David H. Tuggy Turner

Field Linguist

All Affix and No Stem:

Orizaba Nawatl tlahtia

David Tuggy

This paper was presented at the 20th anniversary celebration of the linguistics department of UC San Diego about 1986 (I don’t remember the date). It was accepted for publication in the volume of papers from that meeting, but that worthy project, unfortunately, either was stillborn or is undergoing an extraordinarily long gestation period. I have updated the manuscript only slightly from the version for oral presentation, changing the name Orizaba Náhuatl to Orizaba Nawatl, to follow the most widely-used current orthography, and correcting a few minor errors and infelicities. A well-populated Unicode font such as Gentium or Doulos SIL is recommended for optimal viewing.      The particular example of tlahtia has been discussed in some of my subsequent  publications, most extensively in Tuggy (1992).


Linguistics is attractive because languages surprise us, yielding data which, though reasonable, are beautiful in unpredictable ways, delighting those who discover them. The Orizaba Nawatl verb stem tlahtia is improbably constructed of a prefix tlah- ‘unspecified object’ and a suffix -tia ‘causative’, with no stem between. Its meaning is equally surprising: “give a baptismal or wedding garment to a godchild”. Several patterns combine naturally, though unpredictably, here, including (1) the development of tlah-’s meaning to include “culturally expected/obvious object”, (2) -tia’s use to mean “cause to have” when suffixed to a noun stem, and (3) a gradation from affixal to stem status, such that either affix may also be correctly viewed as a stem.

1. Introduction

Ronald Langacker, in his 1972 book Fundamentals of linguistic analysis, said that he hoped that working through the discussion and examples in the book would give his readers some notion of why anyone would choose linguistics over more lucrative professions like jewelry robbery or TV repair. Although some of our number have succumbed to the lure of such greener pastures, I’m sure you all have experienced the sort of thing Langacker was talking about: the pleasure of discovery, the joy of finding the perfect example to shoot a great gaping hole in someone else’s theory (or even your own), or of seeing a brand-new generalization emerge from a range of data you’d never put together before, or of finding an example that combines two or three patterns in an unexpected way to produce an astounding and completely delightful result. For me, that’s what linguistics is all about.

I’d like to tell you about one such case: a form that, when I found it in Orizaba Nawatl, kept me happy for several weeks. It combines several Nawatl patterns I previously had noted, in as obvious yet unexpected a way as a good pun, and at the same time it exemplifies a number of general principles of language which I have come to think important.

The form is a transitive verb stem, tlahtia, and it means, more or less, “to give someone a baptismal or wedding garment, as this is your duty as the person’s godparent.” It appears in sentences like (1).[*]

(1)  Ihkwak ō-ni-mo-nāmiktih, ō-nēch-tlahtih  n  Severino Ernandes.
  When  past-I-refl-marry.pret,  past-me-TLAHTIA.pret   the Severino Hernández.
  “When I got married, Severino Hernández, as my godfather, gave me my wedding garment”


There are in Nawatl two other nearly homophonous transitive verb stems, one (tlatia) meaning “burn something” and the other (tlātia) meaning “guard/hide something.” When I first heard tlahtia, soon after I started work in Orizaba, I thought it was a variant of one of these, and tried to relate the meaning I was given to one or other of them, but without much success. Then, as I puzzled over it, I suddenly saw “Oh, that’s what it is!”, and started to laugh. What it is is a combination of the prefix tlah- “unspecified object” and the suffix -tia “causative”, with no verb root between them. To appreciate the beauty of the construction, you will need to know a little more about tlah- and about -tia.


2. tla-

The prefix tlah- (with its more common variant tla-) is ubiquitous in Nawatl. Its central, or prototypical, meaning is to designate an unspecified object of a transitive verb stem. It functions in the same slot as any other object marker, following the subject marker and preceding the verb stem. In this usage, a verb stem with tla- can usually be translated by an English verb with things or something as its object, or by an intransitive verb related to the transitive verb. The verbs in (2) exemplify this pattern.

(2)  ni-k-nemaka  “I sell it”     ni-tla-nemaka “I sell (things)”
  I-it-sell     I-TLA-sell  
  ni-k-mati “I know it”   ni-tlah-mati / ni-tla-mati “I am knowledgeable, wise”
  I-it-know     I-TLA-know  
  ni-ki-htowa “I say it”   ni-tla-htowa “I say things, talk (disrespectfully)”
  I-it-say     I-TLA-say  
  ni-ki-hkwilowa “I write it”   ni-tla-hkwilowa “I write (something)”
  I-it-write     I-TLA-write  


Now there are various reasons why people would choose not to specify what the object of a transitive verb is. All of these reasons become part of the meaning of different sub-cases of tla-, by a process so omnipresent it should be a central feature of every semantic theory (though it isn’t of most). For instance, people may choose not to specify an object because they don’t care to let their hearers know what it is, or perhaps because they do not know themselves. These reasons lead to different kinds of non-specific readings of tla-; i.e. tla- comes to mean “unspecified because unknown to speaker and/or hearer”. Sometimes there are many objects, too many to be worth enumerating, or to be enumerable at all. Such cases give rise to “general” or “diverse object” usages. Sometimes the object is unspecified because it is insignificant or unimportant; tla- takes on an “unimportant object” nuance through such usages.

