Hopi and the Deseret Alphabet*

In the fall of 1859, Jacob Hamblin and six other men left their homes in what is now southwestern Utah and traveled to the Hopi villages on the southern tips of Black Mesa. The purpose of the trip was to establish good relations with the Hopis and learn their language with an eye to translating the Book of Mormon and bringing the Hopis into the fold of the Church of Jesus Christ of Latter-day Saints (the “Mormons”). Two of the men, Thales Haskell and Marion Jackson Shelton, ended up in Oraibi, a village on what is called the Third Mesa. They would spend a total of four months there, living and working with the Hopis.

English Phonotypic Alphabet 1847

We know about their time in Oraibi because Thales Haskell kept a journal. It’s pretty interesting reading, provided you can read the Deseret Alphabet.[1] The Deseret Alphabet was all the rage at the time. Or at least spelling reform was. Sir Isaac Pitman (of shorthand fame) developed several versions of a “Phonetic Alphabet” with the goal of reforming English spelling. The Pitman orthographies were based on the traditional Roman alphabet, with modified characters added to fill out the range of symbols needed to adequately represent English sound structure.

The Deseret Alphabet

Brigham Young and other Church leaders thought that orthography reform would be a good idea for the Church as well. In fact, they were on the verge of adopting a Pitmanesque Phonetic Alphabet when Willard Richards called a halt to it, declaring it to be “new wine in old bottles.” His objection was that using traditional characters (however modified) would be confusing, and what was needed was a clean break; nothing less than a brand new set of symbols would do. George D. Watt took up the challenge and produced the Deseret Alphabet.


All of this is well known by historians who have an interest in the Deseret Alphabet. What is not as well known is that the Deseret Alphabet was also used to transcribe foreign languages for pedagogical use. Haskell’s companion, Marion Jackson Shelton, used the Deseret Alphabet to write out an English-Hopi vocabulary list during their time in Oraibi. There are 486 entries, and it shows some interesting things about the state of the Hopi language at the time.

For me the most interesting thing that Shelton’s Vocabulary shows us has to do with a fine phonetic distinction made in Third Mesa Hopi. Unlike the other dialects of Hopi, that of the Third Mesa has developed tone; it thus joins other world languages like Mandarin, Thai, and Hausa in using the pitch at which words are pronounced as important cues to their meaning. It’s been the source of some speculation as to how tone originated in Hopi. Using traditional linguistic reconstruction methods, Alexis Manaster-Ramer in a 1986 paper[3] posited the existence of a set of consonants that caused the pitch of the voice to fall on the vowel just before the consonant. One of these consonants is /h/. This /h/ was eventually dropped in Third Mesa speech, leaving behind only the falling pitch as evidence that it was ever there. At least, this is Manaster-Ramer’s story.[4]

Hopi girl (photo by Edward Curtis)

In Shelton’s Vocabulary, we clearly see many entries with the syllable final /h/ that have since disappeared.[5] Two entries in particular are especially important—the entries for ‘little boy’ and ‘little girl’. In modern Third Mesa Hopi, these words are tiyòoya and manàwya, respectively (the grave accents show the falling tone). In Second Mesa Hopi, they are tiyohoya and manahoya. From these examples and others like them, we can reconstruct a diminutive suffix *-hwya for a pre-modern version of Hopi, which through various sound changes gives us the range of variation we see in the modern language (-hoya ~ -wya ~ -ya).

In the Vocabulary, ‘little boy’ and ‘little girl’ are given as <ti.o.hwi.yʌ> and <mɑ.nɑ.hwi.yʌ>. The diminutive ending is clearly <-hwiyʌ>, which is as good a match as you can hope for to the hypothetical *-hwya. And that’s what makes my toes tingle. The hypothetical reconstructed form isn’t hypothetical anymore: it’s right there in the Vocabulary where we can read it. And it clearly has an /h/, so Manaster-Ramer’s story about the origin of Third Mesa Hopi falling tone isn’t just a convenience to explain a divergent dialect: it’s rooted in fact. And we know about it now because Shelton wrote it down, and he used the Deseret Alphabet to do it.

