bradrn wrote: ↑Sat Apr 18, 2026 3:08 pmRichard W wrote: ↑Sat Apr 18, 2026 2:18 pm'S' = subject (a purely grammatical aspect))
'O' = object (a purely grammatical aspect)
'A' = agent (semantic, though languages can be idiosyncratic in what is the agent and what is the patient)
'P' = patient (semantic)
The usual notation is S = intransitive subject, A = transitive subject, O = transitive object. All three of these are purely syntactic, not semantic. But normally people don’t use P as well; I have no idea what that’s supposed to mean.
Using P alongside S, A and O is a sign that someone has mixed up the language of morphosyntactic argument roles (used to talk about morphosyntactics, not semantics) with the language of thematic relations (used to talk about semantic roles independent of morphosyntactics in order to then compare how languages handle them morphosyntactically). They're from two different systems.
Richard W wrote: ↑Sat Apr 18, 2026 2:18 pmAt this point, I fail to understand 'S=O is mandatory'.
It just means that the objects of transitive sentences are marked the same way (in the morphosyntactic position, whether case or word order) as the subjects of intransitive sentences.
Richard W wrote: ↑Sat Apr 18, 2026 2:18 pmI presume it does not refer to Latin impersonal verbs with accusative of person, e.g.
Me civitatis morum piget taedetque '
I am
sick and
weary of the morals of the state'. (Latin
me is in the
accusative case in the sample sentence;
morum 'morals' is in the genitive. Does 'S=O' perhaps relate to what can cross-connect between clauses (e.g. by being taken as implicitly duplicated)? Does a typical ergative, transitive sentence have something other than a subject (S) and object (O)?
I don't speak Latin either and you should really gloss that unless you're coming from the era where
All learnèd people speak Latin, darling! In any case, I suspect it is something like these German examples:
Code: Select all
Mir ist kalt
1S.DAT be.3S.PRES.INDIC cold
"I'm cold."
Mir ist übel
1s.DAT be.3s.PRES.INDIC nauseous
"I feel sick."
These are a rare case of German not having an overtly marked subject rather than a dummy subject. The verb
ist still agrees with the 3rd person singular dummy subject. The pronoun is in the dative case because
some concepts are expressed in German with the EXPERIENCER (a thematic relation) in the dative. (The majority are in the nominative, as in English:
Ich sehe 'I see';
ich mag 'I like';
ich höre 'I hear'.) It doesn't really say anything about the morphosyntactic alignment of German, which is quite squarely nominative/accusative.
By the way, the sentences above can have their omitted nominative dummy subjects expressed:
Es ist mir kalt. Es ist mir übel. (The latter feels a bit stiff to me.)
In languages that never have dummy subjects (like
It's raining. Es regnet.), for example Spanish (
Está lloviendo, lit. "Is raining.") (but not ones that express the sentence as something like "Rain falls"), there are frequent examples of sentences without subject or object — in Spanish, verbal agreement indicates an unexpressed 3s subject just as
Mir ist kalt does— but in languages without subject-verb agreement, without a dummy subject, the entire concept of a subject can be missing from sentences such as this. This is not a challenge for morphosyntactic alignment tests — unless the whole language somehow
never expresses subjects and objects.
Richard W wrote: ↑Sat Apr 18, 2026 2:18 pmI assume this level of exposition ignores awkward cases resembling
The dog was given a bone by its owner and
Has this hall ever been sung in?.
There's nothing awkward about that at all. The English passive can simply be used in ways that are impossible for a passive in many languages; it can promote not only a direct object but also an indirect object or a prepositional complement to the subject role. One of my little quirks is that I like making passive sentences that most people would call ungrammatical or borderline, like "
How dare he be taller than me? I don't like being been taller than!
German can't do anything like that with its passive construction, but it can do
impersonal passives, which English can't do. It can passivise intransitive verbs! (These can have a dummy subject
es, but it can be dropped if something else is fronted before the finite verb ... and this can even be the past participle.)
Code: Select all
Jetzt wird getanzt!
now becomes danced(PP)
"Now it's time to dance!"
Gestorben wird immer.
died(PP) becomes always
"People are always dying."
"People will always die."
(German title for the HBO series "Six Feet Under")
Bei diesen Veranstaltungen wird immer getrunken.
at these events becomes always drunk(PP)
"People always drink at these events."
The fact that English and German passive sentences have different bounds regarding what is and isn't possible is also not a challenge for morphosyntactic alignment because both languages do the normal "core passive" thing with their passives: promoting a direct object to a subject. Grammatical labels like "passive", "genitive" or "ergative", when applied cross-linguistically don't always mean the exact same thing. They are just labels and the real world is messier than the neat little boxes we make our words mean. There's generally a core meaning associated with each linguistic word, and when linguists are describing a language, they have to make choices about the best words to describe particular structures within the language, or the language itself as a whole. Saying "X language" has a passive doesn't tell you anything about the boundaries of its use, just that it has a structure that can do all (or most of) the things canonically described as passive. It's the same with ergativity. There's a core meaning of ergativity, but not everything fits it neatly, which is why linguists have come up with finer and finer shades of meaning such as split ergativity, fluid-S and Austronesian alignment etc. and defined tests and boundaries for each of these, and there will always be things that just refuse to fit neatly into one category.
Some of the categories are really sloppy. For example the word "predicate" can be used both for the entire syntactic part of the sentence that is not the subject, but it's sometimes also used more or less as a word for "verb" in the sentencial sense (i.e. excluding objects) or even as a label for a part of speech. Words like "imperfect", as already discussed here, are another example. Look at how the word "aorist" is used in different languages. It's a mess.
It's often a case of simply which analysis you go with: as bradrn mentioned, many Polynesian languages can be regarded as NOM/ACC and others as ERG/ABS and the lines between them are fine and linguists describing the same language can come up with competing analyses; more than half of all transitive sentences in Māori dialogues are passive — which could also be analysed as active in an ERG/ABS structure, but then the fact that the passive is morphologically marked means that regarding the absence of this marking as a marker of the antipassive would be a bit weird. Then there's another part of speech called statives which cannot be marked for the passive, but which always inherently function in an ergative/absolutive way (except with the
accusative marker
i introducing the agent instead of the
e ergative marker used with other verbs.) Māori is generally explained as NOM/ACC and Sāmoan is generally explained as ERG/ABS and my understanding is that, in Sāmoan, the passive suffix on verbs need not always be present.
My point is just that real languages (and conlangs) are under no obligation to fit neatly inside categories defined by words. The linguist's job is to choose the best description of the language and choose words that get close and then explain where the words chosen fail to explain the behaviour of the language. (A less parsimonious analysis will require a lot more explaining and anyone reading it may just go "Why don't you just describe it as XYZ then!?" or "This is so genuinely weird that you should just coin a new word for it!" or "Why did you bother coining a new word for that? It's just X with a few fringe cases of Y!?") If you draw some lines on the ground to represent the abstract definitions of linguistic words and then drop cans of paint on the ground, most languages will have a few splatters that go across lines, or they won't land right in the centre of their category. Some will land pretty squarely across both sides of a line and if that happens enough, new lines may be drawn (new words coined) to explain these languages in the simplest terms. English and German's differing bounds for their passive constructions or the presence of some experiencers in the Dative in German and the accusative in Latin are not huge hurdles to get these languages to fit the label NOM/ACC.