Sometimes the opposite reason leads to non-specification: the object, far from being insignificant, is too well known to need specification. This is paralleled in English: note the examples in (3), where the established intransitive usages of the transitive verb shave imply the shaving of an obvious object, or where the strongly transitive verbs throw or hit can be used intransitively in a context where it is obvious what the object is.

(3)  He shaved.  (I.e. “he shaved his face/*legs”)       She shaved.  (I.e. “she shaved her legs/??face”)
  *He threw.     (OK) He threw to first base just in time.
  *He hit.     (OK) He hit safely in four of his first five at-bats.


Nawatl uses tla- in cases like these, as the examples in (4) show. The first example under (4) is somewhat subtle, but the point is that although the object of ni-k-kwa will normally be edible, it needn’t be. It would be natural to use the word if you wanted to describe eating, say, a rock, whereas it would be very strange if not impossible to use ni-tla-kwa to describe the same situation. tla- doesn’t mean just “something”, but rather “the expected thing”, not just “unspecified object”, but rather “canonical, normal object (which therefore need not be specified)”.

(4)  ni-k-kwa “I eat it”       ni-tla-kwa “I eat (food)”
  I-it-eat      I-TLA-eat  
  ni-k-mati “I know it”   ni-tla-mati “I am literate, I know how to read and write”
  I-it-know     I-TLA-know  
  ni-k-kaki “I hear it”   ni-tla-kaki “I understand (speech)”
  I-it-hear     I-TLA-hear  
  ni-k-payana “I grind it”   ni-tla-payana “I grind tortilla dough”
  I-it-grind     I-TLA-grind  
  ni-k-chipāwa  “I clean it”   ni-tla-chipāwa  “I weed the cornfield”
  I-it-clean      I-TLA-clean  


What is the normal or canonical object of course depends very much on the culture. Grinding tortilla dough is not an important part of the American culture, so a reading of grind parallel to ni-tla-payana would be surprising in English, just as a parallel to hit in (3), meaning “hit a baseball”, would be odd in Nawatl. Other examples can be found where the obvious Nawatl object is even more surprising to us, such as those in (5).

(5)  ni-k-tolowa “I swallow it”     ni-tla-tolowa “I partake of the host (at mass)”
  I-it-swallow     I-TLA-swallow  
  ni-k-totōnia  “I heat it”   ni-tla-totōnia “I sun-dry gladiola bulbs or coffee /I reheat tortillas”
  I-it-heat     I-TLA-heat  


Another aspect of tla- that needs mention is that it is used in cases where it is not the primary object of a transitive verb stem. It may be a secondary object of a transitive verb stem, or an object or subject of an intransitive verb stem. In many of these cases tla- takes on strong adverbial nuances, which do not particularly concern us here. The examples in (6) are of such usages.

(6)  ni-k-maka “I give him (something)” (*“I give it (to someone)”)
  ni-k-tla-maka  “I feed him (i.e. give him the expected object, namely food)”
  ni-kīsa “I emerge”  *ni-k-kīsa *“I emerge from it (etc.)”
  I-emerge    I-it-emerge  
  ni-tla-kīsa “I go outside”
  nēsi “it appears”  tla-nēsi “it dawns (i.e. things appear all over)”
  appear    TLA-appear  


It is significant that in all these usages, tla- parallels the behavior of incorporated nouns. Such nouns, as exemplified in (7), (a) prototypically are direct objects of the verb stem which incorporates them, (b) appear directly in pre-stem position, (c) may also be secondary objects of transitive or intransitive stems, or subjects, or may have adverbial effects of various kinds. Furthermore, (d) some verb stems have specialized meanings when they appear with incorporated nouns, and have those same specialized meanings with tla-. For instance, pia means “have” in its normal transitive usage, but when used with an incorporated object, it means “guard (a crop)” or “herd (animals)”; it has that same specialized meaning when used with tla- as well. The differences between tla- and the class of incorporated nouns can be attributed to its highly schematic (i.e. unspecified) meaning, and to its prefixal status.

