--- In [log in to unmask]
, Aidan Grey <taalenmaple@Y
> I want to incorporate middle voice into my conlang, but seeing as
> I'm an L1 English speaker, it's hard for me to see it. ...
> Any additional info or description of the middle voice, to help me
> clarify, is great. ...
(In my previous post I had forgotten about the term "diathesis".
Thanks to Ray Brown for reminding me -- he's the one who taught it to me
in the first place.
Everywhere I said "semantic voice", read "diathesis"; wherever I said "semantic middle voice", read "middle diathesis".
On the other hand, everywhere I said "grammatical voice" or "grammatical middle voice" or "morphosyntactic voice" or "morphosyntactic
middle voice", read it as-is.)
I said in my previous post that "middle voice" was part of the morophosyntax, mostly, of languages with a "Basic Voice System" rather than a "Derived Voice System" or a "Pragmatic Voice System" or a "Hierarchical Voice System". That notion came from M. H. Klaiman's "Grammatical Voice".
These "super-types" of voice systems aren't absolutely pure. The major languages he uses as examples of Basic Voice Systems, do have "derived voices" as well -- it's just that the "basic" voices are more "dominant" than the "derived" ones.
His examples included languages that were accusative, languages that were ergative, and languages that had "split" alignment systems.
Concentrating for the moment just on the accusative languges with a basic voice system:
In these languages, nearly every bivalent verb had a "home voice" (that verb's "basic" voice). For almost all of these bivalent verbs, that "home
voice" was either Active or Middle -- almost no verb (in most languages, no verb at all) was at "home" in the Passive* voice. But on the other hand, nearly every verb could be "coerced", morphologically, into the Passive* voice -- there were usually a very small minority of verbs which could not be caste into the Passive.
As for the other bivalent verbs; many verbs at home in the Active could be coerced into the Passive, but not into the Middle; whereas many other verbs at home in the Active could be coerced into both the Middle and the Passive. Likewise, many verbs at home in the Middle could be coerced into the Passive, but not into the Active; whereas many other verbs at home in the Middle could be coerced into both the Active and the Passive.
*For most languages, coercing a verb into the Passive is a valency-reducing operation; a verb, bivalent before passivization, becomes univalent as a result of passivization. For these languages, therefore,
it makes no sense to speak of a bivalent verb being at home in the passive voice.
If we look at the ergative languages with basic voice systems, what we get is this;
Most bivalent verbs would be at home in either the "Active" (what else would I call it? Maybe there's a better name someone else knows -- but I seem to recall "active" is what it is called) or the Middle voice. Also, most of them can be "coerced" or cast into the Antipassive -- whereby the erstwhile absolutive argument is demoted to an oblique (such as an instrumental) or omitted altogether, while the erstwhile ergative argument is promoted to absolutive; note that that is a valency-reducing operation. Furthermore, many "basically active" verbs can be coerced into the middle voice, while many can't; and many "basically middle" verbs can be coerced into the active voice, while many can't.
Thomes E. Payne in "Describing
Morphosyntax"(1997) says Joan Bybee in "Morphology"(1985) says that valence-adjusting and voice are the most common morphology marked on verbs. 84% of languages have derivations marking valency or voice, and another 6% have inflections marking valency or voice, so 90% of languages mark their verbs with valency or voice somehow. Aspect is second at 74%, and mood, mode, and modality is third at 68%. (I have not had a chance to read Bybee's work yet, so I do not know whether she includes retrospective ("perfect") and prospective as aspects, which is traditional, or as moods or modes or modalities, which I believe is correct; nor whether she includes evidentials and miratives and mediatives among the moods and modes and modalities. That could reverse the order of Aspect and Mood, or maybe not.) Tense is only seventh, at 50%. However AMT (usually written TAM) is usually inflectional rather than derivational, while voice and valency are usually
derivational rather than inflectional, so TAM dominate inflection of verbs, even though voice and valency dominate morphology of verbs overall.
