Pitch Tone and the "Saltillo" in Modern and Ancient Author(): Benjamin . Whorf, Lyle Campbell, Frances Karttunen Source: International Journal of American Linguistics, Vol. 59, No. 2 (Apr., 1993), pp. 165-223 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/1265154 Accessed: 24/02/2010 19:39

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=ucpress.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to International Journal of American Linguistics.

http://www.jstor.org PITCH TONE AND THE "SALTILLO" IN MODERN AND ANCIENT NAHUATL

tBENJAMINL. WHORF EDITED, WITH COMMENTARYAND ANNOTATIONS,BY LYLE CAMPBELLAND FRANCES KARTTUNEN

LOUISIANA STATE UNIVERSITY AND UNIVERSITY OF TEXAS, AUSTIN

1. Introduction.1 The purpose of this article is to make available an apparently finished, but unpublished, paper by Benjamin L. Whorf on Nahuatl. The manuscript is in the Boas Collection at the American Philo- sophical Foundation Library in Philadelphia. The paper is of interest for several reasons. First, since Whorf is an important figure in American linguistics, this article has historical significance. It documents Whorf's thinking and ex- emplifies linguistic practice at the time. It contributes greater understand- ing to the history of the study of Nahuatl, Uto-Aztecan, American Indian , phonology, and linguistic anthropology. In this paper, Whorf spans two traditions, the prephonemic and the phonemic. Thus, while he speaks of phonemes, most of his description (as will be seen below) is not phonemic but is presented in phonetic terms, sometimes broad, occasion- ally quite narrow. Second, Whorf presents text material from two dialects, Milpa Alta and Tepoztlan (which he abbreviates and , respectively), to illustrate his points. Since these record a degree of phonetic detail concerning stress and intonation seldom seen in American Indian studies, they are quite valuable. They are also important primary documents, given that the lan- guage in Milpa Alta and Tepoztlan is fast approaching extinction. Third, Whorf's hypotheses concerning the origin of "saltillo" (basi- cally ) are an important stage in the development of Uto- Aztecan linguistics, and in the study of Nahuatl in particular. On the whole, Whorf's ideas about its origin were in the right direction, though

1 We wish to acknowledge with gratitude the permission granted by the American Philo- sophical Society to photocopy Whorf's manuscript and to publish it here. We also thank William Bright for helpful comments and advice. This paper was originally prepared for publication by Lyle Campbell, but Frances Karttu- nen's comments on an earlier version were so extensive and important that it seemed only appropriatethat for her extensive input she should be recognized as jointly involved in the editing, commentary, and annotations.

[IJAL, vol. 59, no. 2, April 1993, pp. 165-223] ? 1993 by The University of Chicago. All rights reserved. 0020-7071/93/5902-0003$01.00 165 166 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS understandably a more complete picture is now available (see Campbell and Langacker 1978, Canger 1980, and Dakin 1982:26-28, 65-66). Finally, Whorf's 1946 paper has been extremely influential in Nahuatl studies, and it persists in being one of the most frequently cited refer- ences on Nahuatl in the linguistic literature. To cite just one example, it was the basis for Greenberg's (1963) information regarding word order in Nahuatl (cf. also Campbell, Bubenik, and Saxon 1988). The current paper brings perspective to the 1946 article. While the 1946 article is more fully phonemic and thus more easily grasped, its section on "accent" (1946: 369) was always mysterious to readers. Whorf's ideas about "accent" are presented in much fuller form in this present paper, which clarifies greatly what Whorf (1946) was trying to say. There he summarizes his view as: M[ilpa]A[lta] has a stress accentwith associatedpitch differences.Words over one syllablehave primaryaccent on the penult.This accentconsists of loud stress togetherwith one of two varietiesof tone-pattern.In "normal tone-pattern"these varietiesare: 1) words endingin -, including?, have high tone on penultand low tone on ultima,.. i.nbn siwa- ['thatwom- an'];2) wordsending in - have on penulta tone fallingfrom medium high (less high thanthe high of patternone) to medium(higher than 'low') and remainingmedium on ultima, e.g. kisaya ['was leaving'], kwep6.ni ['to burst/bloom'];except that the ultimamay show a furtherslight fall beforea pause.Words ending in -lli e.g. kalli ['house']tend to show a compromise with patternone by often having a nonfallinghigh tone on the penult,the fall seemingto occuron the long 1. One-syllablewords usually have stress and a moderatelyhigh tone, but a few, markedthus, ka, have low tone and optionalstress. This tone-featurealone distinguisheska "with,by" fromka (high tone) "is": ka no-t6min"with my money",ka no-t6min"it is my money". Secondaryaccents and unaccentedsyllables have a mediumtone and the formera louderstress thanthe latter,nearly approaching the stress of the primaryaccent. Secondarystress occurs on: 1) an ultima ending in -C; 2) the firstsyllable of a long word;3) alternatesyllables before the primary accent except that a short-Vopen-syllable is usuallyhurried and does not count, and two in successioncount as one, e.g. .'nonokWep7ia"ya,where no, no, kwe,and pi are thus hurried.(Whorf 1946:369-70.) As will be evident below, this confusing summary is fully spelled out in the current paper, which in turn reflects Whorf's interpretationof colonial grammars.The lack of clarity in the intended phonemic status of Whorf's (1946) tone patterns was always puzzling, but the present paper makes this less mysterious. Here, in this paper, Whorf seems to have both a descriptive and histori- cal goal in treating Nahuatl pitch accent and "saltillo." While his phonetic PITCHTONE AND THE "SALTILLO"IN NAHUATL 167 description and historical explanation of "saltillo" can both be read with profit to this day, his treatment of tone, pitch, length, and accent is quite misleading. On a strictly phonetic level, his description is probably quite useful. However, Nahuatl does not (nor did it ever) have contrastive tones or pitch accent (i.e., tone is not phonemic, organic, or underlying). It is not clear why Whorf thought tone/pitch to be so, but there are abundant grounds for speculation about why Whorf reached such conclusions. First, Nahuatl vowel length is notoriously difficult to perceive correctly in many modern dialects (though it can also be relatively straightforward in others), as pointed out by Whorf himself (Whorf 1946:370). Moreover, length is only one of several vowel properties that enter the picture. An- other is stress, which generally falls on the penultimate syllable. Other factors include whether the vowel is initial or not (some dialects automat- ically lengthen initial vowels), whether the vowel is followed either by two or by a single syllable-finally (such vowels are routinely shortened in some dialects), and whether "saltillo" is present. The interaction of length, stress, and syllable patterns (including presence or absence of glottal stop) can make the analysis of vowels difficult. Second, Whorf seems to have been influenced by colonial grammarians, especially Rincon and Carochi, both in these matters and others. For ex- ample, for voiceless L and , essentially word-final allophones of /1/ and /w/, Whorf writes an in union with these sounds (hl, hw); see below. It is tempting to speculate that in this Whorf was influenced by colonial de- scriptions, where these sounds were written as (lh) and (uh). More to the point, since these grammarians of Classical Nahuatl wrote of "accents," and some presented a rathercomplex description (see below), one suspects Whorf's analysis with pitch accents may be influenced, at least in part, by a misinterpretationof these grammarians'descriptions (see below). For example, Rinc6n (1595), Whorf's major inspiration in his interpre- tation of these matters in Classical Nahuatl (see below), had essentially four to distinguish two contrasts: presence vs. absence of vowel length, and presence vs. absence of glottal stop (see below). Carochi (1645) had the most complete and accurate analysis, though his diacritics were never generally adopted by Nahuatl scribes or grammarians.Nevertheless, Whorf does rely on material from Carochi (see below).2 Carochi employed an acute accent for short vowels (notex [no-tes] 'my flour') and a for long vowels (notex [no-te.s] 'my brother-in-law').Carochi described a

2 For most Classical Nahuatl writers, vowel length was simply not indicated, although the Cantares Mexicanos (Bierhorst 1985) sometimes doubled the vowel to indicate length. Mo- lina, Olmos, and others occasionally wrote h for "saltillo," while a circumflex is written for this sporadically in the Florentine Codex (Sahagdn 1979). (Karttunenand Lockhart 1987:4, 67-74.) 168 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS difference in the pronunciation of "saltillo" depending on whether it oc- curred within an utterance or at the end, using a grave accent mark for phrase-internal saltillo and a circumflex for the phrase-final one (Bright 1960 and Karttunen and Lockhart 1987:69). Carochi (1945[1759:3]) de- scribed his system of notation in a section entitled "De Los Accentos," as follows: De quatroaccentos usaremos en este Arte,para distinguir quatro generos de tonos, con [ue] se pronunciala vocal de cada syllaba, son estos, a, a, a, a. El accento(') nota, y serialde syllababreve como tetl [te-tl], piedra: tletl [tle-tl],fuego. La (-) es accentode syllabalarga, como atl [a.-tl],agua: teotl [teo.-tl], Dios. La (') es serial de la pronunciacion, q[ue] suelen llamar saltillo;por que la vocal sobreque cae este accentose pronunciacomo con salto, 6 singulto,o reparo,y suspension:ver.gr. tatli [ta'-tli],padre: patli [pa'-tli],medicina: mototli [mo-to'-tli],ardilla. Del accento (^) usaremos solamenteen las ultimasvocales de todos los pluralesde verbos, y nom- bres, q[ue] acabarenen vocal quandono se pronunciareimmediatamente otradiccion.3 Rinc6n (1595:63) distinguishes length from pitch/stress: "Hay syllaba larga, y syllaba breve, y syllaba igual, que ni es larga ni es breve.... Pues conforme a esta diversidad de la cantidad dela syllaba, se hallan cinco diferencias de accentos, porque en la syllaba larga hay acento agudo y grave.... Acento agudo es el que fuera de alargar la syllaba le aniadeun tono que levanta la pronunciacion con sonido agudo."4 For Rinc6n, the acute accent is employed for a long vowel with high pitch/stress. His grave accent is for a long vowel with low pitch/stress- with all his examples of grave accent falling on word-final syllables. Rincon's "acento moderado" falls on short vowels of penultimate sylla- bles. As Bright (1960:67) points out, Rinc6n's grave and acute accents are in complementary distribution, with grave only on final syllables and

3 [This section is] about four accents we will use in this grammar to distinguish four kinds of tones, with which the vowel of each syllable is pronounced, and they are these: a, a, ai, a. The accent (') is the notation and symbol of a short syllable, such as tetl 'stone', tletl 'fire'. The (-) is the accent of a long syllable, such a atl 'water', teotl 'God'. The (') is the sign of the pronunciationwhich they tend to call 'saltillo', because the vowel upon which this accent falls is pronounced with a jump or hiccough or stop, and suspension: e.g., tatli 'fa- ther',patli 'medicine', mototli 'squirrel'.We will use the accent (^) only on the final syllables of all the plurals of verbs and nouns that end in a vowel when no other utterance is pro- nounced immediately [thereafter]. [Our -LC/FK.] 4 There are long syllables, short syllables, and equal syllables which are neither long nor short.... According to this diversity of the quantity of the syllable, five different accents are found, because in the long syllable there are acute accent and grave.... Acute accent is that which apart from lengthening the syllable adds to it a tone which raises the pronunciation with an acute sound. [Our translation-LC/FK.] PITCH TONE AND THE "SALTILLO" IN NAHUATL 169 acute elsewhere, where both represent long vowels. Rincon's "breve" and "moderado"are both for short vowels, although "moderado"is less clear, due to lack of examples. Since it is said to be found in sequences with -alli, Rinc6n's perception may be colored by the long 1 (a pattern also dis- cussed by Whorf; see below) (Bright 1960:67). Therefore, while colonial grammariansspoke of "accents" and "tones," it is clear from their descriptions that vowel length and glottal stop were at issue, not tones or pitch accent, as has been clearly shown by Bright (1960). Whorf, however, is not the only one to have been mislead by his reading of colonial grammars into thinking that Nahuatl has tones; see, for example, Barritt (1956); cf. also Schoembs (1949). The date of Whorf's manuscript is not certain. As Whorf indicates in his opening line, the material was collected in 1930, when Whorf went to with the support of a Social Science Research Council research fellowship. Whorf worked on these Mexican materials for several years after his return (Carroll 1956:14), and his posthumous Milpa Alta sketch (Whorf 1946) is a result of this work. The date of this manuscript is, therefore, uncertain, though we suspect it may have been written soon after the trip, judging from its lack of a mature application of the phone- mic principle. We present Whorf's manuscript without change, but with some emen- dations and annotations. Since Whorf did not use brackets ([ ]) in his text (with one exception, so identified), we have enclosed all our emendations in brackets. For the most part, these are limited to an occasional word or punctuation for clearer reading, to added glosses, or to relevant forms for comparison or for clarification. More extended comments are presented in the footnotes. In the manuscript, Whorf often did not capitalize headings, employ single quotations marks to indicate glosses, or employ commas in some cases necessary to an understanding of his text. These have been added here without further indication. The manuscript had no bibliogra- phy; all the references are to sources we cite. (Citations of Proto-Nahua and Proto-Uto-Aztecan are from Campbell and Langacker 1978; Cora forms are from McMahon and McMahon 1959.) The manuscript is essentially in two parts: pages 1 to 35 written in a very clear hand, and pages numbered 36 through 54 in typescript.5 Whorf presented his texts and many examples with both English and Spanish , though these are not always equivalent. The Spanish

5 The only change we have introduced in Whorf's notation is to write a for his symbol which is roughly like a 9, or a backward e, or an upside-down and backward schwa, which we suspect its intended value to have been. We note that Whorf varies in his use of terms, with Aztec and Nahuatl being synonymous. 170 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS appears to be from Whorf's informants.6 We have made no changes in the Spanish glosses; only occasionally we have inserted material in brack- ets to relate these glosses to more standardor more grammatical forms.7

2. Whorf's paper.

Pitch tone and the "saltillo" in Modern and Ancient Nahuatl by Benjamin L. Whorf

The material for this paper was collected in Mexico in the winter of 1930 with the assistance of a grant furnished by the Social Science Re- search Council.

