Why the oddness of reflexives isn’t odd


David Tuggy

Summer Institute of Linguistics

[This paper was originally published in 1985, in the Southwest Journal of Linguistics VIII.1 44-59. This version for the Internet, prepared in 2003, contains, at least in intent, the original text unchanged except for typographical adjustments, a few corrections, and redrawing of the figures. A well-populated Unicode font such as Gentium or Doulos SIL is recommended for optimal viewing. For a much fuller discussion, in the Cognitive grammar framework, of the issues surrounding Spanish se, see Maldonado (1999).]

Reflexives behave in the same sorts of odd ways in many languages. In particular they are often used in a way that must be translated into English by passive constructions; a form that literally says “he hit himself” really means something more like “he got hit”. Consideration of the intrinsic nature of reflexivity and its implications can make this odd behavior seem more natural and understandable.

1. Introduction.

Reflexives are the structures used to code situations in which a person or thing does something that he would normally do to someone or something else, but does it to himself. The reflexive usually is a kind of object marking, either marking the object of a verb directly as a 1st, 2nd, or 3rd person like the subject, or simply marking it as being the same as the subject. However, reflexives in many languages do things that other object markers do not. In particular, they often change the meaning of the verbs they are used with, in funny ways. One pattern that shows up repeatedly in many languages is using a reflexive to make a passive form of the verb: the form seems like it ought to mean “he hit himself”, but in fact it means something more like “he got hit”.

I would like to investigate why reflexives are odd in these ways. Besides English, we will be paying particular attention to reflexives in Spanish and Náhuatl,[1] two unrelated languages with remarkably similar patterns which I happen to know something about. We will lay out a plausible series of connections between the straight reflexives and more deviant usages, including the reflexive passives. The analysis will be approached from the perspective of Cognitive Grammar (Langacker 1982, 1985, 1986, 1987a, 1987b, Casad and Langacker 1985, Lindner 1982, Tuggy 1981).[2]

2. Preliminaries.

2.1 Active Zones.

There are a number of concepts which do not bear immediately on reflexivity, but which will be useful in our later discussion.

One is the concept of Active Zones. When two things interact, it is almost always the case that not all parts of them are equally involved in the interaction. When John hits Bill, not all of John actually comes in contact with all of Bill, but only, say, John’s hand (and in fact a subportion of his hand) actually hits a subportion of, say, Bill’s face. When John sees Bill, his mind, that is, the cognitive, perceptive and recognizant aspect of him, and his eyes, as organs of sensation, are the parts of him that are directly involved in the process. Similarly, it is the outer surface of Bill’s body, typically including clothing, and, interestingly, his eyes, but only their outer surface, that enters directly into the seeing relationship. That subportion of a participant in a process which is directly involved in the process we will call an Active Zone. When John hits Bill, John’s hand and Bill’s face are Active Zones; when John sees Bill John’s visual sensory system and his mind, and the outer surface of Bill, are the Active Zones.

2.2 Variations in scope.

Another important notion is scope. The scope of a concept is the extent of what it designates or names. A given structure may designate a greater or smaller amount of semantic material under different circumstances. For instance, the idea of jumping prototypically includes in its scope an upwards and a downwards motion of its subject, but it often designates just an upward or just a downward motion. It may also designate salient horizontal motion, or include as a salient feature the area jumped over. It may include (or exclude) the muscular movements which precede (and some of which cause) the jumping, or muscular movements and body posture following the jumping.

All of these aspects of a typical jump may or may not be judged important in a given communication situation. We can put the word jump into different constructions which will have the effect of slightly changing its scope to include or exclude different aspects. For instance in jump up or jump (up) on to (s.t.) downward movement is not important, probably not to be construed at all, in parachute jump or jump down off of (s.t.) upward movement is not included, in jump (s.t.) we expect some particular object to be within the space jumped over, and so forth. We can expand or contract the scope of the word, by using it in different constructions.

