Balog: a strictly monovalent conlang

[Vardelm]

I hope this language is not dormant?

I wouldn't think so. It's only been 1 month since Imralu posted. Conlangs can take years to develop, so I'm sure there will be more posts when there are more developments. I haven't posted in my scratchpad thread for almost 4 months, and i can assure you my project isn't dead either. 1 month is nothing.

[Imralu]

I hope this language is not dormant?

It was, but not anymore.

So, after a long time of dormancy, I'm working on Balog again.

I've added a few more consonants to the inventory. The ejective <pq> is new, as is the series of lateral affricates: <dl>, <tl>, <tlq>. They can either be dental-alveolar or velar.

Code: Select all

                   LABIAL    DENTAL/   PALATAL(ISED)  VELAR/    GLOTTAL
                             ALVEOLAR                 UVULAR
NASAL:             <m>       <n>                      <ŋ>/<ng>
PLOSIVE VOICED:    <b>       <d>                      <g>
PLOSIVE VOICELESS: <p>       <t>                      <k>       <q>
PLOSIVE EJECTIVE:  <pq>      <tq>                     <kq>
AFFRICATE VOICED:            <dz>      <dž>/<dzh>    
AFF. VOICELESS:              <ts>      <tš>/<tsh>    
AFF. EJECTIVE:               <tsq>     <tšq>/<tshq>
LAT. AFF. VOICED:            <dl>
LAT. AFF. VOICELESS:         <tl>
LAT. AFF. EJECTIVE:          <tlq>
FRICATIVE VOICED:  <v>       <z>       <ž>/<zh>       <ğ>/<gh>
FRIC. VOICELESS:   <f>       <s>       <š>/<sh>       <x>       <h>
TRILL/FLAP:                  <r>
LATERAL:                     <l>
SEMIVOWEL:                             <y>            <w>

Body parts: Number and possession:
For body parts, the basic root is a verb meaning to use the body part in question. For example:

• b 'use hand'
• dev 'use beard'

The suffix -ed, indicating a tool or body part needed in order to carry out the root verb, is added to indicate the bodypart itself.

• bed 'be hand'
• deved 'be beard'

For paired bodyparts, the suffixes -ak 'use one' and -uv 'use both' may be added as needed. When neither is added, the number used is ambiguous. The combination -ak-ed becomes -ad.

• bak 'use one hand' > bad 'be a hand'
• buv 'use both hands' > buved 'be a pair of hands'

In spite of the singular translations I've given above, this is independent of actual number, which is also optionally indicated by a suffix:

• -an SINGULATIVE
• -im PLURATIVE
• -iŋ COLLECTIVE

E.g.

• badan 'be one hand'
• badim 'be plural individual hands'
• buvedan 'be one pair of hands'
• badiŋim 'be plural collections of individual hands' (morphemically: b-ak-ed-iŋ-im use.hand-use.one-tool-COLL-PL

The regular means of creating a possessive involves the essive/equative proclitic we:- and either the dominant or subordinate possessive proclitic, nu:- or ni:- respectively. Note that subordinate, in this case, does not mean grammatically subordinate, but semantically subordinate, as in the nature of possession. Body parts, with few exceptions, require the subordinate possessive clitic ni:- because having them is not a person's choice. Body parts such as hair and beards, however, can be marked with the dominant possessive clitic because their existence may be the bearer's choice. Using the dominant possessor emphasises this.

• buved wenniiž 'be my two hands' (b-uv-ed we:-ni:-ž use.hand-use.pair-tool ESS-SUB.POSS-1S)
• deved wenniiž or deved wennuuž 'be my beard' (the latter emphasises that the speaker has a beard by choice)

The morpheme -ed, however, is frequently elided to -e- and then takes the possessive clitic directly on the end as a suffix.

• buveniiž 'be my two hands' (b-uv-e(d)-ni:-ž use.hand-use.pair-tool-SUB.POSS-1S)
• deveniiž or devenuuž 'be my beard' (the latter emphasises that the speaker has a beard by choice)

Compounds/Serial Verbs:
Verbs may be compounded but require a vocalic linking element. This preserves the strict CV(CV(CV(...)C structure of clauses. Consonant clusters (other than those where the second element is a semivowel) are only found at the boundaries between clauses.

There are two linking elements for compounding:

• -o- gloss: CN1
• -e- gloss: CN2

Each of these can also be a standalone word, which may be used for extra emphasis or clarity, particularly in freely created compounds that are not yet lexicalised:

• -o- = au gloss: CN1
• -e- = ai CN2

The difference between them is that CN1 is used when both verbs refer to the same entity and CN2 is used when they refer to different entities. Here are a couple of instances of each one that I have already used in the fluency thread.

• ğiwušonubbalog
  

  ğ<iw>uš

  speak<GDV>

  -o-

  -CN1-

  nu:-

  DOM.POSS

  balog

  balog

  
  be the language of the Balog (basically "be Balog's language" - one entity is a language and is of the Balog)
• buweŋekanenuuž
  

  b<uw>eŋ

  heal(ITR)<APASS>

  -e-

  -CN2-

  kan

  use.tooth

  -e(d)

  tool

  nu:-

  DOM.POSS

  ž

  1S

  
  be my dentist (basically: "be my tooth doctor" - this entity is a doctor, but is not a tooth)

The use of -o- or au is not quite as strictly equative as we:-. For example, "try-o-sleep" means "tries to sleep" (one entity performing both actions, regardless of success, i.e. regardless of whether the second verb is understood as realis or not), whereas "try wes-sleep" would mean "tries and sleeps", "tries while sleeping", "is one who tries and is asleep" etc. I think I"ll allow CN1 to be used for partitive and material relationships too, meaning "coconut oil" would be "be.oil-o-be.coconut".

The CN2 connector essentially just introduces a modifier that is associated with the head in any undefined way. For example, "baby oil" would be "be.oil-e-be.baby" ... destroying any chance for jokes about baby oil being made from babies.

