In their study of the Drexel Digital Museum project, Olson & Javins (n.d.) find no adequate thesaurus for use in cataloging the costume collection and recommend that one be written especially for it. Goodrum & Martin (1999), however, suggest several authoritative sources for representing items in the collection, and I have relied on them in compiling the following list:
For terms describing garments and their construction:
- Art and architecture thesaurus (Getty Vocabulary Program, 2000a)
- Dictionary of costume (Wilcox, 1969)
- Oxford English dictionary (1989)
For terms describing fabrics and textiles:
- Art and architecture thesaurus (Getty Vocabulary Program, 2000a)
- The encyclopedia of textiles (Jerde, 1992)
- Fabric glossary (Humphries, 2000a)
- Fabric reference (Humphries, 2000b)
- Oxford English dictionary (1989)
None of these sources is adequate on its own, but each has much to offer. The Art and architecture thesaurus (AAT), though flawed, has become a standard work for content representation in museums, and I have relied heavily on AAT in my work. Though the ICOM vocabulary (ICOM International Committee for the Museums and Collections of Costume, 1982) is the standard for costume collection in the Louvre and the Victoria & Albert Museum (K. Martin, personal communication, May 19, 2002), I recommend against using it unmodified for reasons listed below.
I have not had access to two sources mentioned by Goodrum & Martin:
- Longhouse Reserve classification
- Women's wear daily suppliers guide
Some kind of standardized vocabulary is necessary to ensure the value of data. A museum database full of inconsistencies is confusing to maintain, hard to search, and difficult to export to other databases. While it is possible for each museum to standardize the terms used in its own database, the International Committee for Documentation of the International Council of Museums recommends that all museums work together to adopt universal standards (ICOM-CIDOC, 2001).
There are several good reasons for relying on AAT as a source for terms describing a museum collection. It has already been created--by a wealthy institution--and has been in use for several years. It is large, having over 125,000 terms. It frees staff from the extremely time-consuming, specialized work of creating a thesaurus for every subcollection and from the ongoing maintenance such thesauri require. Birdsey (2000) points out that large-scale projects can never really be finished. A successful museum project would be "part of a jigsaw where the face of each piece differed but the physical structure was constant, thus enabling any piece to fit together regardless of when it was produced and in what cataloguing software." Because almost all repositories will have a mix of records from various time periods, stored on paper and in various digital formats, created over time by different staff with varying levels of training, a flexible but consistent scheme is necessary. Relying on an outside standard such as AAT encourages long-term consistency, allows staff to concentrate on other matters, and also increases the likelihood of being able to share records with other repositories.
Use of AAT should ease the interchange of data between institutions. Besser & Yamashita (summarized in Besser, 1999) suggest that standardization is crucial to the long-term success of digital projects. In their study of digital image distribution networks, they find that different repositories are using different vocabulary and different metadata schemes and this makes creation of an integrated, multi-institutional database very difficult.
There are many advantages to an integrated database. A consortium of repositories can pool resources, needing only one set of programmers, one set of staff to maintain vocabulary and authority files, one set of copyright experts. More importantly, each repository gains access to the others' digital images. Besser (1999) points out that a small image set has very little pedagogical value. Teachers need a "critical mass" of images from which to select--he suggests about 250,000, the equivalent of a moderate-sized slide library--and further, teachers expect the focus of the collection to change constantly to match their research and teaching needs.
Since it is a work in progress, I cannot be too harsh with my predecessors' classification scheme, entitled CATEGORY (available as a separate file). It has a confusing layout, and while it does have a good array of available terms, they are not well standardized. For instance, the word cardigan appears twice; I find this confusing till I realize that one appearance is in the dress hierarchy and the other is in the sweater hierarchy--cardigan sweater and cardigan dress.