2.3 Non-causative transitivity.

Notice the transitive usage of jump as in jump the fence. When jump is used in such a construction, there is within the concept of jumping the notion of an obstacle over which the jumping takes place. (Note that an obstacle may be assumed to be present even without a syntactic direct object, as in He’s jumped 2" higher than anyone else on the track squad.) This obstacle over which the jumping takes place is very salient or important in the meaning of this version of jump; we will be calling it an internal object, and the Thing that is actually doing the jumping, which is even more salient, we will call the internal subject.[3]

This is a very common construal for concepts involving motion: the moving object is taken as the most salient Thing in the situation, the internal subject; and if there is another Thing relative to which the movement of the internal subject is readily calculated, it is taken as the second salient Thing in the situation, the internal object. This particular kind of internal object we will call a landmark.

Figure 1: jump (with landmark object)

2.4 Causative Extensions.

A particularly important kind of scope extension is extension to include a causing agent. With the word jump we accomplish this extension when we use structures like jump a horse (over a fence), or jump a checker down the board. We will call this kind of extension a causative extension.

Figure 2: jump (with causative extension)

Several things are worthy of note here. One is that there is a strong tendency to count the causing agent as more salient or important than the person or thing that is most prominent in the caused action. That is, it is the causing agent that is typically picked as internal subject of a causative structure, and the person or thing most prominent in the caused action is construed as having the second degree of prominence, that of the internal object. Where there is not a causative extension of the concept jump, the person or thing jumping is naturally the most salient person or thing construed, and is thus the internal subject; but when the concept is construed with a causative extension, the causing agent tends to be given higher salience than what actually does the jumping, and thus to become internal subject.

Secondly, note that many if not most of the concepts that we typically code by transitive verbs already come complete with a causative extension of some simpler notion. In the verb throw, for instance, there is a causing agent who causes something else to go flying through the air. That causing agent is the internal subject, and the thing that goes flying is the internal object. Similarly, in kill you have as internal subject an agent who causes the internal object to die.

Finally, note that the effect of passives, at least if you leave off the by-phrase, which many linguists have argued is a distinctly secondary part of the passive construction, is to de-specify the causing agent, leaving you with a picture of an event being caused, but it not being made clear by whom. Thus be thrown codes the concept of something (now the most salient thing in the picture) flying through the air, presumably under external causation, but without it being specified who is doing that causing. Similarly, be killed is almost synonymous with die; the only difference being that we know that the dying is saliently caused by something or someone. Thus, passives come close to undoing a causative extension.

2.5 Multiplicity of causing factors.

One further important idea is that in many situations there are a number of factors that might properly be considered causal in varying degrees. In a given accident, for instance, one might without contradiction blame driving too fast, d. u. i., cornering too late, bald tires, an improperly banked corner, the lack of adequate warning, rain on the road, a too-slippery road surface, and who knows what all else. It is rarely the case that something happens for only one reason. And any such cause can be singled out or ignored, according to how the speaker chooses to construe the situation.

3. We cannot do things to ourselves as we do them to others.

3.1 Mild differences.

The basic idea of reflexivity is doing something to yourself instead of to somebody else. In a reflexive you are both the subject and the object. Now in one sense that is nothing but a simple substitution of one kind of situation for another one. If you see someone, or hit someone, and that person happens to be yourself instead of somebody else, what’s the difference? I want to claim that a lot of the oddness of reflexives comes from the fact that you can’t hit yourself, or even see yourself, in exactly the same way as you would someone else.

