Conlang Random Thread

[/ˌnɐ.ˈɾɛn.dɚ.ˌduːd/]

But can someone help me understand what the imperfect is? When I was a kid it just meant past tense for the Swedish language, but this term is no longer in use. And I don't think that's what I've intended here either. Wikipedia says this:

The imperfect (abbreviated imperf) is a verb form that combines past tense (reference to a past time) and imperfective aspect (reference to a continuing or repeated event or state).

Is this true? Is it the same as a past imperfective? However, I feel like "a continuing or repeated event" isn't quite a correct definition of an imperfective. Is this actually the same as a past continuous/progressive? The rest of the Wikipedia article seems to say so.

the past imperfective, in my mind, conveys something was happening continuously, (kinda like "I was sleeping,") while the imperfect tense conveys something that is actively happening ("I am sleeping").

[Lērisama]

But can someone help me understand what the imperfect is? When I was a kid it just meant past tense for the Swedish language, but this term is no longer in use.

While people have been good at explaining what a (past) imperfective is, the specific term ‘imperfect’¹ is a lot more confusing. It was originally used for a specific Latin tense that happened to mostly match a past imperfective, and then got borrowed across Europe for various past tense/aspect combinations, including cognate ones where the meaning had diverged², and so it is a mess with no clear meaning. Because of this is has fallen out of use except as a name for specific verb forms in languages that used it, and the clearer (im)perfective terminology was introduced.⁴

¹ And it's cognates
² While I don't know what the Swedish imperfect does, the German imperfect had/has³, imperfective uses, and so the word ‘imperfect’ was used for that tense and then for its cognate tense across the Germanic languages.
³ Depending on dialect, I think
⁴ Except ‘perfect’ survives for a use, of course not really matching really any of its namesakes well, although at least the Latin perfect could be used as a perfect, even if it was mostly a past perfective⁵.
⁵ Was that clear? Talking about aspect names tends to cause semantic bleaching much too quickly.

[Travis B.]

To me 'imperfect' highlights a perfectivity contrast in the past, with the 'imperfect' being largely synonymous in general usage with the past imperfective. However, in practice this is rather language-specific.

As for what the imperfective and perfective themselves are, the imperfective is an aspect that emphasizes the internal structure and/or duration or iterativity of a process while the perfective is an aspect that emphasizes a singular event as a whole. Of course, the details here tend to be rather language-specific.

Note that this contrast is particularly relevant in the past, as the past lends itself to contrasts between processes and singular completed events. Many languages lack such a contrast in the present, as anything happening right now cannot be easily considered as a completed event. Languages which do morphosyntactically have such a contrast in the nominal 'present' often attach a special meaning to the 'present perfective', such as referring to future events or for use as performatives (e.g. "I pronounce you man and wife.").

[jal]

Note that this contrast is particularly relevant in the past, as the past lends itself to contrasts between processes and singular completed events. Many languages lack such a contrast in the present, as anything happening right now cannot be easily considered as a completed event.

Iirc correctly, some languages (was it Slavic ones?) have an unmarked imperfective present and a marked imperfective past, while the unmarked perfective defaults to the perfective past.


JAL

[hwhatting]

Iirc correctly, some languages (was it Slavic ones?) have an unmarked imperfective present and a marked imperfective past, while the unmarked perfective defaults to the perfective past.

Slavic languages have varying systems, but the two I know best, Russian and Polish, are very similar in this regard:

Russian / Polish:
imperfective aspect, present tense forms - present tense ya pishú, (ja) piszę - I write, am writing
imperfective aspect, past tense forms - imperfective past ja pisál, (ja) pisałem - I wrote, was writing
imperfectve aspect, future tense forms - imperfective future tense ja búdu pisát', (ja) będę pisał / pisać - I will write, will be writing

perfective aspect, present tense forms - perfective future tense ja napishú, (ja) napiszę - I will write, will have written
perfective aspect, past tense froms - perfective past tense ja napisál, (ja) napisałem - I wrote, have / had written

As you can see, there is a gap - there is no perfective present. Old Church Slavic still had that, for out-of-time or hypothetical events, but Russian and Polish see present tense activity as something that simply cannot be complete. And, conversely, the periphrastic future cannot be formed from perfective verbs; *ja búdu napisát', *(ja) będę napisał / napisać don't exist.

