That wasn't his point. His point was that systematic sound changes show descent, not borrowing. Maybe Robertson has now convinced him that this, instead, was his point and he'll get his testimony back?
It has long been observed that language borrowing involves a process,“nativization,” which may be described as bringing the phonological content of forms of the donor language into systematic alignment with the phonological disposition of the recipient language. There is a huge literature on this with thousands of articles treating the predictability of the phonological passage from donor to recipient language. The only possible explanation for this long-observed consistency is to be found in the assessment of the phonological content of the donor and the inherent and rigid phonological laws that govern the phonological production of the receiving language.
Hmm. Well, I guess we can consider it progress that Robertson has moved beyond his fallacious appeal to the "comparative method," which is a discovery tool of the genetic relationship between one language and another. Still, whether it's a deliberate suppression of the full picture or incompetence—personally, I hope it's not the latter—what he presents here is not the full story. It is simply not the case that borrowing follows laws in the way that Stubbs has laid out his evidence in order to advance his argument. Borrowing involves alignment, as Robertson correctly terms it, but it is not systematic in that it follows a system-wide pattern. Otherwise, to pick some obvious examples, you wouldn't get doublets in English like guarantee/warranty or guard/ward, and you wouldn't have the diversity of phonological realizations that occur with the Spanish suffix -illo/-illa (flotilla and guerilla and armadillo on the one hand vs. tortilla and quesedilla on the other). It is inconsistent and non-systematic, which is one indicator of borrowing (among others). The issue is not the predictability of phonological transfer but the predictability conditioned by constraints.
For example, English phonology has one kind of unvoiced "s" sound, which we spell with "s." Arabic has two kinds of "s" sounds (called unvoiced sibilants). Now, English is more constrained than Arabic in this regard, because it only has one unvoiced sibilant in its phonological system. Therefore, we can predict that English speakers, when borrowing a word from Arabic and aligning Arabic phonology to English, will use the English unvoiced sibiliant for both the Arabic unvoiced sibilants. Thus, English uses the same sound in sofa
, even though both of these reflect different "s" sounds in Arabic (the history of these words in English is a bit more complicated, but the point still stands because the languages that served as intermediaries, Turkish and Swahili, have the same phonological constraints as English in this regard). On the other hand, how would you predict things the other way, English to Arabic? In certain cases, the proximate vowels as perceived by Arabic speakers will condition how they perceive the English "s" sound, but in other cases (particularly with mid back vowels and consonants), there is variation, and you really can't predict. All you can say is that it will be perceived as some kind of "s" sound. Another instructive example is the Arabic realizations of "p," a phoneme which it doesn't have. In many words borrowed from Persian, it was realized as an "f" because Persian is "p" aspirated (hence, the province of Pars—where we get the word "Persia" from—becomes Fars in Arabic, which was re-borrowed into Persian, so that you get Farsi as the name of the language of Persia). But English, too, has an aspirated "p," so why do Arabic speakers today perceived it as "b" but not an "f" like borrowings from Persian? As in the case of borrowings from Spanish into English, the reasons are often extra-linguistic or at least socio-linguistic, and it would be a mistake to structure these as laws in the way that Stubbs does with UA and Semitic, or to deduce the existence of dialects based on them. If we were to apply Stubbs's method, we would imply that people who say quesadilla and tortilla speak a different dialect of English from those who say armadillo and guerilla.
In any case, the constraints I refer to are not only the limits of a given phonological system but also the constraints of the phoneme being transferred. In order for a speaker to adapt non-native phoneme to his/her native phonology, there must be some element of the phoneme that already has some phonetic linkage that a speaker perceives. For example, there is an old joke among Arab-speaking English learners: an Egyptian in New York, desperate to park his car, rolls down the window to ask a pedestrian, "where I can bark?" to which the ____ New Yorker says "wherever you want, just don't bite." Arabic speakers often articulate an English "p" as a "b" because "p" is foreign to Arabic phonology, as I mentioned, but English "p" and Arabic "b" are distinguished only by two features, voice and aspiration, and they are otherwise articulated in about the same way (they are both labial plosives). The nature of one phoneme evokes that of another because of similar features. On the other hand, when there are few or no shared features, you get variety that is not predictable. Arabic, for example, has three sounds that English speakers perceive as some kind of "h" sound, but one of them is particularly difficult for English speakers especially before consonants, the voiceless pharyngeal fricative. It occurs in the name Muhammad, and in that case English speakers just treat it as an English "h," with which it shares some characteristics. But it also occurs in the name Ahmad, and in that case you get it realized as a "kh" sound (a sound not native to English but easy to imitate for English speakers for various reasons) or as a "k" (so one used to hear of some terrorist named "Akmad"), or it is deleted (i.e., it is simply ignored). Can Robertson tell me what "systematic" law operates here? I doubt it. The point is not that you can predict how a phoneme in language A will be realized in language B but that you can isolate a range of possibilities derived from the constraints of the phonology against the constraints of the phoneme: if language A has "t" sound then any "t" sound in language B will likely be realized with that and not, say, as an "f" sound (leaving aside, for the moment, constraints imposed by syllable structure, prosody, and morphology). Even then it can be tricky, because historical or social forces can intervene, or it is also possible that the phonemes will simply be borrowed into the new language without any "alignment." This happened with Berber, which borrowed several phonemes from Arabic which it did not previously contain.