The concept of Active Zones (2.1) here becomes relevant. When we see ourselves our active zones as internal subjects are the same as when we see others, but our active zones as objects are not the same. Others can see our body as a whole, including as perhaps the most salient part of us our face; but we cannot, without the aid of mirrors or pictures, see our whole body at once, and in particular we cannot see our face (or any other part of our head, or our back). Besides this difference in the completeness with which we can see ourselves, there are other differences: perspective, for instance. We can, with difficulty, see our shoulders, but we do not normally see others’ shoulders from any such angle. Another difference is that we can see others only when they are present with us (and most people are more often absent than present), whereas we ourselves are (hopefully) always present, and thus available to be seen. Or, again, since we can so readily see ourselves, we relatively rarely do so; we are conscious of seeing others perhaps hundreds of times in a day, but not nearly so often of seeing ourselves.

And here we are, I think, at the starting point of why reflexives are odd. It is built into the nature of our experiences that we cannot and do not interact with ourselves as we do with others; in this case we cannot and do not see ourselves as others see us or as we see them. This amounts, on the Cognitive Grammar view, to a difference in the meaning of see as construed reflexively and non-reflexively.[4] It may seem a small difference: one might respond, “Yes, there are differences, but isn’t it much more important to notice the similarities? Isn’t it the sameness of the experience that is being coded by using the same verb in a reflexive and a non-reflexive structure?” And of course that is right. But the differences are there nonetheless, and they often prove to be important.

I will only give one example here, though several could be made for see. By a natural semantic extension (attested in English, Spanish, Náhuatl, and who knows how many other languages) the form see can come to designate not only the prototypical process of seeing, but the whole interaction with another human being that typically happens when we are most conscious of seeing them. It is the meaning of see in sentences like I’m going to S.C. to see my folks, or Have you seen a doctor about that?, or See you tomorrow. But the naturalness of that extension depends on the fact that we typically can see people when we interact with them, and that when we cannot see them it is usually because they are not present, and thus are not accessible for interaction. In contrast, when we interact with ourselves, seeing ourselves is not typically an important part of the interaction. It is no accident than that reflexives with this sense are strange, if not downright unacceptable. Who would ever say I’m off to see myself or anything of the kind? (Parallel sentences in Spanish and Nahuatl are just as bad.)

It makes little difference whether the non-reflexive usage changes meaning while the reflexive stays put, or whether the reflexive changes meaning while the non-reflexive stays put: the non-reflexive construal of almost all verbs is more salient and will be thought of as normal, so the reflexive will look deviant.

3.2 Drastic differences.

In the case of seeing, the differences are minor, but we have seen that they have consequences. There are many other concepts for which the differences are much more pronounced. You cannot in any prototypical sense throw a rock at yourself, much less give birth to yourself, or marry yourself, or plow furrows in yourself, or carry yourself on your back or sit on your lap; you do not sign contracts with yourself or give yourself Mass or accompany yourself or snub yourself or any of dozens of other things that you might do to other people or things.

One consequence is that with some verbs a reflexive may virtually never be used, and may be rejected by native speakers as impossible. I have never to my recollection heard anyone speak of either a farmer or a field plowing himself or itself; nor do I recall hearing se ara in Spanish, nor mo-kʷenƛāliya in Náhuatl.[5] Similarly we do not talk of carrying yourself on your back, and neither se carga en la espalda nor mo-māma is used (except as “reflexive passives”.

Another common result is that the reflexive is used, but with its meaning adjusted so as to “make sense”. People write words, not people, and thus not themselves. However, in Náhuatl the verb “write” can have the reflexive form mo-hkʷilowa, but it has the meaning “sign oneself up” rather than just “write”. This is a very natural construal (though one we do not have in English): for Náhuatl speakers a person’s name is the aspect of himself that can be written, so when a person writes himself his active zone as object is his name. The same thing happens in the Spanish apuntar-se “put your name down”. In English we take a different tack: we expand the scope of write to include a resultant state of someone receiving a message through what is written, and take that person as object instead of the written message; at least that is how most of us would interpret (and use) I wrote myself.

