-
Notifications
You must be signed in to change notification settings - Fork 0
Bitterroot Classification System (BCS)
The Bitterroot Classification System (BCS) is a library classification system whose main function is to categorize research material in a very specific way, using relatively few characters. It originates from the work of a linguist and a programmer (as most good things do). It's named after the bitterroot, mostly in reference to where it was invented (Montana).
Bitterroot is very good at defining hyper-specific research in a relatively low amount of characters. Consider this example:
LN/TU 838.2.V7 H318
This call number references the paper "Vowel Harmony and Disharmony in Tuvan and Tofa" by K. David Harrison, a linguistics conference paper from 1999. There are a few components of this number, and we'll go through each part in detail, as well as detailing what features this identifier features in comparison to the Dewey and Library of Congress approaches.
The first two letters of the call number is the domain which signifies the general research area of the work - in this case we have LN, which is for works of Linguistics or Language. Domains can also have subdomains, particularly when the nature of the domain varies greatly depending on geographic area or changes in a gradient fashion. You can think of the domain and subdomain as the "root" of the call number (haha). Subdomains are commonly featured in fields like Anthropology (AT), Literature (LT), Language/Linguistics (LN), and Music (MU). In our example, /TU denotes "Turkic", thus works prefixed with LN/TU will highlight Turkic Languages.
Although 5 characters may not seem optimal for describing this, let's look at alternate approaches to describing this relatively broad category.
DDC: 494.3...
In Dewey, this is also done in, at minimum, 5 characters. However, it's of note that not all language groups are represented equally or fairly under Dewey. Although all Dewey numbers beginning in 4 are language, certain blocks of 400 are reserved entirely for a single language, typically a language of Europe. Thus, French requires only 2 digits of specificity (44...), but Chinese, a language with 10 times as many speakers as French, requires 5 (495.1...). Not to mention that Dewey separates both languages from their larger language families (languages related to Chinese exist at 494.4... and 494.8..., separated by other languages such as Japanese 494.5...)
It's also worth mentioning that languages with small speaker counts tend to require a much larger amount of specificity in order to be reached within Dewey, and large families outside of Europe will also tend to require more than 5 characters worth of specificity in order to be described. Some of these include Siouan languages (497.52...) Kwa Languages (496.337...) Kartvelian Languages (499.968...).
Overall, this approach is disorganized and doesn't reflect the research done on these languages nor does it give fair representation of language families.
LCC: PL21-PL396
In Library of Congress Classification, this is done in 4-5 characters, depending on the language. "P" denotes language, and "PL" denotes languages of Asia, Africa, and Oceania. This does mean that LCC suffers a similar specificity issue to Dewey, but this is not as consequential for length, as most LCC call numbers tend not to exceed 6 characters. These numbers are typically grouped together as well, eliminating that issue which Dewey carried. However, other language families take up a large amount of PL as well, and one must distinguish works of Turkic languages from these other families, such as Japanese (PL501-PL889), Chinese (PL1001-PL3208) and Dravidian (PL4601-4797)
This is overall not a bad approach to representing these languages, although the section of PL needs to be memorized in order for Turkic languages to be indentifiable.
BRC: LN/TU
Most language families in this solution can be represented in 5 characters. Siouan languages can be represented as LN/SX, Kwa languages as LN/KW, Kartvelian Languages as LN/KT. Very little memorization is needed here, and works are organized directly by language family.
The first whole number after the domain or subdomain is a subject code, which can identify a certain subject from a domain or subdomain. It's very possible that a call number in BRC can be made up of solely a domain and a subject code, though it's likely that this will be at the cost of specificity. We'll explore the two cases of subject codes in BRC, those with subdomains, and those without.
