bradrn’s scratchpad

Conworlds and conlangs
bradrn
Posts: 6257
Joined: Fri Oct 19, 2018 1:25 am

Re: bradrn’s scratchpad

Post by bradrn »

Notes on SVCs

Serial verb constructions (SVCs) are a somewhat vexed topic. Too many good linguists have wasted too much ink trying to pinpoint exactly what is and isn’t an SVC. Whatever they might be, Proto-‘Savanna’ certainly has a lot of them. For these reasons, this post will focus on trying to give some overview notes on the more common types of SVCs in Proto-‘Savanna’, rather than trying to write a full and complete listing of the status of every possible SVC in the language.

I will adopt a working definition of SVC as ‘any construction involving two verbs in the same clause, in which there is no intonational break and each verb shares at least one argument with another’. This is a very non-rigorous definition, and is also quite wide: it includes quite a lot of things which other linguists disqualify from SVC-hood. In fact, the recent literature tends to refer to this category using another term entirely, as ‘multi-verb constructions’, rather than the narrower subset of ‘serial verb constructions’. Nonetheless, I will continue to call them SVCs because that seems to be the more usual term amongst conlangers, and also because quite a lot of them are, in fact, SVCs.

Firstly: what is not an SVC? Obviously, two independent clauses are not an SVC. This becomes less obvious when the subjects are coreferential, since in this situation the subject of the second clause can be elided:

Be
1s
walhali
go.IMPF-CONT
tsagif,
running,
lhiisi
perceive.PFV
di
can
meŋul
eye
kasip
star
tlii.
INDEF.PL

As I ran, [I] could see stars.

In this case, the difference in inflectional aspect between the verbs is a giveaway — SVCs in Proto-‘Savanna’ always have their verbs agreeing in inflectional aspect. The other obvious giveaway is the comma between the clauses: this has an intonational break, so cannot be an SVC.

More puzzling is the case of complementation, which is usually unmarked: e.g. be siwe tsisa qi walha, lit. ‘1s have desire 3s go’ = ‘I want him/her to go’. Is this an SVC or not? It certainly looks like one — but on the other hand it just as easily could be biclausal, in support of which the inner verb can freely change in aspect. Even more puzzling is the case where the inner subject can be deleted, to give e.g. be siwe tsisa walha ‘I want to go’, a construction which looks even more SVC-like. At some point we are forced to conclude that ‘is this an SVC or not’ is not, in general, a revealing question.

Now, onto the actual SVCs. These are a rather heterogenous group, and can be classified along several axes, the most important of which are probably:
  • Symmetry: Are all verbal slots unrestricted, or are some verbs drawn from a smaller subset in an asymmetric fashion?
  • Contiguity: Are the verbs necessarily placed immediately next to each other, or may other material (e.g. objects) intervene?
  • Argument sharing: Which arguments are shared, between which verbs?
In Proto-‘Savanna’, most SVCs are asymmetric and non-contiguous, and usually the subject is shared across all verbs. Furthermore, most SVCs (though not all) are composed of two slots: longer strings of verbs may be created by nesting SVCs inside SVCs. Thus, for instance, yusaye tsagif yagi lalef, lit. ‘come running return accompany’ = ‘run back with’ is parsed as [[[yusaye tsagif] yagi] lalef] ‘[[[come running] return] accompany]’.

Perhaps the most common SVCs are motional SVCs, which have the following summary structure:

Starting pointDirectionMannerPathDestination
fas ‘stand’ + nounwalha ‘go’,
yusaa ‘come’
coverb or ideophone, e.g.
naŋa ‘walk’,
tsagif ‘running’
verb, e.g.
qaathan ‘fall’,
yagi ‘return’,
qumum ‘go about randomly’
noun, possibly preceded by walha ‘go’

All of these elements are optional, although some have dependencies on others (e.g. the manner words require a preceding directional verb). The motional SVC as a whole may further be followed by a main verb, to indicate how the subject travels to perform that action:

Be
1s
fasi
stand.PFV
sasay
home
thaŋ
DEF.SG
waalhi
go.PFV
tsagif
running
qaathan
fall.PFV
fawetli
speak.PFV
feqe.
cry

I ran down away from home and cried.