4. Internal vs. External Causation.

4.1 We do not move ourselves as others move us.

Depending on their shape and orientation, Náhuatl has three main verbs to denote placing objects. If you place a relatively long object with its long axis horizontal, you can use the verb tēka, which we can translate “lay”; if it is long and you place it with its long axis vertical, you can use kec̸a, or “stand”, and if the object is not saliently long and you put it in a stable position, you can use the verb ƛāliya, which we can translate “set”. What happens if you try to reflexivize these verbs? Can you lay yourself (down), or stand yourself (up), or set yourself (down)? Yes, but not in the same way as you do it to other objects.

Notice that the transitive notions lay, stand, and set all have causative extensions. To lay something is to cause it to lie, to stand it is to cause it to stand, to set it is to cause it to sit. Also notice how that causation operates. You come from outside the object and, typically using your hands and arms, lift it from outside and place it in the position designated. However, when you do it to yourself, the causation is all internal. You are both the causer of the positioning and the thing that is positioned. You cause by internal means, moving your muscles in appropriate ways to cause your body (including those same muscles) to achieve the designated position. We can call the one case external causation, and the other internal causation.

Figure 3: stand / parar / kec̸a (externally causative)

Náhuatl and Spanish both have reflexive verbs to code these and a number of other position and movement verbs, where in English we use intransitives. parar-se, acostar-se, and sentar-se correspond to mo-kec̸a, mo-tēka, and mo-ƛāliya, and to English stand, lie, and sit. mo-hkʷeniya and mover-se both mean “move (intransitive)”; mo-ƛankʷakec̸a and arrodillar-se both mean “kneel down”, mo-ƛamuiƛa and aventar-se both mean “dive” (literally “hurl oneself”), and so forth. Where we in English do not explicitly code any causation, both Spanish and Náhuatl systematically construe the situation as involving internal causation, using a reflexive form of a transitive verb.

Figure 4: mo-kec̸a / parar-se (internally causative reflexive); stand (intransitive)

4.2 Extension of internal causation to less obvious cases

Interestingly enough, the absence of movement can also be construed as internally caused and thus coded by a reflexive, the assumption being that it takes effort to remain still, and that a person in order to remain still must cause his body not to move. Spanish uses the form quedar-se; Náhuatl uses mo-kāwa.

Other types of verbal notions besides movement (or its lack) or position can enter into the same paradigm. A change of emotional state, for instance, can be viewed as caused internally and thus coded by a reflexive. Náhuatl mo-pasolowa “get irritated” and Spanish enojar-se “get angry” may be viewed as of this type.

Such a change can be viewed as internally caused when the process is not even conscious, much less volitional. Why do we grow? It is not because we choose to, but yet it does seem to be from some internal cause. It is thus not irrational to code the notion by a reflexive; Náhuatl mo-wepāwa is an example. (Spanish usually uses an intransitive, crecer, here). Similarly, sickness may be viewed as internally caused, and thus coded by a reflexive: this is probably at least part of the explanation for mo-kokowa and enfermar-se “get sick”.

And, of course, there are many things that phenomenologically parallel actions that we know to be internally caused when they happen to ourselves. For instance, water runs. What makes it do so? There is no obvious external cause; it seems to do so under its own compulsion. Is it any wonder, then, that its motion can be construed as a reflexive? (Náhuatl uses mo-ƛalowa here; Spanish uses an intransitive, correr.) When you throw a ball in the air, it returns (mo-kʷepa) seemingly of its own accord, just as a person might return (mo-kʷepa) after starting away. Plants and animals grow (mo-wepāwa) much as humans do; the sun rises (mo-kīxtiya) apparently of its own accord, and in an earthquake the earth trembles (mo-hkʷeniya) even though there is no obvious external causer shaking it.