[Travis B.]

perfective aspect, present tense forms - perfective future tense ja napishú, (ja) napiszę - I will write, will have written

I was precisely thinking of that in the case of Polish and Russian.

[Richard W]

How does one decide if it's split ergativity?

If sometimes an ergative construction is used, and sometimes an accusative construction, it's split no? One doesn't "decide", one checks the grammar .

So when is a construction 'ergative' as opposed to passive? How is answering this not making a decision? Or are you in fact relying on a grammarian to make the decision?

[Travis B.]

How does one decide if it's split ergativity?

If sometimes an ergative construction is used, and sometimes an accusative construction, it's split no? One doesn't "decide", one checks the grammar .

So when is a construction 'ergative' as opposed to passive? How is answering this not making a decision? Or are you in fact relying on a grammarian to make the decision?

One should consider if there is an active to choose (which indicates that a usage is passive rather than ergative) whether an antipassive voice is needed for omitting the patient (which indicates that a usage is likely ergative).

Also, split ergativity can operate separately on each argument, e.g. you can have an unmarked nominative-absolutive (a.k.a. 'direct') case that contrasts for the agent of a transitive clause with a marked ergative case (e.g. for inanimate agents) and contrasts for the patient of a transitive clause with a marked accusative case (e.g. for animate agents).

[Travis B.]

I'm thinking of making a sister language to Rihalle Kaafi, and am wondering what I can do somewhat differently yet somewhat similarly.

Phonologically I am thinking of the following vowel system:

/i u/
/e ə o/
/a/

/ĩ ũ/
/ẽ ə̃ õ/
/ã/

/iː uː/
/eː əː oː/
/aː/

(The protolanguage, for the record, will have /i æ ɑ u iː æː ɑː uː/ like Rihalle Kaafi.)

with /i u/ deriving from the protolanguage's /iː uː/, /e o/ deriving from the protolanguage's /i u/ and /iː uː/ in proximity to uvular consonants (and /e/ from the protolanguage's /æ ɑ æː ɑː/ in proximity to pharyngeal consonants), /ə/ from the protolanguage's /æ ɑ/ not in proximity to uvular or pharyngeal consonants and the protolanguage's /u uː/ in proximity to pharyngeal consonants, and /a/ from the protolanguage's /æː ɑː/ not in proximity to the pharyngeal consonants and /æ ɑ/ in proximity to uvular consonants.

/æj ɑj æw ɑw æːj ɑːj æːw ɑːw/ not before a vowel in the protolanguage will become /e aj aw o ej aj aw ow/ when not preceded by uvular or pharyngeal consonants, /a aj aw o aj aj aw ow/ when preceded by uvular consonants, and /e ej ew ə ej ej ew əw/ when preceded by pharyngeal consonants.

The consonant system will be more complex than that of Rihalle Kaafi largely because, as mentioned later, palatalization will have become phonemic (note that a number of contrasts are preserved that are found in some Rihalle Kaafi dialects but not in standard Rihalle Kaafi):

/m n ɲ/
/b d̪ d dz dʒ/
/pʰ t̪ʰ tʰ tsʰ tʃʰ cʰ kʰ qʰ/
/t̪ʼ tʼ tsʼ tʃʼ cʼ kʼ qʼ/
/s ʃ ç x χ h/
/β ð z ʒ ɣ ʁ/
/r l ʎ/
/w j/

(Aspirated stops derive from tenuis stops in the protolanguage, which are also reflected as aspirated stops in some Rihalle Kaafi dialects. Also, like some Rihalle Kaafi dialects, ejectives are only weakly ejective and may in fact be tenuis.)