Before I go, have a Frankenstein word. devenubbuweŋekanedeğiwušonubbalog = 'be the beard of the/a Balog-language (i.e. Balog-speaking') dentist'. Divided up into its full form it would be deved wennubbuweŋ ai kaned ai ğiwuš au nubbalog.

[bradrn]

Compounds/Serial Verbs:
Verbs may be compounded but require a vocalic linking element. This preserves the strict CV(CV(CV(...)C structure of clauses. Consonant clusters (other than those where the second element is a semivowel) are only found at the boundaries between clauses.

There are two linking elements for compounding:

• -o- gloss: CN1
• -e- gloss: CN2

… The difference between them is that CN1 is used when both verbs refer to the same entity and CN2 is used when they refer to different entities.

Technically speaking, if you have overt dependency markers it can’t be called a serial verb construction (at least in the usual definitions). I’m not sure what this would be, though.

[Imralu]

Technically speaking, if you have overt dependency markers it can’t be called a serial verb construction (at least in the usual definitions). I’m not sure what this would be, though.

Yeah, just compounds then with two different types of modifiers. I don't know. Words are just words.

[Namdio]

: Kqal uvvaŋ magaz iddauz.
→ "The hunter cut the tree down (with an axe).

I notice the presence of Kq in the verb to cut, which you also said was a negative marker. Does that imply that the verb to cut is more like to reduce? Or like to not(something)? I'm not sure if I'm understood.

B'uvvaŋ, magaz iddauz wettsik. b use.hand u= DEF(a)= vaŋ hunt | | magaz fall.over i= DEF(e)= dauz be.tree we= ATTR= tsik be.small The hunter pushed/pulled the small tree over (by hand).

I also noticed a change in the verb 'to use hand' from 'b' to 'd'. Is that a mistake or is it some construction that appears in certain context?

[Namdio]

One thing about the internet though: in nearly all cases, things can't be settled physically and there is no actual danger of ranking someone lower than you when they could actually kick your arse, so I imagine, if this language were used on something like the internet, there would just be a lot of people arrogantly ranking themselves higher than everyone else and either this would result in flame-wars or it would just be ignored and used to form judgements about who is arrogant and who isn't. Potentially, there would be a politeness standard where second person is placed above first person, always, and third persons could be placed even higher or lower depending on what you want to say about them. Maybe placing 2nd person above 1st person would even become a rule of politeness on some forums. I don't know. I think something like that would change the language. When it's spoken mostly face to face, this system seems like it could work well (although horribly, lol), but in fairly anonymous, written communication, I think a change would come about.

Perhaps speaking with oneself in the highest formal rank would become the norm when talking on the internet. (I remember reading something about the Chinese using the arrogant(?) I pronoun when speaking online (can't validate though)

Maybe in the beginning it might have started of as a pissing contest, 'You cannot verify my rank so therefore I will put myself higher than you (or make myself the highest'. Then as new users get introduced to the medium, they just assume that it is the norm for speaking when on such a platform and adopt it as well (only for internet speak, though there might be some spill over).

I also think that the people who make use of the internet first would heavily determine the official formal rank used when speaking on the internet. And it may also differ from site to site, as different people use different sites and certain purposes call for certain speech modes. For example you might find the the u= rank I= rank might be used a lot more in competitive websites or online programs, like online games (mostly of the sports and action genre) much like trash talk.

Then in more casual sites like chatrooms, there may be some semblance of the spoken rank ordering (mods having higher ranks, and most people just defaulting to the neutral-ish o= rank (I hope I got this right).

Then perhaps on sites for businesses, you might see the use of a slightly more prestigious a= rank to refer to either the targeted user or the company making the product (either in an effort to butter up the customers (as in, 'wow a strong person like you would definitely want to patronize us') or in an effort to paint themselves in a much better light ('we are much better than the competition so chose to do business with us'). Such spaces may avoid the usage of the u= rank, in order to avoid coming off as arrogant or proud.

Also I can't seem to get this idea out of my head. Two orcs writing scholarly papers, and one referring to himself with a higher rank than the other.Then in retaliation, the slighted orc goes all the way to where the other orc stays and beats him up.

[Imralu]

: Kqal uvvaŋ magaz iddauz.
→ "The hunter cut the tree down (with an axe).

I notice the presence of Kq in the verb to cut, which you also said was a negative marker. Does that imply that the verb to cut is more like to reduce? Or like to not(something)? I'm not sure if I'm understood.

Hi. I haven't really thought of an etymology for kqal. The kq is just there because of a bit of sound symbolism. [kʼ] represents sharpness. I hadn't intended it to have anything to do with the negative morpheme kq, but I like the way you're thinking, like it could mean "reduce to nothing". Kqal doesn't mean "cut" though. It means "use/wield an axe". (Whence kqaled 'be an axe'.) It doesn't specifically mean cut with an axe, for example:

Do kqal aaž ŋaf offuweŋ.

do

PST

kqal

use.axe

a:=ž

DEF(b)=1S

ŋaf

flee

o:=f<uw>eŋ

DEF(c)=be.stolen<APASS>

I chased the thief off with my axe.
I waved my axe and the thief ran away.

If needed, you can add extra clauses to the sentence if the jump is too big to be followed. I use axe, tree is cut through, it falls over. Then I use axe, thief use eye, he gets scared, he runs away. Paratactic chaining of unmarked clauses is always understood with a causal/contextual connection between them. To break this, a particle like tšqe "unrelatedly" needs to be used. It can be used humorously as well, like Tšqe dohi kqal aaž tšqe da tak oo leg aa tak iwwa diž o med. ("So, I was doing something with my axe and then, unrelatedly, they gave me all of this money here.")

The verbs I've already made for "cut" are:

• tsquwak = cut into something hard quickly with a blow, e.g. stone, wood, bark, stick - not for slow cutting, as with a saw or cutting through/off)
• suwax = cut something soft but crisp, e.g. hair, paper, dry leaves

There will be more. Dividing soft things such as cake, flesh. Sawing or grinding into something. Cutting something off or cutting right through something (tree, branch, limb etc.).