Many in and out of the Church have regarded the Deseret Alphabet as nothing more than a historical curiosity—something that certifies the weirdo status of those wacky 19th century Mormons. But in Shelton’s Vocabulary we see it providing part of serious documentation of an indigenous language now in decline; a 150 year old aural snapshot of a lost time.

Contemporary view from Oraibi (photo by JaumeBG)

[*] This post is based on a forthcoming book that I coauthored with Kenneth R. Beesley. (Beesley, Kenneth R. and Dirk Elzinga. 2015. An 1860 English-Hopi Vocabulary Written in the Deseret Alphabet, University of Utah Press.) My thanks to Jeremy Grimshaw for his careful reading and suggestions that improved this essay.

[1] He later transcribed (and lightly edited) his journal into standard English spelling, which was then edited and published by Juanita Brooks in 1944. That’s probably the version you’ve seen—if you’ve seen it. Haskell’s original DA journal is in Special Collections at the Harold B. Lee Library at BYU.

[2] This is the later, 38-letter version. The version discussed here is older, and had two additional letters: OY (boy) and YU (use).

[3] Manaster-Ramer, Alexis. 1986. Genesis of Hopi Tones. International Journal of American Linguistics 52.2. pp 154-60.

[4] When linguists don’t have written records to help illuminate earlier stages of a language, we have to resort to inferences based on evidence from contemporary language usage, often with an assist from diverging dialects. This is what Manaster-Ramer did in his 1986 paper.

[5] There are also many words where we might expect an /h/ but don’t get one; Shelton was a decent field linguist, but even the best of us miss stuff. The Deseret Alphabet also got in his way. It was intended to write English, after all, and Hopi is very different from English.

Timpanogos and Numic borrowing

Today’s Ute Word of the Day is nukwi [nuˈkʷi], a verb that means ‘flow, run’. It is typically used to describe running water. I have always thought that the name ‘Timpanogos’ had this word as one of its elements, but now I’m not completely sure (I’m still pretty sure). If ‘Timpanogos’ is in fact from Ute, it would break down as: tïm(-pi) ‘rock’ + pa(a) ‘water’ + nogo-tsï ‘running’, yielding tïmpanogotsï “rock water running”.[1]

The differences between nukwi and nogo are pretty standard Numic stuff. In all of the Numic languages there is some indeterminacy regarding the quality of the vowel found in the first syllable of words like nukwi; sometimes it’s heard as [u], and other times it’s heard as [o]. So the variation between nu- and no- is expected.

The variation between the voiceless stop [k] and the voiced fricative [ɣ] (which I’m writing as g in the practical orthography) is also found in all of the Numic languages and has been grammaticized in some of them as verbal aspect, with one alternant signifying progressive or durative and the other something like momentaneous.

The variation between wi and final o is also not uncommon and is found in Central Numic (Timbisha, Shoshone, Comanche) as well as in Ute. I assume that o is the result of coalescence of wi, so wi should be considered the older form. The w, which really is just lip-rounding on the k, is in turn the result of coarticulation from the preceding u. This means that the form nukwi would have originally been *nuki. Central Numic has just this form (usually written nukki) with the more generic meaning ‘run’.

The suffix -tsï is probably an active participle suffix with a function similar to that of English -ing. It has various forms in Ute: -tï, -rï, and -chï/-tsï, with the latter variant found following a stem-final i. So ‘Timpanogos’ lines up very neatly as a word that comes from Ute.

However, all of its parts are also found in Shoshone (Central Numic). The word for ‘rock’ in Shoshone is tïm(-pi), just as in Ute; the word for ‘water’ is also paa, and the word for ‘run’ is also nukki (as discussed above). Shoshone also has a verbal suffix -tï, with a meaning similar to that of Ute. (The Shoshone -tï doesn’t have -tsï as one of its variants, however.) So the name ‘Timpanogos’ could also claim a Shoshone origin, though the claim isn’t perhaps as strong.[2]

This hints at a larger problem encountered by those of us working with Numic languages. Historically, there was extensive contact between speakers of neighboring Numic languages, and as all historical linguists know, one of the results of contact is borrowing. In this case, it is perfectly reasonable to assume borrowing between Shoshone and Ute. It is well established that the Goshutes, who speak a variety of Shoshone, interacted regularly with the Utes around Utah Lake. But how would we know that such borrowing took place? As I’ve shown above, all of the elements of the name ‘Timpanogos’ are present in Ute and Shoshone. While I think it most likely that ‘Timpanogos’ is a Ute word, I can't be 100% sure. And this isn’t the only case; ask me sometime about Panguitch and sego lilies.


