that time, the principle on which the Index is structured is of multiple entries for multiple representations.  In other words, if a song is published in three books – even if we know that each is a representation of the same performance - it is entered three times.

A Note on Tunes:

The Indexes are almost exclusively concerned with texts, as music is not amenable to this type of indexing.  Tunes are, however, mentioned in three different ways:
  1. The presence/absence of a written tune is noted in the CONTENTS field.
  2. Named Tunes (as for broadsides which say 'To the Tune of…') are entered in the NAMED TUNE field.
  3. For research into early (i.e. 17th and 18th century) songs, an important body of information exists in song titles used for tunes in Ballad Operas and other musical-theatre productions and in dance manuals such as Playford's English Dancing Master.  These sources have therefore been included in the remit of the Indexes, and the tune/song names are entered in both the NAMED TUNE field and the TITLE1 field, distinguished in the latter field by the words [TUNE ENTRY].

Spelling:

Variant spelling is always problematic for textual databases which have such literal minds, but is particularly difficult in an area of vernacular usage such as traditional song.  There is no internationally-accepted standard for spelling in the English language.  In particular, British and American usage differs, and even within sources from one country there is variation, while spelling has also changed considerably over time.  Even more problematic are the deliberate renderings of dialect or non-standard pronunciation in spelling.  Editors' motives in these cases vary.  Many believe that it is an important service to the community whose songs are being recorded that their language is represented.  Speakers of Scots, for example, would understandably object if Auld Lang Syne was entered as Old Long Since.  However, in other cases, such as Music Hall 'country bumpkin' songs, or Blackface Minstrelsy, the rendering of 'dialect', while taking a similar form to that of the caring editor, is hardly done from such altruistic motives, and, indeed, can be seen as racist.  In many cases, editors' attempts to replicate spoken dialect are idiosyncratic and inconsistent even within the same song text.  Black traditions seem particularly victim to this type of handling, and many printed titles do not contain a single standard spelling: Foller de drinkin' gou'd is a relatively mild example.

The answer would seem to be to standardise and be damned, but unfortunately this is not as simple as it seems.  Indeed, those who argue in favour (favor) of standardisation (standardization) always assume that it is their own standard which will be used. 

There is also a long-standing principle in cataloguing that catalogue entries should faithfully record the details of the material, even if these are known to be wrong.  Thus, a variant spelling on the title-page of an early book should be reflected in the catalogue entry, and even a patent spelling-mistake should be retained.  The reason is not difficult to ascertain.  Not only are particular editions or even copies of early works often only distinguishable by such differences, but the expert user must always be able to find an item using its actual details and not need to guess what the cataloguer has made of them.  The apparent 'mistake' may indeed have been deliberate anyway.  Many 19th century books are deliberately entitled Scotish Songs, for example.

The adopted procedure has thus been, somewhat reluctantly, to leave the spelling as it is in the source in most cases, and users are hereby warned to use their intelligence and the truncation facility built into most databases to find what they want.

Two-Way Traffic

Most users will be focussed on the desired end-product of their search, which will normally be a particular source (book, recording, manuscript, etc.) which contains the information they seek.  They will thus regard such sources as the end-result of their journey.  However, another user may already possess that source, and be using it as the starting point of their search ("that's an interesting song, I wonder if there are other versions…").  Any one source can thus be the start or the desired end, and this has ramifications not only for the way the database is structured, but also, crucially, what sources get included.

Levels of Use:

A database designer cannot entirely predict how his/her creation will be used, but at least a broad idea of users' needs and characteristics is necessary at the design stage.  In the Folk Song field, a combined matrix of users' needs and attributes (including ability, level of interest, and so on), can be plotted on a continuum running from simple to complex:

Simple     >     Complex
Amateur     >     Professional/Academic
Passing interest     >     Enduring interest
Minimal access to resources     >     Maximum access to resources
Specific item needed     >     Comparative material needed

Perhaps the simplest enquiry is someone who wishes to find a text of a song, the title of which they already know, without setting conditions about the song's provenance or other details of its history.  This user will be happy to find one accessible source – probably a 'popular' one.  At the other extreme will be a researcher who requires the widest possible range of texts of a song, to gain a historic-geographic overview before narrowing the search by inclusion of relevant detailsof provenance, or other criteria.  The latter will wish to be informed of all known examples of a song, however physically difficult to access, but will him/herself normally have well-developed networks (such as national or university library collections) on which to draw.

Most users will be found at points between these two extremes, and will, in fact, move about on the scale, depending on their current research needs and interests.

The Indexes seek to cater for all these different types of users and needs, but this very universality can become a barrier to simple use.  A basic search for some songs can bring up hundreds of 'hits', and, as users of the Internet will know, too many hits is almost as bad as too few.

Suggested Formats:

The more user-friendly database packages allow the user to design different formats (sometimes called lay-outs or other similar names) in which selected fields can be presented in different ways on the screen, and as print-outs.  The key advantage of a shorter format is that several records can appear on the screen at the same time, and key data is presented more clearly when stripped of less important fields.

List of Fields:

TITLE1
TITLE2
NAMED TUNE
FIRST LINE
TYPE
SOURCE
PREVIOUS SOURCE
PERFORMER
PLACE COLLECTED
DATE COLLECTED
COLLECTOR
CONTENT
DATE ADDED
ROUD NUMBER
OTHER NUMBERS
SUBJECTS
PRINTER/PUBLISHER
COPYLOCATION
NOTES1
NOTES2

Top of page Home Page Articles Reviews News Editorial Map

Site designed and maintained by Musical Traditions Web Services   Updated: 5.1.02