Zompist Bboard Again

since the original corpus includes not just the mainland united states but also alaska, straddling canada, i was curious to see how incorporating the analogous data from canada would help the project. i included the names of provinces/territories and their capitals, the largest cities, and the longest rivers (for the latter two lists, i set the cutoff point at around the same population/river length as the smallest item in the corresponding u.s. data, meaning the city of london and the winnipeg river respectively). i am also assuming throughout that we only have phonetic data to go on, not spellings

even with the new canadian data, we only have two examples of /wɛst/ (west virginia, northwest territories), which isn't enough to reconstruct its meaning. it's tempting to construct /nu/ as "east" given the preponderance of the prefix in the eastern part of the continent (new brunswick/hampshire/jersey/orleans/york/york city, newfoundland and labrador, nunavut), but this is sketchy given a) there are no corresponding nu-less forms such as *brunswick, *west hampshire, etc., and b) new mexico is in the western part of the continent. (there is still not a single instance of the word "east" in the entire expanded corpus.) on the other hand, the meanings of /norθ/ and /sauθ/ are confirmed by the saskatchewan/north saskatchewan/south saskatchewan rivers, all three of which are in the canadian corpus

it's still unclear exactly what /lənd/ means, as it appears in the names of cities, states/provinces, and rivers; on coasts as well as inland; in the north, south, east, and west; and in lowlands as well as mountainous areas. however, the apparent compound form /ailənd/ seems to be a term for a coastal state or province (rhode island, prince edward island)

if we ignore the generic term "river", almost all features in the entire corpus consist of one or two words. one of the only exceptions is "newfoundland and labrador", with two longer names (one of which bears already postulated roots /nu/ and /lənd/) separated by the monosyllabic /ænd/. could this word be a locative preposition indicating "on", "at", "upon", or something similar?

other potential word elements:

• /-æskə/ (alaska, nebraska; athabasca river) is a strikingly distinct suffix, but it's difficult to figure out what its usage or meaning would be. the great size of alaska and the length of the athabasca suggest an augmentative, but the relatively average size of nebraska casts doubt on that theory
• /-hɪl/ is only found in names of rivers (smoky hill river in the US, and two canadian rivers named churchill) and may be a suffix meaning "river", "water", etc.
• /-vɪl/ is found in three city names (jacksonville, nashville, and louisville), all in the southeastern u.s. and all located a short distance away from their respective states' border with a neighboring state. perhaps these were once defensive outposts in a pre-unification period and /vɪl/ is a regional word for "fort"
• /-əs/, meaning unclear, but almost all instances are found west of the mississippi river (las vegas, dallas, los angeles, kansas; the rivers pecos and brazos; only columbus is in the east); possible eastern cognate is /-ɪs/, found in memphis and in the compound /-æpəlɪs/ (annapolis, indianapolis, minneapolis)
• could the /-pəl-/ in "minneapolis" be related to the /pɔl/ in neighboring "st. paul"?
• the city name /ʃarlət/, located some 450 km inland in the southern US, is repeated in the city /ʃarləttaun/, which is an island harbor in the far northeastern part of canada. as noted above, the latter name appends an element that is probably cognate with the /-tən/ suffix, but the connection between the two settlements is a mystery

Ares Land wrote: ↑Wed Oct 21, 2020 9:09 am The first thing we pick up is 'River'. 'Rio Grande' is a weird outlier that is best ignored for now.

the apparent incongruity of rio grande may be explained by "la grande river", which appears in the canadian data. the word /grænd~grand/ is probably a synonym for "river". given how uncommon it is in the data, it's possible that the word fell out of common use in the early period of the language; once lost, the word ceased to be applied to newly discovered rivers, and it's even possible that features that previously used this element were renamed, a process that may have been partially completed with the name "la grande river" itself

Ares Land wrote: ↑Wed Oct 21, 2020 9:09 am Of course, we see the suffix -on, as found in Yukon, Oregon, Washington, Cimarron, Trenton, Carson City, Jefferson City, Jackson, Boston, Baton Rouge, Houston, Arlington
It seems to have three allomorphs: -ton after s or a nasal, -son after s, r (possibly assimilated in Cimarron).

if we look at the vowel, we can break this into at least two different suffixes: one with a clear vowel and one with a schwa.