[1] It probably referred to the Provo River rather than the mountain itself, but that’s just my opinion.

[2] One thing strengthening the claim (if only by a little bit) for a Shoshone origin of ‘Timpanogos’ is the fate of syllable-final nasals in Ute. In Ute, syllable-final nasals (like the m in ‘Timpanogos’) were lost; the word for ‘rock’ is actually tïpwi. (The w of tïpwi is the result of coarticulation from ï. Really.) If ‘Timpanogos’ came from Ute, it must have done so before the loss of syllable-final nasals. Alternatively, the word is actually from Shoshone, which has not lost syllable-final nasals.

UPDATE (31 March 2014): I’ve been meaning to get to this update for quite a while now. I was reminded not long after publishing the original post that Northern Ute still retains syllable final nasals that southern Ute varieties have lost, so tïm-pi ‘rock’ is a perfectly reasonable Ute word and pretty much clinches the case for the Ute origin of Timpanogos.

UPDATE 2 (21 August 2014): I was browsing the Domínguez/Escalante journal (edited by Ted J. Warner and translated by Fray Angelico Chavez) this morning and came across the name that the expedition members gave to Utah Valley: Valle de Nuestra Señora de la Merced de los Timpanocuitzis. What was interesting in the context of this blog post was the name ‘Timpanocuitzis’. The verb stem nokwi is clearly visible through the Spanish spelling conventions and provides an extra data point regarding the variation between nogo and nukwi I discussed above.

I love being a linguist.

Piimihchi

Today’s Ute Word of the Day is piimihchi [ˈpiːmi̥tʃi] ‘love’. Another way to say ‘love’ in Ute is tïï asïhti’i [tɨː ˈasɨ̥tiʔi], which is literally ‘well’ + ‘like’, or “like a lot”. These words give me an opening to talk about voiceless vowels in Ute and how I’ve chosen to represent them in the practical orthography I’m working on. But first a confession: in both of these words I cheated a little; the voiceless vowels aren’t really there (as far as I can tell), so the transcriptions are more accurately [ˈpiːmtʃi] and [tɨː ˈastiʔi]. This would lead to orthographic forms piimchii and tïï asti’i. So why put in the voiceless vowels if they’re not pronounced? The answer has to do with syllable structure and phonotactics, and the history of the language.

Ute is a Numic language, most closely related to Paiute, and more distantly related to Shoshone and Comanche. All of these languages have fairly simple syllable structure: a syllable consists minimally of a single vowel or diphthong. This vowel or diphthong may be introduced by a single consonant. It may also be followed by a single consonant, provided that 1) that consonant is /ʔ/ or /h/, or 2) it is a nasal consonant (/m, n, ŋ/) that shares its place of articulation with a following consonant. These syllable structure restrictions seem to apply to the Numic languages generally.

One place where these restrictions seem to relax a bit is with voiceless vowels. A voiceless vowel is, as its name suggests, a vowel without any vocalization or voicing. Voiceless vowels are variants of regular vowels. Certain conditions must apply for a vowel to become voiceless. Typically, the vowel must be unstressed and short. Additionally, voiceless vowels show up in word-final position or before voiceless consonants like /p/, /t/, /k/, and /s/. Voiceless vowels are not universal in Numic languages, but they are common; indeed, Paiute and Comanche are well known among linguists for having voiceless vowels.