4.3 Using internal causation as an excuse to ignore external causation.

In a number of these cases there may be external causing factors present as well as the internal causing factors. However, by coding the situation with a reflexive you in effect ignore any such external causes. A man may become irritated, and a fly buzzing around his head may be a contributing cause. If you want to say that the fly is irritating him, you can of course say so (ki-pasolowa n sāyōli), but if you want to ignore the fly and other possible external causes, you just say mo-pasolowa. Thus the reflexive construction provides a way for ignoring external causes, by choosing to construe the situation as internally caused. This can lead, by only a slight extension of meaning, to its being used to undo any causative extension.

5. The self as a “landmark” direct object.

There is another type of construal that I think important for understanding some of these same forms. Recall the discussion of non-causative transitivity in 2.3, where we claimed that a moving thing is canonically construed as internal subject, and a landmark relative to which its motion is calculated as internal object. Consider the concept move or move over. Is there any landmark relative to which the motion of the internal subject can readily be calculated? Similarly, in curl up or expand or contract is there a single salient landmark? No, and yet we clearly perceive the motion. In fact, the only landmark there is is the internal subject’s previous position. What then if we take that as a landmark, and make it the internal object? A person’s position can be taken to be an aspect of himself, just as we have seen a person’s name be so taken; it can, in fact, be an active zone entering into just such a relation. If the person moving is also construed as the landmark against which his own movement is calculated, then a reflexive construal is very natural. I believe that this construal is also involved to some extent in many of the reflexive movement forms of Spanish and Náhuatl, including some or those we have already looked at.

This “landmark reflexive” construal is exemplified in the English phrase double back on yourself, where yourself is pretty clearly yourself at your previous position. Similarly surpass yourself and its Spanish counterpart superarse both refer to exceeding one’s previous position on some scale of achievement, and both are reflexive.[6] Similarly parar-se and mo-kec̸a, which were given an “internally causative reflexive” analysis in Figure 4, are diagrammed with a “landmark reflexive” construal in Figure 5.[7]

Figure 5: mo-kec̸a / parar-se (landmark reflexive construal); stand (intransitive)

It may strike those accustomed to other theoretical frameworks as incoherent to provide more than one accounting for the same form. Cognitive Grammar, however, explicitly denies the “exclusionary fallacy”, the assumption “that one analysis, motivation, categorization, cause, function, or explanation for a linguistic phenomenon necessarily precludes another” (Langacker 1987a:28). Thus alternative analyses may be invoked by language users on different occasions or even simultaneously without theoretical contradiction. The extent to which one account predominates over others is ultimately a matter of degree.

Not just physical movement, but any change of state may be construed in these terms. If a person gets angry or irritated (enojar-se, mo-pasolowa) he himself in his former emotional state may be taken as a landmark relative to which the change can be calculated, and thus the reflexive usage is natural. In fact, change or becoming can itself be viewed in these terms, and both the Spanish and Náhuatl verbs for the concept are typically reflexive: mo-kʷepa and cambiar(se) or volver-se.[8]

When a reflexive construal of this type is used with a transitive verb which normally has a causative extension, the net effect is again to ignore the causative extension. If the stems hkʷeniya and mover normally mean “cause to move”, but the reflexive forms mo-hkʷeniya and mover-se mean “move with respect to one’s previous position”, we have in effect undone the causative extension. This can be done even in cases where it is obvious to anyone why something moved (e.g. someone moved it) but the speaker chooses to confine his attention to the movement itself; this reflexive construal, like the internal causation construal, can be used to ignore causation and simply focus on the event.

6. Indirect causation.

A third way in which a reflexive can be used to ignore a causative extension is a rather paradoxical one; rather than contracting the scope to exclude the causative extension, you can expand it to include a more tenuous or remote kind of causation, and use the reflexive to code that causation. As an example, consider again mo-hkʷilowa or apuntar-se “sign up”. These verbs can be used when referring to illiterates, who are unable to write their own names, when they have someone else write their name on a list. Note that what we have done here is to expand the scope to include a causative extension, with the internal subject as the causer, causing an intermediate person to cause the internal object (identical with the internal subject) to be written (i.e. to have his name, as active zone, written). As opposed to “signing up”, this concept might be translated “getting signed up”.