Those verbs above are both the antipassive/causative derivations of tsqak and sax, both meaning "be cut" (action, not state - dammit English! ... "get cut") with the same connotations as above. The perfect derivation indicating a resulting state is tsqwak and swax ... which also translates to "be cut".

B'uvvaŋ, magaz iddauz wettsik. b use.hand u= DEF(a)= vaŋ hunt | | magaz fall.over i= DEF(e)= dauz be.tree we= ATTR= tsik be.small The hunter pushed/pulled the small tree over (by hand).

I also noticed a change in the verb 'to use hand' from 'b' to 'd'. Is that a mistake or is it some construction that appears in certain context?

Oh yeah, I changed it. I go through my lexicon and change things from time to time and then try to look for all instances of it and change them too. It was d and now it's b, but I guess I didn't change it everywhere in this thread. I guess it's still d in the list of instrumental verbs, and wherever anyone has quoted me.

[Imralu]

Perhaps speaking with oneself in the highest formal rank would become the norm when talking on the internet. (I remember reading something about the Chinese using the arrogant(?) I pronoun when speaking online (can't validate though)

Maybe in the beginning it might have started of as a pissing contest, 'You cannot verify my rank so therefore I will put myself higher than you (or make myself the highest'. Then as new users get introduced to the medium, they just assume that it is the norm for speaking when on such a platform and adopt it as well (only for internet speak, though there might be some spill over).

I also think that the people who make use of the internet first would heavily determine the official formal rank used when speaking on the internet. And it may also differ from site to site, as different people use different sites and certain purposes call for certain speech modes. For example you might find the the u= rank I= rank might be used a lot more in competitive websites or online programs, like online games (mostly of the sports and action genre) much like trash talk.

Then in more casual sites like chatrooms, there may be some semblance of the spoken rank ordering (mods having higher ranks, and most people just defaulting to the neutral-ish o= rank (I hope I got this right).

Then perhaps on sites for businesses, you might see the use of a slightly more prestigious a= rank to refer to either the targeted user or the company making the product (either in an effort to butter up the customers (as in, 'wow a strong person like you would definitely want to patronize us') or in an effort to paint themselves in a much better light ('we are much better than the competition so chose to do business with us'). Such spaces may avoid the usage of the u= rank, in order to avoid coming off as arrogant or proud.

Yeah, that's all possible and I really don't know how it would go and also can't be bothered coming up with the canon for how it does go in world, because it's not necessary. Their world is internetless for now. It would definitely cause a bit of language change, at least in the variety used online. Once the rank is more or less meaningless, people would need to keep explicitly mentioning first and second person with ž and l, rather than mentioning them briefly and then letting the rank make it clear.

Also I can't seem to get this idea out of my head. Two orcs writing scholarly papers, and one referring to himself with a higher rank than the other. Then in retaliation, the slighted orc goes all the way to where the other orc stays and beats him up.

LMAO! Could happen! Disputes over rank can turn into arm-wrestling, full on wrestling or even all-out combat ... the two involved typically agree on what it will take to make clear who is higher ranked first. Both will of course try to suggest things that they are best at ... e.g. "Let's arm wrestle." "No, let us try to push one another out of this circle." (There should be a word for that.) If there is a clear winner, but one still won't acquiesce to the ranking in speech, then it may turn into real violence.

[bradrn]

I just got three notifications from old posts in this thread. Does anyone know what happened?

[Imralu]

I just got three notifications from old posts in this thread. Does anyone know what happened?

Ah, I edited some old posts to update some changes in vocabulary. Weird that it gave you a notification without indicating that it was for an edit. I guess because I had quoted you in those posts, it triggered the "quote" notification again. Sorry!

[bradrn]

I just got three notifications from old posts in this thread. Does anyone know what happened?

Ah, I edited some old posts to update some changes in vocabulary. Weird that it gave you a notification without indicating that it was for an edit. I guess because I had quoted you in those posts, it triggered the "quote" notification again. Sorry!

No worries! That sounds about right, since it gave me all ‘quoted’ notifications.

(In fact, I don’t think I ever have gotten an ‘edit’ notification…)

[Imralu]

Technically speaking, if you have overt dependency markers it can’t be called a serial verb construction (at least in the usual definitions). I’m not sure what this would be, though.

I've realised that these are essentially converbs. They subordinate the following verb to the preceding one.

Here are another couple of examples.

• ŋel = be unique, be the only one, be the sole one
• nam = be liked, appeal
• tlev = be a name
  
  → ŋel au nam | ŋel o nam | ŋelonam = be exclusively liked
  (E.g. Ŋelonam ool m'aaž. = You're the only one I like. | I like only you.
  → ŋel ai tlev | ŋeletlev = be uniquely named
  (E.g. Ŋeletlev aaž. = I'm the only one with my name.

The reason for au|o|-o- (CN1) in the first example and ai|-e- (CN2) in the second is that someone who is exclusively liked is both unique in some regard (ŋel) and liked (nam). Someone who is uniquely named is unique in some regard (ŋel), but is NOT, themself, a name (tlev). The ai tlev or -etlev merely gives more information about what kind of uniqueness the referent has.

The CN1 converb comes up a lot with the kind of verb chains that we often use in English.

Hi tsal-o-tuwem aa=ž.
CONT try-CN1-eat DEF(b)=1S
I'm trying to eat.

Adverbial particles (such as hi in the above sentence) can be converted into verbs by appending the suffix -q) and this can be used with CN1 for emphasis.

Bo beg aa=ž.
want(ADV) leave DEF(b)=1S
I want to go/leave.

Bo-q o beg aa=ž.
want-VB CN1 leave DEF(b)=1S
I want to go/leave.

But this leads me to the next topic. The sentences above are all a single clause (with the exception of [Ŋelonam ool] [m'aaž] further up, which is two). Rather than subordinating one verb to another within one clause, it is possible and common to do it across two clauses (provided it is the kind of verb pair where the subject applies to both, as is the case with CN1 constructions but not CN2). For example:

Hi tsal aa=ž tuwem aa=(ž).
CONT try DEF(b)=1S eat DEF(b)=(1S)
I'm trying to eat.
(Literally: I try, I eat.)