Alternately, a motion verb can also be placed after the main verb. This construction is more restricted than that shown above, and at most one or two verbs are allowed. Semantically this can represent motion performed after the action, or some other form of associated motion done during the action: for instance, while walha fawetl feqe in the above example is ‘go cry’, fawetl feqe walha means something more like ‘go crying’. If the main verb has an object, often the motion applies to that, giving a switch-subject argument sharing pattern: e.g. be ŋandi ŋay yusaye, lit. ‘1s deform 3s come’ = ‘I pulled it towards me’. (Note that the pronoun cannot be reduced to its subject form here.) Sometimes the motion verb may be there purely to give a direction or path, rather than to express physical motion: e.g. lhiise meŋul qaathan is ‘look down’, rather than ‘look and fall’.

The two forms of motion SVC may of course be combined: walha fawetl feqe yagi, for instance, might mean ‘go and cry and come back’. The general principle here is that the verbs tend to be placed in iconic order, following the temporal order of events.

Aside from motional SVCs, there are also numerous constructions with applicative meaning, which add arguments. We have already seen one in passing: fas X, meaning ‘at X’. These are basically preposition-equivalents, performing a role very similar to that of prepositions in other languages. (Sinitic linguists tend to call these things ‘coverbs’, but we’re already using that word for something different.) A summary table (where X is an arbitrary noun and V is an arbitrary verb):

ConstructionGlossSemantics
fas X V /
V fas X
stand X V /
V stand X
V at X (locative)
siwe X Vhave/take X VV with X (instrumental)
tlaquf X Vaccompany X VV with X (comitative)
paatli X Vgive X Vhelp X with V (benefactive)
V paatli XV give XV to X (dative)
wedeŋ X Vresemble X VV like X (similative)

The instrumental construction is particularly productive, and can also be put together with other verbs to give constructions with semantics of disposal or despatch. For instance, ‘give X to Y’ is expressed as siwe X paatli Y, lit. ‘take X give Y’. Motion verbs can also be used: ’bring X to Y’, for instance, is walha siwe X yusaa paatli Y, lit. ‘go take X come give Y’. Siwe is furthermore able to introduce objects, if they are reasonably concrete: thus a sentence like Be qisni qi ‘I cut it’ can be rewritten as Be siwe ŋay qisni ‘I took it and cut’.

Similar to the applicative SVCs are causative SVCs. These take the general form of X waq Y V (Z), lit. ‘X do Y V (Z)’ = ‘X makes Y V (Z)’. Other verbs with similar meaning may be substituted for waq here: for instance, Be ŋandi qi waalhi qitsim ‘I squeezed it through the hole’, where ŋandi means ‘deform’.

The above were all asymmetric SVCs. However, Proto-‘Savanna’ also has some symmetric SVCs, in which we cannot as easily identify one main verb and another modifying verb. In general, these come in two types: simultaneous SVCs, in which two adjacent verbs share a subject and represent events occurring at the same time, and resultative or cause-effect SVCs, in which the object of the first verb is the subject of the second verb, and in which the second verb gives a result of the first verb’s occurrence. These SVCs have a strong tendency to lexicalise into compound verbs with an idiomatic meaning; the dictionary contains numerous such examples. I will not go into any further detail about symmetric SVCs here — partly because Proto-‘Savanna’ uses asymmetric SVCs far more often, partly because in any case such SVCs are often listed individually in the dictionary, but mostly because I don’t want to bother figuring out their details just right now.
Conlangs: Scratchpad | Texts | antilanguage
Software: See http://bradrn.com/projects.html
Other: Ergativity for Novices

(Why does phpBB not let me add >5 links here?)
bradrn
Posts: 6257
Joined: Fri Oct 19, 2018 1:25 am

Re: bradrn’s scratchpad

Post by bradrn »

Eŋes: Basic phonology and clause structure

Yes, I know I really should have been working on the grammar for my reconstruction relay language Jinyɛ́y. But I got an idea for a new language and it’s gotten lodged in my brain, so I guess this is what you’re getting instead. Phonologically, it’s inspired by Dorig, Japhug and Coptic. Grammatically… well, I’m not sure what this is grammatically. You’ll see in due course, I suppose.