There are a number of verbs which apparently involve this sort of construal. For instance, mo-kʷāpihpi means “get a haircut”, including as the prototypical case having someone else do the actual cutting, but literally it is “head-shave yourself”; Spanish cortar-se el pelo is parallel. Some English speakers can say cut your hair in such a situation, though many more easily say get your hair cut.

This kind of construal is important in structures such as mo-hta or ver-se “be visible” (literally “see oneself”) (Figure 6). When an object is visible, we can conceive of those aspects of it that are visibly salient (its active zones as internal subject) causing that we (the intermediate actors) see it (the internal object). The construal is paralleled by English structures like that house almost built itself, or this paper pretty much wrote itself.

Figure 6: see / ihta / ver (transitive); mo-hta / ver-se (indirectly causative reflexive)

The net semantic effect of this kind of construal is to ignore or at least downplay the role of a causer or other potential internal subject, by attributing causality to the internal object, construing it rather than some other eligible entity as the subject of the process.

7. Reflexive passives.

We have seen three different construals that will let us use reflexives to undo or ignore causative extensions. Where such causation is not salient, Spanish and Náhuatl speakers will use these construals, and in English we use intransitive verbs. In other cases the external causation is quite prominent, and in order to ignore it we would have to use a passive, but in Spanish or Náhuatl you can still use the reflexive.

For instance, when an inanimate thing mo-ƛāliya or se pone (literally “sets itself”) it does not do it to itself. In most instances a person places the thing somewhere. Thus there is a salient causative extension. Yet the reflexive form can be used, in either Náhuatl or Spanish, with the result that only the change of the inanimate thing’s position is designated. The scope of the stem ƛāliya or poner is restricted so as to exclude the causative extension, even though the speaker and hearer may both be quite aware of the existence of such an extension. As a sort of side effect of this restriction of scope, the Thing that would naturally be taken as internal object if the causative extension, with its salient causer, were included in the scope, is now taken as the internal subject, since it is now clearly the most salient Thing in the structure. The only way we can get near such a construal in English is to use a passive; we say that the inanimate Thing "is placed".

This, then, is a reflexive passive. The reflexive form, which started out coding the identity of the internal subject with the internal object, has, through various means, proved useful for the undoing or ignoring of causative extensions, which then means that the erstwhile internal object becomes the most prominent Thing in the semantics of the verbal structure and thus its internal subject. By whatever path such usages have gotten established, whether by construing internal causation, or by construing the self as landmark for a non-causational change, or through construing an extended causation in which the internal object is responsible for what happens to it, the important fact is that the usages have become established.

Cognitive Grammar claims that you cannot ultimately divorce meaning from usage. Constant usage to achieve a certain end automatically implies that the achievement of such an end becomes a central part of the meaning of the structures involved.[9] If the reflexive forms mo- and se are used constantly to ignore a naturally construable (causative) subject and instead to structure a verb so that its internal object is the most salient Thing and is therefore taken as internal subject, that can and will become a central part of their meaning. Thus there is established in both Spanish and Náhuatl a meaning (corresponding to and produced by common usage) of the “reflexive” morphemes which is essentially “passive”. There is a pattern of using reflexives in this way, and new forms can readily be put in and used thus, even in cases where the steps by which the usage presumably became established may not be particularly relevant. mo-mati and saber-se (both literally “know oneself”) mean “be known”, even though knowing is not a saliently causative process and thus the internal causing is probably irrelevant, and being known is not a change of state, so the landmark construal is irrelevant, and I would not judge that any extended causality is attributed to the thing that is known, as if it had caused people to know it; in some instances it even resists being known. Cases like this are rightly considered to be passives.