(7)  ni-k-nemaka “I sell it” ni-toma-nemaka “I sell tomatoes”
  I-it-sell   I-tomato-sell   
  ni-k-melawa “I straighten it” ni-k-īx-melawa “I plane it”
  I-it-true   I-it-face-true  
  ni-pātz-miki “I suffocate” ni-kēch-pātz-miki “I choke”
  I-pressure-die   I-neck-pressure-die  
  ni-k-pia “I have it” ni-chiboh-pia “I herd goats”
  I-it-have   I-goat-have  
  ni-toma-pia “I guard the tomato crop” ni-tlah-pia “I herd/guard (a flock/crop)”
  I-tomato-have    I-TLA-have  


Such affixal status, I believe, amounts to the expectation (produced by usage) that the affix will be used with a stem. Constantly finding tla- used with a stem gives rise to the expectation, even the requirement, that it be used with a stem the next time, and that is what it means to be affixal. The incorporated noun kwā- “head” provides a direct parallel (see examples in (8)): through the historical adoption of a different noun (tzon-tēko) for “head” when used independently, kwā has come to be used only in incorporational constructions. Since it is always so used, it is always expected to be so used, and is thus a prefix, as well as (in other respects) an incorporated noun.

(8)  no-tzon-tēko “my head” *no-kwā *“my head”
  my-hair-gourd   my-head  
  no-kwā -tix “my brains” ni-kwā-totōnia “I think too much”
  my-head-dough   I-head-heat  
  ni-k-kwā-ā-tēkia “I baptize him”    


One other example of tla- that is quite relevant to our concerns is in (9). In ni-tla-kōwa tla- again means “normal, expected object”. In ni-k-tla-kōwi-lia it means “normal, expected secondary object”, much as it did in ni-k-tla-maka “I feed him” ((6)). But this time the object that was so obvious that it didn’t need to be mentioned is another surprising one (to us), though not so to the Nawatl speakers, who rarely buy gifts for others except in the ritual of giving clothing for a godchild’s baptism or wedding.

(9)  ni-k-kōwa “I buy it” ni-tla-kōwa “I do the shopping”
  I-it-buy   I-TLA-buy  
  ni-k-kōwi-lia “I buy him (something)”    
  ni-k-tla-kōwi-lia “I buy him baptismal/wedding clothing (I being his godparent)”


This finishes our survey of tla-. The important points to remember are that (a) it prototypically designates an unspecified object of a verb stem, though it may also designate other nominal arguments as unspecified, or be adverbial. (b) In many ways it can usefully be considered to be a schematic (i.e. unspecified as to detail) incorporated noun, though (c) since it is always in construction with a following verb stem it is naturally prefixal. (d) One of several meanings of tla- which have arisen through usage is “object so obvious as not to need mentioning”, or “canonical, expected object”. Finally (e) some things that are obvious or expected for a Nawatl speaker are not so obvious to us: in particular, the notion of a godparent buying clothing for his godchild’s baptism or wedding looms much larger in their culture’s gift-giving practices than in ours’.

3.           -tia

The suffix -tia is also very common in Nawatl. Its prototypical or central meaning is “causative”: it is used on verb stems to denote the causing by a subject of the stem’s execution by a direct object, as the examples in (10) illustrate.

(10)  ni-miki “I die”     ni-k-mik-tia “I kill him”
  I-die      I-him-die-TIA  
  ni-kalaki “I enter”    ni-k-kalak-tia “I stick / bring it in”
  I-enter      I-it-enter-TIA  
  ni-k-mati  “I know it”   ni-k-mach-tia “I teach him”
  I-it-know     I-him-know-TIA   


-tia, like tla-, has a range of different, closely related meanings that correspond to different usages. Most of these do not particularly concern us, but one group of usages that does is that in which -tia is suffixed to a noun. Some examples of this construction are given in (11).

(11)  ni-k-kal-lō-tia “I have him as guest in my home”
  ni-k-ama-tia “I issue him documents”
  ni-k-koton-tia “I put his shirt on him”
  ni-k-misah-tia “I celebrate a mass for him”
  ni-k-pah-tia “I put insecticide on it (a plant)”


As can be seen from the examples, in every case the noun -tia combination produces a transitive verb stem, whose subject causes some relation to obtain between a direct object and the noun in question. In the first example that relationship is a “being” relationship: the host causes his guest to “be” (in an attenuated sense) the owner of his home. The other examples involve variations on the notion of “having”. In the second example the object comes to possess certain papers and to benefit both from them and from the fact that the subject has a record of having given them to him. In the third case the relationship is that of wearing the article of clothing, in the fourth that of profiting (through mitigation of purgatorial suffering) from the celebration of the mass, and in the fifth that of the plant having the insecticide physically on it, and profiting from its bug-killing powers.