Note: this one is very much going to be a scratchpad thread. There will almost certainly be inconsistencies, so feel free to ask if you get confused.

Consonants

Let’s start with the consonant inventory, shall we:

LabialAlveolarPalatalVelarGlottal
Nasalmnŋ
Voiceless stopptt͡ʃ ⟨c⟩k(ʔ)
Voiced stopbdg
Fricativefsx
Rhoticr
Liquidlj ⟨y⟩w

There are some notable cases of allophony:
  • The stops are unreleased syllable-finally: e.g. isip [isip̚] ‘star’. Voiceless stops can be mildly aspirated elsewhere, particularly before another consonant: e.g. tno [tʰno] ‘rope’.
  • The sole rhotic is the most variable consonant. At the end of a syllable it is realised as [ɾ] (e.g. waraʔr [waraʔɾ] ‘be big’), while in clusters with voiced consonants and at the end of a phrase after a vowel it is often [ʒ] (as in rmo [ʒmo] ‘the tree (accusative)’). It is a trill [r] in other cases.
The status of /ʔ/ is unclear. It has a highly restricted distribution, and in many ways it acts more as a vowel feature than as a consonant, but in other ways it seems quite consonantal. I’ll say more about this when I talk about phonotactics.

Vowels

There are five phonemic vowels: /a e i o u/. I don’t really have very much to say about them.

Phonetically, a sixth vowel [ə̆] is quite common. However, it appears only in the preverb as an epenthetic vowel: e.g. nd-bi-lis-i is realised as [ndə̆bilisi] ‘did you see it?’. Generally I’ll leave this unwritten, though if I come across a case where it’s really ambiguous I might write it as ⟨ĕ⟩.

Phonotactics

This is where it starts to get a little interesting. The general syllable structure is (C)(C)V(ʔ,y,w)(C). Hiatus is forbidden, so vowels must be separated by at least one consonant. The only constraints of note are that /ŋ/ is disallowed at the beginning of a syllable, and /ʔ/ is disallowed in the onset entirely. In particular, there are no constraints whatsoever on initial consonant clusters.

And when I say ‘no constraints’, I mean it. Syllables like msiŋ andlbuʔ and wgi and rmuk and fgis are all completely licit. Eŋes laughs at your ideas of the sonority hierarchy.

That being said, it’s not as bad as it could be, thanks to extensive resyllabification processes which mean most syllables are just CVC or CV, even when individual morphemes are more forbidding. For instance, to-wrob ‘it was black’ is syllabified as the comparatively more manageable /tow.rob/.

Another minor issue is the glottal stop. Since it’s forbidden from the onset, the end result is that /ʔ/ can appear only immediately after a vowel. This would seem to suggest that it’s more like creaky voice than a ‘proper’ consonant. On the other hand, it counts as breaking hiatus, the same as any other consonant. For now, I’ll treat it as a consonant, with the understanding that this identity is a little uncertain.

Basic clauses

I’ll finish off this post by writing a little about basic clause structure. (I won’t cover anything complicated because, firstly, I haven’t figured out all the intricacies of the grammar yet, and, secondly, because as fun as they are to say, sentences like inab bitowramumpefwesfeʔban? ‘why did it seem my dog was barking everywhere to help me?’ are just a little too complex to start off an introductory guide with.)

Anyway… the simplest clause is just a single verb:

Wal. ‘It goes.’
Asan. ‘It falls.’
Pesef. ‘It shines.’
Sa. ‘It dies.’
Isa. ‘It comes.’

The subject can be changed using a bound pronoun:

Basan. (b-asan) ‘I fall.’
Bawal. (ba-wal) ‘We go.’
Nisa. (n-isa) ‘You come.’
Ndasan. (nd-asan) ‘You all fall.’
Impesef. (iŋ-pesef) ‘They shine.’

A whole NP can also be used as subject. Note that if it’s definite (and singular, but more on that later), the formative to- must be inserted before the verb:

Inab sa. ‘A dog died.’
Nun topesef. ‘The water shone.’

For transitive verbs, bound pronominal objects can be placed after the verb, for instance:

Blismundin (b-lismun-din) ‘I see you.’
Nsiwi. (n-siw-i) ‘You have it.’