Bo-q aa=ž beg aa=(ž).
want-VB DEF(b)=1S leave DEF(b)=(1S)
I want to go.
(Literally: I want, I go.)

The final ž is in brackets because once a referent has been asigned a rank (in this case rank B, as indicated by aa) explicit mention of the referent can be dropped, including 1st and 2nd person pro-verbs.

I've just realised that this extra-clausal structure meets the requirements for some definitions of serial verb constructions. One of the links you posted in the SOV help thread said this:

For
some researchers, serial verb constructions must contain only one inflected verb, while the
others are not or less inflected (Dol 1996, 1999), while for others all verbs must be
inflected (van Staden 2000). Even so, whether all verbs must have the same inflections
(Baker 1989) or not (Durie 1988, 1997) is an issue.

In Balog, each verb (or verb phrase where there is subordination) must be inflected inasmuch as it must at the very least have a subject (with few exceptions), so this fits van Staden's definition. (TAM marking can be marked once on one verb phrase and then understood for the others. There are two types of negation: with the adverbial particle kqa, only the immediate clause is negated. With the verb kq followed by the complementiser ä(ä)=, negation can spread across the whole sentence.) Verbs do not have to have have the same inflections, in line with Durie's definition. The examples above do, but "switch-subject serialisation" is common:

Boq aa=ž, beg oo=l.
want-VB DEF(b)=1S | leave DEF(c)=2S
I want you to leave.
(Literally: I want, you leave.)

Kqal aaž magaz iddauz.
kqal aa=ž, magaz ii=dauz
use.axe DEF(b)=1S | fall.over DEF(e)=be.tree
I chop the tree down with an axe.
(Literally: I use an axe, the tree falls down.)
(Also possible are more specific sentences leaving smaller semantic gaps when necessary: I use an axe, the axe is swung repeatedly, the tree is struck repeatedly, it (the tree) falls down.)

There are, however, several properties that are mentioned almost universally: (i) a
single intonation contour covers the entire construction, (ii) no conjunction can be inserted
between the verbs, and (iii) the entire construction represents a single notional event. The
first two criteria together are taken to mean that the construction is in fact a single clause,
and this single clause analysis is also what almost all linguists agree on.

(i) Yes.
(ii) Depends how you define a conjunction. The equivalent of conjunctions tend to work a bit more like conjunctive adverbs in Balog (I class them with the adverbial particles) and they can be more or less thought of as inflections in the verb phrase. In the Lawarim dialect, these are mostly present as unstressed infixes within the verb itself.) E.g. kavu "but" can be more or less thought of as "however" or "in contrast to the previously mentioned context". In Lawarim, this is -akuv-.
(iii) Yes.

If that means that whole clusters of what I'm calling clauses count as one clause together, then I'll need to find new terminology for them. Or alternatively, I'll just ignore this and continue to describe it as I've been describing it because that's nice and parsimonious.

[bradrn]

Technically speaking, if you have overt dependency markers it can’t be called a serial verb construction (at least in the usual definitions). I’m not sure what this would be, though.

I've realised that these are essentially converbs. They subordinate the following verb to the preceding one.

Yes, that sounds a lot more accurate to me. It also means your CN1/CN2 distinction is basically just switch-reference — CN1 would be the different-subject converb while CN2 would be the same-subject converb.

The sentences above are all a single clause (with the exception of [Ŋelonam ool] [m'aaž] further up, which is two).
…
If that means that whole clusters of what I'm calling clauses count as one clause together, then I'll need to find new terminology for them. Or alternatively, I'll just ignore this and continue to describe it as I've been describing it because that's nice and parsimonious.

Traditionally, subordination is considered a relationship between two clauses, so ‘subordination in a single clause’ would be a contradiction. This all leads me to suspect that the concept of a ‘clause’ is in fact completely inapplicable to Balog, at least without substantially redefining the term.

I've just realised that this extra-clausal structure meets the requirements for some definitions of serial verb constructions.

I’d say that what you have here is a construction which is very SVC-like in some ways, but very different in others — which, incidentally, is often what tends to happen with converbs. (Bisang wrote a chapter comparing the two in ed. Haspelmath and König’s Converbs in Cross-Linguistic Perspective.)

One of the links you posted in the SOV help thread said this:

For
some researchers, serial verb constructions must contain only one inflected verb, while the
others are not or less inflected (Dol 1996, 1999), while for others all verbs must be
inflected (van Staden 2000). Even so, whether all verbs must have the same inflections
(Baker 1989) or not (Durie 1988, 1997) is an issue.

What they’re getting at there is, essentially, that no-one can agree on the criteria for defining SVCs. Personally I consider inflection more or less irrelevant for defining the concept.

There are two types of negation: with the adverbial particle kqa, only the immediate clause is negated. With the verb kq followed by the complementiser ä(ä)=, negation can spread across the whole sentence.

This immediately disqualifies it from SVC-hood, though, since almost all definitions I’ve seen forbid negation of only one verb in an SVC.

(Also possible are more specific sentences leaving smaller semantic gaps when necessary: I use an axe, the axe is swung repeatedly, the tree is struck repeatedly, it (the tree) falls down.)

This reminds me strongly of how things are phrased in Kalam — which incidentally has both SVCs and converbs, if you want to compare the two constructions. Kalam has a small and closed verb class, though, which makes it nearly the opposite of Balog.

There are, however, several properties that are mentioned almost universally: (i) a
single intonation contour covers the entire construction, (ii) no conjunction can be inserted
between the verbs, and (iii) the entire construction represents a single notional event. The
first two criteria together are taken to mean that the construction is in fact a single clause,
and this single clause analysis is also what almost all linguists agree on.

(iii) Yes.

But you just said it’s common to subordinate verbs across two clauses!

[Imralu]

Yes, that sounds a lot more accurate to me. It also means your CN1/CN2 distinction is basically just switch-reference — CN1 would be the different-subject converb while CN2 would be the same-subject converb.