Phonology Stops: p, t, k, q, (in qw, qW), (,).8 All un-aspirated. [The] t [is] dental; k before a, o, u [is] nearer q than is English k; it is the unvoiced correlate of Eng[lish] g in 'go'. : ts, tc [= c] unaspirated Spirants: s, c [= s], The last is a variant for c [= s] in certain localities. It is an untrilled unvoiced r with [the] end of [the] tongue raised and al- most touching [the] hard palate just back of the c-position [= s], giving a hollow spirant quality.9

6 One of Whorf's informants was Dona Luz Jim6nez, who went on to an illustrious career as informant for others and Nahuatl teacher with Fernando Horcasitas and as the author of several publications, including Life and Death in Milpa Alta. 7 For example, Nahuatl has the cross-referencing object pronoun prefix k(i)- on transitive verbs, followed by the full noun phrase object, e.g., ni-k-ciwa tlaskalli [I-it-make tortilla] 'I make tortillas'. The Spanish translations in Whorf's texts frequently translate both the pro- noun prefix and the noun phrase object, though both are not necessary in Spanish and in fact their cooccurrence is ungrammatical in most instances, e.g., for this example: lo hago tor- tilla, where standard Spanish would require hago tortilla(s) or la(s) hago, or perhaps in strained emphatic form la hago a la tortilla, but even in this doubtful form, the pronominal object la must agree in gender with the object noun, tortilla, both feminine, which is not the case in many of the Spanish translations from Whorf's texts. These glosses were probably provided by Whorf's informants and as such are useful examples of the Spanish produced by Nahuatl speakers. 8 In the handwritten section of the paper, Whorf's symbol for glottal stop seems to be about halfway between ['] and [7], whereas in the typescript section he uniformly employs [']. Therefore, except in the introduction (1), we employ ['] throughout, both for Whorf's symbol and for cases where we cite Nahuatl forms which contain glottal stops. 9 Whorf employs the phonetic symbols [c] "voiceless alveopalatal " and [tc] "voiceless alveopalatal " of the older system of representing American Indian lan- guages, which in later usage were replaced by [s] and [c] respectively. We maintain his sym- bols throughout, though in forms we have inserted for comparison, we utilize the newer, more familiar symbols. PITCHTONE AND THE "SALTILLO"IN NAHUATL 171

sr A variant of the last differing but very slightly in sound, tongue-tip in s-position, slightly more hissing and less hollow than r, but with dis- tinct r-quality. occasional only- in Ger[man] ach, without scrape. c Occasional only-between c [=s] and x-resembles ch [of] Ger[man] ich ['I']. ic = faint c heard after vowel.10

Nasals: voiced m, , q unvoiced N, 1I nasal breath 1 nasalizing of vowel 6 [All these nasals except m are variants of the phoneme n.] (This is the only case where Whorf used brackets.) Liquids: voiced 1 varies in tongue position from "light" to "dark" but generally "lighter" than Eng[lish] 1 with a . With expiring voice 1 As in Eng[lish] 'melt', 'else'. Unvoiced L, I [The] second [is] a variant of [the] first with 1 nearer ly and c-like [=s] spiranti- zation. Sounds like cl (shl) [=sl], or still more like c & 1 simultaneously. (I find that i does not occur in this paper.) Occluded liquid: tl t [is] supra-dental, 1 [is] produced in [the] same po- sition as the t, lips retracted. When tl is final of a word t is only faintly audible and 1 is fully unvoiced; elsewhere t is distinctly audible and 1 not wholly unvoiced (tinged with incoming voice). Aspirate: h As in Eng[lish,] but mostly heard in the combinations LhW and 'h, the last being the "saltillo", to be described more fully herein. Semi-vowels: Voiced y, w (& w) Like Eng[lish] except that in w the back contraction is stronger and more like that in the vowel o. Thus, awa

10 Actually, s and s exhibit very little allophonic variation in most Nahuatl dialects. Whorf heard r-like sounds for s presumably because there is a clearly audible retroflexion of /I/ in much of Morelos (where Tepoztlan is located) and parts of Puebla. Nahuatl has no x (velar fricative) in native words. There is considerable variation in "saltillo," however. This was apparently originally h, which changed to ' (glottal stop) in central dialects, precisely those upon which the colonial descriptions of Nahuatl are principally based, often called Classical Nahuatl (or Classical Aztec-Whorf uses both terms). (Going even further back in Nahuatl's prehistory, we find that some of these "saltillos" [h/'] derive from certain syllable- final consonants, e.g., -w, -t; cf. Campbell and Langacker 1978). In some dialects ['] and [h] are in complementary distribution as variants of a single underlying segment. In dialects which have h, today one can sometimes hear phonetic [x] (velar fricative), probably due to influence from Spanish x ("jota"), since Mexican Spanish has no h. 11 Modern dialects may typically vary between full final n and a nasalized vowel, but most do not exhibit voiceless nasals. 172 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS resembles aoa. Where I have heard the w pronounced with this guttural quality especially strong I have recorded w. Unvoiced W Fully unvoiced, not partially like w in English hw (wh).12 Faintly heard consonants are written superscript. Doubled Consonants 11occurs in all dialects heard and mm, nn, tt, yy, [and] ww [occur] in Tepoztlan. These are not merely lengthened or pro- longed consonants; the speech organs undergo a double wave of compres- sion and relaxation in pronouncing them. tt [and] kk = doublings with one part faint.13 Vowels: a (father), a between a in father and u in cut, nearer [the] latter. e (Sp[anish] pero) E (met) t (it)14 i (machine) o (home) u (pool)15

12 Nahuatl has the rule that final nonnasal sonorant consonants (i.e., 1, w, y, [and r in Spanish loans]) are devoiced. Historically, the final y went on to become s. Thus, in tradi- tional phonemic terms, Nahuatl has the phonemes // and /w/ which each have a voiceless al- lophone ([L], [W]) in syllable-final position, and voiced allophones elsewhere. Whorf's LhW is thus apparently /lw/, i.e., syllable-final /V followed by syllable-initial /w/, where the syllable-final 1 is phonetically voiceless, i.e., [Lw]. 13 Basically the only doubled consonants typical of Nahuatl are 11. These are found only across a morpheme boundary and come about through the rule whereby the combination of a morpheme-finalI followed by a morpheme-initial -tl changes to 1I(tl -, 1 / I _ ) (cf. Whorf 1946; Campbell and Langacker 1978; and Dakin 1982). Otherwise, the only other double consonants commonly found are in certain dialects, in- cluding variants of Classical Nahuatl, which have itta 'to see'. This is from PUA **tiwa, which became *ihta in Proto-Nahua by regular rules, vowel loss (twa), epenthesis (itwa), metathesis (iwta), and change of final w to h (ihta); the vowel /i/ changed to /i/ (ihta) in all Nahua varieties except Pochutec. This form remains in many dialects, though in some others it became i'ta and then itta, producing a somewhat anomalous geminate tt found only in this and a couple of other words (Campbell and Langacker 1978). The other double consonants Whorf records are probably features of emphasis and intonation, not underlying geminates (though one expects some geminates at morpheme boundaries, often due to assimilation, e.g., m-p > p-p, and Classical Nahuatl had many geminates at morpheme boundaries created by assimilation, e.g., es-yo' > es-so' 'blood'-InalienablePossession Suffix and ok-sep-pa 'once more' < ok-sem-pa). 14 This actually occurs very rarely in Whorf's manuscript.He writes it i in handscript,but wrote a note in the margin with an arrow pointing to this sound, saying: Note In printing change to t (Greek iota). Therefore, we have written it t in its few occurrences in the paper. Between this "iota" and the (e) immediately above it in the line above, Whorf has a bar (-); it is difficult to tell whether this is intended as an underlining of the (e) or as a macron over the "iota," but in either case, it seems to play no significant role in his later usage in the paper. 15 Whorf does not actually employ u, and it is not part of the Nahuatl phonological system, save in Spanish loanwords. There is, however, some phonetic variation in the realization of PITCHTONE AND THE "SALTILLO"IN NAHUATL 173

Diphthongs & Combinations: ai, ei occasional au, eu, eu common ao and ae, the first vowel modulating smoothly into the second. Length: a-, o. = long vowel16 a, a = short vowel ti, cim?pa = ultra-short vowel wE, WI = ultra-short whispered vowel. Accent: Stress accent is indicated by underlining the vowel: ko-mitl It is not strong.17 Pitch accent and the saltillo are treated as the subject of this paper.

PITCH TONES AND THE SALTILLO

Dialects Studied. Except for scatterings of phrases and sentences heard in other places, this study is based on two dialects; that of Milpa Alta, Fed- eral District, and that of Tepoztlan, Morelos. These dialects will be de- noted M and T. M is probably the best preserved remnant of the old Aztec of Tenochti- tlan. It is the least corrupted phonetically of any dialect I found, but on the other hand it shows a large proportion of Hispanisms. It preserves faint w and W sounds which have disappeared in other dialects, and re- tains pure c [=sl.18 T is not descended from Aztec but from the speech of the Tlahuica tribe,19 who came from the valley of Mexico into the valley of o and o.; o before I is often u in Colonial texts, and the long vowel sometimes tends toward a higher quality. Similarly, Whorf scarcely uses e in the paper, except when repeating forms from Colonial sources. His handwrittene is quite clear, although in the typed part of the pa- per it is sometimes difficult to distinguish E from e, since he apparentlyhad typed e, and then added a bit by hand to the top to make it appear as e; unfortunately, the added portion is not always easy to see clearly. 16 Also e- occurs very frequently in the paper, though Whorf did not list it here. 17 Nahuatl stress is predictably on the penultimate syllable. 18 The w is quite well preserved in most modem Nahuatl dialects; voiceless w ([W])- which occurs only syllable-finally-is also preserved in many, though it frequently changes in many other dialects as well, often to h (see Lastra de Suarez 1986). 19 Tlahuica or Tlalhuica is the name applied to the Nahuatl-speaking group that settled Morelos and established centers at Cuernavacaand elsewhere. It should not be confused with the Tlahuica that is the name modem Ocuilteco speakers prefer for their and ethnic identity. Ocuilteco is a moribund Otomanguean language spoken near Ocuilan de Arteaga, in San Juan Atzingo, Santa Lucia, and Colonia Gustavo Baz-to the west of Mexico City, be- tween Toluca and Cuernavaca. 174 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

Cuernavaca in pre-Aztec times. Grammatically it is purer Nahuatl than M and shows less importation of Spanish words, but it shows greater pho- netic leveling. It largely lacks W and in place of c [=s] has r. Pitch Tones Medium: The voice [is] neither raised nor lowered. [It is] to be under- stood when no other tone is designated for a syllable. Slightly Raised: The beginning of the syllable is slightly raised above me- dium tone and almost instantly drops back to medium tone, usually while on the same vowel. It could therefore be called a slightly raised falling tone, except that in certain cases as noted below the fall is deferred till [sic] the next syllable. [This] resembles [the] tone on [the] last syllable of "That is right."-said in a manner showing lack of interest. Designated a, 6, etc. M. noka tso.yo-ni 'mientras se freia' 'While it was frying' (Don't drop the voice below medium on ni.) When two short vowels come together, especially in the combinations ia, oa, and the first has slightly raised tone, they are so quickly pro- nounced that the resumption of medium tone does not occur till [sic] the second is reached. Designated ia, oa. M. niktEcolowla 'I am mashing it.' High: [There is a] high tone on a syllable like that on "know" in "Didn't you know it?" [There is] no falling or wavering of pitch on the syllable. Denoted a, 6. M. ka 'is' cito6ma'tl 'tomato' . pan 'in' tla'qwalli 'meal' to'min 'money' nikmanas tla.ckall 'I shall pat out tortillas.' Low: [There is a] low tone on a syllable, like that on "so" in "I should say sb!" said with finality. [It] almost always comes after high tone on [the] preceding syllable. This frequent jump from high to low continually occurring in the midst of a little-varying level of medium tone is the out- standing trait of Nahuatl accentuation. Denoted a, . See above examples. Where I have heard the height or depth of the high and low tones mark- edly exaggerated I have recorded a double accent mark a', o. M. tcawe'! (for tiyawi') "iVamonos!" PITCH TONE AND THE "SALTILLO" IN NAHUATL 175

Saltillo: All early writers describe the "saltillo" as an accentual phenome- non, and some refer to the holding of the breath that accompanies it. It has nothing to do with tone however, and the true or strongly pronounced saltillo is more than a simple glottal stop as some have described it. This true or strong saltillo always occurs after a vowel and before a consonant. The vowel is pronounced with considerable force of breath. A sudden glottal closure cuts short the vowel before the chest has fully contracted and for a moment, about the duration of a single stop consonant, the breath is pent up behind the closure. Then the glottis opens and the chest completes its contraction, expelling the remaining breath as a sort of h, resembling a pant, which is cut off or broken into by the next consonant, and then comes the next expulsion of breath on the next vowel.20 This sound can be designated with fair accuracy as 'h. M. tla'htla 'it blazes.' nota'htsiN 'my father' o-ki'ht6'k&''dijeron' 'they said' tepo'hco'hflN 'los botes' 'the tubs' o.kinti'htlanilf'kl' tl'ros'dispararon unos tiros' 'they discharged some shots.' The early grammarians say that the saltillo is placed on the final vowels of plural endings and of the preterite tense. What is heard in these posi- tions to-day [sic] is either a simple glottal stop and usually a pretty faint one, or nothing. If however the next word after such a vowel-ending be- gins with a vowel, an ordinary h (as in Eng[lish]) is usually prefixed to it. T. ok'wikilihkEh-i-tlak 'le llevaron junto a el' 'they took him over to him' The saltillo is sometimes inserted into a foreign word. M. kiteikilla ha'hseite blarko 'le echa aceite blanco' 'one pours for her white oil'21 But I also heard from the same informant ha-seite.

In addition to the saltillo, Nahuatl talk frequently inserts faint glottal catches or light hesitations after certain vowels of a word. The usage may vary with respect to the same word with the same individual. Unlike the saltillo, these do not pre-aspirate the next consonant. I record these slight

20 This is a good description of the production of the saltillo and its variation in pronun- ciation. It is essentially a glottal stop in these dialects. However, final stops have aspirated allophones (a widespread phenomenon found in many Mesoamerican Indian languages), and this includes glottal stop: [], [kh], and ['h]. (Cf. Lastra de Suarez 1986.) 21 From Spanish aceite 'oil'. 176 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS checks as simple glottal stops (see above instances) or when they sound more like faint anticipated closure upon the consonant I record itta, okka, etc. instead of i'ta, o'ka. The uses of the tones and the saltillo in the Milpa Alta dialect may next be observed from the following text. M. Pwes ona'hsito ipam pwebllto Pues llegue en un pueblito Then I arrived at a village i-to-ka lallhs. llamado La Luz. called La Luz. onikitt'aksentetl teopaNtrl vide una capilla22 I saw a chapel wan ontc'htc-mo.lyAya y le buscaba and he asked about it23 fe.tca 68 antes la fecha de los antiguos the ancient date i'hqwak okitci'htc^lckE (this tone is anomalous) cuando lo hicieron when they made it pero amo nikne'ctil'k. pero no le encontre. but I did not find it. sani-ma ompa o-walaya luego venia soon yonder there was coming

22 In colloquial Spanish, archaic vide 'I saw' occurs in many rural dialects of Amer- ica in place of standardvi. 23 A note in the margin indicates Whorf's intention to add a footnote: As footnote * (i.e., I, the writer, asked the informant). PITCH TONE AND THE "SALTILLO" IN NAHUATL 177 s_nt/tl tlak'atl un hombre a man wan o-titla'htlani'hke' y le preguntamos and we asked him, "amo .titctcmaka rason ",No nos das raz6n "Can't you give us information tlaon ciwitl o.motci'htcilwi en que ano se fund6 about what year was made ini tsopaNtll?" esta capilla?" this chapel?" wan oki'hto inon tlakatl y nos dijo ese hombre and said that man

6c ompa ftca, de alla del pueblo, from that place his home,

"pos, amo nimitsmaka rasbn "iPues, no le doy raz6n "Why, I can't give you the information porkEamo nikpilia qwiab6o porque no le he tenido cuidado because I have no interest in it pero nicmati kdakali'htik pero se que esta adentro but I know that it (the date) is inside ka sFntstl qwadro, pero cka-wa! con un cuadro, pero deje! on a tablet, but wait!24 24 A closer Spanish translation would be 'en un cuadro, pero deje'; English 'in a tablet, but stop!'. Since the Nahuatl verb seems to be transitive, it probably should be glossed 'leave it, drop it!'. 178 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

ni'tla'htlanis i.ntlaqWE cnkargaos preguntare con los mayordomos I will inquire among the caretakers

6e ino teo.kalli de ese templo of that temple

wan nimits'iLhWis y te lo he de decir and I will tell you

qwa-k ohss by_hc tiwa'las." (viaje) cuando otra vez vengas." when you come again."

