Spurious languages are languages that have been reported as existing in reputable works, while other research has reported that the language in question did not exist. Some spurious languages have been proven to not exist. Others have very little evidence supporting their existence, and have been dismissed in later scholarship. Others still are of uncertain existence due to limited research.
Below is a sampling of languages that have been claimed to exist in reputable sources but have subsequently been disproved or challenged. In some cases a purported language is tracked down and turns out to be another, known language. This is common when language varieties are named after places or ethnicities.
Some alleged languages turn out to be hoaxes, such as the Kukurá language of Brazil or the Taensa language of Louisiana. Others are honest errors that persist in the literature despite being corrected by the original authors; an example of this is Hongote, the name given in 1892 to two Colonial word lists, one of Tlingit and one of a Salishan language, that were mistakenly listed as Patagonian. The error was corrected three times that year, but nonetheless "Hongote" was still listed as a Patagonian language a century later in Greenberg (1987).
In the case of New Guinea, one of the most linguistically diverse areas on Earth, some spurious languages are simply the names of language surveys that the data was published under. Examples are Mapi, Kia, Upper Digul, Upper Kaeme, listed as Indo-Pacific languages in Ruhlen 1987; these are actually rivers that gave their names to language surveys in the Greater Awyu languages and Ok languages of New Guinea.
Following is a list of ISO 639-3 language codes which have been retired since the standard was established in 2006, arranged by the year in which the change request was submitted; in most cases the actual retirement took effect in the beginning of the following year. Also included is a partial list of languages (with their SIL codes) that appeared at one time in Ethnologue but were removed prior to 2006, arranged by the first ion in which they did not appear.
The list includes codes that have been retired from ISO 639-3 or languages removed from Ethnologue because the language apparently does not exist and cannot be identified with an existing language. The list does not include instances where the "language" turns out to be a spelling variant of another language or the name of a village where an already known language is spoken; these are cases of duplicates, which are resolved in ISO 639-3 by a code merger. It does include "languages" for which there is no evidence or which cannot be found. (In some cases, however, the evidence for nonexistence is a survey among the current population of the area, which would not identify extinct languages such as Ware below.)
SIL codes are upper case; ISO codes are lower case. Once retired, ISO 639-3 codes are not reused. SIL codes that were retired prior to 2006 may have been re-used or may have reappeared as ISO codes for other languages.
Removed from Ethnologue, 12th ed., 1992
Itaem (PNG) [ITM]
Marajona (Brazil) [MPQ]
Nemeyam (PNG) [NMY]
Nereyama, Nereyó (Brazil) [NRY]
Numbiaí (Orelha de Pau) [NUH]
Oganibi (PNG) [OGA]
Tijuana Sign Language (Mexico) [TJS] - added to Ethnologue 1988 by mistake due to a misunderstanding, removed in 1992. No evidence that it ever existed.
Tyeliri Senoufo [TYE] – the Tyeliri are a caste of leather workers, and do not have their own language
Zanofil - name of an ethnic group that speaks Yongkom [yon]
Removed from Ethnologue, 13th ed., 1996
Bibasa (PNG) [BHE] – described as "isolate in need of survey" in the 12th ed.
Removed from Ethnologue, 14th ed., 2000
Alak 2 [ALQ] – a mislabeled fragment of a word list
Dzorgai [DZI], Kortse [KBG], Pingfang [PFG], Thochu [TCJ], Lofuchai (Lophuchai) [LFU], Wagsod [WGS] – old names for Qiangic languages, some of uncertain correspondence to currently recognized names