Full NPs can be placed before the verb, after a full subject if there is one. If the NP is definite, it gets the accusative casemarker si-:

Inab bsiw. ‘I have a dog.’
Sinun mbilismun? ‘Did you see the water?’

…which incidentally leads me to note that bi- before the verb marks a polar question.

A rather interesting feature of Eŋes which may now be apparent is that definiteness is marked on all nouns, but never with an article. Subject definiteness is marked on the verb, while object definiteness is represented on the casemarker.

And that’s all I have for now, I’m afraid. I’ll post more when I get time and/or I’ve figured out more about this language.
Last edited by bradrn on Wed Aug 09, 2023 8:22 am, edited 2 times in total.
Conlangs: Scratchpad | Texts | antilanguage
Software: See http://bradrn.com/projects.html
Other: Ergativity for Novices

(Why does phpBB not let me add >5 links here?)
chris_notts
Posts: 682
Joined: Tue Oct 09, 2018 5:35 pm

Re: bradrn’s scratchpad

Post by chris_notts »

bradrn wrote: Tue Aug 01, 2023 11:08 am This is where it starts to get a little interesting. The general syllable structure is (C)(C)V(ʔ,y,w)(C). Hiatus is forbidden, so vowels must be separated by at least one consonant. The only constraints of note are that /ŋ/ is disallowed at the beginning of a syllable, and /ʔ/ is disallowed in the onset entirely. In particular, there are no constraints whatsoever on initial consonant clusters.

And when I say ‘no constraints’, I mean it. Syllables like msiŋ andlbuʔ and wgi and rmuk and fgis are all completely licit. Eŋes laughs at your ideas of the sonority hierarchy.

That being said, it’s not as bad as it could be, thanks to extensive resyllabification processes which mean most syllables are just CVC or CV, even when individual morphemes are more forbidding. For instance, to-wrob ‘it was black’ is syllabified as the comparatively more manageable /tow.rob/.
It reminds me of various languages with iambic stress and historical vowel reduction processes. Ojibwe has some weird CC- initial clusters for similar reasons, and then there are all the Asian languages with sesquisyllabic words and weak limits on clusters. Khmer, Semelai, ... maybe old Tibetan, although I've always been confused to what extent those pervasive clusters were real, and how much they were an artifact of the writing system and/or transliteration.

The difference, of course, is that in Ojibwe the iambic reduction is iterative because it has long words, whereas most Asian languages with weak initial syllables don't show much iterativity because they don't have much morphology, or long roots.

This pattern also seems to be more pervasive than unrestricted final clusters? I get the impression that final clusters tend to be resolved over time by either consonant dropping, assimilation, ..., whereas some languages can preserve weird initial clusters for some time, presumably because initial position is more prominent than final position.

The other thing worth mentioning here is that often these are phonological clusters, not phonetic clusters. In many of the languages mentioned above, schwa-like transition vocoids occur in difficult clusters, but since they are short, invisible to phonological processes, and/or the languages in question otherwise lack centralised vowels, they're generally not written.
bradrn
Posts: 6257
Joined: Fri Oct 19, 2018 1:25 am

Re: bradrn’s scratchpad

Post by bradrn »

chris_notts wrote: Tue Aug 01, 2023 1:03 pm
bradrn wrote: Tue Aug 01, 2023 11:08 am This is where it starts to get a little interesting. The general syllable structure is (C)(C)V(ʔ,y,w)(C). Hiatus is forbidden, so vowels must be separated by at least one consonant. The only constraints of note are that /ŋ/ is disallowed at the beginning of a syllable, and /ʔ/ is disallowed in the onset entirely. In particular, there are no constraints whatsoever on initial consonant clusters.

And when I say ‘no constraints’, I mean it. Syllables like msiŋ andlbuʔ and wgi and rmuk and fgis are all completely licit. Eŋes laughs at your ideas of the sonority hierarchy.