86 ino pweblito De ese pueblito From that village

san'ima o-nitla-ictimbk luego me baje soon I went down i.tFtc sEntFtlwe.i tla'htEmantli i. ctEtc en la superficie de un empedrado grande over a stony road (lit. over a big expanse of stones) kampa ka tl_'hkb pwsblo tla-kia. por donde suben los del pueblo. the way where the pueblo people go up.

sani-ma o.nikitskik luego tome Soon I took

ohscnt6tl o.'hpitsa'htli, otra vererra [=vereda], another narrow road, PITCH TONE AND THE "SALTILLO" IN NAHUATL 179 sani-ma o-nitlc'htl&'hkbk en seguida me subi en una cerca and then I went around up para onikitskik para tomar so that I took ohs'ntetl qwE'hqwFntlah'b.'htli otra vererra [=vereda] de milpa another road through the fields para onikisato (or onikisato) para ir a salir in order to come out i-pan ohs~ntEtl -.'htli para otro camino upon another road pero ws'hkap'a pero estaba [=estuvo] muy alto but it was very high kampa ka o-nitsmoskia. por donde iba yo a bajar. for where I would be going down. entonsEsonitmbok entonces me baje Then I went down ka itqjko snntetl mpa_mitl; por la orilla de una inglera de magueyes; by the edge of a row of magueys; san-ma onikisaski.a luego iba yo a salir soon I should be coming out (on the main road) pero no-i.hki o-tstsauhWt'aya pero tambien estaba cerrado but also closed was 180 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS ino kaLte-mitl ka tEtl. ese portillo con piedras. that gateway with stones. san1-maonika bo'htli luego agarre otro camino soon I reached [=took] the road para pwEblito i-to-ka san agustin para el pueblito llamado San Agustin to the village called San Agustin kampa o-nikimfttak(or o-nikimittak) en donde los vide [=vi] where I saw mi.yak kaLto'htbn muchas casitas, many little houses, wan ka ik'ampa tohtitikdat y atras estaban sembrados and behind these were (lit. are) being sown kalw'a-samiltin; muchas milpas de chicharos; fields of peas; se-ki ys kipla kalwa-sa unos que ya tenian el fruto some already have peas wan seki apenas co.tciqwepontika. y otros apenas estaban en flor. and some are just bursting into bloom. wan i-pan i-pa.tyo ino tsokalli* y en el patio 'esa [=de esa] capilla And in the courtyard of the chapel there25

25 Added later above the line (* as footnote): *(This phrase was spoken rapidly and in rather a monotone, so that I did not distinctly hear pitch differences. Normally these words would take the tones ipan, inb, tsokdlli.) PITCHTONE AND THE "SALTILLO"IN NAHUATL 181 okimpyaya mi-yak qwauhWti.N tenia muchas [=muchos] arboles it was having (there were) many trees

6e tlatska. de cedro. of cedar. (Cupressus benthami). itsintla'hyb qwauhWt~ Abajo de esos arboles Under the trees o-totla'htlalf'hkc' nos sentamos we sat down para o-ti'tla'hqwilo'hks' para escribir to write (lit. for that we wrote) wan tosskatitla y en un lado de nosotros and on one side of us o-motla'htlali'hk' se sentaron sat down na_wi ko-k6ne'. cuatros nifos. four children. o-k'wi.kdya o-me me-t6tbn Llevaban dos borrejos [=borregos] They were leading two sheep wan otikintla'htla.ni'hk? y les preguntamos, and we asked them,

"ka-mpan a-nyawE'?" [",]adonde van[?"] "where are you going?" (nan, written n an, is for classical [Nahuatl] in an, one of the few survivals of the classical [Nahuatl] particle in) 182 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS wan otcetcnankili'hke' y nos respondieron and they answered us

"tiya'w' ti'qwa'hqwawisks." "Vamos a lefiar." "We are going to gather wood." ([The] tone of [the] last word [is] anomalous). sani-ma ohsp'a onika o.'htli Luego tome el camino Soon I reached [=took] the road again, ka tlacitla. por abajo. going down. oni.kqws'tcaLhWik Lo pase I went over26 sentetl pwentc 68 qwawltl un puente de madera a wooden bridge o-yo.yot'aya. que estaba atravesado. that intervened. sanima ona.'hsfhtb ipa.n barrio Luego llegue en el barrio Soon I arrived in the parish [= district/ward of town] i-to-ka lakonsepsi'tsin. llamado La Concepci6n. called La Concepcion.

26 Added above the line (as footnote): *(An interesting survival of an old meaning of *qwetca > vb. qwetcawa > applicative qwet- caLhWia; qwetca = "go over" > qwetcawa "flood" and "relentecer o humedecer algo"- Molina [.]) PITCHTONE AND THE "SALTILLO"IN NAHUATL 183 sani-ma ona'hsi'hto Luego llegue And soon after I arrived i-panbn kaLt'.mitl en ese patio there in the yard

8E _i.noteokalli. de esa capilla. of the chapel there. onikittak miyak t_.mc' Los vide muchas piedras [=vi muchas piedras] I saw many stones mo'hmontojka para ka^kisnjkaw'aske amontonadas para acabar de construir piled up in order to finish [building] inMtEo-kalli ese templo that temple porke ko-sa tsi-tsikitsi"ka. porque estaba muy reducido. because it is rather dilapidated. (lit. "reduced") wan a-caNbarrio tla-ka Y ahora los del barrio And now the parishioners yo.kipensaro'hkgkiw_yillskc' ya pensaron agrandarlo, have decided to enlarge it, porke ayohmb kimpa-tia porque ya no les parece because it no longer looks to them (lit. fits, corresponds or is worth to them) [= 'they don't like it'] kc okitci'htci^hke como lo hicieron, they way they made it, 184 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS atcit6N tla'htiwanimc'. los antecesores. the original authorities.

Discussion of Tones in the M Dialect

The above text is a fair sample of most of the tone phenomena encoun- tered in a much more extensive body of tonally recorded texts collected by me in the M dialect. The results of a study of this whole body of texts will next be summed up. This discussion of tones will be subdivided ac- cording to the following general types of tone which are found in these texts and also in conversational talk.

1. typical tone (where the tone depends on the phonetic type of the word, that is, goes with certain arrangementsof vowels and consonants). 2. special tone (where typical tone seems to be varied according to some special condition, such as a desired emphasis or an effect of rhetoric). 3. morphological tone (where the tone indicates the grammatical nature of the word). 4. semantic tone (where the tone is essential to meaning).27 1. typical tone. a. The medium tone. This is the tone of the majority of the syllables in the stream of talk. The syllables tend to run in successions presenting a uniform level of the medium tone, interruptedby occurrences of the other tones chiefly on the last two syllables of importantwords. Generally all of a verbal complex up to the final two syllables is in medium tone. Mono- syllables with rare exceptions are in medium tone. Many connective words though of more than one syllable have only medium tone as they would normally be pronounced in the midst of a sentence. Ex[ample:] ka- mpa o-nikimittak['where I saw them'] b. The slightly raised tone. This occurs regularlyon the next to the last syl- lable of words ending in a vowel, with certain exceptions to be studied.28

27 As pointed out above, Nahuatl in actuality has/had no contrastive tone. Thus, Whorf's "typical type"-where the tone depends on the phonetic shape of the word-would seem to reduce to Nahuatl's penultimate stress pattern. His "special tone"-used for emphasis or rhe- torical effect-would seem best treated as part of intonation. The third and fourth types turn out to correspondto various permutationsof vowel length, stress, and glottal stop (see below). 28 This is apparentlyWhorfs perception of penultimate stress in open syllables. PITCHTONE AND THE "SALTILLO"IN NAHUATL 185

Ex[amples:] cka.wa ['drop it, leave it!'] o-motci'htci-wi [o--mo-ci'-ci.wi (PAST-REFLEXIVE- REDUPLICATION-make)'was made']29 i.to.ka ['his/her/its name'] nimitsmaka ['I give it to you'] nikpilla ['I have it'] o-walaya. ['(he/she/it) was coming'] c. The high tone. This occurs regularly: (1) on the penult of words ending in consonants. sentEtlor s6ntetl ['one'] tla-katl ['person'] tiwa-las ['you will come'] o-nikimittak[o.-ni-kim-itta-k (PAST-I-THEM-see-PAST) 'I saw them'] mi-yak ['many, much'] kaLto'htbn [kal-to'-to.n (house-REDUPLICATION/PLURAL- DEROGATORY) 'little houses']30

(2) M tends to pronounce final n very lightly, sometimes as barely audi- ble n (voiced), sometimes as N (unvoiced), sometimes as nasal breath or nasalizing of [the] vowel, while sometimes it disappears entirely. Never- theless these cases, even the last, are regularly toned as in (1).

M Form Classical [Nahuatl] min inin ['this'] ino inon ['that'] onika "agarre" onikan (o-ni-k-an) v[er]b ana ['I grabbed it'] mlltin miltin ['fields'; cf. mirl- 'field'] tosckatitla tosekatitlan ['on one side of us'] itca ichan (i-chan) ['his/her/its home'] From another M text I have i.paN ['on it'], i'hwaN [i.-wa.-n ? with him/her/it'], etc.

(3) on the penult of words ending in a glottal stop, or words whose final vowel takes the saltillo in classical Aztec, even when the glottal stop is absent in M.

29 The initial o-- of this example (and of the last in this list), which we have labeled "past," may occur with preterit, imperfect, and even future forms to indicate an event after some other event; some grammariansprefer to call it the "antecessive." 30 Other varieties of Nahuatl do not have "saltillo" in such forms; this makes us wonder whether Whorf heard correctly here. 186 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

o.titla'htlani'hke' ['we asked him/her'] tohtitikata ['were being sown' (?)] o.motla'htla.lihkk ['they sat down'] ko-ko'n' ['children'] itsintla'hyb ['under it'] From other texts I have plurals of this toning, corresponding to classical [Nahuatl] plurals in -me', -ke' o.'htldmE'"otates" 'bamboos' But in this one t-.me' is out of line.

(4) on the penult of verbs having suffixed -to and -ka ona'hsito ['I arrived';literally: 'I went to arrive there'; o*-n- a'si-to 'PAST-I-arrive-PURPOSIVE'] co-tciqw?pontika ['is bursting into bloom'] (5) on the penult of nouns in -lli tco.kdlli ['temple'] From other M texts I have numerous other instances, e.g. to-nalli 'day' ma.qwilli bora'five hours' d. The low tone. This occurs regularly on the last syllable of a word whose penult has the high tone. See the above instances of the high tone. Rarely the drop to low tone is deferred till the next word (see above); when this occurs the intervening syllable remains high. Occasionally the drop from high is only to medium tone. 2. Special tone. Occasionally the normal toning as outlined above is al- tered, apparentlyoften in the interest of emphasis or rhetorical expression. a. Negatives. Emphatic negative words of which the first element is the negative particle a- usually take high tone on this a-. In the above text there is one case- ay6hmb 'no longer'. From other M texts I note, e.g. antle-yi tomi-n 'no hay dinero' 'there is no money' ayMmo 'not yet!' (in reply to question). aybhmo 'ya no' 'no longer!' aw11i 'no pudieron' '(they) could not' a.ihwlli 'ya no pudieron' '(they) could no longer'. b. Thanking. The expression "thank you!" is usually tlaso'hka'-matior tlaso'hka'-matior tlaso'hka.mafi c. Impressiveness. etc. To give impressive emphasis the low tone is sometimes anomalously used, especially for ideas of depth or height. PITCH TONE AND THE "SALTILLO" IN NAHUATL 187 wchk_apa 'muy alto' 'very high' Also heard: whhka! 'far off!' wcli-hki! 'harder!' tc'hwatsfn n6o-hk 'you too!' d. Questions & Misc. In questions the first syllable or first few syllables of the sentence are in slightly raised tone and the last syllable [is] also usually slightly raised without [a] following drop. If the last syllable would normally be low it often is low and rises at the very last. Failure of the normal drop to low to occur after a high penult seems to be often because the word is being uttered quickly and unemphatically. There are also of course occasional anomalies of tone for which I can give no explanation.