Some languages have some verbs which have both a bivalent, "transitive" sense, and a univalent, "intransitive" sense. In some of these languages, some such verbs are what are called "labile" verbs.
In an accusative language, such as English, the subject of an intransitive verb usually aligns with the agent of a transitive verb.
Thus, because "eat" is not a labile verb, "I" am the agent of both of the following sentences:
"I ate a sandwich" (transitive)
"I ate" (intransitive)
However, "break" is a labile verb; in its intransitive "incarnation", its subject is its patient, not its agent.
"The stupid mover broke my vase."
"My vase broke." (intransitive)
According to Payne, English's intransitive uses of labile verbs are in the middle diathesis.
The semantic roles of Agent and Patient have to do with Control and Affectedness. The Agent is the one that has Control, and the Patient is the one that is Affected. But in many necessarily two-participant clauses, the participant with the most control is also the one that is most affected; and neither participant has complete control.
Verbs of emotion (or judgement or ... etc) and verbs of perception are the two main kinds of verbs like this.
Take "fear". "I fear the Greeks". How does that affect the Greeks? It doesn't. They probably don't even know. Do they have any control over it? Probably not a bit. Do I? probably not a whole lot. Who is
most affected? Me. Who is most conscious of it? Me. So, I am the most agent-like (conscious), but I am also the most patient-like (affected).
Similarly for "love". "I love my daughter". How does that affect her? Only indirectly. Does she know? Only indirectly. Does it affect me? Yes. Do I have control over the fact of my love? Yes. Do I have control over the outcome of my love? Yes. So I am completely the agent, and much more directly the patient than she is -- she is only indirectly a patient.
What about "see"? "I see that Chafe and Nichols's book on Evidentiality is red." How much and what kind of control do I have over this? I have agenda control, but not outcome control -- that is, I can choose whether or not to look at the book, but, having done so, cannot choose what color to see it as. My agenda control is both conscious and
voluntary. The book has outcome control but not agenda control -- whether or not it gets looked at is beyond its control, but if it gets looked at, its going to be red. Its outcome control is completely involuntary, and unconscious. Is it in anyway affected by being seen? No. Am I affected by seeing it? Yes. So I am a bit more agentive than it is, and also a bit more patientive than it is.
Verbs of Emotion are sometimes said to have Experiencer/Target roles instead of Agent/Patient roles.
Verbs of Perception are sometimes said to have Perceiver/Stimulus roles instead of Agent/Patient roles.
Some natlangs, according to Blake's "Case", distinguish Agents from Experiencers from Perceivers (from some fourth thing I've forgotten).
Voice systems and Case systems are often discussed together or one after the other. Some of the
same semantics that can be expressed by one can be expressed by the other. In particular the distinctions of control and affectedness I just discussed can be expressed either by "Middle Voice" or by a case.
Blake says that, cross-linguistically, the case that ought to be called "dative" should be the one used for "targets or focuses of activities which do not affect them". Examples are not only verbs of perception and emotion, but also verbs such as "seek" and "find". If "I seek a copy of Joan Bybee's Morphology", and then "I find a copy of Joan Bybee's Morphology", the book is relatively unaffected by this activity.
The dative is the most common "adverbal" case after the nominative/absolutive and the accusative or ergative as the "case" may be; it is the most common after the two (or three, if the language is tripartite) of them and the genitive.
Note that in many languages, in verbs of emotion and perception and judgment, and even in verbs such as "seek" and "find", either the Experiencer or the Stimulus gets put in the Dative instead of the ---
Nominative (Experiencer, Accusative language)
Ergative (Experiencer, Ergative or Tripartite or Split Language)
Accusative (Stimulus, Accusative or Tripirtite or Split Language)
Absolutive (Stimulus, Ergative Language)
In some languages, these verbs would be put in the Middle Voice.
Tom H.C. in MI