That being said, it’s not as bad as it could be, thanks to extensive resyllabification processes which mean most syllables are just CVC or CV, even when individual morphemes are more forbidding. For instance, to-wrob ‘it was black’ is syllabified as the comparatively more manageable /tow.rob/.
It reminds me of various languages with iambic stress and historical vowel reduction processes.
Bingo! In this case the inspiration is Dorig, as mentioned, though that had historic trochees rather than iambs. Refer to François (2005) for more.
The other thing worth mentioning here is that often these are phonological clusters, not phonetic clusters. In many of the languages mentioned above, schwa-like transition vocoids occur in difficult clusters, but since they are short, invisible to phonological processes, and/or the languages in question otherwise lack centralised vowels, they're generally not written.
In the case of Dorig at least, I’ve listened to some samples (on PARADISEC) and they sound like phonetic clusters to me.
Conlangs: Scratchpad | Texts | antilanguage
Software: See http://bradrn.com/projects.html
Other: Ergativity for Novices

(Why does phpBB not let me add >5 links here?)
User avatar
foxcatdog
Posts: 1662
Joined: Fri Nov 15, 2019 7:49 pm

Re: bradrn’s scratchpad

Post by foxcatdog »

sonority hierarchy breaking isn't very aesthetic
bradrn
Posts: 6257
Joined: Fri Oct 19, 2018 1:25 am

Re: bradrn’s scratchpad

Post by bradrn »

foxcatdog wrote: Tue Aug 01, 2023 8:21 pm sonority hierarchy breaking isn't very aesthetic
Aesthetics are subjective. I like it, and for my conlang, that’s the only thing that matters.
Conlangs: Scratchpad | Texts | antilanguage
Software: See http://bradrn.com/projects.html
Other: Ergativity for Novices

(Why does phpBB not let me add >5 links here?)
User avatar
foxcatdog
Posts: 1662
Joined: Fri Nov 15, 2019 7:49 pm

Re: bradrn’s scratchpad

Post by foxcatdog »

bradrn wrote: Tue Aug 01, 2023 8:44 pm
foxcatdog wrote: Tue Aug 01, 2023 8:21 pm sonority hierarchy breaking isn't very aesthetic
Aesthetics are subjective. I like it, and for my conlang, that’s the only thing that matters.
Pretty defensive for a semi ironic comment but its okay. And a good *wsketk sounds tempting
keenir
Posts: 948
Joined: Fri Apr 05, 2019 6:14 pm

Re: bradrn’s scratchpad

Post by keenir »

foxcatdog wrote: Tue Aug 01, 2023 9:23 pm
bradrn wrote: Tue Aug 01, 2023 8:44 pm
foxcatdog wrote: Tue Aug 01, 2023 8:21 pm sonority hierarchy breaking isn't very aesthetic
Aesthetics are subjective. I like it, and for my conlang, that’s the only thing that matters.
Pretty defensive for a semi ironic comment but its okay.
And you made it crystal clear that you were being semi ironic...how/when/where ?
User avatar
foxcatdog
Posts: 1662
Joined: Fri Nov 15, 2019 7:49 pm

Re: bradrn’s scratchpad

Post by foxcatdog »

this isn't reddit
keenir
Posts: 948
Joined: Fri Apr 05, 2019 6:14 pm

Re: bradrn’s scratchpad

Post by keenir »

foxcatdog wrote: Tue Aug 01, 2023 10:53 pm this isn't reddit
That is true.

Also

That answers exactly nothing.
User avatar
foxcatdog
Posts: 1662
Joined: Fri Nov 15, 2019 7:49 pm

Re: bradrn’s scratchpad

Post by foxcatdog »

reddit uses /s's to denote sarcasm
bradrn
Posts: 6257
Joined: Fri Oct 19, 2018 1:25 am

Re: bradrn’s scratchpad

Post by bradrn »

foxcatdog wrote: Tue Aug 01, 2023 11:15 pm reddit uses /s's to denote sarcasm
It’s a very helpful convention, and I’ve seen it used outside Reddit too. Sarcasm is famously hard to convey over text, and this is especially true when your comment is only six words long.
Conlangs: Scratchpad | Texts | antilanguage
Software: See http://bradrn.com/projects.html
Other: Ergativity for Novices

(Why does phpBB not let me add >5 links here?)
keenir
Posts: 948
Joined: Fri Apr 05, 2019 6:14 pm

Re: bradrn’s scratchpad

Post by keenir »

foxcatdog wrote: Tue Aug 01, 2023 11:15 pm reddit uses /s's to denote sarcasm
i've never seen /s used to denote it on any of the reddit sites I've been to...though I've seen /sarcasm used on nonreddit sites.