3. Morphological Tone. The above text does not happen to contain the proper verb forms for showing the use of tones as part of the grammatical apparatus, yet the functioning of tones as such is not uncommon, occur- ring in the following ways. a. In the present and imperfect tenses of the verbs, classical Aztec distin- guishes the plural from the singular by a saltillo on the final vowel. In M where even a faint glottal stop to represent the saltillo may have disap- peared, the difference in tone pattern between a vowel-ending and a con- sonant-ending word may remain the only distinctions. ki'hqwilOa 'he[/she] writes it' ki'hqwiloa 'they write it' [cf. Classical Nahuatl ki-'kwiloa-' 'it-write-PL'] otikisaya 'you were leaving' otikisaya 'we were leaving' [cf. Classical Nahuatl o.-ti-ki.sa-ya-' 'PAST-we-leave-IMPERFECTIVE-PL'] (In Aztec "we" is the formal plural of "thou", not of "I").31

Tone alone would distinguish the continuative present in -ti-ka of verbs from adverbials in -ti-ka of certain nouns from the same stem. kaLtika 'he is making a house' kaLtika 'with (or by) houses' toktika 'he is sowing' toktika 'with (sown) maize' nokcnt'ka 'I am clothing myself' noksntika 'with my garment'

31 Here Whorf is reacting to the formal similarity between the markers for 'you singular' (ti-) and 'we' (ti-... -'), where the final glottal stop marks 'plural'. 188 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

4. Semantic Tone. The most common case is ka '(he, she, it) is' (Spanish estar) [cf. Classical Nahuatl ka. stem, ka' 'he/she/it is'] ka prep[osition] 'with, by', conj[unction] 'that' [cf. Classical Nahuatl -ka 'with, by (means of)', ka 'conjunction'] Of the first, we have in our text nicmati ks ka kalfhtik se que esta adentro I know that it is inside. ko-sa tsi-tsikitsinka. (muy) (reducido) (estaba) it is very dilapidated (is used for was) A very common expression is qwalli ka! 'estg bueno' 'it is good! (=all right!)' Sometimes ka is enclitic as regards stress accent; qwallika. I was cor- rected when I said qwalllka.32 In the first and second persons nika ['I am'], tika ['you are'] are generally heard. The plural, kats, has typical tone. The other ka is also ka ('), meaning that it is regularly followed by a ten- dency to depress the tone, so that the accented syllable of the next impor- tant word has a lower tone than the normal. Our text shows the following examples: Expressions with ka () Typical tone would be ka scntetl qwadro sentetl ['one'] ['with a tablet, picture'] ka tl'hkb pweblo tl'aka tl^'hko pweblo tla-ka ['by (where) the townspeople ascend'] ['townspeople ascend'] ka itejko ['by the edge of'] iterko ['the edge of'] ka tEtl ['with stone(s)'] tetl ['stone'] ka i.kampa i.kampa ['behind (it)'] ['throughbehind it' (por atras)'] I have recorded one case where the tone drop does not occur: ka tlacitla tlacitla(n) [in Classical Nahuatl this form meant 'room in palace where nobles and leaders heard people's petitions and resolved criminal cases']

32 That is, Whorf should have stressed this as required by its two constituent words, kwdlli ka [good is] 'it's OK'; his pronunciation as a single word with typical penultimate stress, *kwallika, was not accepted. PITCH TONE AND THE "SALTILLO" IN NAHUATL 189

The word sani-ma in my texts when toned anomalously san'lma, san'ima or more rarely sani-ma, is always translated "luego"-'soon', or 'immedi- ately'. When toned typically sanl-ma, it is translated "en seguida"-'next'. The present text contains only one example of sanl-ma, but other texts [ex- hibit] numerous ones. The other cases of semantic tone that I noted are: patla 'change, exchange' patla 'knead, stir, mix up' tcitci or tcitci 'dog' titci33 'nurse, suck (baby, subject)' t6ka 'follow' toka 'sow (seed)'34

Discussion of the Saltillo in the M Dialect

The saltillo occurs 1. After the first of two reduplicated syllables in reduplicated verbs. o-motla'tlalf'hke35 verb tlalia ['sit down'] red[uplicated] tla'htlalia okitci'htclc6kE verb tciwa ['to do, make'] red. tci'htciwa o-nitlc'htlh'hkbk verb tl'hko ['to ascend'] red. tlc'htle'hko ti'qwa'hqwawisks verb qwawia ['to chew'] red. qwa'hqwawia 2. Similarly [saltillo occurs] with nouns reduplicated to express plurality or generality, but with numerous exceptions. qwe'hqwFntla'fields, country' qw6ntla 'field' but ko.k6ni' 'children' k6ni-tl 'child' Note: Where a reduplication is part of the stem there is usually no saltillo. M. kokoa 'it hurts'

33 Whorf mistakenly wrote (titci) where tcitci is expected. 34 It is instructive to compare Whorf's forms with the Classical Nahuatl equivalents: Whorf's forms Classical Nahuatl pdtla 'change, exchange' [patla] pitla 'knead, stir, mix up' [pa-tla (?)] tcitci or tcitci 'dog' [cici] t[c]itci 'nurse, suck' [ci-ci.] t6ka 'follow' [toka] toka 'sow (seed)' [to-ka] 35 Whorf gives o-motla'tla-l'hkl here, where from the way he represents other forms we might expect o.motla'htla-lf'hke(i.e., not .. 'C, but .. 'hC). 190 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS

3. [Saltillo appears] after prothetic i-, except when i- is followed by 1, and the saltillo remains after i- has been eclipsed by the vowel of a prefix.36 okf'hto [o.-ki-'to 'he said it'] verb i'htoa < *to [actually (i)'to-] tla'htiwanimE' tla'htiwani = tla'htoani < i'htoa ni'tla'htlan's [ni-'tla'tlani-s 'I will inquire']37 verb tla-'htlania < vb. i'htlania < *tla o-ti'tla'hqwilo'hk&'[o.-ti-tla-'kwilo'-ke-' PAST-we- UNSPECIFIED OBJECT-write-PAST-PLURAL'we wrote (something)'] tla- + v[er]b i'hqwiloa < *qwilo [actually *(i)hkwVlo-] Note: Pronominal i- (i.to-ka, itca, i-pan) does not take saltillo except possibly in i'hqwak [i'kwa.k 'when'], fhwan [i.-wa.n 'with him']; it is a question whether i- in these words is pronominal or prothetic.38 4. [Saltillo occurs] before -ke of past tense plural. o-titla'htlani'hkc' ots-tcnankili'hk&'

5. Let y = any final vowel of a verbal stem. Classical Aztec shows groups each of practically synonymous and interchangeable verbs from the same stem, of the forms y-wa, y-qwa, y-ka, y-na, and corresponding to these verbs may be one or more nouns of the forms y-oh-tli, y-cu-tli (y-qw-tli), y-k-tli, y-n-tli; with y-oh-tli and y-k-tli preferred, and not necessarily fol- lowing the consonantal suffix of the verb. In M such nouns tend to be of one form, y-'h-tli. pitsa'htli 'contracted, narrow' Class[ical Aztec] pitsauhtli and pitsaktli pitsawa 'to contract, narrow, make thin' < *pitsa 6. In the stem of words the saltillo occurs as in classical Aztec and gen- erally must be learned. o'htli 'road' tl'hko 'ascend'

36 Actually, the saltillo is not added because of prothetic i, but rather is part of the root in such cases. Historically, certain C1VC2V roots lost the first V (C1C2V) and then later pro- thetic i was added (iC1C2V) (e.g., *kasi > ksi > iksi 'foot'); the prothetic vowel does not ap- pear when the root is not word-initial (e.g., no-ksi 'my-foot'). When the second consonant was one which eventually became h (and changed further to glottal stop in dialects such as the principal ones recorded as Classical Nahuatl), as syllable-final w did, the two consonants metathesized (e.g., PUA **tiwV > two > (i)two > (i)wto > (i)hto- [ > (i)'to- 'to say, tell']; contrast tla-htoaltla-?toa 'he says something' [tla- 'unspecified object marker']). (For details, see Campbell and Langacker 1978.) 37 Actually, there should be no glottal stop between the subject prefix ni- and the verb stem tla'tlani-. It seems to be an error on Whorf's part. However, if the verb is transitive tla'- tlania, then Whorf's r/ would be the phonetic realization of the object prefix which otherwise is k-, and not a "saltillo" at all. In this case, the gloss should be 'I will interrogate him/her'. 38 Whorf apparentlyconfused matters here; the initial i of i-wa-n has no "saltillo." PITCH TONE AND THE "SALTILLO" IN NAHUATL 191

Final Saltillo. The final saltillo required by classical [Nahuatl] grammar after all vowel-ending plurals, preterites, imperatives in -ti and -ki, and adjectival derivatives in -wa and -yo, is in M either lacking or represented by ('), a faint glottal stop, sometimes with h- or h- prefixed to an initial vowel following in the next word.

Text in the T Dialect

We may now compare the somewhat different occurrence of tones in the T Dialect from the following sample of texts secured at Tepoztlan. scnts to6tctli o.ne.m.-ya Un conejo andaba A rabbit was going along

I.pan irtl_awatl. en el campo. in the country. iniN to-tctli o-kina_mik Este conejo se encontro This rabbit met sents tlaka&tltla'mo.tla'n [a] un hombre tirador a man going shooting ika it?pbs. con su escopeta. with his gun. (lit. [with] his "iron")39 inin totctli -.wanon tlak'atl Este conejo y ese hombre This rabbit and that man o.p_.'hke mo-no-no.tsa ompa. comenzaron a platicar alla. began to converse there. yF'h-kima.ihtaLhWlatla.yo-lli Le ofrece maiz The farmer offers him maize

39 Spanish escopeta means 'shotgun'; Nahuatl tepos means 'iron, ax', and apparently by extension also 'gun' in this dialect-hence Whorf's literal gloss with 'iron'. 192 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

6e tlkn okit6okak de lo que sembr6 from what he had sowed

_imiL totctli. en su milpa del conejo. in the rabbit's field. iwan tlamo6tlak Y el tirador And the hunter o.klLhWi t6otctli le dijo al conejo, said to the rabbit,

"kE*kitct1otikma.i'htohwa ",qu6 tanto ofreces "for how much are you offering motlayb-L?" o.ki^'htoto-tctli[,] de tu maiz?" Dijo el conejo[,] your maize?" Said the rabbit[,]

"ma.tlaktli kairga "Diez cargas "Ten loads y_'hysi p?-so ka.rga. a tres pesos [la] carga. at three pesos a load. ri-ktci-wa qwe-nta- Haz tu cuenta- Make your reckoning- ke.kitc tine-tcmakas?" ,cuanto me has de dar?" how much will you give me?" inl'.tlamo-tlak okP.hto Este tirador dijo, This hunter said, PITCH TONE AND THE "SALTILLO" IN NAHUATL 193

"nimi.tsmka&s "he de dar "I will give you scmpwalli wan_ma-tlaktli p?-so." treinta pesos." thirty pesos." (lit. twenty and ten) iniN to-tctli o-kise-li tomiN Este conejo recibi6 [el] dinero This rabbit received the money i-wa.n oklLhwi tlamo.tlak[,] y le dijo al tirador[,] and said to the hunter[,]

"tikse'-is tla-yo.lli "has de recibir el maiz "you will receive the maize i-maN ni.pfrkas." cuando he de cosechar." went I harvest." o.qkaN o-m?nawatl-'kF Asi se avisaron, Thus they took council, tlamo'tlak iwan to.tctli, el tirador y el conejo, the hunter and the rabbit, jokan o.ya.hk&i.paN iNtcahtcN. despues se fueron en sus casas. then they went to their homes. iniN to-tctli okipe-waLti Este conejo le comenz6 This rabbit began

.-miLtla'yb'Lk6oni-tl kl.'hqwa; su milpa la planta chica a comer; to eat the young maize plants in his field; 194 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS a'hmb okikaN mo.ws'paw'astlayo.lli, no lo dej6 crecer el maiz, he did not let the maize grow, san ni-ka o-kiqwa'htaiya. no mas seguia comiendo. but still kept eating. o.-ahsk pirkai.tla; Lleg6 la cosecha; The harvest time arrived; a.mitl' n-.si 6e m'.lli pirkistli. no [a]pareci6 nada de milpa para cosechar. nothing appeared from the field for harvest. o-walla tlamo'tlak Vino el tirador The hunter came ye_ki'htlanil.hko tlay6olli a pedirle el maiz to ask him for the maize

56 tlEn o-kitlartla'-wi.i-nin to-tctli. de lo que ofreci6 este conejo. for which he had paid this rabbit. (The Spanish here does not seem to be a literal translation.) ayo-kmo o.kimati-a tlci ki-'htbs Ya no sabia que decir He no longer knew what to say ya-N am'itlatlayo.lli porque no hay nada de maiz- because there was no maize at all- inin tlamo6tlak. este tirador. this hunter. no-ka kino-no-tst'ka Mientras que estaba platicando While he was talking PITCH TONE AND THE "SALTILLO" IN NAHUATL 195 oqka-N to6tctli o-kistikis, tcolo'w_a, este conejo sali6 corriendo right there the rabbit darted away on the run, ojka.n tlamo-tlak kit6k'a. de ahi el tirador lo sigue. right there the hunter chases him. okipo'lo'h. Lo perdi6. He lost him. asta airan kitcmo'htine-mi. Hasta ahora lo anda buscando. Until now he is going searching for him.

In order to give a wider view of T tones and saltillo I shall add a frag- ment from another tale.

.fiN so_w'a-tl katc iqwalli o.ka-m piLtsintli; Esta mujer mejor lo cogio el nino; This woman took a better hold on the baby; o-ks_ppa o.k'wi-kfli i.na-na. otra vez l llev6 su mama. again she brought it to its mother. tler kitclwa i.na.na.? Lo que hace su mama[?] What does its mother do? oki'kbo0s_ntc ka.ka-ra Compr6 una cajita She bought a box l-tik okitsahtsak piLtsl.ntli, adentro lo encerr6 al niiio, in which she shut the baby; orkaN bksEppa.kiLhWia inon so'.*'atl, y otra vez le dice a esa mujer, thereupon again she says to that woman, 196 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

"arkitta! tsi-kdmme am^itlaokitciwilihkc! "Mira! Las hormigas no le hicieron nada! "See! The ants did nothing to it! ia-ran shiwia sriktlakalfhti Ahora vete! Anda, tiralo Now go! Go throw it k_ani-rqWEyi a.mandlll endonde es grande pozo; [=en donde es grande el pozo] where the big water reservoir is; [=where the water-reservoir is big] inscpantla sriktla.kdli inir kaka-ra. en medio lo tiras esta cajita. [=en medio la tiras (a) esta cajita] into the middle of it throw this box. ompa mfkkis piLts_.ntli. Alla se muere el nino. There the baby will die. l.ni] ka.kai-ra Esta cajita This box irko ya.tinemla inoa a.mandlli. andaba encima de la agua de ese pozo. [de la = del] went floating on the surface of that reservoir. o-moka-wd'htbi.tcs.ko a-manalli. Fue a quedar en la orilla del pozo. It came to rest at the edge of the reservoir. oqk'a-noki-sahtb se?nts. tla'-katl. Despues fue a salir un hombre. At that time a man was about to go forth. o.kfttak i.nborkaka-ra, Vi6 a esa cajita, He saw that box, o.mo.'patcb '-tlak, se arrim6junto a el [=ella], he approached close to it, PITCH TONE AND THE "SALTILLO" IN NAHUATL 197 o.kiktWif. la sac6. he drew it out.

1-ma'nyl_o?kik"ilrti. [Tan] luego que la sac6 As soon as he had drawn it out o.kitlahpboinou ka*ka*ira. la abri6 esa cajita. [=abrio esa cajita.] he opened that box. i-yEkitsfi;'1.htik,ka' sFntF piLtsi-ntli, Lo va mirando; adentro esta un nino, He goes looking; inside is a baby, i*wa*nn6hsb pi*p'LtoQntli. y es muchachito. and a little boy too. [=and it is a little boy, too.]

-jka.N tl&Nkitci*wa 'Ini*Ntla*katl Este [=esto] lo que hace este hombre; Then what does this man do but [=Then what this man does is] ok'wilkak i1'tcain. lo llev6 en su casa. bring it home. - ka.N kiLhWia i-so'"*a.n Luego le dice [a] su mujer, There he says to his wife,

66 vivitt' I "i Mira! "Look! o-nonohphp nni snts piLtsj.ntfl Me encontre [a] un nifio I found a baby ompa. kdni.1)toNto'pahpa.kka. alla en donde nos vamos a bafiar. there where we go bathing. 198 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS a.ran40 tik'wEppa-waskFl.n' konnF-tl; Ahora lo hemos de criar [a] esta ninio; Now we shall rear this child; to_.tl y_o-tstctitla.-nil-hkE. Dios ya nos lo mand6. God has sent him to us. i.niq piLtsl.ntli o-mowsPpa't'a; Este niiio se fue criando; This baby kept on growing; o-ka-HiLti tcikko.ms sr1iwitl; ajusto siete afios; he reached seven years; ifniJ pipiLto.ntli ihi-wiya.n. este muchachito ligero viv6 [sic]. this little boy went along merrily.