Damn, I didn't think of it that way. Thanks! They're the other way around though. As a mnemonic, the 1 and the 2 refer to how many entities are mentioned. With CN1, there's only one entity. CN2 references two different entities. I suppose I could also gloss them as CNV.SS and CNV.DS, but that's longer.

Traditionally, subordination is considered a relationship between two clauses, so ‘subordination in a single clause’ would be a contradiction. This all leads me to suspect that the concept of a ‘clause’ is in fact completely inapplicable to Balog, at least without substantially redefining the term.

Hmm, sure, but a subordinate clause is within its matrix clause. But yeah, I'm happy to just keep using the term clause to describe the PREDICATE-SUBJECT structure in Balog, regardless of whether someone else might look at it and identify other structures that they call clauses. Describing grammar, like phonology, is often just about finding a way of describing it that words.

For example, I'm calling the content words verbs because they are unmarked in predicates and marked by a nominaliser in subjects, but another way to look at it would be to call them nouns. When appearing in a subject, they are preceded by an article. The lack of an article indicates the predicate and one could posit an underlying, unexpressed copula. The end result is exactly the same structure though, and it's simply easier to call them verbs (for me anyway), although there are follow-on effects. Once I call them verbs, then things like VERB CN1 VERB become descriptions of converbs, whereas if I were calling these nouns, they would just be compounds or analagous structures.

At the end of the day, they're just words. Verbs in one languages are not the same as verbs in another, etc. etc. They're just labels that we apply to classes and the classes can vary quite a bit. I could have gone the route Lojban took and just use the language's own terms for these words, but I'm sure 70% of the reason I don't understand Lojban grammar better is because my eyes glaze over when I start reading bridi, lujvo, sumti, selbri, brivla, gismu, fu'ivla, rafsi. I remember brivla is short for bridi valsi, but what that is .... predicate word? The only one I really remember is gismu because the gismu list is pretty good as a word list for conlanging purposes. If I did this with Balog, I'm sure I'd even confuse myself and use the wrong word all the time. I do that enough as is.

What they’re getting at there is, essentially, that no-one can agree on the criteria for defining SVCs. Personally I consider inflection more or less irrelevant for defining the concept.

Yeah, I got that. Languages literally just do what they do. Things are how they are. We come up with names for phenomena and then also have to define where the boundaries of what is covered by that name are. Some people seem not to get that; as soon as there is a name for something, they assume that it's a natural phenomenon with hard boundaries, as if all we've done is labelled a platonic ideal that has its own edges rather than create a label for a area of land that blurs into its neighbouring areas without any hard boundaries until we decide on them ... and where those boundaries are can be arbitrary as fuck and different according to different authorities or in different contexts.

There are two types of negation: with the adverbial particle kqa, only the immediate clause is negated. With the verb kq followed by the complementiser ä(ä)=, negation can spread across the whole sentence.

This immediately disqualifies it from SVC-hood, though, since almost all definitions I’ve seen forbid negation of only one verb in an SVC.

So "lie roll not-sleep" (for "have insomnia") would not be an SVC, but "lie roll fail sleep" or "lie try sleep fail roll" would be?

(Also possible are more specific sentences leaving smaller semantic gaps when necessary: I use an axe, the axe is swung repeatedly, the tree is struck repeatedly, it (the tree) falls down.)

This reminds me strongly of how things are phrased in Kalam — which incidentally has both SVCs and converbs, if you want to compare the two constructions. Kalam has a small and closed verb class, though, which makes it nearly the opposite of Balog.

Yeah, but if I call the content words nouns instead, tada! Now there are no verbs.

But you just said it’s common to subordinate verbs across two clauses!

Did I? I'm struggling to understand how that applies to Balog. I feel like I probably said something worded slightly differently which I imagine very differently from what I imagine reading those words.

There are verb phrases subordinated inside other verb phrases (which is something that happens entirely inside something that I'm calling a clause), as in the o nam (the VP labelled on line 5) in the following sentence, which is inside the verb phrase ŋel o nam (labelled on line 4).

Ŋel o nam ool maaž.
You're the only one I like. | I only like you.

Code: Select all

1 [SENTENCE________________________________________]
2 [CLAUSE_____________________] [CLAUSE____________]
3 [PRED__________] [SUBJ______] [PRED__] [SUBJ_____]
4 [VP____________] [NOM__] [VP] [VP____] [NOM_] [VP]
5 [VB] [VP_______] |       [VB] [VB____] |      [VB]
6 |    [CN] [VB__] |       |    |        |      |
7 ŋel  o    nam    o:      =l   m        a:=    ž
8 sole CN1  please DEF(c)  =2S  perceive DEF(b) =1S 

Then there are clauses and groups of clauses ("sentences") which are subordinated inside subject phrases by the complementiser ää=, which causes them to take on the role of a verb phrase as the complement of the nominaliser in the subject phrase. (As yet, I don't have a verbal complementiser ... something that allows a complementiser phrase to appear in the predicate, and maybe I won't.) The subordinated sentence, below, labelled SENTENCE2 (on line 4), which consists of two clauses (labelled CLAUSE3 and CLAUSE4 on line 5) is entirely within the subject of CLAUSE2 (on line 2).

Mul aa ŋeb ähhe beg eel das ibbanuun.
He said (that) you should leave his house.

Code: Select all

1  [SENTENCE1_________________________________________________________________________________]
2  [CLAUSE1_______] [CLAUSE2__________________________________________________________________]
3  [PRED___] [SUBJ] [PRED_] [SUBJ_____________________________________________________________]
4  [VP_____] [NOM_] [VP___] [NOM___] [SENTENCE2_______________________________________________]
5  [VB_____] |      [VB___] [COMP__] [CLAUSE3_______________] [CLAUSE4________________________]
6  |         |      |       |        [PRED______] [SUBJ_____] [PRED] [SUBJ____________________]
7  |         |      |       |        [VP________] [NOM_] [VP] [VP__] [NOM_] [VP_______________]
8  |         |      |       |        [ADV] [VB__] |      [VB] [VB__] |      [VB__________] [VP]
9  |         |      |       |        |     |      |      |    |      |      |              [VB]
10 mul       aa     ŋeb     ää=      he    beg    ee=    l    das    ii=    ban  -uu=      n
11 use.mouth DEF(a) be.said DEF.COMP=SUBJ  depart DEF(d)=2S   origin DEF(e)=house-DOM.POSS=3S

I'm struggling to understand how a verb could be subordinated across two clauses, so I think I've just misunderstood what you've said.