According to Stubbs, the Uto-Aztecans did not borrow any phonemes from the Egypto-Hebraeo-Aramaic of the Nephite settlers, despite borrowing a great deal of their core vocabulary from the Nephites. That in itself typologically improbable: I'm supposed to believe that the mixed language that became Uto-Aztecan was in such deep contact with Nephitish that Uto-Aztecans borrowed the Nephitish words for "head" and "finger" but they could never articulate the Nephitish "b" despite its differing from Uto-Aztecan "p" by only a single feature and despite the fact that there are Uto-Aztecan languages with that phoneme that Stubbs culls evidence from? Nothing is impossible. Certainly not for him. But there is a reason for that: the basic inventory of phonemes that he reconstructs for Uto-Aztecan contains about 18 phonemes. Combining Egyptian, Aramaic (at any period!), and Hebrew yields between roughly 35 phonemes, depending on how you treat certain consonants and how you treat vowels and diphthongs. It's roughly twice the size of the Uto-Aztecan set (in addition to that, Stubbs isolates clusters of consonants, as well as consonants and vowels in different positions, in the Egypto-Hebreao-Aramaic hybrid spoken by the Nephites, but we don't need to get into this to illustrate the point). What this means is that each of his U-A phonemes gets multiple possible matches with each of the Nephitish phonemes: it's the equivalent of double dipping. So consider what his U-A phoneme "t" does: it was the realization of 1) Proto-Semitic "d-h," 2) Semitic "d," 3) Semitic initial "r," and 4) Egyptian initial "r" (the articulation of which is actually unclear and changed over time). So, the Uto-Aztecans realized all 4 Nephite sounds as a "t."
But, wait—there's more! In the very section of the introduction where Stubbs summarizes his rules/laws, he uses Uto-Aztecan words that he reconstructs with an "r." If that were so, why didn't the Uto-Aztecans simply realize Nephite "r" as their "r"? Thus, we see the very first entry:
Stubbs wrote:The other voiced stops also devoice, that is, Semitic b, d, g > UA p, t, k; also Semitic q > k:
(606) dubur "buttocks, rear" > UA tupur, "hip, buttocks"
That looks compelling, doesn't it? Well, firstly, you'll notice he doesn't tell you here that "dubur" is an Arabic word, and he simply assumes that it existed in Hebrew or Aramaic, which it most certainly did not, since this structure did not exist in Hebrew or Aramaic (there are syllabic and prosodic constraints that would prevent it; Stubbs ignores it here but elsewhere makes phonological rules/laws based on the structure of certain Syriac words—if he had picked the Arabic form, his law wouldn't have worked; see his use of Syriac kawkba
—ignoring spirantization!—rather than Arabic kawkab
in section 1.17). But as everyone knows, the root d-b-r in Hebrew and Aramaic, while distantly related to the Arabic, has no semantic relation to it and generally refers to 1) speaking and 2) grazing, and only the word debīr
means anything like "behind": it is the word used for the inner sanctum of the temple. I wonder how many of the pious Nephites who built that temple would appreciate the connection with "buttocks" that Stubbs has hit on here by appealing to the Arabic. But I digress: because the real issue, as you can plainly see, is that he has used a word in Uto-Aztecan with an "r" phoneme, which makes us wonder how the Uto-Aztecans heard it as "t" sound. One could wiggle out of that in all sorts of ways, but what it illustrates the inherent meaningless of the exercise, since we don't even know what "r" means in some of these contexts (Egyptian) and Stubbs isn't really clear whether it existed in Uto-Aztecan as such. No matter—just lump it all in with a "t" as is convenient. Except of course when it isn't—as in the case of an "r" at the end of words, since no "t" existed in that environment in Stubbs's reconstruction of Uto-Aztecan, he decides that the expected "r" just disappeared rather than becoming a "t." All this despite the fact that "r" apparently existed in some words in final position in Uto-Aztecan ("UA tupur, 'hip, buttocks'").