Discussion of Tones in the T Dialect

1. Typical tone. Except in the case of certain grammarforms, the kind of raised tone used on the penult of a word seems to be determined not by whether the word ends in a vowel or consonant as in M, but by whether the penult vowel is short or long. If the penult vowel is long its tone is slightly raised, even though the word ends with a consonant; 1.pan irtla-watl ['in the countryside'] .wan ['with him/her/it'] a.ran ['now'] I miL ['his/her/its field'] -or is vowel-ending of the type taking high tone in M. T. tlayo-lli M. tlaylli ['corn'] T. pirka-tla M. pickatlia (class[ical Aztec] -tlan) [cf. piska 'to pick, harvest'] If the penult vowel is short it very generally takes high tone even though the word ends in a vowel. When the penult takes high [tone] the ultima takes low tone.

40 In Whorf's manuscript there is an r in this form; given that other occurrences of this word have r, probably he intended that here as well. PITCHTONE AND THE "SALTILLO"IN NAHUATL 199 tla'mo.tlani ['shooter'] tcolo'w'a ['flee'] tsi-kamm~ ['ants'] We find both ami-tla and amitl'a['nothing']. Often the vowel length is such as to give the same tones as in M; inin ['this'] to-tctli ['rabbit'] a-mandlli ['pool of water'] But again it is such as to give different tones as above. In s6o'w*'atl ['woman'] and tlaka'tl ['man, person'] the penult vowels may be etymo- logically short, though the second vowel of the first-named is etymologi- cally long [=*siwa.-tl] and that of the second etymologically short [=*tlaka-tl]. Grammar Forms. However in certain grammatical forms we find a quite different state of things, namely the M type of tone-usage, and this condition constitutes a connecting link between typical and morphologi- cal tone. Preterites and futures take the high-low tone pattern as in M (though the penult vowel may be long and the final saltillo or glottal stop of the pret- erite quite lacking) except where they bear the plural suffix -ks, and then a long penult produces the slightly-raised tone pattern as usual in T. preterites o-kinam'ik ['he/she encountered him/her/it'] o-kise-li ['he/she received it'] o-wall ['he/she came'] o-kitlaitla'wi ['he/she paid it'] futures tins-tcmdkas ['you will give it to me'] tiks.l'is ['you will receive it'] kf.'htbs ['he/she will say it'] tik'w_ppa-waskS[cf. Classical Nahuatl ti-k-kwepa-ske-' (we-him/her/it- return-FUT-PL)'we will return him/her/it'] But, with a long penult vowel before -ke: o.p_.hkc ['they began' (root pe.wa)] o-m?nawati.'hke[o.-mo-na-wati'-ke-' (PAST-REFLEXIVE-advise- PAST-PL) 'they sent/advised/ordered themselves'] o.ya.hk ['they went'] tlam6otlak,which alternates with tlamo-tlak, has the grammatical form of a verb preterite, though used as a substantive-"he who shot". 200 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS

The preterite o-kito-kak 'he sowed it', and the present kitoka 'he follows him', are of course cases of fixed semantic tone. The association of the high-low tone patterns (' ') with the preterite is marked in T, and when we find the present (more properly the indefinite tense) used in a dependent clause after an independent clause in the pret- erite (and then often translated by our infinitive or participle) we find this present taking the tone pattern characteristic of a preterite. o-kistikis tcolo'wa 'he darted away running' In the following case the preterite pattern is repeated on every word of the dependent clause, though it places the high tone on the long vowel of the nouns: o.kipe,,waLtif'imiL tlayb-L-ko.n-.tl kl.'hqwa 'he began to eat the young maize in his field.' Presents (true present tense) and imperfects however, regularly take the same tone pattern as in M, that is slightly raised on [the] penult, and [they] often make an etymologically short penult long so that this tone pattern conforms to [this] type. Thus, according with phonetic type [we have]: o-nc.ml.ya (imperfect) M. o.ne.m aya['he/she/it was living'] o-kiqwa'hta'ya (imperfect) M. o-kiqwat'aya ['he/she/it was going along eating'] class[ical Aztec] okiqwatiaya41 We find both kitcl.wa and kitciwa (present) ['he/she/it makes/does it'.] While differing from phonetic type and still in accord with grammartype and with M [we have]: kima.ihtaLhWla (present) kiLhwla (present) ['he/she says it'] ya.tinsmia (imperfect) [cf. nemi-ya 'was living'] The imperative follows [the] phonetic type, but whether the penult vowel is short or long seems to be optional with the speaker; in sharp, quickly spoken commands it is usually short.

41 There appears to be a misunderstandinghere. The forms Whorf lists are not parallel to the Classical Aztec form he lists. That is, these are not based on kwa. 'to eat' + -tia 'caus- ative'. The form (quatiaz.nic) 'yo yre comiendo' ['I will go eating'] in Molina's (1571) dictio- nary is a compound verb, kwa?-ti-ya- [eat-LIGATURE-go-FUT]. Whorf's Q-kiqwa'hta-yais an imperfect of this, not the simple imperfect of kwa-tia. Also, it is to be assumed that Whorf's M. form, o-kiqwat'aya,whatever its actual phonetic realization in that dialect, is structurallyas the form in the other dialect, o-kiqwa'hta-ya,that is, where the "saltillo" (') belongs with the root kwa and does not occur after the following t but before it. PITCHTONE AND THE "SALTILLO"IN NAHUATL 201 arkfitta!['look at him/her/it!']42 sriktla.kdli! ['throw it!'] sriktlakalihti! ['go throw it!' (i.e., 'go for the purpose of throwing it')] While in a more languid style of speech it is drawled- riktc^lwaqwe-nta [literally 'make count!', probably a calque on Spanish dar cuenta 'to notice, pay attention to, be aware of']. ia-ranshiwia ['now go!']. There seems to be also a grammar form taking precedence over phonetic type in the case of monosyllabic words with the 3rd person possessive pronoun prefix i.-; though this vowel, which is long, is then in the penult, the high-low (' ') pattern agreeing with M (for these words generally end in consonants) seems to be general: i-'tca.n ['his/her/its house'] i.tlak ['close to it' (?), cf. Classical Nahuatl i.-tlok] i.man ['now, time' (?)]43 (and both i.miL and i-miL ['his/her field']) and so by analogy i-nin ['this'], i.nbn ['that'] (where i is not a prefix).44 2. Special Tone. Little can be said yet as to this in T. The rule therein be- ing more complicated, it is harderto detect special rhetorical exceptions to them than in M. We have tlaso'hkamati pronounced about as in M.

3. Morphological Tone. In general [morphological tone is] the same as for M, and the penult vowel-length tending to alter accordingly, though my texts do not give clear examples of plurals distinguished in this way. To the M data we may add for T: -ka adverbial suffix, -'ka 'be... ing', and -Aka or -^ka 'come to . . .', M -'kIuhwI which is similar to -^t'a 'go to . . ', M -'tiuhWI. toNto'pahpa.ka lit. 'we come to bathe' (for tontopa'hpa.k-ka,stem pa.k)

42 The initial a of Whorf's ark tti! is unexplained. Since the parallel forms he gives be- gin with (sr), perhaps it is a transcription or copying error for an intended s. In any event, Nahuatl dialects do not really have [r], save perhaps in Spanish loanwords. Where Whorf writes (i), others have [s], e.g., Whorf's (a.-an) 'now' for others' /a.-an/ or /a-skan/. The ini- tial (si) which Whorf writes in these forms is the imperative marker /4-/ in other dialects. Whorf's transcriptions probably reflect the retroflexion of /1/ that is common in Morelos dialects. 43 Compare Classical Nahuatl imman 'to be high time'; notice that the vowel is neither long nor does it represent the possessive prefix. 44 The initial vowel of these forms is short in most Nahuatl dialects. 202 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS

[Morphological tone] also [makes the] distinction of a dependent clause verb equivalent to [an] infinitive or participle after a preterite from a true present tense by its taking the preterite tone pattern as treated above.45 4. Semantic Tone. In general [semantic tone is] the same as for M. I have an idea that [a] systematic search would find in T quite a few more examples of semantic tone than I am now able to give-viz. toka, tcitci, patla and ka the verb "is" [are] the same as in M. On the other hand ka 'with, by', needs separate treatment for T.

The present texts show only: ika itepbs 'with his gun'. We need more cases to study, and the rest of my T text material shows the following: o.tlamotlaya ika tlakpak tiraba por arriba he was shooting upward (lit. "by above") ka'itlawihtbL con su flecha (not literal) [Spanish flecha 'arrow'] with his bow (lit.) ka inon con eso with that i.nir na.ntli ika taihtli46 esta mama y el papa this mother with the father. amo ka_nipatciwwis j.niN piLto.Ntli no me [he] satisfecho con este muchacho I am not satisfied with this boy. (Prep[osition] with before the verb, the invariable classical [Nahuatl] syntax, and still common). ika tlaLpa.N w6iwe-tl con el bandolon with the native (or, local) drum.

45 Saltillo (glottal stop), not tones, distinguishes these forms. 46 Whorf is almost certainly mistaken here; the vowel of na-ntli 'mother' is long, but the vowel of ta'tli 'father' is short. Also i-ka ta'tli would be instrumental; one would expect i.-wa.n ta'tli. PITCHTONE AND THE "SALTILLO"IN NAHUATL 203 o-kikakk~ tlattsohtsonna tlawi.t?kki oyeron que tocaba y le pegaba they heard him playing and beating (indefinite present in dep[endent] clause [tla- 'unspecified object']) ika tlaLpanw6iwetl, con el bandolon, with the native drum, ika tlaLt'po-na-stli. con el teponastle. with the native teponaztli.47 ka tle-n okirjnala-nilihtel] con lo que les habia quitado with what he took back to them omawlLtincmia ika a't148 jugando con la [=el] agua playing with the water. ka in6o o.pano-ya ka i-so6wwa- con eso pasaba con su mujer with that he was passing with his wife. tla ajkiqwa'skE l-ka a-mikiske so lo han de comer se mueran if you eat it thereby you shall die (you plural) i-ka' i-pattfil por lo que vale by what it is worth (lit. whereby it is worth) [=for what it is worth]

ik_.so-.wwa. con su mujer with his wife

47 The tepona-ztli is a log drum. 48 The vowel of this word is simply long, with no saltillo, i.e., a-tl 'water'. One might imagine that Whorf's (a'tl) is simply a transcription/copying error of (') for (.); however, later Whorf discusses this in connection with compounds involving 'water' and insists the saltillo is there (see below). 204 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS

san yehWa. i-ka tldhqw'atikatte nomas estaban comiendo thereupon they are only eating.49 This may seem at first like a mere confusion of every sort of toning for ka and ika; but I believe the explanation to be as follows, and it is one that reduces the seeming confusion to a fair degree of order. In classical Aztec ka is found alone only as a conjunction 'for, because, that'. Suffixed to a noun stem it makes the noun an adverb and so may be considered a postposition, usually translated "with" or "by". Every postposition may become a preposition by prefixing to it (suffixing it to) the pronomial base i- [i-], and in actual use such prepositions are far more common than the original postpositions. As a preposition then, ka appears in classical Az- tec always as ika, 'with' or 'by'. We may next assume that this was toned ika, following the toning which we still find in both M and T to be gener- ally used for prepositions, perhaps because with the exception of this ika they almost all end in consonants. In modern Nahuatl ika has tended to become contracted to ka. In T all stages of the process may be seen; in this dialect one may say at his own option and as best fits into his sentence either ika, ika or ka, much as in English we have the option of saying either electric or electrical, bril- liance or brilliancy, cannot or can't, thirteen or thirteen.50 In M however we find that ka has won out and is the only form in use. In the course of this contraction and diversity of forms what happens to the tone pattern (' ') of the old ika? In T there has been a tendency so long as the word remains ika to change the tone to ika or ilka, following the influence of [the] vocalic ending that is so strong in M and not without manifestations in T. But in the case of ika contracted to ka the tone pattern (' ') or some part of it tends to persist and serve as a means of distinguishing the ka that means ika from the ka that means is. The last syllable of the preced- ing word would often be preempted by a typically necessary low tone so that the high [tone] or (' ') could not fall there. It would be displaced for- ward and fall on ka, while the low [tone] of (' ') would fall on the first syllable of the next word, which in nearly all cases would be capable of being lowered. This gives the ka (') that we find among the above quota- tions. Once this type has been established it tends to affect ika, giving us ika (') and ika (') as alternatives to ika. Then there is a further tendency for the high [tone] on ka to disappear in order to further the distinction

49 In Whorf's text "were" is crossed out before "are." 50 Presumably here Whorf means the two variants to be "thirteen" with stress either on the first or on the second syllable. PITCHTONE AND THE "SALTILLO"IN NAHUATL 205 from ka ('is'), giving ka ('), which we have seen to be the regular form in M. The above texts contain, for example, ka in6o', tone depressed from inbn. Note also the cases of ika (') and ika () followed by the word tlaLpan or tlaLpan, the typical tone of which is tlaLpan. There remains an unaccountable ka inon and a few other lesser anomalies.