[bradrn]

Yes, that sounds a lot more accurate to me. It also means your CN1/CN2 distinction is basically just switch-reference — CN1 would be the different-subject converb while CN2 would be the same-subject converb.

Damn, I didn't think of it that way. Thanks! They're the other way around though. As a mnemonic, the 1 and the 2 refer to how many entities are mentioned. With CN1, there's only one entity. CN2 references two different entities. I suppose I could also gloss them as CNV.SS and CNV.DS, but that's longer.

I prefer just plain SS and DS for glossing.

At the end of the day, they're just words. Verbs in one languages are not the same as verbs in another, etc. etc. They're just labels that we apply to classes and the classes can vary quite a bit. I could have gone the route Lojban took and just use the language's own terms for these words, but I'm sure 70% of the reason I don't understand Lojban grammar better is because my eyes glaze over when I start reading bridi, lujvo, sumti, selbri, brivla, gismu, fu'ivla, rafsi. I remember brivla is short for bridi valsi, but what that is .... predicate word? The only one I really remember is gismu because the gismu list is pretty good as a word list for conlanging purposes. If I did this with Balog, I'm sure I'd even confuse myself and use the wrong word all the time. I do that enough as is.

On the other hand, there’s a reasonable argument that if something differs enough from the prototypical ‘verb’ or ‘noun’, then it’s best to find a different term to avoid confusion. ‘Content word’ seems a better label here.

More generally, I strongly feel it’s important to be accurate in the terminology you use, especially in linguistics where definitions are vague to begin with. Doing otherwise invites confusion: not just when communicating to other people, but also on one’s own part when thinking about things. I can think of times when I’ve gotten myself quite muddled about particular topics, only to suddenly realise they make a lot more sense when you restrict a term or two to have more focussed and coherent meanings.

What they’re getting at there is, essentially, that no-one can agree on the criteria for defining SVCs. Personally I consider inflection more or less irrelevant for defining the concept.

Yeah, I got that. Languages literally just do what they do. Things are how they are. We come up with names for phenomena and then also have to define where the boundaries of what is covered by that name are. Some people seem not to get that; as soon as there is a name for something, they assume that it's a natural phenomenon with hard boundaries, as if all we've done is labelled a platonic ideal that has its own edges rather than create a label for a area of land that blurs into its neighbouring areas without any hard boundaries until we decide on them ... and where those boundaries are can be arbitrary as fuck and different according to different authorities or in different contexts.

Yes, you get it! And this applies to SVCs especially, which blend into all sorts of other categories. (In fact, some recent writers have given up on the term altogether.) I feel there’s still a coherent nucleus of SVCs with shared properties, though.

There are two types of negation: with the adverbial particle kqa, only the immediate clause is negated. With the verb kq followed by the complementiser ä(ä)=, negation can spread across the whole sentence.

This immediately disqualifies it from SVC-hood, though, since almost all definitions I’ve seen forbid negation of only one verb in an SVC.

So "lie roll not-sleep" (for "have insomnia") would not be an SVC, but "lie roll fail sleep" or "lie try sleep fail roll" would be?

That’s not quite right. What I meant was that a particular construction in a language can only be termed a ‘serial verb construction’ proper when clausal negation has scope over the whole construction. (Lexical negation via e.g. an ‘un-’ prefix would be fine, though.)

But you just said it’s common to subordinate verbs across two clauses!

Did I? I'm struggling to understand how that applies to Balog. I feel like I probably said something worded slightly differently which I imagine very differently from what I imagine reading those words.

I was referring to this:

Rather than subordinating one verb to another within one clause, it is possible and common to do it across two clauses …

So almost literally what I said you said — ‘it is … common to [subordinate verbs] across two clauses’.

[Imralu]

I prefer just plain SS and DS for glossing.

Hmm. If I see that though, I probably assume that it’s finite. I guess it’s clear from the context though. I'm not sure it’s classic switch reference though. I came up with them just because I wanted to prevent consonant clusters within clauses, so it was first just a linking element. I couldn’t really find an English word for that and only know Fügen-S in German, so I just decided on “connector”. Then I decided I wanted to mark a distinction there so it’s not just a whole syllable doing nothing at all.

On the other hand, there’s a reasonable argument that if something differs enough from the prototypical ‘verb’ or ‘noun’, then it’s best to find a different term to avoid confusion. ‘Content word’ seems a better label here.

More generally, I strongly feel it’s important to be accurate in the terminology you use, especially in linguistics where definitions are vague to begin with. Doing otherwise invites confusion: not just when communicating to other people, but also on one’s own part when thinking about things. I can think of times when I’ve gotten myself quite muddled about particular topics, only to suddenly realise they make a lot more sense when you restrict a term or two to have more focussed and coherent meanings.

Yeah. Content word is just too long and annoying though. AFAIK, there’s no diagnostic that shows that these aren’t verbs (or nouns).

That’s not quite right. What I meant was that a particular construction in a language can only be termed a ‘serial verb construction’ proper when clausal negation has scope over the whole construction. (Lexical negation via e.g. an ‘un-’ prefix would be fine, though.)

If a prefix like un- is completely productive and can be used with every root, does that disqualify it? I distinguish the negative kqa from the opposite/reversive dze, but both share the same scope of negating only a single verb-phrase. These contrast with the negative verb kq that can negate an entire sentence, depending on where it’s placed.

But you just said it’s common to subordinate verbs across two clauses!

