Secondary features of ergative languages.

[Legion]

I know that ergative languages tend to have Other Things associated with them, things that are not directly tied to morpho-syntactic alignment but result from it, various syntactic features and constructions… but I'm not quite sure what those are, and the wikipedia article on ergativity only covers the basics. Basically, what does ergativity imply?

[finlay]

Like antipassives?

[cromulent]

Have you checked the Universals Archive?

This should take you to a search on the term "ergative."

I think it looks promising.

EDIT: Oops, that link didn't do what I thought it would. Guess you'll have to type in "ergative" yourself.

[Legion]

Like antipassives?

Like that kind of thing, yeah.


cromulant > that looks promising, thanks.

[TomHChappell]

Have you checked the Universals Archive?

This should take you to a search on the term "ergative."

I think it looks promising.

EDIT: Oops, that link didn't do what I thought it would. Guess you'll have to type in "ergative" yourself.

Having done so, I found:

• 3. Grammatical systems with ergativity tend to have ergative-accusative splits.
• 44. Assuming that grammatical relations are ordered from left to right according to their decreasing semantic concreteness and correspondingly increasing grammaticalization as in the diagram (where arrows indicate attested diachronic grammaticalization), then IF there is noun-phrase external agreement with a relation of any column of this diagram, THEN there will be noun-phrase external agreement with the relations of all columns to its right. IF there is overt case marking for a given column, THEN there will be overt case marking for all columns to its left.
• 64. In nominative-accusative languages the causee of causative verbs derived from transitive verbs is an oblique object, whereas in ergative languages it is a direct object
• 134. Ergative languages favour SENT and SENT-POSS ANCs and disfavour POSS-ACC and NOMN ones. (whatever the hell that means -- thc)
• 172. In languages with ergative/absolutive distinction, the frequency prediction holds: if the verb agrees with the ergative, then it agrees with the absolutive.

And those are just the first five.

Here's another sublist:

Result No. 1
Number 64
Original In nominative-accusative languages the causee of causative verbs derived from transitive verbs is an oblique object, whereas in ergative languages it is a direct object


Result No. 2
Number 134
Original Ergative languages favour SENT and SENT-POSS ANCs and disfavour POSS-ACC and NOMN ones.

Result No. 3
Number 206
Original The most important implications of ergativity are the following: the opposition of ergative and absolutive series in personal affixes in the verb conjugation and the opposition of the ergative and absolutive cases in the noun declension, similar to each other in their function.

Result No. 4
Number 225
Original Difference of case inventories in the different systems:
the strong active system is characterized by an absence of case inflection; the opposition of active and inactive cases occurs in its later periods. The main cases of the ergative system are the ergative and absolutive, added to a number of locatives. The case system of the nominative system is characterized by the opposition of a nominative and an accusative case, to which can be added genitive, dative, instrumental, and a number of locatives.

Result No. 5
Number 226
Original Languages of the class type have no case paradigm. In rare cases when languages of the active type have an elementary noun declension, its fundamental components (active and inactive cases) have no subject-object orientation. Ergative and absolutive cases in the ergative type are characterized by a fused distribution of subject and object functions. Nominative and accusative cases of the nominative system are transparently oriented towards subject-object relations.

Result No. 6
Number 258
Original In ergative languages the subject is characterized by its autonomous position in the sentence:
first of all, subject and predicate are in marginal positions: the former left, the latter right; between subject and predicate there is a special rhythmic-intonational break; and in contrast to the almost unmarked direct object, the subject has a special morphological marker.

Result No. 7
Number 259
Original In ergative and active languages only secondary topicalization, by means of changing the word order in the sentence, is possible. Primary topicalization or subjectivization – choosing either agent or patient as subject – is not possible in these languages. This is due to the fact that in ergative languages transitive verbs lack the opposition of an active and passive voice and in active languages the version opposition is of a non-voice character.