Discussion of the saltillo in the T dialect

The occurrence of the saltillo is practically the same in T as in M, though we may add the following notes: 1. In T the saltillo is often, perhaps usually, reduced to simple h (pre- aspiration of next consonant), losing the sharp glottal click heard in M. 2. Final saltillo is even more completely eliminated from T than from M, except when the next word begins with a vowel and then it is perhaps more generally heard as ('h) or (' h-) or (h-). oa.sihki hompa (for ompa) 'llegaron alla' 'they arrived there'. 3. Occasionally internal saltillo completely disappears, as in the preced- ing quotation a-si 'arrive', M a'hsi. 4. Sometimes saltillo or h is inserted without apparentreason into places where there is no apparentetymological or historical justification for it. ki.'hqwa 'he eats it' M & class[ical Aztec] kiqwa a'hmo 'not', is found beside amo (M. amo)51 o.kiqwa'ht_aya,M o.kiqwat'aya, class[ical Aztec] okiqwatiaya.52

Tones and the saltillo in old Aztec

In the oldest texts of classical Nahuatl there is no written indication of the saltillo beyond rare occasional h's, and none whatever of the tones. Some of the old grammarianshowever give us some scanty information on these matters, under such headings as "los accentos" [modern Spanish acentos] or "las cantidades". The best treatment of these "accents" and "quantities" and the one that shows most clearly that they are of a piece

51 This is not, however, a case of insertion, but rather the conservative form, as found, for example, in Carochi's spelling of it. 52 The Classical Aztec form is more appropriatelyo-kikwa-tiya-ya; Whorf may have mis- understood its structure-cf. the discussion in n. 41 above. 206 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS with the modern tones, is a rare "Arte Mexicana" by Antonio del Rinc6n, 1595, published by Peniafiel in 1885. The only copy I have ever seen I found in the library of Mrs. Zelia Nuttall at Coyoacan, Mexico. Carochi's grammar also gives a list of cases in which differences of accent change the meaning of the word, but this list can be best understood with Rincon beside it.53 Rinc6n's "accento agudo" is the high tone. His "accento grave" is the low tone. His "accento moderado" is the slightly raised tone. He marks it with a circumflex, but he also calls it "largo" and marks it with the ma- cron. Here he confuses two things that very likely to his ear simply com- bine in one effect of emphasis or distinctiveness concerning a certain syllable; length of vowel and the slightly raised falling tone on the vowel. It would seem that very often his macron mark indicates simply the slightly raised tone, and the same applies to Carochi's macron. Rincon next refers to the "accento breve predominante"and to "silabas breves". He says that all syllables that do not bear one of the other ac- cents named are pronounced "breves", from which it follows that "breve" cannot mean "short vowel" or cannot mean simply that. A study of him shows that "breve" means medium tone at least where it applies to the syllables without stress accent, and that "accento breve predominante" is either or both of the following conditions, undistinguished by him: (1) medium tone + stress accent, (2) high tone on short vowel. To the ex- tent that the last is the case, his "accento agudo" is distinguished from it [high tone on short vowel] by being not merely high tone but high tone plus some other emphasizing feature, vowel length or stress accent or both. Finally his "accento saltillo" is the modern saltillo, but it is clear that he regards it as a phenomenon wholly coordinate with the other "ac- cents" which are tones. It follows that under his system he cannot record any other "accent" on a vowel which has the saltillo, and consequently he gives us no information as to the tone of all such vowels [that occur be- fore saltillo]. From now on I should translate his notation and nomencla- ture into that of the following table, and give his picture of the situation in classical Nahuatl. For convenience I shall usually write as uhw the sound which in Modern M I often hear as UhWI.54

53 Whorf writes "Rincon" without an accent throughout the manuscript;here it has been corrected to "Rinc6n" throughout. 54 At this juncture Whorf's handwritten text ends, and his typescript begins. The hand- written pages number 1-35; typed pages are 36-54. PITCHTONE AND THE "SALTILLO"IN NAHUATL 207

My notation Rinc6n's Rinc6n's probable value in the (for Rinc6n terminology notation previous notation of & Carochi) this article a "silaba breve" a (no mark) a (medium tone on unaccented vowel or on monosyllable) al "accento breve a (no mark a usually short predominante" but on penult also a short? vowel) a2 "accento a a , a, a (and a- ?) moderado" a3 "accento a a, a., a. agudo" a4 "accento a, or a grave" unmarked vowel in ultima a'h "accento a (other a'h (with any tone) saltillo" writers who mark the saltillo but not the tones, may use any accent mark for saltillo, e.g., a, a or a for a'h)

In what follows I shall use the notation of the left-hand column for reportingRinc6n and Carochi,placing afterthe word and in parenthesis[sic] a renderinginto the tone notationused above for modem Nahuatl,as I conjecture the probable sound of the ancient word to have been. Then the modem pronunciationof the word, when it is known, will be given not in parenthesis [sic] and precededby "mod." 1. Typical tone As Rinc6n treats it, it is mostly according to the type of morphologyrather than type of phonology. a. Medium tone A level of tone applying to most syllables, to be understood where other "accents"are not indicated. Etl, metl, sctl melkatl (mekatl- or mekatl ?) mod. mekatl ['vine, string'] salkatl (sakatl- or sakatl ?) mod. sak'atl ['grass'] tolkatl (tokatl- or tokatl ?) mod. t6k'atl ['spider'] 208 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS b. Slightly raised tone 1. passives in -a2lli (-alli). But mod. -alli. 2. frequentives in -ka and -tsa from verbs in -ni take it (i.e., on penult?) 3. nouns in -ka2yotl (-kayotl) not derived from verbals in -ki and -ni;- but mod. -kayotl. 4. all words compounded with the "ligature" -ti-, e.g., nikotcti2ka (nikotctika) 'I am sleeping' qwauhwti2tlan(qwauhwtitlan) 'places of trees' but mod. nikotctika, qwauhwtitlan. c. High tone 1. never on last syllable except for vocative -e3 (-e). [The] modern [is the] same barring special tone, [with] special conditions like T. ika, and a few monosyllables, e.g., ka, ya. 2. penults of verbs ending -wa,-but not so modern, e.g., M tc^wa, not tciwa. 3. although given as a separate rule by Rincon, this one would follow from (2) viz. impersonal verbs in -tiwa; pilti3wa (piltiwa). I have no mod- ern examples of such verbs. 4.(a) on penult of verbs in -tiuhw and -kiuhw; nitematcti3tiuhw (nite- matctitiuhw) mod. M. nitcmatctitiuhw, T nitcmatctiftin. (b) exception: disyllabic verb stems have "accento breve"; makatiuhw (makatiuhw) mod. follows (a). 5. gentile names in -s3katl (-4katl); modern same.55 6. words ending -i3lli, -o311i(-illi, 611i);[the] modern [form is the] same. 7. nouns in -o3yan (-6yan) formed from passives; [the] modern [form is the] same. 8. nouns in -ka3yotl (-kayotl) from verbals in -ki and -ni, e.g., teopicka3- yotl (tcopickayotl); modern [is the] same. 9. "the ligature ka in all composition always has accento agudo"-e.g., nitcipawaka3nEmi(nitcipawakanemi). There is one clear modern example; tlaso'hkamati, 'thank you'. 10. all verbals in -o3ni, -o3ka (-6ni, -6ka); but modern -6ni, -6ka. d. Low tone 1. it is found on last syllables, unlike the high tone. Modern [is the] same. 2. all endings -ya4n, -ka4n (-yan, -kan). Modern [is the] same.

55 The reference here is to the grammatical category found in colonial (and modern) grammars of Nahuatl called in Spanish "gentilicios," sometimes in English "gentilics" or "gentile names" (cf. Andrews 1975:331-34); "ethnonyms" seems more appropriate. This means specifically forms derived by the morpheme -e-ka-(tl), as in to-lt-e-ka-tl 'the Toltecs', people from to-l-la-n 'Tula' (to-l + tla-n), a-st-e-ka-tl 'the ', people from Aztlan (a-s + tla-n), etc. PITCH TONE AND THE "SALTILLO" IN NAHUATL 209

Rinc6n makes only these two brief statements about the low tone. Most of the old grammariansmake even less mention of it. This comprises the total of Rinc6n's references to regular placing of tones according to pho- netic and morphological considerations (typical tone), while neither he or any other grammarian gives us data on special toning or morphological toning (tone used by itself as a morphological agent). But when it comes to semantic distinctions involving tone, Rinc6n and Carochi both give us material.

2. Semantic Tone In quoting the following list of semantic tone distinc- tions from Rinc6n I shall insert remarks on points that seem worthy of some attention and make a few additions from Carochi. In point of fact all the examples beginning with a- are from Carochi, since while studying Rinc6n, to save time I simply noted that Rinc6n's list of a-'s was the same as Carochi's. The rest of the alphabet is from Rincon, except additions from Carochi noted specifically. Some distinguished by saltillo are occa- sionally included. a2tctli (atctli) 'older brother or sister' [a.c-tli] atctli (atctli ?)'seed' [ac-tli] awa'he scolds' [a'wa] a2watl, a2wa- (awa-) 'oak' [a.wa-tl] a'hwatl, a'hwa- 'prickly down of cactus, fine prickles' [a'wa-tl] a2wa'h (awa'h) 'possessing water' [a.-wa'] a2wa2tl, a2wa2- (awa-) 'downy caterpillar' [a.wa--tl] Note that awatl ['downy caterpillar'] and a'hwatl ['fine prickles'] appear to be different developments of the same meaning "fuzz, down", yet they have different tones and one has internal saltillo. a2wik (awik) 'toward water' [a.-wi.k] a'hwik 'from here to there' [a'wi-k] a2o'htli, a2yo'htli (ayo'htli) 'water cane' [cf. a.yo' 'containing water'] ayo'htli calabash [ayo'-tli] In modern Nahuatl a[]ltl 'water' is quite generally pronounced with saltillo, a'htl, but in the above cases of words compounded of its stem a[l- there does not seem to be a saltillo connected with this stem and serving to distinguish it.56 a2ma2k(amak) 'on the river-brink' [a.-ma.-k] a2mak (amak) 'on paper' [a-ma-k] patla (patla ?) 'change, exchange' mod. patla [patla]

56 It is not clear why Whorf insists that there is a saltillo in the word for 'water'-the vowel of this word is simply long, with no saltillo, i.e., a.tl 'water', in all known dialects. Whorf's version would go against the Nahuatl prohibition against final consonant clusters in words, here -'tl. 210 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS

pa2tla (patla) 'knead, stir, mix', mod. same [pa-tla ?] piloa 'hang' [piloa] pi2loa (piloa) 'make thin or fine like a shoemaker waxing a thread' [pi.loa] In this case and several cases below the tone distinction is on a syllable that in modern [Nahuatl] would have no stress accent and would regularly take medium tone. Possibly many former tone distinctions on such sylla- bles disappeared. piltik (piltik ?) 'pertaining to the nobility' [pil-ti-k] pi2ltik (piltik) 'thin and long' [pi.l-ti-k] poloa 'lose' [poloa] po2loa (poloa) 'stir around, mix up, like bran or mud' [po.loa] matia imperfect tense of mati 'know' [mati-ya] ma3tia (matia) 'put one's hand on' [ma.tia] The last [ma.tia] is a demonstrative verb from stem ma- [=ma.-] 'hand'. This stem does not ordinarily carry a high tone in composition, but ap- pears to do so here. The next couplet is inserted from Carochi. mimiloa, mo- 'topple over, be overturned' [mimiloa] mi2miloa, mo- (mimiloa) 'wallow like an animal' [mi.miloa]57 These two are from the same stem, milo 'role', etc. and are distinguished by the tones and possibly also [by] the lengths of their first syllables. mctstli 'thigh, leg' [mets-tli] me3tstli (metstli) 'moon, month' [me.ts-tli] tcctli 'meal, dough' [tes-tli] tc2ctli (t'ctli) 'brother-in-law'[te.s-tli] tc2piltsin (tEpiltsin) 'son of someone' [te.-pil-tsi-n]58 tspiltsin is called a "verbum turpe" [tepil-tsi.n 'vulva' (diminutive)] What tones teka 'lay, pour' and tcki 'cut' had are not stated; but in cer- tain of their derivatives the distinctive vowel-endings are lost and then these semantemes are distinguished by tone or saltillo, not however in a regular manner, but as follows: a the passives tc3ko (t~ko) 'be laid' [te.k-o.] teko 'be cut' [tek-o-] b the applicatives ts2kilia (tEkilia) 'lay for' [te.ki-lia] tskilia 'cut for' [teki-lia]

57 The mo- after these verbs is the 'reflexive' prefix, e.g., mo-mimiloa 'it rolls'. 58 Throughout Morelos, in particular, the nonspecific possessive prefix te.- is used with kin terms as a sign of respect and deference. Whorf's gloss 'son-of-someone' might seem to suggest an English translation of the Spanish hidalgo 'noble', from hijo de algo/alguno 'son or daughter of someone (important)'which was still used in sixteenth-century Spanish, and in colloquial versions. This, however, is apparentlyjust fortuitous in this instance. PITCHTONE AND THE "SALTILLO"IN NAHUATL 211 c the compounds with ma[.]- 'hand', which are intransitive in form and therefore both take the regular transitive ending -i (tcki 'cut', a transitive, is out of line in taking it): ma'htcki 'pour on hands, wash one's hand' [ma'-tekia] ma2tcki (mateki) 'cut one's hand' [ma.-teki] Thus the stem tEk-'lay' [=te.ka] is characterizedonce by high tone, once by slightly raised, and once in a compound by saltillo on the other member of the compound; while tek 'cut' [=teki] is characterized twice by medium tone and once by slightly raised on the other member of the compound.59 This is not quite what one might naturally expect. It might be supposed that of two stems tsk and tek of different meaning each would have a cer- tain tone and would carry that tone through its various inflected forms and compounds. This is not the case, for tck lay has ?3 in one inflected form and E2 in another. It might next be supposed that certain tone patterns would be associated with certain morphological patterns, but here we are confronted with the fact that tsk 'cut' has e and e for the same two mor- phological forms in which tek 'lay' has e2 and ?3. This is typical of Rinc6n's whole list, and as I give that whole list in the course of this arti- cle the reader will see that nothing in the way of regular rules of tone in- cidence can be made out from it, and also that it would be unsafe to etymologize using as main arguments the presence or absence of certain tones or of the saltillo. tsma 'bathe in steam chamber, roast in oven' [tema] te2ma (tEma)'pile up' [te.ma] toka 'pursue, follow' mod. t6ka [toka] to3ka (t6ka) 'sow, plant, bury'-mod. toka or t6oka [to.ka] Here there has been a change in [the] tone of the second at least, and it is possible that the real distinction was one of vowel length, and Rincon should have written t6ka and t6oka, [of] which [the] last became to.ka. Note next how three different semantic developments of toka 'pursue' are distinguished: toto3ka (tot6ka) 'hasten' [toto.ka] to2toka (totoka) 'go running or skirmishing behind' [to.toka] to'htoka 'chase away, scare off, banish' [tohtoka] toloa 'swallow' [toloa] to2loa (toloa) 'bow the head' [to.loa] tlatia 'burn' [tlatia] tla3tia (tla'tia) 'hide' [tla-tia] tcitci 'dog'-mod. tcitci [cici ]

59 Whorf seems here to miss the fact that 'hand' has two different compounding stem forms: ma- and ma'. 212 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

tci2tci (tcitci) 'suckle, nurse'-mod. same [cici-] tci'htci 'spit'-mod. same [ci'ci-] tci'htci 'mend' (from Carochi) [ci'ci ] The following couplet from Rinc6n is interesting: csloa "divide something continuous like a log" [seloa] cE3loa (celoa) "divide and scatter" [se.loa] Note that here tone distinguishes two nuances of the same word. ciwitl 1. 'herb' 2. 'year'-mod. ciwitl [siwi-tl] ci2witl (ciwitl) 'comet' [siwi-tl] yckoa 'try' [yekoa 'to taste, sample'] ys2koa (y~koa) 'finish what one is doing, finally accomplish' [ye.koa] From Carochi: kalli 'house'-mod. kalli [kal-li] ka2lli (kalli) 'tongs for drawing toasted maize out of fire' [ka.l-li] kokok 'peppery, "hot" to taste'-mod. kokbk [koko-k] koko2k (kokok) 'afflictions; toil, work' [koko.-k] Continuing with Rincon: qwawitl, qwauhw- 'tree'-mod. M. qwawitl [kwawi-tl] qwa2uhwtli, qwa2uhw- (qwduhw-) 'eagle' [kwa.w-tli] Note the manner of distinguishing the following three derivatives of qwa 'eat'. qwaqwa 'chew' [kwakwa.] qwa'hqwa 'bite' [kwa'kwa.] qwa2qwa (qwaqwa) 'eat again and again' [kwa.kwa-'to eat continuously']