The short of it is, each of the reconstructed Uto-Aztecan phonemes gets to play multiple roles in the play. Just as English has one phoneme of "s" to fill in for the two of Arabic, Uto-Aztecan apparently has one to capture four phonemes of Nephitish. That is very convenient. And when it is not, there are further convenient rules to explain whatever inconveniences that have arisen. It would be amazing if it weren't so typical of every single argument made by the apologists.
Gadianton wrote:Was this taken on previously? I don't remember. Grover and Robertson both gave examples (in response to my sock puppet* at Interpreter and to others) in linguistics literature that sort-a could possibly fit the bill for what Stubbs is doing by analogy, but my impression then was that Stubbs wasn't doing anything strictly textbook, nor has he alluded to anything textbook that one could simply look up to get a primer on, and then delve into Stubbs' work. Does Stubbs ever say, "Hey guys, I'm talking about nativization here as you should all be aware!"
Stubbs is not very clear in his 2015 work, except for a few phrases here and there work what he is arguing for; at section 1.18 he says that Uto-Aztecan is a language mix, and then you get the closest thing to a thesis:
Nevertheless, the intensity of the contact during French rule in England caused English to change rapidly, and to end up as quite a language mix of Old English and French. Yet that kind of mixing of languages and peoples happens regularly. In fact, the Norman French themselves were a mixture of at least four peoples: the Viking (Germanic) Norseman (source of Norman) who settled their area of France, and they mixed with the French, who descend from the Celtic Gauls, the Germanic Franks, and the Romans who brought the Latin language which in that area became French. UA is also a language mix, as shall be seen later.Such mixing happens often among Native Americans as well. In my classes, I ask my Navajo students how many of them have all four grandparents’ being Navajo. Few raise their hands. Then I ask how many have one or two grandparents who are of another tribe or ethnic group. Most raise their hands. Most have one or two grandparents who are Ute or Hopi or Walapai or Sioux or Hispanic or Irish, etc.
Besides words being borrowed, language influences alter the grammar of a language as well. These grammatical changes are sometimes harder for native speakers to identify or even perceive, because, as we said previously, we mostly do grammar subconsciously, and so when bilingualism is prevalent in a border area between languages, the subconscious grammatical patterns of the two tongues can and do influence each other slowly enough that native speakers are hardly aware. For example, English whom, as accusative (object) form of who, is nearly dead as a last survivor of the Old English case system, yet most English speakers do not know how to use it and so do not, or if they do, they often use it incorrectly, because the case system in which it fits or which used to be part of the language, has all been lost for centuries.
This is all very applicable to a hypothesized arrival of Mediterranean speakers in ancient America, because the languages would differ enough that it is to be expected that such an arrival in a very different language environment would change very much. The derivational detail being lost would not be surprising, just as the Germanic case endings were lost in Middle English. The simplification or loss or fossilization of some verb conjugations would be expectable, just as English lost most of its verb conjugations.
There is a thesis buried in this fluff somewhere. Some of this is simply incorrect: the constant appeal to the example of English and French among Stubbs and his admirers reveals just how ignorant they are of the history of English. French had very little influence on English until the 14th century, three hundred years after the Norman conquest, as anyone who has ever read Middle English earlier than Chaucer knows well (see Layamon's Brut
, for example, or Ancrene Wisse
or any other number of English texts before the mid fourteenth century). in fact, the greatest influx of French and Latin spanned the 15th century to the 18th century because French held the dominant place in international life that English holds today. Excuse me for thinking that a linguist like Stubbs or John Robertson should remember that historical linguistics involves knowing history as well, and that they should know more about the history of English than the average History Channel viewer, especially if they are going to use it as a touchstone for their own claims.
Anyway, I truly wonder how many of his Navajo students speak a mixed language of Sioux and Irish—which is what we should logically expect, if this is supposed to be a corollary to his thesis—but in any case you will notice how he slides between discussing the mixing of human beings and the mixing of languages. The one does not necessarily imply the other: I myself am married to someone from land quite distant from Parowan whose native language is not English or even Indo-European, yet our offspring does not speak a mixed language derived from my native English and my spouse's native language, even though both are spoken at home.
To be clear, a mixed language is not the same thing as borrowing; one is the result of language contact, the other is but one of many processes that occur when languages are in contact. What I have seen is Stubbs/Robertson playing around with these terms as it suits them.
"As to any slivers of light or any particles of darkness of the past, we forget about them."
—B. Redd McConkie