The main problem with CATEGORY, however, is that it is based on the ICOM vocabulary (ICOM International Committee for the Museums and Collections of Costume, 1982), which does not comply with international thesaurus standards. ANSI/NISO standard Z39.19-1993 (National Information Standards Organization, 1993) sets forth useful guidelines for creating useful thesauri. The guidelines support of ease of use and compatibility between computer systems. Among the problems with ICOM:
- creates its own terms rather than following general usage; inverts terms rather than presenting in natural order (COMBINATIONS Camiknickers; OVERSHOES Galoshes Snowboots; UNDER WAISTCOAT Waistcoat borders)
- uses multiple terms for the same, or almost the same, item (a women's T-shirt is a BODICE Pullover (Tee shirt), a men's T-shirt is a PULLOVER Tee shirt)
- numbers some terms but not others; provides scope notes or French and German equivalents for some but not others; in general has a confusing layout
- assigns duplicate numbers to terms (term 1.21 in Women's garments is BODICE; in Men's garments, COAT)
- provides line-drawing illustrations for many terms, but does not provide labels for all, making it difficult to match an illustration with a term
- uses a mix of plural and singular forms (CASE Spectacle; SUPPORTS Busks; TRIMMINGS Bead)
While the classifications in the ICOM vocabulary can be very detailed, I suspect that their current structure will make thesaurus-building more difficult. An example: the term DRAWERS Knickers Directoire, with parent term Below waist with shaping for legs; with ICOM's classificatory system, a panties researcher interested in comparing the museum's panties with other garments with legs would have to rely on keyword searches or previous knowledge of the classification scheme to retrieve all garments with shaping for legs.
With a faceted classification and a hyperlinked thesaurus, a panties researcher could search the costume catalog and immediately see, for instance, all female costume or all garments with shaping for limbs. Furthermore, with the faceted system, the researcher could move up the hierarchy to underwear worn next to body and retrieve both male and female undergarments. The researcher could exclude men's underwear by combining the facets underwear worn next to the body and female costume. (See my sample thesaurus entry.)
A faceted scheme encourages a much richer thesaurus with more related terms (RTs), while a non-faceted scheme actually discourages connections between related terms, by separating them in the hierarchical display. For example in ICOM, Underwear for men is several pages away from Underwear for women. Likewise, each section of the scheme requires repetition of terms like above waist and below waist or shaped and unshaped, making an unwieldy display.
Though AAT generally complies with Z39.19-1993, it is not perfect. Many of its terms have no scope notes. Most of the terms have few or no RTs. Many terms are not included at all. For instance, AAT does not include the terms spandex or dashikis. And because AAT is so huge, it separates terms that in a smaller museum are closely related. For example, in AAT costume has its own hierarchy, but various types of fiber are scattered throughout the materials hierarchy.
AAT does lack one major strength of the ICOM vocabulary, which is the consideration of "garments in their relation to the human body, the one constant factor in all dress" (A. Buck, in ICOM International Committee for the Museums and Collections of Costume, 1982). While AAT does not completely obscure a garment's relation to the body, it could be stronger.
The biggest problem I find with AAT is that it occasionally attempts a faceted scheme, but fails. For example, in the costume hierarchy, AAT has created two facets: costume by form and costume by function. Logically, the name for a garment is its form (dress, shirt, skirt) and should be found under costume by form, while the garment's category of use (ceremonial costume, protective wear) should be found under costume by function. However, AAT places the names of some garments under costume by function and the uses of some under costume by form.
Because I based my work on AAT, I inherited its weaknesses. I have tried to mitigate those with my modifications. I have also tried to accommodate the terms found in CATEGORY.
I have some fear that my scheme will not be intelligible to researchers. I showed a draft of it to fiber expert David Stensland of A Craftsman's Touch Alpaca Co., and he replied, "The whole textiles section confused me." Regarding the costume by function facet, he said, "The proper terms are day wear, casual/sportswear, evening wear, formal wear, sleepwear, foundation" (personal communication, Apr. 12, 2002). I hope that careful use of synonyms (UFs) in the thesaurus, including ICOM vocabulary, will resolve these problems.
A separate file (entitled thesaurus.xls) contains my faceted scheme. The advantages of this scheme are its compatibility with AAT and its facets. As much as possible I have retained AAT's classificatory structure, but have improved it with faceting for qualities such as gender of wearer and age of wearer. I have also tried to carry over ICOM's emphasis on the garment's relationship to the body. Deviations from AAT are marked in the notes column.
next page: metadata
back to project home page
Donny
Smith
May 2002
[email protected] or [email protected]