Figure 7: Summary: links between simple reflexive and reflexive passive

8. Conclusion.

Thus we have seen that reflexives can be naturally extended by small, motivated but not determined or predictable steps, from simply identifying the internal subject with the internal object, automatically changing the meaning of the verb in subtle ways, to changing the meaning of the verb in more drastic ways, to ignoring causative extensions, including the naturally salient causer in such an extension, to ignoring other potential subjects, and making the otherwise less prominent internal object be the subject. At each step there are Náhuatl and Spanish forms that can be construed under either of the two analyses. Thus the whole complex of relationships not only offers a diachronically plausible progression, but is synchronically attested.

I would like to make it clear that I do not think that these relations that I have set forth between “normal” reflexives and reflexive passives are the only ones possible and relevant (I know of some others that I believe relevant); but in a model that recognizes the multiplicity of causal factors, that is what one should expect. I think this account is very relevant, and might be enough by itself to account for the existence of the phenomenon, at least in Náhuatl and Spanish; that is all I want to claim for it.


[1]The Náhuatl data are from the Tetelcingo and Orizaba dialects.

[2]Cognitive Grammar differs from traditional models such as Transformational Grammar in a number of non-trivial ways. Among those important to this study (even if not overtly invoked) are

(1) (Linguistic) meaning cannot ultimately be separated from (general) cognition. As a sub-point, meanings are encyclopedic, comprising in their degree all cognitive structures (concepts) conventionally associated with a form. Pragmatics differs from semantics only in degree, not in kind.

(2) Only three kinds of linguistic entities are considered to exist (all of them cognitive constructs): phonological entities (which for oral speech link more or less directly to auditory and articulatory systems), semantic entities (which may be of any cognitive type), and symbolic entities which consist of the pairing of a semantic with a phonological structure. Lexical, morphological, and syntactic structures are all symbolic; they differ not in kind but in degree (along several parameters including complexity, abstractness, productivity, etc.)

(3) “Grammatical” constructs such as subject, object, noun, verb, etc., are given cognitively motivated characterizations.

(4) Any linguistic structure is grammatical (i.e. included in the grammar of a language) to the extent that it is conventionalized (established as shared by speakers) through usage. This is true whether or not the structure is redundant, largely explainable by independently established structures, or not. Established structures can sanction other structures which they subsume. Such sanctioned structures may be grammatical in their own right (e.g. big toe is sanctioned by an Adj + N structure, but is also part of English in its own right), or they may be novel usages, which will then have a degree of grammaticality in the sense of being allowed for / predicted by the grammar (e.g. purple colander is grammatical in this sense.)

(5) A given form may be sanctioned by more than one pattern. The relative strength of its sanction from each may vary from speaker to speaker or from occasion to occasion.

     Linguistic categories of all sorts are generally organized on the prototype model, with some members being better exemplars of the category than others, and with no absolute boundaries between membership and non-membership, or between one category and another.

[3]“Internal subjects” are what are called “trajectors” in the Cognitive Grammar tradition, and the “internal objects” are what are called “landmarks”. What we will call landmarks in the next paragraphs are in fact a prototypical kind of internal objects, and the name is most appropriate to them: the Cognitive Grammar tradition uses the name of this prototypical case to name the whole class, but it is clear that the definitional characteristic of the class is salience rather than functioning as a reference point for calculating change.

[4]As noted in Note 2, a cognitive structure is part of the meaning of a form to the extent that it is, through usage, conventionalized. I am claiming that English (and Spanish and Náhuatl) speakers share the knowledge that reflexive and non-reflexive seeing differs in the above-mentioned ways, and usage differences such as those mentioned ahead further establish the conventionality of that difference in meaning.

[5]The fact that plow, arar, and kʷenƛāliya are not used reflexively is of course related to the fact that we do not speak of plowing people or of fields plowing anything. There is a large class of naturally useful concepts in which human subjects act on inanimate objects, and verbs designating those concepts (including plow / arar / kʷen-ƛāliya and write / apuntar / hkʷilowa in the next paragraph) tend to treat reflexives and transitives with human objects or inanimate subjects alike, either not using them (the first case) or adjusting the meanings (the second). Nevertheless there remains a difference between the transitive usages and the reflexive: the transitive usage will be inoperative or adjusted only in a minority of cases, while the reflexive will be in all cases.