Did I? I'm struggling to understand how that applies to Balog. I feel like I probably said something worded slightly differently which I imagine very differently from what I imagine reading those words.

I was referring to this:

Rather than subordinating one verb to another within one clause, it is possible and common to do it across two clauses …

So almost literally what I said you said — ‘it is … common to [subordinate verbs] across two clauses’.

Omg, LMAO! I get where the confusion is from. By “do it”, I didn’t mean “subordinate verbs”. I just meant do the thing, achieve the same goal, express that. I didn’t even realise that would be interpreted as anaphora. In “I want, I leave”, there’s no subordination (semantically sure, but not syntactically), which is kind of the whole thing I was trying to say.

[bradrn]

I prefer just plain SS and DS for glossing.

Hmm. If I see that though, I probably assume that it’s finite. I guess it’s clear from the context though. I'm not sure it’s classic switch reference though. I came up with them just because I wanted to prevent consonant clusters within clauses, so it was first just a linking element. I couldn’t really find an English word for that and only know Fügen-S in German, so I just decided on “connector”. Then I decided I wanted to mark a distinction there so it’s not just a whole syllable doing nothing at all.

Some languages do have an ‘SVC marker’ which is functionally useless — I seem to recall Khoekhoe, Yimas and Hup here, though at the moment I can’t access my resources to double-check.

AFAIK, there’s no diagnostic that shows that these aren’t verbs (or nouns).

Yes, but only because Balog has no distinction between ‘verbs’ and ‘nouns’ in the first place! I don’t see why you need to assign a specific name to a word class when there’s no need to be that specific.

That’s not quite right. What I meant was that a particular construction in a language can only be termed a ‘serial verb construction’ proper when clausal negation has scope over the whole construction. (Lexical negation via e.g. an ‘un-’ prefix would be fine, though.)

If a prefix like un- is completely productive and can be used with every root, does that disqualify it?

This is interesting… at that point, I’d say you have to use language-internal factors to determine whether that counts as ‘inflection’ or ‘derivation’. But then again, my impression is that languages with morphological negation are also the ones with single-word SVCs.

Did I? I'm struggling to understand how that applies to Balog. I feel like I probably said something worded slightly differently which I imagine very differently from what I imagine reading those words.

I was referring to this:

Rather than subordinating one verb to another within one clause, it is possible and common to do it across two clauses …

So almost literally what I said you said — ‘it is … common to [subordinate verbs] across two clauses’.

Omg, LMAO! I get where the confusion is from. By “do it”, I didn’t mean “subordinate verbs”. I just meant do the thing, achieve the same goal, express that. I didn’t even realise that would be interpreted as anaphora. In “I want, I leave”, there’s no subordination (semantically sure, but not syntactically), which is kind of the whole thing I was trying to say.

Oh, whoops, sorry! I struggle to read that without anaphora.

[Imralu]

Yes, but only because Balog has no distinction between ‘verbs’ and ‘nouns’ in the first place! I don’t see why you need to assign a specific name to a word class when there’s no need to be that specific.

It would be nice not to have to assign names to anything, but to talk about it, it needs a name and "content word" is such an annoying name. And they're words that can be used alone to form a predicate and must be preceded by something (a nominaliser) to be used as a subject, so why not verb?

I might just be an absolute fuckwit and call that part of speech veb in Balog and then just call them vebs

This is interesting… at that point, I’d say you have to use language-internal factors to determine whether that counts as ‘inflection’ or ‘derivation’. But then again, my impression is that languages with morphological negation are also the ones with single-word SVCs.

What's a single word SVC? Isn't that kind of an oxymoron?

Oh, whoops, sorry! I struggle to read that without anaphora.

Yeah, don't be sorry, lol. I think it's just a weird idiolect thing of mine. I can totally see how your interpretation makes sense and I was just like "But people use 'do it' like that all the time ... and by people I mean ... maybe just me."

[bradrn]

Yes, but only because Balog has no distinction between ‘verbs’ and ‘nouns’ in the first place! I don’t see why you need to assign a specific name to a word class when there’s no need to be that specific.

It would be nice not to have to assign names to anything, but to talk about it, it needs a name and "content word" is such an annoying name. And they're words that can be used alone to form a predicate and must be preceded by something (a nominaliser) to be used as a subject, so why not verb?

Hmm, I didn’t realise they required nominalisers… though then again, even the name ‘nominaliser’ presupposes a distinctively ‘nominal’ word class! You could just as well call them ‘articles’ and on that basis call the content words ‘nouns’, which can be used alone in a monovalent ‘copular construction’.

(Also, I didn’t mean that I wanted to ‘not … assign names to anything’; by ‘specific name’ I was referring to words like ‘noun’ and ‘verb’.)

I might just be an absolute fuckwit and call that part of speech veb in Balog and then just call them vebs

This may actually be the best option.

This is interesting… at that point, I’d say you have to use language-internal factors to determine whether that counts as ‘inflection’ or ‘derivation’. But then again, my impression is that languages with morphological negation are also the ones with single-word SVCs.

What's a single word SVC? Isn't that kind of an oxymoron?

No, since you can have multiple roots within a word. You did it yourself with words like ŋelonam and ŋeletlev. Here’s a bigger example from Hup (A Grammar of Hup, Epps 1973:389):

yɨ́t

then

tɨh

3sg

hi-jʼɨ̃p-ʔé-w-ǎn

FACT-tie-PERF-FLR-OBJ

tɨy-hi-jʼap-bʼuy-dʼəh-ye-yɨ́ʔ-ay=mah

push-FACT-snap-throw-send-enter-TEL-INCH=REP

baʔtɨ̌bʼ-ɨ́h

evil.spirit-DECL

Then, to that which she had caused to be tied up (i.e., the door), (he) pushed it until it snapped, threw it out of the way, and entered, they say, (did) the evil spirit!

(Sometimes people call these ‘verb compounds’, though multi-word SVCs occasionally get called that too.)

[Imralu]

Hmm, I didn’t realise they required nominalisers… though then again, even the name ‘nominaliser’ presupposes a distinctively ‘nominal’ word class!