though its all moot, as there was exactly 000 /s in your post of
foxcatdog wrote: Tue Aug 01, 2023 8:21 pmsonority hierarchy breaking isn't very aesthetic
...which you later said was semi ironic.

so, as I said before, how exactly was anyone supposed to know you had intended your statement to be ironic in part or in full?
chris_notts
Posts: 682
Joined: Tue Oct 09, 2018 5:35 pm

Re: bradrn’s scratchpad

Post by chris_notts »

Bingo! In this case the inspiration is Dorig, as mentioned, though that had historic trochees rather than iambs. Refer to François (2005) for more.
Of course, right to left trochees can produce the same effect more intermittently. :D

My reworked Qummin, unlike Ojibwe and Doric, has left to right trochees and syncope, so this doesn't happen and only intervocalic clusters are produced.

I thought, although with no evidence, that left to right footing should be more stable as a productive synchronic basis for a syncope process in a morphologically complex language, because it means you can iteratively syllabify with little look ahead or backtracking, whereas when you've got a long morphologically complex word, you'd need to know the entire word in advance to do right to left syncope correctly. Of course, this doesn't matter if this is either a historical, levelled process or the words aren't that long anyway.

I guess this is programmer brain: thinking about the algorithm in terms of how local its decision making is.
Last edited by chris_notts on Wed Aug 02, 2023 2:57 am, edited 1 time in total.
chris_notts
Posts: 682
Joined: Tue Oct 09, 2018 5:35 pm

Re: bradrn’s scratchpad

Post by chris_notts »

Also, I think #CC- clusters are aesthetically fine. It doesn't stop Tibetan from being a cool kid language.
bradrn
Posts: 6257
Joined: Fri Oct 19, 2018 1:25 am

Re: bradrn’s scratchpad

Post by bradrn »

chris_notts wrote: Wed Aug 02, 2023 2:43 am I thought, although with no evidence, that left to right footing should be more stable as a productive synchronic basis for a syncope process in a morphologically complex language, because it means you can iteratively syllabify with little look ahead or backtracking, whereas when you've got a long morphologically complex word, you'd need to know the entire word in advance to do right to left syncope correctly. Of course, this doesn't matter if this is either a historical, levelled process or the words aren't that long anyway.
This has more to do with footing than with syncope, but indeed I believe I’ve heard this suggested with regards to footing.

(Also, ‘Doric’ and ‘Dorig’ are different languages…)
Conlangs: Scratchpad | Texts | antilanguage
Software: See http://bradrn.com/projects.html
Other: Ergativity for Novices

(Why does phpBB not let me add >5 links here?)
chris_notts
Posts: 682
Joined: Tue Oct 09, 2018 5:35 pm

Re: bradrn’s scratchpad

Post by chris_notts »

bradrn wrote: Wed Aug 02, 2023 3:19 am (Also, ‘Doric’ and ‘Dorig’ are different languages…)
Oops...
User avatar
quinterbeck
Posts: 394
Joined: Sat Jul 21, 2018 12:19 pm

Re: bradrn’s scratchpad

Post by quinterbeck »

I have a soft spot for funky consonant clusters, so I love the aesthetic of Eŋes!
bradrn wrote: Tue Aug 01, 2023 11:08 am A whole NP can also be used as subject. Note that if it’s definite (and singular, but more on that later), the formative to- must be inserted before the verb:
Looking forward to when you explain why this is/isn't a ditropic clitic :P
bradrn
Posts: 6257
Joined: Fri Oct 19, 2018 1:25 am

Re: bradrn’s scratchpad

Post by bradrn »

quinterbeck wrote: Thu Aug 03, 2023 8:35 am I have a soft spot for funky consonant clusters, so I love the aesthetic of Eŋes!
bradrn wrote: Tue Aug 01, 2023 11:08 am A whole NP can also be used as subject. Note that if it’s definite (and singular, but more on that later), the formative to- must be inserted before the verb:
Looking forward to when you explain why this is/isn't a ditropic clitic :P
For one thing, it comes after other morphemes of the verbal complex, e.g. the question marker:

Nun bitopesef? ‘Did the water shine?’

(Not that that question makes much sense, but at least it shows the principle.)