Ute also has voiceless vowels, and the conditions under which they can appear in speech are roughly those described above. However, in Ute voiceless vowels are sometimes dropped altogether and not pronounced at all. Whether a vowel is dropped or not depends on a number of factors, including the quality of the vowel (it happens to /ɨ/ more frequently than to /a/, for example) and the dialect or speech habits of the speaker. Final voiceless vowels are more frequently dropped than voiceless vowels in other positions, though as today’s Word of the Day shows, it can happen in the middle of a word as well. The dropping of a final voiceless vowel may occur so frequently with a particular word that it becomes impossible to know which vowel was originally there. In that case, the vowel is lost. An unrecoverable dropped vowel can’t be written in a practical orthography.

But other dropped vowels can be recovered; here’s an example. The object form of nouns in Ute and Paiute were originally formed by adding the suffix *-a. When voiceless vowels began to be dropped word-finally, the suffix disappeared. But the vowel at the end of the noun stem was still pronounced. Historically, the object form of the word ‘boy’ was *aːpatʃi-a. When the final vowel became voiceless and dropped off, the object form became /aːpatʃi/. The subject form of the same word was historically *aːpatʃi. After its final vowel became voiceless and dropped off, it became /aːpatʃ/. Some speakers still pronounce the final vowel, so for them it’s [aːpatʃi̥]. In the practical orthography, this would be written as aapachih. The final vowel of ‘boy’ isn’t always pronounced, but it is recoverable because it pops up again in the object form.

We’re not always so lucky to have object forms to help recover dropped vowels. But sometimes comparison with other dialects and closely related Paiute can show us what vowel was there originally. But first, a question. How do we know that the underlined vowel in piimihchi or tïï asïhti’i was there to begin with? We know that because of what we know about syllable structure constraints in Ute and other Numic languages. Recall that syllables have a simple structure: there’s a vowel and maybe one consonant before it and one consonant after it. The consonant following the vowel could only be /h/, /ʔ/ or a nasal consonant that agreed in its place of articulation with a following consonant. Now look again at asïhti’i, pronounced as [ˈastiʔi]. The consonant /s/ agrees with /t/ in place of articulation, but it isn’t a nasal, nor is it /ʔ/ or /h/. So this is a likely place for a voiceless vowel to have been dropped. Looking in Sapir’s dictionary of Southern Paiute, we find ’ac·ïntu’i, ‘to like, to want’, which we would transcribe today in IPA as /ʔaʃːɨntuʔi/. The resemblance with /astiʔi/ isn’t accidental; these words are cognates. Southern Paiute has the vowel /ɨ/ in the second syllable. This vowel apparently became voiceless and then dropped out in Ute. But it’s recoverable if we look at the Paiute cognate. For this reason, I represent [ˈastiʔi] as asïhti’i in the practical orthography.

For piimihchi, pronounced as [ˈpiːmtʃi], it’s a little trickier. The nasal consonant /m/ doesn’t agree with the affricate /tʃ/, so this is also a likely place for a voiceless vowel to have been dropped. This word is unique to Ute, which means that there are no cognates in related languages to help recover the vowel. In Givón’s 1979 dictionary of Southern Ute, he shows pi?ímhci under the entry for ‘love’. This would be transcribed in IPA as [piˈʔimV̥tʃi]. The underlined h represents a voiceless vowel of undetermined quality. So there is evidence of a voiceless vowel; it's just a guess on my part that it’s [i̥].

These might seem like weak reasons to insert vowels that aren’t pronounced at all. And it might turn out that speakers who take the time to learn to read and write using this orthography will simply not write the vowels. If that’s the case, then fine. But I hope that by including the vowels now, the orthography may become useful beyond White Mesa in places where these vowels (and others like them) are still pronounced. We’ll have to see what happens.

UPDATE (20 February 2013): I found the verb sųtí?i [sɨ̥ˈtiʔi] ‘feel, sense’ in Givón’s 1979 dictionary. I don’t think it’s too much of a stretch to relate this verb with [ˈastiʔi] ‘like, want’, giving a little more support to the inclusion of the voiceless vowel ïh in the spelling of [ˈastiʔi] as asïhti’i.