Stress will be phonemic, because stress will have been conditioned by syllable weight prior to loss of phonemic vowel length, but the stress position afterwards will be preserved. Stress will always fall on one of the last three syllables of a word, including any clitics attached.

The syllable structure is C(C)V(ː)(C)(C)(C), with the exception that vowel-initial syllables are permitted in initial position.

Nasal codas will be lost, nasalizing preceding vowels. Note that nasal consonants may occur after vowels, due to geminate nasals nasalizing preceding vowels while simultaneously being shortened (resulting in nasal vowels followed by nasal consonants) and vowel elision after following onset nasals closing syllables (resulting in oral vowels followed by nasal consonants).

Now where do the long vowels come from, considering the vowel length contrast in the protolanguage has collapsed? Well... after vowel length was lost, stressed vowels lengthened in open syllables, and then gemination was lost.

Also, like some dialects of Rihalle Kaafi, there will be a system of vowel elision operating on historic short vowels where short vowels at odd distances from the stressed vowel will be elided provided that a consonant cluster of four or more consonants does not result or the vowel is word-final or is in an initial syllable. Also, this will make palatalization phonemic, because elided vowels will have already conditioned palatalization prior to elision. Note that this applies to clitics, just like how stress position is affected by clitics.

The above operated after the loss of gemination, and after subsequent rationalization of consonant clusters, new geminates were formed. Particularly, clusters of stops, affricates, and pairs of nasals took the POA and MOA of the second elements therein and became new geminates.

Like standard Rihalle Kaafi I plan on merging uvular and pharyngeal consonants, but only after vowel affectation and palatalization take place (which further will make palatalization more phonemic, because a palatalized uvular consonant will be velar, but a pharyngeal consonant will become uvular regardless).

I have not thought as much about morphosyntax, but one key change I want to make relative to Rihalle Kaafi is to lose the ergative and accusative cases while simultaneously not relying on word order to indicate grammatical roles. There will still be agreement clitics, marked for S/A versus P (the protolanguage, while having split ergative marking of nominal arguments, has accusative marking of agreement clitics), person, number (singular, dual, plural), and gender (masculine animate, feminine animate, masculine inanimate, feminine inanimate), attached to the verb. I am thinking of introducing an inverse marker on the verb, e.g. -/(ə)r/, which will have been inherited from the protolanguage (originally -/ɑr/) but lost in Rihalle Kaafi.

Another change I want to make is to introduce serial verb constructions, which are lacking in Rihalle Kaafi. I am undecided as to whether coverbs should receive agreement clitics for S/A, or whether they should only receive agreement clitics for P.

[Travis B.]

I have also decided to eliminate genitive case, with non-personal position being solely expressed via construct state.

[Travis B.]

One thing I realized is that the sound changes above will result in the merger of many pairs of perfective and imperfective verb forms. I am not sure how I want to go with this...

[jal]

One thing I realized is that the sound changes above will result in the merger of many pairs of perfective and imperfective verb forms. I am not sure how I want to go with this...

If perfective/imperfective is an important distinction you do not want to lose, I'd think it likely a periphrastic perfective or imperfective would emerge. Otoh, the contrast could just disappear altogether, contrasting it with the sister language.


JAL

[jal]

So when is a construction 'ergative' as opposed to passive? How is answering this not making a decision? Or are you in fact relying on a grammarian to make the decision?

A passive is typically marked on the verb, while in split ergativity the ergativity is either marked on the noun or the verb (note that there's a difference in marking preference between split-S and fluid-S, where split-S has a tendency to be marked on the verb, while fluid-S has to be marked on the noun).