It also comes after the object:

Wnus inab tosiw. ‘The man has a dog.’
Conlangs: Scratchpad | Texts | antilanguage
Software: See http://bradrn.com/projects.html
Other: Ergativity for Novices

(Why does phpBB not let me add >5 links here?)
bradrn
Posts: 6257
Joined: Fri Oct 19, 2018 1:25 am

Re: bradrn’s scratchpad

Post by bradrn »

The preverb

With this post I’m going to make a start on the verbal complex and its somewhat intricate structure (for lack of a better term). It has a generally layered form: this post focusses on the outermost portion, which I’m calling the preverb.

To summarise, the preverb generally marks adverbial, discourse and participatory categories. As the name suggests, it is the first element in the verb complex. To a first approximation, it can be analysed with the following template-and-slot formalism (with some subtleties):

AdverbialSubject 1ConnectivityEvidentialSubject 2

No examples are known with all five slots occupied, however. (And in fact some slots are nearly mutually exclusive, e.g. Subject 1 and Subject 2.)

I won’t be focussing much on phonology here, but it is worth noting that the preverb as a unit is somewhat distinct from the verb complex proper. For instance, hiatus avoidance does not apply at the end of a preverb: e.g. to-asan ‘it fell’. Similarly, when they contain more than a single subject marker, preverbs tend to get final stress independently of the main verb. On the other hand, processes like nasal assimilation (on which I’ll say more later) do apply across the boundary.

(Formatting note: to keep things comprehensible, I’ll try to underline morphemes from the slot being covered. If I happen to miss any, well, there’s a challenge for you to figure out yourself!)

Adverbial

The first slot in the preverb is the adverbial, which marks notions such as place, time and manner. This is by far the largest slot in the preverbal paradigm: there are many tens of adverbial markers, ranging from the highly general (e.g. sey- ‘like this’, nan- ‘somewhere’, isaʔ- ‘certainly’) to the highly specific (e.g. say- ‘at home’, sik- ‘with cracking’). I won’t bother trying to make a complete list now, though maybe I will later.

The adverbial slot is peculiar in a couple of ways. Firstly, it’s the only slot which can in some cases admit more than one formative: e.g. sayseybwa ‘at home I do it like this’. However, this tends to be avoided.

Secondly, the presence of an adverbial alters the order of the following slots. Specifically, the ‘Subject 1’ slot is metathesised ahead of the connectivity marker when an adverbial is present: compare e.g. iŋmigiʔsam ‘they were eating’ with saymiŋgiʔsam ‘at home they were eating’. This is the only situation in which the preverbal slots occur in a order different to that outlined above.

Subject 1

The first subject slot indicates both pronominal subject and interrogative status with a single mildly fused marker. The declarative series is:

PersonSingularPlural
1b-ba-
2n-d-
3(none)(i)ŋ-

And the interrogative series is:

PersonSingularPlural
1bi-babi-
2mbi-mbi-
3biʔ-(i)mbi-

Some phonological notes on this slot:
  • Eŋes has a consistent process of nasal assimilation in which a nasal assimilates to the place of articulation of a following stop. This process is particularly relevant for this slot: e.g. +pesef is impesef, not *iŋpesef.
  • In the 3p forms, the initial vowel is omitted after another vowel. Note there are some tricky cases here: e.g. isaʔ++pesef becomes isaʔmpesef, not *isaʔimpesef.
The usage of the 1/2 forms vs the 3 forms is somewhat different. The former function as true agreement morphemes: they are included even when a full subject pronoun is used. By contrast, if a nominal subject is present, the third-person forms must be omitted in declarative sentences, or replaced with bi- in interrogatives.

In a similar way, the Subject 1 and Subject 2 slots are nearly mutually exclusive. In declarative sentences, if one is occupied the other must be empty; in interrogatives, Subject 1 can only be bi- if Subject 2 is filled.