UPDATE (8 September 2022): In Givón's 2016 dictionary (a revision and expansion of his 1979 dictionary) I find: piyɵmichi- [piˈjœmi̥tʃi-] v-tr 'love', 'care for', 'cherish'. So now I know. The voiceless vowel is in fact an /i/. I don't hear the stretch -iyɵ- as anything but a long /iː/, though.

Ute Word of the Day

In my previous post, I alluded to the "Ute Word of the Day." Every weekday I post a different Ute word from my files on Facebook and Google+. Most of them are words that I have collected and recorded during field work in White Mesa, Utah. Some of them are from a preliminary version of the White Mesa Ute Dictionary, prepared by Brian Stubbs working with elders from the White Mesa Ute community. This is a little something I do to motivate myself to keep working on the sound files containing these words. Documenting minority and endangered languages is important work, but it can also be tedious. Posting a Ute Word of the Day lets me try out and refine ideas for my practical orthography, and the posts usually generate comments concerning pronunciation and etymology; the conversation keeps me excited about doing the work.

In case you've missed them, past Words of the Day include:

11 January 2013: pahsagwov [pḁˈsaɣʷɔv̥] noun eye boogers
14 January 2013: sihkuchih [s̩ˈkutʃi̥] noun Abert’s squirrel (Sciurus aberti) 
15 January 2013: muhkwi’ya [m̥u̥ˈkʷiʔjæ] verb come to a sharp point (inanimate object)
16 January 2013: magwei’ih [maˈʁʷeiʔi̥] noun blanket
17 January 2013: Nuuchiun [ˈnuːtʃiũ] noun (pl) Ute people; Indian people
18 January 2013: komö’ni [qoˈmøʔni] verb turn around
21 January 2013: togwei [toˈʁʷei] verb be right, correct; be well
22 January 2013: kwïhchikuhchapïh [kʷɨ̥ˈtʃiku̥tʃapʰ] noun headband
23 January 2013: pagï [paˈɣɨ] noun fish
24 January 2013: aapachih [ˈaːpɐtʃi̥] noun boy
25 January 2013: sakïi [saˈkɨi] verb limp; be lame
28 January 2013: tïhkakunav [tɨ̥ˈkakunav̥] noun lunch sack
29 January 2013: chïhkwi’napïh [tʃɨ̥ˈkʷiʔnapɨ̥] noun key
30 January 2013: kanavïh [kaˈnav̥] noun willow; riverbank rush
31 January 2013: pö’öi [pøˈʔøi] verb write
 

Neyanwahke

Today's Ute Word of the Day (1 February 2013), Neyanwahke [neˈjãw̃ḁkɛ] 'Handgame', presents a puzzle for a practical orthography of Ute. In Sapir's description of Southern Paiute, the stem naiyáŋwi- 'handgame' contains a labiovelar nasal (ŋw = /ŋʷ/). In Ute, the labiovelar nasal corresponds to (heavy) nasalization of the first vowel in a vowel cluster. So vowel nasalization is important in Ute and ought to be represented in a practical orthography. In the practical orthography for White Mesa Ute that I'm developing, my first solution for representing vowel nasalization was to follow the vowel with n (e.g., Nuuchiun [nuːtʃiũ] 'Ute people; Indian people', the Ute Word of the Day for 17 January). This parallels my treatment of voiceless vowels as a vowel followed by h. The trailing h for voiceless vowels works because /h/ is otherwise unattested as a consonant phoneme of Ute, so there's no ambiguity: Vh is always and only a voiceless vowel. However, /n/ is amply attested as a consonant phoneme in Ute, so ambiguity could arise in sequences of orthographic VnV: is it [ṼV] or [VnV]? However, in all of the words that I can find with vowel nasalization, the vowel is back and round (/u/ or /o/), or, as in the present case, it is followed by /w/--either nasalized [w̃] or plain [w]. So my current thinking is to use Vn(V) when the vowel is round and Vnw(V) elsewhere. Alternatively, I could use Vnw(V) in all cases (except word-finally; in those cases a trailing n seems sufficient). 

Do my Numicist friends have any thoughts?