Now with split-S it's clear it's not a passive, as there's afaik no languages that have constructions we call passives that have these passives mandatory based on the verb. So that leaves fluid-S. To answer the question whether a language uses fluid-S or passives, you'd have to answer the following questions:
1) Is the S of a detransitivized transitive that's the former P of a transitive, marked the same as the S of an intransitive where S is an experiencer (though different from an S that is an agent)? If so, it's ergative.
2) Does the language have a grammatical voice that is used to detransitivize a transitive that marks the former A as S? If so, it's ergative.
3) Does the language have a grammatical voice that is used to detransitivize a transitive that marks the former P as S, which is marked the same as a former A as S? If so, it's a passive.

There's of course cross-linguistically a lot of tendencies for languages to cluster a number of constructions that make it easy to label it as split ergative rather than accusative with a passive. No doubt you could come up with a conlang (or even natlang) where it's not so clear-cut, but I'd say in 99.9% it's quite clear.


JAL

[Travis B.]

One thing I realized is that the sound changes above will result in the merger of many pairs of perfective and imperfective verb forms. I am not sure how I want to go with this...

If perfective/imperfective is an important distinction you do not want to lose, I'd think it likely a periphrastic perfective or imperfective would emerge. Otoh, the contrast could just disappear altogether, contrasting it with the sister language.

In Rihalle Kaafi and the protolanguage perfective versus imperfective is a core part of the verbal system, because the unmarked usage of the perfective is specifically the past perfective and the unmarked usage of the imperfective is specifically the present imperfective, and extra affixes and word order are needed on the verb to construct the past imperfective or the 'present perfective' (which is used for the future and for performatives). In most situations these extra affixes are not used, which makes the basic contrast the past perfective versus the present imperfective. So losing this contrast would be like if English lost its distinction between the present and the simple past.

While the sound changes in question by no means affect all verbs, it affects enough verbs that a substantial portion of active verbs would no longer distinguish the past perfective from the present imperfective (stative verbs by their basic nature only distinguish tense and not perfectivity). I could ditch the perfectivity contrast altogether and rely solely on the verbal affixes and word order to indicate past versus present, but I do not exactly like this.

Another solution would be to make the past perfective by default and the present imperfective by default even when the verb stem does not distinguish perfectivity, and require auxiliary verbs to construct the past imperfective and the 'present perfective' in addition to the main verb, which would also be marked for perfectivity if it still distinguished it along with tense. I like this idea better, as it does not get rid of this basic distinction while at the same time not requiring special exceptions to sound change and still preserving the historical distinction between perfective and imperfective stems for verbs where these would not merge. It also can be justified in terms of that people really wanted to keep around such a basic distinction (like how many High German varieties adopted the present perfect to replace the preterite rather than simply merging the present and preterite when the preterite disappeared for most verbs), so they adopted periphrasic means of doing so.

[Travis B.]

Because agreement clitic position ambiguously marks the present imperfective affirmative in the protolanguage except when there is no agreement (e.g. "It rains"), I could actually dispense with affixes marking tense except for impersonal verbs with no direct object, and I could extend placing agreement clitics before the verb to mark the present imperfective in the negative too.

[Richard W]

A passive is typically marked on the verb, while in split ergativity the ergativity is either marked on the noun or the verb (note that there's a difference in marking preference between split-S and fluid-S, where split-S has a tendency to be marked on the verb, while fluid-S has to be marked on the noun).

Now with split-S it's clear it's not a passive, as there's afaik no languages that have constructions we call passives that have these passives mandatory based on the verb. So that leaves fluid-S. To answer the question whether a language uses fluid-S or passives, you'd have to answer the following questions:
1) Is the S of a detransitivized transitive that's the former P of a transitive, marked the same as the S of an intransitive where S is an experiencer (though different from an S that is an agent)? If so, it's ergative.
2) Does the language have a grammatical voice that is used to detransitivize a transitive that marks the former A as S? If so, it's ergative.
3) Does the language have a grammatical voice that is used to detransitivize a transitive that marks the former P as S, which is marked the same as a former A as S? If so, it's a passive.

Is a translation to English available?

Why would a language mark ergativity?

I couldn't parse ""mandatory based on the verb".