• /ɑn~ɔn/ would seem to be a suffix for a river (cimarron, saskatchewan, thelon, tucson, yukon); tuscon is the name of a city, but that city is located on a river that may well have once been called tucson itself
• /ən/ doesn't have as obvious a meaning, appearing in the names of two rivers (canadian, nelson) and two states (michigan, oregon, both with /g/), as well as several cities (the ones you list plus london, lincoln, madison, and probably jacksonville); it's possible that wisconsin (/-ɪn/) and/or assiniboine (/-oin/) are related
• the allomorphic variation suggested for /ən/ doesn't seem to be borne out by the corpus, but we can distinguish a clear suffix /tən/, which seems to only be used in names of cities (except for washington, which is the name of a city in the east and a state in the west, the latter probably named for the former). cities added from the canadian corpus are fredericton, hamilton, edmonton, and brampton; charlottetown and yellowstone may also be related, although the latter is a river

2+3 Clusivity wrote: ↑Fri Oct 23, 2020 7:56 pm /*monT/ > Rich-mond, Ver-mont, mont-pelier, mont-ana, mont-gomery .... probably researchers can guess something with hills/mountains with the exception of Montgomery. Richmond seems to be a bit of an outlier but it's one of the fewer items in final place.

GreenBowtie wrote: ↑Wed Dec 23, 2020 1:56 pm

dhok wrote: ↑Sun Oct 25, 2020 11:37 am Manhattan looks like it has the -tən morpheme. Probably the Man- would be taken as cognate to the mi- of Minnesota and Missouri--maybe Massachusetts as it's much closer. (All of these connections are incorrect--Manhattan is from PA *menahanwi.
...
-chester is likely to be recognizable: Manchester (NH) and two Rochesters (NY and MN) possess it. Not so clear is whether it would be connected to Lancaster (CA) and the particularly impenetrable [wʊstə(r)], Massachusetts.

it's tempting to link Man- in Manhattan and Manchester to the reconstructed mont, though the loss of -t in the former is harder to justify than in the latter. Sacramento might reflect this root as well

looking closer at the corpus, neither manhattan nor manchester are present. but with the expanded canadian corpus we can add montreal, edmonton, and possibly manitoba to the mont list

2+3 Clusivity wrote: ↑Fri Oct 23, 2020 7:56 pm Building on the form pointed out above, I'd note a broader /*məS ~miS ~ wiS/ mar-yland, mas-sachusetts, mich-igan, mis-sissippi, mis-souri, wis-consin, ?min-nesota, ?win-nipeg ??wyo-min-g. ... Maybe something to do with coastland, rivers, deltas. Interestingly, this form itself seems to be followed by /*-s(V)-/ or /*-ni- ~ -ne-/ frequently.

mississauga is another clear example of this morpheme

Rounin Ryuuji wrote: ↑Wed Dec 23, 2020 10:29 pm On which note, some such prefix or word "car-~cal-" (appearing in Carolina, California) seems to have had something to do with temperate coastal regions, probably having originally been something like */karʲɵ/, differentiating into */kaʎɵ~kaʎɪ/ > */kalɪ/ in the West, but */karɵ~karo~kero/ (there is evidence of an /e/ pronunciation) in the East. Perhaps it referred to coniferous forests (the redwoods of California and the pine forests of Carolina being thought of as containing the same sorts of trees, or variants of them, with the name being applied to various similar species along the way).

colorado is almost certainly built on this root, as is calgary in the canadian corpus. could this root be related to the columb- morpheme identified by Ares Land?

u.s. corpus canadian corpus (not including rivers already in the u.s. corpus)

I like adding the Canadian data. One set of names that struck me: Nevada, Nunavut, Nova Scotia, maybe Niobrara. In line with previous suggestions, I assume the common element means "river".

(I'm pretty sure looking at Mexican states would make it clear that another language was involved.)