Connectivity

I feel quite sure there’s some more ‘proper’ linguistic term for this, but if so, I can’t remember it. Whatever it may be called, this slot has a strongly discourse-focussed role: it marks the relationship between the current clause and an adjacent one (usually the previous one), often across speakers. Four possible morphemes can occupy this slot (aside from the null marker, of course):

MorphemeGloss
dan-‘so’, ‘thus’
bu-‘however’, ‘but’
man-‘because’
mbi-‘changed state’
m-‘backgrounding’

I’ve listed the first three with a fairly straightforward English gloss or two, which matches their usage fairly closely. However, the last two may require a bit more explanation.

mbi- highlights a situation with a changed state. In some cases, it might appear to correspond directly to English ‘but’, for instance:

Simwaskun, bmbiwraw.
sim-waskun, b-mbi-wraw
little-pain, 1s-CS-good
It was hurting a bit, but I’m better now.

However, there is more to the story: saying … bwraw, without the connectivity marker, could equally well mean ‘… and I was fine’, with no implication of state change. Furthermore, mbi- can be easily used in cases where ‘but’ would be inappropriate: e.g.

Anmiŋwal, say mbitowraŋwasnar.
an-m-iŋ-wal, say mbi-to-wraŋ-wasnar
now-BG-3p-go, house CS-SG-seem-empty
Now they’ve all gone, (*but) the house looks so empty.

Here the state change is implied: earlier the house was full, but now it’s empty.

As it happens, the previous example also shows the use of the other somewhat tricky connectivity morpheme: namely m-, glossed as ‘backgrounding’ above. In fact it could instead have easily used the state change marker (Anmbiŋwal …), without much change in meaning, but the use of m- acts to de-emphasise the first clause and instead focus on the second, as the speaker wants.

This points to the other very common use of m-: in narratives, as a marker of what has been called a ‘secondary storyline’. This basically refers to the events which surround the primary sequence of events in a story. For now, it’s hard to give examples of this; perhaps one day I’ll try translate a story into this language, when it’s more developed.

Evidential

The second-last preverbal slot can optionally be filled by only one morpheme: tow- ‘indirect evidence’. This indicates a proposition which is known from another person’s report, from hearsay, from folklore, or from inference.

Note that the absence of tow- does not necessarily indicate direct evidence! Even though tow- cannot for an event which was directly seen or heard, the converse is not true: a lack of evidential marker merely creates an implicature of direct evidence. The contrast is not ‘direct vs indirect evidence’, but ‘unmarked vs indirect evidence’.

There’s not much more to say on this, so let us turn to the final element of the preverb…

Subject 2

The final slot can be occupied by one of two markers: to- ‘singular’ and taŋ- ‘plural’. (A third morpheme ti- is attested in some old expressions, where it may possibly refer to some kind of generic: rnis timownesuf ‘the bee never hungers’ — or more idiomatically, ‘waste not, want not’. But it’s fossilised and entirely unproductive in the modern language.)

As the glosses suggest, these mark plurality of the subject. Importantly, they only mark definite, third-person subjects: when the subject is indefinite, the Subject 2 slot must be unoccupied. And, as already mentioned, other persons are cross-referenced in Subject 1, not Subject 2.

At some point I may end up writing a whole post about these very common markers, but for now let me summarise the main uses. Obviously, if a definite subject NP is explicitly specified, it needs a Subject 2 marker:

Say mbitowraŋwasnar. ‘The house looks so empty.’ (from above)
Tam taŋma. ‘The food is done.’

But in fact, most Subject 2 markers are used without an explicit subject NP at all. In those circumstances it seems to be acting as a topic-tracking device: the definite NP being referred to is implied from context, and may only have been specified many sentences ago, if at all. Many sentences may continue using to- or taŋ- sequentially, with the occasional bound pronoun thrown in for those clauses where the subject changes briefly. (Alas, I’m too tired at the moment to be bothered with making up an example which shows this really well.)

[EDIT: revisiting this point, the situation turns out to be more or less the opposite: to- and taŋ- are for non-topical NPs, with topical NPs using the Subject 1 markers.]

…and that’s it for the preverb, I think! Next up: the first portion of the verbal complex proper. (Or at least, when I figure out how on Earth it’s supposed to be structured. Don’t hold your breath.)
Last edited by bradrn on Wed Aug 30, 2023 3:53 am, edited 1 time in total.
Conlangs: Scratchpad | Texts | antilanguage
Software: See http://bradrn.com/projects.html
Other: Ergativity for Novices

(Why does phpBB not let me add >5 links here?)
Post Reply