One of the meanings of -tia, then, is “cause to have”, or, as we would more normally say it in English, “give”. -tia differs from the verb stem maka “give”, however, in that it occurs only following a noun in this meaning. As I claimed for tla- that it is prefixal because it always has, and therefore always must have, a verb stem following it, so I can claim for -tia that it is suffixal because it always has, and therefore always must have, a stem preceding it (in this usage, a noun stem). Thus it makes perfectly good sense to consider -tia to be a verb stem which only occurs as the head of incorporational constructions (verb incorporations in (10), and noun incorporations in (11)), much as kwā- (in (8)) is a noun stem that occurs only as the incorporated member of such constructions.

4. tlahtia

It seems unusual to have a verb that is all affix and no stem, but it all depends on your point of view. If you assume that the universal grammar, eternally enthroned in the heavens, or perhaps hardwired since time immemorial in the mind of man, has laws saying “once an affix, always an affix,” and “thou shalt not permit stemless verbs”, then a form like tlah-tia will certainly be disconcerting. However, if, as we have suggested, affixes like tla- and -tia may be viewed as stems which are invariably used in incorporational-type constructions, the problem goes away. We can correctly place tlah-tia in the paradigm of prefix-stem constructions (along with e.g. tla-htowa and the other examples in (2)), or put it with the stem-suffix structures (such as pah-tia et. al. in (11)), or with the stem-stem incorporational structures (examples in (7)). -tia is a verb stem which always incorporates a stem before it, and which means that a subject causes a direct object to do the stem (examples in (10)), or to be or have the stem (examples in (11)). tlah-, similarly, is a schematic, i.e. a highly unspecified, noun stem, which always occurs incorporated by a verb stem, and one of whose meanings is “canonical, expected object”. What could be more natural than to combine the two of them? The projected meaning would be “give”, or “cause to have” ( -tia) “the canonical, expected thing” ( tlah-). And, as we saw in the case of ni-k-tla-kōwi-lia in (9), an important gift-giving situation (therefore easily construed as canonical or expected) is that in which a godparent gives his godchild a wedding or baptismal garment. All the pieces fit together nicely, and we suddenly understand what tlahtia is all about.

We have not, of course, predicted, or exhaustively accounted for, tlahtia. There is no reason that it had to occur (it doesn’t in related dialects), or that it had to have exactly the meaning it does. But that is a problem only if we assume a strict dichotomy between the arbitrary and the predictable, and believe that there can be no true explanation or understanding of a phenomenon unless it can be predicted. If we exchange the notion of that dichotomy for the notion of a continuum of reasonableness, with the totally arbitrary and the absolutely predictable as theoretical endpoints, but with most of language falling somewhere in between, it will neither surprise nor upset us that we cannot explain tlahtia away. We do not have to see through it in order to see it.

5. Commercial

So much for the show; now for the commercial. I would like to explicitly mention several aspects of language that we have touched on in our discussion which are perhaps not generally apprehended, and which I have come to believe are quite important.

One is the necessity of avoiding artificial compartmentalization: in this case the splitting up of stems and affixes into strictly separate kinds of morphemes. Many categories we linguists have set up differ not in kind but in degree, and it is important that we recognize that fact, and let our analyses reflect the similarities and the lack of strict boundaries, as well as the differences.

Another is the importance of usage in establishing meaning. Usage and meaning are not the same thing, but each is the major determinant of the other. Words and morphemes and grammatical constructions mean what they do because of how we use them, and we use them as we do largely because of what they mean.

Another matter is what is sometimes termed “encyclopedic meaning”, the ultimate impossibility (if we are not to go wholesale into the business of denying validity to clearly valid insights) of dividing off linguistic knowledge from knowledge of the world. Only in the context of the whole Nawatl culture can meanings such as that of tlahtia be understood and appreciated, and it does little good for us linguists to call it pragmatics or some other impressive-sounding name and shove it into a corner and ignore it.

Finally, there is the unpredictability of language. Language is founded on usage, and until the human choices and actions and thoughts that constitute usage can be predicted, language cannot really be predicted. I rejoice in the conviction that language will always surprise us, that forms like tlahtia will continue to show up, which, although perfectly reasonable, partake more of the spontaneous grace of an artistic creation than of the stodgy predictability of an assembly-line product. I do not believe that we will ever see through language to the point where it becomes dull and boring; rather it will continue to amaze and delight us.


Langacker, Ronald W. 1972. Fundamentals of linguistic analysis. New York: Harcourt Brace Jovanovich.


[*] The following conventions and abbreviations are used in glossing: pret = ‘preterite’, refl = ‘reflexive’. tla(h)-, -tia and tlahtia are glossed TLA, TIA, and TLAHTIA respectively. ‘He’, ‘him’ and ‘his’ are used in their generic senses to represent third person singular pronominal elements when the most likely referent is human; ‘it’ is used when the most likely referent is non-human.