Result No. 8
Number 303
Original If a language has an ergative construction or a passive voice in non-past or imperfective tenses, then it also has them in past and perfective tenses.

Result No. 9
Number 436
Original Ergative alignment is associated with dependent marking , stative-active and hierarchical with head marking.

Result No. 10
Number 438
Original Ergative alignment is associated with high complexity.

This correlation is not direct one, but a consequence of those between ergative and dependent marking, and dependent marking and high complexity - see , .

(Definition of Complexity see in )

Result No. 11
Number 470
Original If in a language there is an opposition “ergative/nominative” in the declension of independent personal pronouns of the 1st and 2nd person (or at least one of them), there is the same opposition in the declension of nouns of this language.

Result No. 12
Number 471
Original If in a language personal and/or demonstrative pronouns of the 3rd person have an opposition “ergative/nominative”, the nouns of this language have the same opposition.

Result No. 13
Number 600
Original If in a language there are two classes of sentences such that there is ergative-type verb agreement in one but not in the other type and the semantic property whose respective presence and lack characterizes the two classes is perfectivity, then it is in the class of perfective sentences that ergative verb agreement will occur, rather than in the other class.

Result No. 14
Number 602
Original No accusative language has only ergative-type verb agreement in person or number or gender.

Result No. 15
Number 895
Original Provided there is actancy variation in terms of tense, accusative constructions will always be found in the future/present/imperfective while ergative constructions occur in the perfect/past/perfective tenses.

Result No. 16
Number 1101
Original Any language that is syntactically ergative will also have some ergative characteristics at the morphological level.

Result No. 17
Number 1128
Original If subjects and objects can be incorporated into verbs, then they are not case-marked when occurring absolutely or they are case-marked and case marking shows ergative alignment.

Result No. 18
Number 1267
Original Ergative systems are found only in SOV and VSO languages. SVO languages are never ergative.

And another:

Result No. 1
Number 3
Original Grammatical systems with ergativity tend to have ergative-accusative splits.

Result No. 2
Number 134
Original Ergative languages favour SENT and SENT-POSS ANCs and disfavour POSS-ACC and NOMN ones.

Result No. 3
Number 172
Original In languages with ergative/absolutive distinction, the frequency prediction holds: if the verb agrees with the ergative, then it agrees with the absolutive.

Result No. 4
Number 240
Original If a language has ergative case marking, it will with far more than chance frequency also have divergent ordering of genitive and adjective attributes (with genitives preceding and with adjectives following their head nouns); if it has divergent genitive and adjective ordering, it will [supposedly with somewhat lower frequency] also have ergative case marking.

Result No. 5
Number 260
Original In ergative languages personal conjugation is of prefixal or prefix-suffixal nature.

Result No. 6
Number 261
Original In ergative languages personal verb affixes are arranged such that the object marker almost always precedes the subject marker.

Result No. 7
Number 262
Original Among the lexical frequentalia of ergativity one can mention: Classes of possessive and affective verbs (verba habendi and verba sentiendi), forming special sentence constructions [and affective verbs can sometimes be implied — ??? ]; at the level of noun morphology the presence of a special affective case; a distribution of nouns into lexical classes; groups of verbs that are not lexicalized according to the feature ‘transitivity vs. intransitivity’, i.e. so-called labile or diffuse verbs; an opposition of inclusive and exclusive forms in the 1st person plural.

Result No. 8
Number 263
Original Among the morphological frequentalia of ergativity one may mention: the morphological category of words that are syntactically connected with nouns (verbs, adjectives, pronouns); the opposition of alienable and inalienable possession with nouns; number agreement of transitive verbs with direct object and of intransitive verbs with subject (or compensating it by suppletion of the verb stem).

Result No. 9
Number 472
Original If in a language nouns have the opposition “ergative/nominative”, then with a high plausibility the same opposition have personal and/or demonstrative pronouns of the 3rd person.