What does detransitivize mean? Does it mean to convert from a sentence with a subject and an object to one with only a subject while referring to the same action? If so, both forming a passive and an anti-passive would count as detransitivizing.

In (2), what does 'it' refer to? It seems to say that Thai is ergative because of its ถูก and โดน passives. Thai has no case marking related to subject and object.

Does the impersonal construction, which converts agent_in_nominative + active_verb to agent_in_instrumental + passive_verb, make Sanskrit fluid-S? I don't know whether direct objects are allowed in this construction.

[Travis B.]

A passive is typically marked on the verb, while in split ergativity the ergativity is either marked on the noun or the verb (note that there's a difference in marking preference between split-S and fluid-S, where split-S has a tendency to be marked on the verb, while fluid-S has to be marked on the noun).

Now with split-S it's clear it's not a passive, as there's afaik no languages that have constructions we call passives that have these passives mandatory based on the verb. So that leaves fluid-S. To answer the question whether a language uses fluid-S or passives, you'd have to answer the following questions:
1) Is the S of a detransitivized transitive that's the former P of a transitive, marked the same as the S of an intransitive where S is an experiencer (though different from an S that is an agent)? If so, it's ergative.
2) Does the language have a grammatical voice that is used to detransitivize a transitive that marks the former A as S? If so, it's ergative.
3) Does the language have a grammatical voice that is used to detransitivize a transitive that marks the former P as S, which is marked the same as a former A as S? If so, it's a passive.

Is a translation to English available?

I understood jal perfectly well.

Why would a language mark ergativity?

Ergativity simply means that there are situations where S=O is mandatory.

I couldn't parse ""mandatory based on the verb".

What jal meant is that in some cases the situation where S=O is mandatory, either based on the identity of the verb, tense or aspect, personhood/animacy/topicality hierarchies, or like.

What does detransitivize mean? Does it mean to convert from a sentence with a subject and an object to one with only a subject while referring to the same action? If so, both forming a passive and an anti-passive would count as detransitivizing.

Well, yes ─ detransitivize simply means to take a clause with A and O arguments and leave it as an intransitive clause with only an S argument by either deleting or demoting to oblique one of A or O and making the other S.

In (2), what does 'it' refer to? It seems to say that Thai is ergative because of its ถูก and โดน passives. Thai has no case marking related to subject and object.

What jal is describing is an antipassive, which is a classic feature of many languages with ergativity.

Does the impersonal construction, which converts agent_in_nominative + active_verb to agent_in_instrumental + passive_verb, make Sanskrit fluid-S? I don't know whether direct objects are allowed in this construction.

Fluid-S tends to be based on factors such as volitionality and like; in cases like these, it probably depends on the particulars.

[jal]

I understood jal perfectly well.

Thanks, I was starting to doubt myself 😅


JAL

[jal]

In Rihalle Kaafi and the protolanguage perfective versus imperfective is a core part of the verbal system, (...) So losing this contrast would be like if English lost its distinction between the present and the simple past.

Well, I can see the old Romans decrying the loss of the distinction between nominative and accusative the same way :D. Kidding aside, a language could easily ditch the imperfective/perfective paradigm for a past/present paradigm, repurposing the past imperfective for preterite. Not saying you should go there, but stranger things have happened.


JAL

[Travis B.]

In Rihalle Kaafi and the protolanguage perfective versus imperfective is a core part of the verbal system, (...) So losing this contrast would be like if English lost its distinction between the present and the simple past.

Well, I can see the old Romans decrying the loss of the distinction between nominative and accusative the same way . Kidding aside, a language could easily ditch the imperfective/perfective paradigm for a past/present paradigm, repurposing the past imperfective for preterite. Not saying you should go there, but stranger things have happened.

The key thing, though, is that the Romans didn't lose the distinction between S/A and P, they just expressed it different ways rather than relying on grammatical case. (Of course, this took longer than commonly alleged, as the French were still distinguishing a nominative and oblique a thousand years after the death of Julius Caesar.)