A subject code with a subdomain, like the one in our example, is usually used to derive a single object from a subdomain. This usually means a single language or variety of something within the subdomain, or at other times a single topic. To refer to our example:
LN/TU 838
Within LN/TU, 838 is the identifier for the Tuvan language. This is not arbitrary, it follows separations within the subdomain, in this case, "Siberian Turkic" falls alphabetically around 83% through the English Language in terms of lexicon - therefore, "Siberian Turkic" commences at LN/TU 830. Each entry following this is entered alphabetically, as so:
830 Siberian Turkic.
831 - Chulym
832 - Dolgan
833 - Duha
834 - Fuyu Kyrgyz
Closely related varieties, what might be called "dialects", are denoted by adding subsequent digits to the whole number.
838 - Tuvan
8381 - Kyzyl Tuvan
8382 - Tere-Xol Tuvan
8383 - Todzhu Tuvan
839 - Western Yugur
840 - Xakas
In larger and more studied language families like Indo-European, the numbers can become quite long.
450 - Germanic
4501 - West Germanic
45011 - English
450111 - American English
...
45012 High German
This can begin to fail the goal of using small amounts of characters, but prioritizes varieties which haven't been studied yet. It also unfortunately has the bias of preferring smaller language families. This is a fallback of BRC, and is a point of address for the future.
Also of note is that these codes are organized decimally, meaning that there is an implied decimal point before they commence. Although in ordinal fashion, 91 would come before 187 or 5237, but in this case, the order would be 187, 5237 and then 91.
A subject code without a subdomain acts a little differently, and acts more like an identifier for a field of research rather than of an individual element of a gradient. For example:
LN 507 O99
This code references the Thesis "Case, Referentiality and Phrase Structure" by Balkız Öztürk, submitted to Harvard University in 2004. The domain, LN, once again references Language/Linguistics, but with the abscence of a subdomain, "507" refers to a specific subject within linguistics. The "5" at the beginning denotes a field of research within Linguistics. Here is a sample table.
LN 1xx - General Studies / Theory
LN 2xx - Phonetics / Phonology
LN 3xx - Morphology
LN 4xx - Syntax
LN 5xx - Semantics
LN 501 - General Semantics
LN 502 - Lexical Semantics
LN 503 - Compositional Semantics
...
LN 507 - Referentiality
These numbers are in fact ordinal by nature.
LN 6xx - Pragmatics
LN 7xx - Sociolinguistics
LN 8xx - Historical Linguistics
LN 9xx - Documentation / Revitalization
LN 10xx - Lexicography
Aspect Codes are meant to complement subject codes in call numbers with subdomains. Aspect codes contain information concerning the domain, and are based off of the numberings of subject codes within the main domain. To return to our example:
LN/TU 838.2
Our aspect code here is "2". If we recall the numbering of aspect codes within LN...
LN 1xx - General Studies / Theory
LN 2xx - Phonetics / Phonology
LN 3xx - Morphology
The 2 here, in this case, corresponds to Phonetics and Phonology. This helps both to shorten call numbers with subdomains as well as to create a more generative system of expanding knowledge. It becomes much easier to identify gaps in literature and to analyze patterns of knowledge accrual within a field to feature this aspect code. Here are some more examples of aspect codes within LN.
LN/TU 636.4 - Turkish Syntax
LN/KT 600.7 - Sociolinguistics of Laz
LN/AL 543.4 - Syntax of Blackfoot
LN/IE 45011.10 - English Lexicography
Topic codes are the most specific portion of the call number. These are alphanumeric codes which specify the topic within the field which is being discussed.
LN/TU 838.2.V7 - Vowel Harmony in Tuvan
These codes are not like aspect codes in that their generation is based less on information directly based on the domain, and more based upon subdomain, subject, or aspect-specific topics. They are bound to whatever portion of the call number precedes it.
LN/TU 838.2.I6 - Intonation in Tuvan
LN/TU 838.2.C6 - Consonant Assimilation in Tuvan
LN/IE 45011.9.V7 - Historical Vowel Shifts in English
Other additions can be added to help categorize these documents, including author information, year of publication, first letter of titles, etc. Many use cases will be explored within this guide. However, the essential portions of these call numbers are the ones mentioned in this article.
To summarize, a call number at its base can be a domain and a subject code. However, the most vital identification information can be stored in the domain, subdomain, subject code, aspect code, and topic code. We will explore more about each one in the following chapters.