Result No. 10
Number 635
Original If a language has Suffixaufnahme, then (perhaps) it is predominantly of ergative or other non-accusative alignment.

The only one that was unconditional, achronic, and absolute, was

Result No. 1
Number 1103
Original If there is a bound vs. free split, the bound prefixes will be accusative, and case marking on free forms will be ergative.

Also look at features 98, 99, and 100 in wals.info.
http://wals.info/feature/98
http://wals.info/feature/99
http://wals.info/feature/100
http://wals.info/feature/description/98
http://wals.info/feature/description/99
http://wals.info/feature/description/100

http://wals.info/feature/combined?id1=98&id2=99
http://wals.info/feature/combined?id1=98&id2=100
http://wals.info/feature/combined?id1=99&id2=100

Note that of the 19 languages in their sample database with ergative person-marking on the verbs, and the 20 languages with ergative-absolutive case-marking of the pronouns, only 2 have both. If the verb agrees with the ergative, the case-marking of the pronouns is usually neutral; if the pronouns' case-marking is ergative/absolutive, the verb's person-agreement is usually accusative or split or active or neutral.

Something similar happens comparing ergative person-marking on the verbs with ergative-absolutive case-marking of full noun-phrases. Of the 32 languages with ergative-absolutive case-marking of full noun, phrases, most have accusative or split or active or neutral person-marking of the verbs; only 3 have ergative person-marking of the verbs. Of the 19 languages with ergative person-marking of the verbs, most have neutral case-marking of full NPs; only three have erg-abs case-marking of full NPs.

But most languages with ergative-absolutive case-marking of full NPs also have ergative-absolutive case-marking of pronouns. 32 languages in their sample database have erg-abs full NPs and 20 have erg-abs pronouns; the overlap is 19 languages.

You can combine any other feature WALS.info maps with feature 98 or 99 or 100 and see what you can see.

[qiihoskeh]

An ANC is an Action Nominal Construction. WALS has an article on those. IIRC POSS-ACC means that the subject of the action nominal is expressed as a possessive while the object stays in the accusative. Possible example:
"John disliked Tom's buttering the honeybun."

[TomHChappell]

An ANC is an Action Nominal Construction. WALS has an article on those. IIRC POSS-ACC means that the subject of the action nominal is expressed as a possessive while the object stays in the accusative. Possible example:
"John disliked Tom's buttering the honeybun."

Thanks.
What's a SENT?

[qiihoskeh]

An ANC is an Action Nominal Construction. WALS has an article on those. IIRC POSS-ACC means that the subject of the action nominal is expressed as a possessive while the object stays in the accusative. Possible example:
"John disliked Tom's buttering the honeybun."

Thanks.
What's a SENT?

agreement.

In this case, SENT means that the arguments of the action nominal are marked just as they would be in a finite clause. I'm guessing that SENT-POSS is like ERG-POSS (see WALS 62).

[TomHChappell]

I see. So "the enemy's destruction of the city" would be double-possessive, or at least double-genitive, since both the agent "the enemy" and the patient "the city" appear in the genitive relative to the nominalized action "destruction".

Combining feature 62 with features 98 or 99 or 100 doesn't make WALS.info seem to back up that universal (134) mentioned in the Universals Archive. Or, at least, not significantly so as far as I can tell, assuming I understand correctly; I have my doubts.

In fact it isn't obvious that WALS.info significantly supports any correlation between any of features 58 thru 64 on the one hand and any of features 98 thru 100 on the other.

But see http://wals.info/feature/combined?id1=126&id2=98.
Also check http://wals.info/feature/24 against features 98 and 99.

Feature 26 also seems correlated with features 98 and 99. Feature 28 may be correlated with feature 98; look at it and decide for yourself.

http://wals.info/feature/combined?id1=29&id2=100 Features 29 and 100 appear to be correlated; if so that's perhaps not surprising.

Oddly enough feature 98 might be correlated with feature 67.

And so on.