     Of course in many cases a concept can be easily construed with both a human object and a human subject, but still cannot be easily construed reflexively: the carry on the back forms mentioned below are of this type. In such cases as well the reflexive will either not be used or will be used with a slightly (or drastically) different meaning.

[6]This example was suggested to me by Barbara Hollenbach. (The corresponding Náhuatl form chikʷa-kīsa “excel” is intransitive and would require an adverbial phrase to introduce explicit comparison with previous performance.)

[7]Although a “landmark reflexive” analysis for these forms is clearly possible, it is not easy to argue that it is necessary. Some speakers, however, report intuitions that they definitely do not mean, when using these forms on certain occasions, that the object is causing itself to stand (or lie or move or whatever), or even that it is acting as if it were doing so. For such cases the “landmark reflexive” analysis, or one that simply says “construe motion as uncaused” (c.f. Section 7) seems nearly necessary.

[8]It is my opinion that this factor is stronger in Spanish than in Náhuatl; that would go some way towards helping explain why Spanish reflexives (which are sometimes put on intransitives with this effect) so often code perfectivity or change of state. (E.g. dormir means “sleep, be asleep”, whereas dormir-se “go to sleep”, callar means “be quiet, keep silent”, whereas callar-se means “shut up, stop talking”.) When the change of state is in fact a change of physical location, some kind of “landmark” construal is almost necessary: ir means “go”, but ir-se means “go away (from where you were)”; volver means “return”, while volver-se often means more emphatically “come/go back (to where you were previously)”.

     The above is not offered as the full explanation for the above types of Spanish reflexives, which are indeed very complex. I believe it is very relevant, but am sure it is not the whole story.

[9]This is, of course, a truism. Yet it has been denied by enough modern theories in the case of “function” or “grammatical” words or morphemes, including reflexives and passive morphology, that it bears saying explicitly. These forms may have very abstract meanings, and they usually do not have a single, all-encompassing, easily expressed, monolithic meaning, but they are meaningful nonetheless, and they acquire their meanings in the same way that other forms do: through usage.


Casad, Eugene H., and Ronald W. Langacker. 1985. “‘Inside’” and ‘outside’ in Cora grammar”. IJAL 51.247-281.

Langacker, Ronald W. 1982. “Space Grammar, Analysability, and the English Passive.” Language 58.22-80.

————. 1985. “Observations and speculations on subjectivity.” In John Haiman (ed.), Iconicity in syntax, p. 109-150. Amsterdam and Philadelphia: John Benjamins.

————. 1986. “An introduction to Cognitive grammar”. Cognitive Science 10.1-40.

————. 1987a. Foundations of Cognitive grammar, Vol. 1. Grammatical prerequisites. Stanford: Stanford University Press.

————. 1987b. “Nouns and Verbs.” Language 63.53-94.

Lindner, Susan. 1981. A lexico-semantic analysis of English verb-particle constructions with OUT and UP. UCSD doctoral dissertation, also published by Indiana University Linguistics Club.

[Maldonado, Ricardo. 1999. A media voz: Problemas conceptuales del clítico se. México: Universidad Nacional Autónoma de México.]

Tuggy, David. 1981. The transitivity-related verbal morphology of Tetelcingo Náhuatl: An exploration in Space grammar. UCSD doctoral dissertation.

© 2003, David Tuggy
All rights reserved. May be reproduced for nonprofit use so long as it is not modified in any way.
Derechos reservados. Puede reproducirse para fines no lucrativos siempre que no se altere en forma alguna.

Back to the top
[Versión en español]

To David Tuggy’s home page
A la página principal de David Tuggy