The saltillo in Old Nahuatl

Rincon gives the following rules for the use of the saltillo in grammar: 1. All preterites ending in a vowel take saltillo on this vowel (onitematc- ti'h) (pret. of nitsmatctia) and in the plural before the suffix -ke'h (otitc- matcti'hkE'h). 2. Possessive nouns in -wa'h, -o'h, -c'h and their plurals -wa'hks, etc. 3. Place-name derivatives of these in -wa'hkan, -o'hkan, -c'hkan. 4. All verbs compounded with the ligature ti if there is a vowel before ti have saltillo on this vowel, e.g., nitcmatcti'hti'hkak(a compound of the verbs matctia ['to learn'] and i'hkak ['to be standing']). 5. All reduplicated frequentatives take saltillo on the first of the doubled syllables (this rule however is not observed in the frequentative of qwa 'eat' cited by Rincon and quoted above). PITCH TONE AND THE "SALTILLO" IN NAHUATL 213

6. The postposition -tla'h and gentile names in -tla'hkatl related to it, e.g., qwauhwtla'h, qwauhtla'hkatl. The situation in modern Nahuatl is about the same, except that the final saltillos are usually at the most faint glottal stops, and even these as often as not are absent. Other old grammarians place the saltillo on the final vowel of all plurals, including plural nouns in -me'h, -ks'h, reduplicated noun plurals (kokonc'h [ko.kone' 'children']), verb plurals otherwise indis- tinguished (titciwa'h, tinsmi'h), [and] verb plurals in -kc'h. Traces of all these final saltillos are to be found in modern Nahuatl as glottal stops or h's. Rincon gives the following short words containing saltillos: tla'htla 'burn', a'hsi 'arrive', tsa'htsi 'cry out', i'hsa 'awake', i'hkak 'stand', i'htitl 'belly', ne'hwatl 'I', tc'hwatl 'thou', yc'hwatl 'he', tls'hko 'ascend, climb', pi'htli 'elder sister', si'htli 'aunt' [actually 'grandmotheror sister of one's grandfather']. Besides the cases that have already been cited in discussing tones, he lists the following pairs distinguished by saltillo: i2kak (ikak) 'his shoe'-mod. M. ikak [i.-kak] i'hkak'stand' (really 'he stands')-mod. M. f'hkak [i'kak] omitl 'bone' [Carochi gives omi-tl, cf. Proto-Nahua *o.mi-] o'hmitl 'down, fuzz' [o'mi-tl] pa2ti (pati) 'melt' [pa.ti] pa'hti 'heal' [pa'ti] ma2sc2walli (mas~walli) 'a plebian' [ma-se.wal-li] ma'hscwalli 'merit' [ma'se.wal-li] ta2tli (tatli) 'thou drinkest' [t-a.tli] ta'htli 'father' [ta'-tli] tlapaloa 'soak, moisten, soften, dip in broth and eat' [tlapaloa] tla'hpaloa 'greet, salute' [tla'paloa] The next instance is particularly interesting: tlancwia 'borrow' [tlane.wia] tla'hncwia "mistake, taking one thing for another." [tla'ne.wia] The basic meaning of tlancwia is probably "take one thing as the equivalent of another, substitute", hence "use a substitute, borrow an- other's to use in place of one's own". Its causative tlaneuhwtia is naturally "lend", also "rent".The tla- is an original morphological element that has become inseparable, as shown by the cognate Cora nehte "prestar",'lend', which would correspond to Nahuatl form ncuhwtia.60 The source seems

60 Cf. Cora ra-ta-niihte 'to lend'. 214 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS to be the monosyllabic semanteme or root - 'same, self' (

Possible origins of the saltillo Before venturing this very tentative sketch of possible origins[,] I ought to re-emphasize the complexity of the situation regarding the saltillo and the need for more study of dialects and cognate tongues. The saltillo may have diverse origins. On the basis of present information it may represent a consonant, or rather various consonants that have dropped out under the influence of certain special circumstances. Cases

61 Proto-Uto-Aztecan contrasted two morphemes. Langacker (1977:46-47) reconstructs PUA **ni 'unspecified human object' and **na- 'reciprocal verb prefix'; however, **ni be- came 'reflexive' and 'passive' in Nahuatl, while **na- 'reciprocal'disappeared in most south- ern languages (including Nahuatl, e.g., ne-pan 'on one another' [-pan 'on']). PITCHTONE AND THE "SALTILLO"IN NAHUATL 215 that are exceptions to the regular Uto-Aztekan phonetic correspondences (and dropping of consonants would be such an exception) will probably be found to be explained by other principles, especially either (1) seman- tic, or (2) rhythmical, accentual and tonal. The semantic principles I refer to would simply enable us not to mistake the apparentdisappearance of a consonant that is not a firm part of the UA semanteme, but rather a quasi- affix, for actual quiescing of one of the original consonants of the se- manteme. We need not consider such questions in this paper, but only the question of reduction of consonant to saltillo under prosodical influences. These influences are often related to the ablaut phenomena of Uto- Aztekan items, which phenomena may be observed in the Nahuatl dia- lects and in some Sonoran tongues, e.g., Heve. For example M tsaqwa, tsaqwi- 'close, shut, hide, make sterile', etc., shows the full grade of a certain stem (with a variation in final vowel that goes back to the Uto-Aztekan period), but this stem also appears in cer- tain tenses and derivations in contracted or narrow-grade forms such as tsaqw, tsak, tsauhW, tsa'h, tsah, in which the final vowel is missing and the last consonant often reduced, the minimum of the reduction being saltillo or h. Thus we have the classical [Nahuatl] spellings "oquitzacu" (Molina) and "oquitzauc" (Sandoval) intended for okitsaqw 'he closed it', M o-kitsak 'he closed it', M o.tctsauhWt'aya 'they made a closure', M tsauhWhIhttikdtE'62and tsa'htikatt', T tsahtikatt, 'they are hiding (or hid- den)'; cf. Heve sov- 'obscure, dark, black' (< narrow grade *tsauw-); Te- pecano sob- 'shield, protect' (< narrow-grade *tsauqw-). The word for "pulque" is pronounced in M sometimes neuqWtli, often n_u'htli (both narrow grade) [phonemically /nekW-tli/]; the full-grade neqw-vowel appears in n_qwa.tl "aguamiel" ('unfermented pulque') [nekw-a.-tl], a compound with a[r]tl 'water'. Again, the full grade tciqwa- "five more than"[,] in M tciqwa.sE 'six' (se 'one'), M tciqw'yi 'eight' (yEyi 'three', at one time eyi answering to Pochutla eyo-m, Heve veidu-m, etc., < UA *(paiyi [**pahayu]), M tciko6me 'seven' (-.mc 'two', but the syllable wo is inadmissible in Nahuatl and becomes o [i.e., cikw + o-me > cikome]) corresponds to the narrow grades tciqw-, tciu'h-; M tciu'hna-wi'nine' (na-wi 'four'), classical [Nahuatl] both thus (chiuhnaui) and tciqwna-wi (chicunaui). In these examples qw becomes 'h (saltillo) but always incidentally to ablaut construction. Numerous examples could also be given in which original w in the same situation becomes 'h. For instance M tciwa, tciwi- (full grade) 'make'; narrow grades, when final M tciuhW, T tcin prob[ably]

62 This form is blurred;it has been typed over and therefore its exact form is uncertain, as is the order of some of the letters in it. 216 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS

< tcim < tciw, but when medial M tci'h, T tcih; M o.kftciuhWI'he made it', T o-kitcin, M o-kitcf'hk', T o.kitci.hks 'they made it'. Full grade iwi 'go', iwa 'send'; narrow grades iuhw, i'h in classical [Nahuatl] iuhw, iuhwki (yuh, yuhqui) 'the way that, how, thus', M i'hki 'how, thus'. Full grade isiwi 'hasten'; narrow grades in cl[assical Nahuatl] isiuhwka (iciuhca) M is^'hka 'quickly'. Again we have the full grade cl[assical] ciwitl 'herb, grass, green, turquoise'; narrow grade ciuhwtli and compounding form ciuhw-. These forms are found written as xihuitl, xiuhtli, xiuh-. While I have not noted the narrow-gradeform in M, by many analogies it should be ci'h-, showing the saltillo. The full grade appears in the Heve nomina- tive case seqwat 'flower' (

63 But see also Cora samwa 'leaf'. 64 That is, these point to PUA **sawa (**sama), the main reflex of which is iswa- in Nahuatl, *iswa in Proto-Nahua, though the Pochutec form is cognate with Nahuatl siwi- 'green, vegetation, grass', also apparentlyrelated to this same PUA root. 65 This sentence was written in hand with a note to insert it here. Where we have placed "cited [above]," Whorf had written "cited on p. 19 (5.)," referring to his manuscript. PITCHTONE AND THE "SALTILLO"IN NAHUATL 217

'h may represent [the] whispered, muffled, or otherwise reduced -ya of tla- lia. When the present is not of the -ia, -oa form, the past ends in a conso- nant other than saltillo and this question does not arise.66 If the saltillo is a weakened or quiesced mora that may have been once either vocalic or consonantal, possibly the weakening went with a low tone on such moras, for in modern Nahuatl the low-toned syllables are rather weakly uttered, and final consonants, which as we have seen are regularly low-toned moras, are especially weak. Final 1 is always un- voiced, and final n is usually unvoiced, sometimes a mere nasalization of the vowel, and sometimes quite silent, without even nasality (ini for ninm).67I hesitate to ascribe either the low tone or the weakening as the original cause of the other, but would prefer at present to regard them as concurrent manifestations of the same tendency. This tendency to weaken the end of a word apparently once took place in certain situations more than in others, and in such situations produced the clipping of final vow- els and the contracted or ablauted forms of stems that are characteristic of certain grammatical forms, especially verb preterites and the closely re- lated nomen patientis and the possessed form of nouns.68 I think we might go as far on the hypothesis that the original principle of the preter- ite formation and of various other ablaut processes was this weakening and its concommittant [sic] drop in tone at the end of a stem, so that the last mora or two was weakened to possible quiescence of a vowel or a consonant or of both, certainly to quiescence of the last mora if vocalic. This occurring without change in the rhythm of the phrase would leave in place of the quiesced sound a "jump" ("saltillo"),69 a gap or interruption,

66 Nahuatl has a clear difference in verbs between those ending in underlying /owa/ and /oa/, similarly between underlying /iyal and /ia/. These underlying contrasts are not necessar- ily phonetically distinct, however, and can be quite complex in the various dialects. It is rarely represented consistently or accurately in colonial sources, since it was not easily heard by native speakers of Spanish where no such contrast could exist. Proto-Nahua had *tapowa 'to open'; the preterit, however, was consonant-final, *tapow, where final w became h (with ' h > in the central dialects). The verbs ending in -oa, however, for the most part appear to derive from *-oha, where *h was lost except syllable-finally. Thus, the 'present' tense of such verbs ends in -oa, while in the preterit, where the h was final and not lost, these end in -oh (or -o' in glottal-stop dialects). The pattern for the -iya vs. -ia verbs is similar, but not identical. These may be difficult to distinguish in present tense, but in the preterit, the *-iya verbs end in -is (final y -- s) or -iya-k, while the -ia verbs were originally *-iha and thus have preterits ending in -ih (or -i'). (For details, see Campbell 1985:70-74, Campbell and Langacker 1978, Canger 1980, and Dakin 1982.) 67 Both forms, inin and ini, are widespread in Nahuatl dialects, and therefore it is probably unnecessary to expect ini somehow to exhibit evidence of a final nasal or vowel nasalization. 68 For details of the vowel reduction and loss rules in Proto-Nahua and how these appear in later varieties, see Campbell and Langacker (1978). 69 Saltillo in Spanish actually means 'little jump'. 218 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS which is how the early writers describe the saltillo, and [this is] a fair de- scription of what it sounds like in M today. Thus tlalia would have formed past tlalia > tlaliYA> tlali'h. The plural with the suffix -ke' would have been tlali'ak' > tlaliYAkU'> tlali'hke'. So also kitciwa, past kitciwa > kitciWA70 > kitcluhW, pl[ural] kitciwa-k'' > kitciWAkG'> kitciUWkF' (Cl[assical] quichiuhque), and by continuation of the same process mod- ern M kitci'hkE'. Agreeably to this supposition we find that to-day [sic] in M the last syl- lable of a preterite is always in low tone. Again there are almost no low- toned final vocalic-ending syllables except those that are reduced from consonant-ending syllables (like -? from -e' and -a from -an). This is in accordance with the hypothesis, which supposes disappearance at an ear- lier period of final vocalic moras in those cases where the end of the word became low-toned. A different occasion for saltillo is in the reduplicated syllable of redu- plicated frequentative verbs. Some verbs have a stem that is always in a reduplicated form, and with these there is no saltillo-e.g., mama 'carry', kokoa 'to be sick', kakali 'shoot', tsotsona 'strike', tcitcitoka 'drip'-and there are some others that are found both unreduplicated and reduplicated without saltillo. This is simply reduplication of the first syllable, and it can probably be explained in every case as a reduplication of the whole se- manteme, originally a monosyllable of the form c[on]s[onan]t-v[o]w[e]l followed by a formative element (-a, -na, etc.). But the reduplication with saltillo, that is with an infixed consonant, is harder to explain. The sim- plest explanation that occurs to me is that the above-mentioned weakening principle has acted upon the reduplication of a disyllabic combination. Thus qwi'hqwika 'he sings and sings' (qwlka 'he sings') could be from qwika-qwika > qwikAqwika> qwi'hqwika, and wi'hwilana 'it creeps and creeps' (wilana 'it creeps') from wila-wila-na > wiAwilana > wi'hwilana. It might be noted here that ololoa 'make round, gather together' is redu- plicated from the less common oloa of similar meaning, and as the se- manteme is certainly lo and not ol (o- is prothetic, cf. i-lo-tl 'rounddrop', i- lo-ti 'turn',i-lo-a 'return',o-lo-tl 'cylinder, corncob') we have here evidence of reduplicationof a disyllabic combination; olo-olo-a with vowel contrac- tion to ololoa. The form elel- in elelti 'care, suffer', elelkictia 'delight', ap- pears reduplicatedfrom el-, but these are the only cases I know of where a consonant in such position in a reduplication has survived.71Usually only

70 Whorf probably intended this A to be superscripted, as well, as in the other examples. 71 This form does not actually involve reduplication as Whorf imagined it. The verbs he cites begin with e-llel-, where the first e-l- apparentlyis the incorporatedstem for 'liver', the seat of emotions and mental states, which appears incorporatedin a number of verbs and de- rived forms. It is not a case of VC-VC reduplication. (Cf. Campbell 1985:82.) PITCHTONE AND THE "SALTILLO"IN NAHUATL 219 saltillo or its vestiges h, ', appears in this position; e.g., ana reduplicates to a'hana, ahana or a'ana, not to anana.72 Reduplication for frequentative meaning is semantically akin to redu- plication for the plural of nouns, which is common in Nahuatl but usually lacks saltillo. However the reduplicated syllable is usually long, e.g., ko-k6on-''children', and is stated to be regularly long by the old grammar- ians. As Sapir [cf. Sapir 1913:421] has pointed out, the Pipil dialect of Nahuatl in Nicaragua, as described by Lehmann [1910] uses a redupli- cated plural containing a palatal fricative x which corresponds to the saltillo and its Tepoztlan equivalent h; e.g., Pipil ta.gat 'man', taxta.ga- met 'men'.73Besides this, I have heard glottal stops and saltillos in Milpa Alta reduplicated plurals-e.g., tla'htlayo.yo "clacloyos" ('tarts' or 'turn- overs'), qwc'hqwFntla'fields'.74 We now come to the classical [Nahuatl] saltillo at the end of every plural form, noun or verb, [which] in modern M and T [is] usually reduced to a faint glottal stop or nothing. The evidence bearing on the point makes it seem like the vestige of a consonant that has disappeared. It appears to me that this consonant often was -tl, which in the historical period ends nouns that are not definitely plural. In one dialect of Pipil the plural suffix answering to Cl[assical] -me'h is me-t, in another dialect -mil, thus point- ing to -metl or -mitl.75 Compare the fact that in classical [Nahuatl] poetic texts the possessive suffix -ke'h sometimes appears as -ketl. A common UA plural suffix is (*mi) -mi, -mu, -m, the last becoming -n in Nahuatl ac- cording to the rules of that tongue. Then -me'h might be from -mi-tl, the plural suffix before the substantivizing -tl. This is the regular plural form of unablauted or vowel-ending stems (kolo-tl 'scorpion', pl. kolo-me'h <

72 Proto-Uto-Aztecan had at least two kinds of reduplication. One is simple CV- redupli- cation, usually associated with 'plural' in some general way (e.g., Nahuatl we-wetska 'he laughs a lot'). The other type is CVC(V)- reduplication, associated with 'distributive'among other meanings. While this needs greater study, in varieties of Nahua the second vowel is lost due to regular sound changes and the second consonant usually gives glottal stop (e.g., Nahuatl te'teki 'to cut up, hack, cut into pieces', cf. teki 'to cut') (Langacker 1977:128-29). Thus, a'ana is quite regular. It is conceivable that the VIVI cases cited by Whorf are of the second kind of reduplication, but where I did not give glottal stop. 73 Pipil actually has tah-ta-ka-me-t'men' (cf. Campbell 1985). As mentioned above, the h typical of the majority of Nahua dialects varies in some of these with [x], a velar fricative, apparentlydue to influence from local Spanish [x], which has no h. This is phonemically /h/ and corresponds to r/ in the dialects where h > '. 74 This second form is a 'distributive'rather than the typical plural. 75 The plural in these forms is -t in Pipil and Pochutec, but not elsewhere in other Nahua varieties, cf. Pipil -me-t. This comes from the Proto-Nahua*ti plural (cf. Dakin 1982:77). It is most unlikely that it involves tl, since, as shown by Whorf (1937; cf. Campbell and Langacker 1978:207-8), tl comes from t before a in Proto-Nahua.With these plurals, there is no evidence of a previous a. There are, moreover, no known dialects with -mil (Lastra de Suarez 1986). 220 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

*kolo-mi-tlo),76 while ablauted or consonant-ending stems differ only in that their plural suffix is placed, with the regular phonetic changes, after the substantivizing -tl, which the ablaut contraction has welded more closely to the stem (qwa-uhW-tli 'eagle', ablauted from, e.g., qwa.wa-tla, pl. qwa.uhW-ti-n, contracted from, e.g., qwa.wa-tl.a-mi, where tl. = *tl regularly hardened before -a-c[on]s[onan]t to a simple stop, t). How then shall we account for the fact that final -tl has disappeared from plurals such as kolo-mi-tl, but not from singulars such as kolo-tl? Here again it is possible to use the explanation suggested for the phe- nomena of the past tense-a weakening at the end of the word involving low tone and occasioned in some way by the usual rhythmical or accen- tual pattern of plurals. This explanation indeed simply accounts for most of the facts of Nahuatl plural noun endings. The superficially different plural endings and the incidence of saltillo and tone can be derived by the usual phonetic rules from two factors: suffix -mi plus low-tone weakening at [the] end of [the] word.77 Examples follow: Unpossessed forms stem unablauted: sing. *kolo-tlo > kolotl plur. *kolo-mi-tla > kolo-mi-tl1> kolome'h stem ablauted: sing. *qwa*wa-tla> qwa-uhWtli plur. *qwawa-tl-mqwawa.w-t-mi > qwa > qwuhWfin Regularly m-moras in this situation[,] instead of completely disappearing or becoming saltillos, become final -n. There is a class of stems which take ablaut under certain conditions but form the singular from the full grade. Nearly all are inanimates and have no plural form, the singular being used in a generic sense in most usages wherein European languages would employ a plural. Their narrow grade appears in the possessed form, to be next considered. Possessed Forms Possessed nouns usually do not take the suffix -tla, but if animate [these] take a suffix *-wa, evidently low-toned when final. Stem unablauted: 'my child' *no[-]kone-wa > no[-]kone.-wA > nokone.uhW

76 Whorf has a footnote here, which reads: Footnote 1: kolotl is not the best instance I could have chosen, as I find it regularly forms plural by reduplication, kokolo'h, but it will serve this purpose schematically: siwa-tl 'woman', with -me'h plural, would have been better. 77 Compare here PUA **-mi, Proto-Nahua *-m(i) 'plural', which became -n word-finally in general Nahua (not in Pochutec). Whorf's -tla- in these forms, which he apparentlyrelates to Nahuatl 'absolutive noun suffix' (Proto-Nahua *-tla), is rather more directly related to Nahua 'ligature' (*-ti-). PITCHTONE AND THE "6SALTILLO"9IN NAHUATL21 221

'my children' *no[-]kone...w'a-.m`i> no[-]kone.-wa-mi > nokone~-w'an In the Tepoztlan dialect final wyeverywhere has apparently first been nasalized to m and then undergone the regular change of final m to n; hence the first form above in T is: 'my child' *no[-jkone..w'a > no[-]kone.mA> nokon'e*n [The] stem [is] ablauted in possessed form only: Possibly low tone on [the] end of [a] word was associated with the ablauting of these forms. 'Pot' *komi-tl-3> kom'itl 'my pot' *no k'o'kmi[>] noQ k'omI> nokbon no plural Stem ablauted throughout:'man (vir)', e.g., *okitci..tlo> okitctli 'my man' *no[-]okitci.w a n[]kitclw > nok'itcW> nok'itc 'men' *okitci-t1.3m`1> okitci-Ci-mi1> okitcfln 'my men' *no[-]okitciw'am`1 > no[-]okitci wa~m>nktw Unusual possessed forms in classical [Nahuatl]: Sandoval's "Arte de la lengua mexicana" gives for "my man" the following alternative forms be- sides "noquich": "noquichhui, noquichtiuh, noquichtihuatzin" (the last reverential). The first may be explained as [a] conversion of the weakened wa to wi by vowel assimilation (or simply by a narrowing effect), fol- lowed by strengthening and persistence of this vowel. The other two are exceptional cases where -tlo was not dropped in the possessed form. *no[j]okitcLwa > no[-]okitciwk> no[-]okitciw > nokitcwi cf. 'apron' *mactla..tlJ> mactliatlhut with ablaut 'my apron' *no[j]mactla> no[-]mactl' > nomactli ['my man'] no [-]okitci-tl.3-w'a > no[-]okitci-ti-wA > nokitct7PuhWand with reverential -ts'in:

The saltillo at the end of plural verbs ending in -k&'h(preterite and fu- ture) may represent '-t15 (-k&'-tl;'),as -ks seems to answer to a UA plural suffix. The plural of the imperative, -k'an, may be from -k'a-mi. The final saltillo[j] which is the sole plural suffix for verbs in the present and im- perfect[,] would seem to be some low-toned suffix with plural signifi- cance, possibly '-kbi. The saltillo with the suffixes forming nouns denoting the possessor of a thing, -w'a'h,-k'c'h, -4', -ybo'h,is probably '-tlo'. All this may lead one to suppose that the puzzling and seemingly very irregular saltillos in the couplets cited by Rinc6'n may be explained by patterns of rhythm, accent, and tone, though we cannot yet make out what these patterns are. I am not aware that writers on any of the other Uto- Aztekan languages have mentioned anything analogous to the saltillo in one of these tongues. However something apparently very similar is 222 INTERNATIONALJOURNAL OF AMERICANLINGUISTICS found in Otomi, a language wholly unrelated to UA but geographically adjoining and in places enclaved by Nahuatl. It is given the same name "saltillo" and described similarly to the Nahuatl saltillo by the Spanish grammariansof this tongue.

Conclusion

Pitch tone is clearly an important linguistic factor in both modern and old Nahuatl. In modern Nahuatl it is generally determined by phonetic form, but also to some extent by grammatical form; also semantic tone exists to a small extent. Tone as a morphological means is also found to a certain extent, apparently as an outgrowth from the association of tones with grammatical forms. In old Nahuatl there would seem to have been less connection between tone and phonetic form, but more between tone and grammatical form, and much more semantic tone, than at present. The saltillo appears to have diverse origins, though possibly all based on a common principle which I cannot yet confidently state. However, in many cases it seems to stand in the place of a former "weakened"- where weakened means unstressed, feebly articulated, and usually low- toned-mora, or moras, which moras have retained the elements of length (duration, or prosodic quantity), glottal action, and breath, but nearly or quite lost those of oral action and sonancy. Intermediate stages between the original weakened moras and the saltillo persist in cases where the moras contained the sonant labials w, qw, m; in the forms of uhWI, uhW, uhW, etc. and n, N. The distinctions between these various intermediate forms and between them and the saltillo is best preserved in Milpa Alta of the dialects I studied, while the least distinction is found at Tepoztlan, where all the forms including saltillo are leveled to either h or n. Both tone and saltillo are bound up with questions of rhythmical and accentual patterning, with the ablaut of stems, and with other contracted forms, and with similar phenomena in the other Uto-Aztekan languages.

9 Wolcott Gill Road, Fairfield, Conn. Benjamin L. Whorf

REFERENCES

ANDREWS,J. RICHARD.1975. Introduction to Classical Nahuatl. Austin: University of Texas Press. BARRITT,WESTBROOK. 1956. The phonemic interpretationof 'accent' in Father Rinc6n's "Arte Mexicana." General Linguistics 2:22-29. BIERHORST,JOHN, trans. 1985. Cantares mexicanos: Songs of the Aztecs. Stanford, Calif.: Stanford University Press. BRIGHT,WILLIAM. 1960. 'Accent' in Classical Aztec. IJAL 26:66-68. PITCHTONE AND THE "SALTILLO"IN NAHUATL 223

CAMPBELL,LYLE. 1985. The Pipil language of . Berlin: Mouton de Gruyter. CAMPBELL,LYLE, AND RONALD W. LANGACKER.1978. Proto-Aztecan vowels: parts I-III. IJAL 44:85-102, 197-210, 262-79. CAMPBELL,LYLE; VIT BUBENIK; AND LESLIE SAXON. 1988. Word order universals: refinements and clarifications. Canadian Journal of Linguistics 33:209-30. CANGER, UNA. 1980. Five studies inspired by Nahuatl verbs in -oa. Travaux du Cercle Lin- guistique de Copenhague, no. 19. Copenhagen. CAROCHI,HORACIO. 1645. Arte de la lengua mexicana con la declaracion de los adverbios della. Mexico: Juan Ruiz. [Frequently reprinted;the most accurate is the facsimile repro- duction of the 1645 version, Mexico: Universidad Nacional Aut6noma de Mexico, 1983.] CARROLL,JOHN B. 1956. Introduction. Language, Thought, and Reality: Selected Writings of Benjamin Lee Whort, ed. John B. Carroll, pp. 1-39. Cambridge, Mass.: The M.I.T. Press. DAKIN,KAREN. 1982. La evoluci6n fonologica del protonahuatl.Mexico: Universidad Nacio- nal Aut6noma de Mexico. GREENBERG,JOSEPH H. 1963. Some universals of grammarwith particularreference to the or- der of meaningful elements. Universals of Language, ed. Joseph Greenberg, pp. 73-113. Cambridge, Mass.: The M.I.T. Press. KARTTUNEN,FRANCES, AND JAMES LOCKHART. 1987. The Art of Nahuatl Speech: The Bancroft Dialogues. UCLA Latin American Studies, no. 65. Los Angeles: UCLA Latin American Center, University of California, Los Angeles. LANGACKER,RONALD W. 1977. An Overview of Uto-Aztecan Grammar. Studies in Uto- Aztecan Grammar,vol. 1; Publications in Linguistics, no. 56. Arlington: Summer Insti- tute of Linguistics and University of Texas at Arlington Press. LASTRADE SUAREZ,YOLANDA. 1986. Las areas dialectales del Nahuatl moderno. Instituto de Investigaciones Antropol6gicas, Serie Antropol6gica, no. 62. Mexico: Universidad Na- cional Aut6noma de Mexico. LEHMANN,WALTER. 1910. Ergebnisse einer Forschungsreise in Mittelamerika und Mexiko, 1907-1909. Zeitschrift fir Ethnologie 42:687-749. .1920. Zentral-Amerika.Berlin: Dietrich Reimer. MCMAHON,AMBRISIO, AND MARIA AITON DE MCMAHON.1959. Cora y espafiol. Serie de Vocabularios Indigenas Mariano Silva y Aceves, no. 2. M6xico: Instituto Lingiiistico de Verano. RINC6N, ANTONIODEL. 1595. Arte mexicana. Mexico: Pedro Balli. SAHAGUN,FRAY BERNARDINO DE. 1979. Florentine Codes, trans. Charles E. Dibble and Arthur J. O. Anderson. Santa Fe, N.M.: School of American Research and the University of Utah. SAPIR, EDWARD.1913-14. Southern Paiute and Nahuatl: a study in Uto-Aztekan, pts. I and II. Journal de la Soci6et des Americanistes de Paris 10:379-425; 11:443-88. SCHOEMBS,JAKOB. 1949. Aztekische Schriftsprache: Grammatik (mit Lautlehre), Text und Glossar. Bibliotek der allgemeinen Sprachwissenschaft; 3. Reihe: Darstellung und Unter- suchungen aus einzelnen Sprachen. Heidelberg: Carl Winter. WHORF,BENJAMIN L. 1937. The origin of Aztec tl. American Anthropologist 39:265-74. ~___ . 1946. The Milpa Alta dialect of Aztec (with notes on the Classical and the Tepoz- tlan dialects). Linguistic Structuresof Native America, pp. 367-97. Viking Fund Publica- tions in Anthropology, no. 6. New York: Viking Fund.