Hmm, yeah, I just called them that because subjects are an archetypally nominal role, just as the predicate is an archetypally verbal role. In languages that have the distinction (i.e. basically every natlang), anything that appears in the subject has to be nominalised (whether that means turning a verb into a gerund or agent noun or a clause into a complement clause - even when this is done through zero-derivation) and even though it's incredibly common to allow parts of speech other than verbs to head the predicate, for some stupid reason, in typology, people uses the labels S, O and V rather than S, O and P.

Of course, the word predicate has its issues because objects are embedded inside the predicate (depending on how predicate is defined), so "verb" means "predicate minus arguments and adjuncts", and generally, if a language has an object in the sentence, the predicate is headed by a verb and, when assigning the typological word order to a language, they're generally explicitly not looking at sentences with a subject plus a predicate adjective or predicate noun. (And I guess there are also possibly languages that have a different word order for subject + non-verbal predicate vs subject + verbal predicate.)

So yeah, it annoys me when people mix up these two categories of things and mix and match words from each:

Code: Select all

Lexical categories:
- noun
- verb
- adjective
- ... 

Code: Select all

Syntactic roles:
- predicate
- argument (subject, object OR agent, patient etc.)
- adjunct

... but it's a very well trodden path, and I do it myself because (a) I'm a hypocrite and (b) when using preexisting words, it's hard not to have the way you use them shaped by the way they are used. It's much clearer if I say clauses consist of "predicate then subject" rather than "verb then subject" and if I don't refer to the subject phrase in any way that indicates it's a noun, but essentially what I've been doing is using the existing words verb and noun to create an interpretation of the grammar. There are two obvious interpretations:

Code: Select all

                       Verbal interpretation             Nominal interpretation
magaz fall(er).over    = verb                            = noun (bare noun indicates underlying copula)

i=    DEF(RANK:e)      = subject marker/nominaliser      = subject marker/article
                         /pronoun (heads rel. clause)      (indicates no underlying copula)

dauz  (be).tree        = verb                            = noun
________________________________________________________________________________________________________
magaz                    falls over                        (is a) faller over
i=                       the one that                      the
dauz                     is a tree                         tree

Both arrive at exactly the same result. Neither is perfect, but I know of no diagnostic test that rules either out because the definitions of "verb" and "noun" are defined within each language and cross-linguistically have very fuzzy, generalised definitions. This kind of thing is routine, for example in phonology, where the choice between regarding something as a cluster or a phoneme on its own may simply come down to which interpretation makes for the most compact description of phonotactics. The words we use are just a framework to describe the phenomenon. Descriptions can differ enormously, but still accurately describe the same thing.

You could just as well call them ‘articles’ and on that basis call the content words ‘nouns’, which can be used alone in a monovalent ‘copular construction’.

Yeah, I said exactly that a couple of posts ago.

For example, I'm calling the content words verbs because they are unmarked in predicates and marked by a nominaliser in subjects, but another way to look at it would be to call them nouns. When appearing in a subject, they are preceded by an article. The lack of an article indicates the predicate and one could posit an underlying, unexpressed copula. The end result is exactly the same structure though, and it's simply easier to call them verbs (for me anyway), although there are follow-on effects. Once I call them verbs, then things like VERB CN1 VERB become descriptions of converbs, whereas if I were calling these nouns, they would just be compounds or analagous structures.

I might just be an absolute fuckwit and call that part of speech veb in Balog and then just call them vebs

This may actually be the best option.

I've tentatively called that part of speech vun.

The problem is, that the two interpretations I mentioned above provide a framework to name other structures, and once I start using the word vun rather than verb, then the whole thing we've been talking about with converbs with ai/-e- and o/-o- needs a new name and something else needs a new name, and I'll either have to end up talking about gismu rafsiing the fu'ivla brivlaically (and lose everyone reading and be accused of thinking my language is a special snowflake that can't be constrained to English words) or I'll have to create additional English-based terminology or use horrendously clunky phrases like "content word" (and lose everyone reading and be accused of thinking my language is a special snowflake that can't be constrained to existing English words).

I can't tell you the number of times I've started talking about one of my conlangs and mentioned there being no lexical distinction between verbs and nouns and people going "ThAt'S iMpOsSiBlE" and others assuming going "Hey, me too!" and when they start talking about their language, they're talking about zero-derivation. The first giveaway to me is often when there's a semantically verb-like word appearing in an argument and suddenly it has the meaning of a gerund instead of an agent noun. That allows you to quickly go through the lexicon and sort out the class of verbs. That is routine in natlangs with extensive zero derivation, but that's not the lack of a noun/verb distinction. I don't remember who it was, but someone a few years ago was writing in other people's languages in the fluency thread and when they wrote in mine, their errors were mostly those kind of assumptions.

Sticking to one of the interpretations I outlined above, it's much easier to make it clear that there are no abrupt, irregular changes of meaning of some lexemes depending on their position in the sentence. Either describing the CWs as verbs that are relativised to be used in the subject, or as nouns with an underlying, unexpressed copula in the predicate, makes it clear what the meaning of any CW will be in the other sentence position and eliminates the room for assumptions about verby words taking on a gerund meaning etc. To me, a description of a language that succinctly makes it clear how it's used is superior to one that doesn't make it as clear or requires more extensive explanations to make it clear.

This is interesting… at that point, I’d say you have to use language-internal factors to determine whether that counts as ‘inflection’ or ‘derivation’. But then again, my impression is that languages with morphological negation are also the ones with single-word SVCs.

What's a single word SVC? Isn't that kind of an oxymoron?

No, since you can have multiple roots within a word. You did it yourself with words like ŋelonam and ŋeletlev. Here’s a bigger example from Hup (A Grammar of Hup, Epps 1973:389):

yɨ́t

then

tɨh

3sg

hi-jʼɨ̃p-ʔé-w-ǎn

FACT-tie-PERF-FLR-OBJ

tɨy-hi-jʼap-bʼuy-dʼəh-ye-yɨ́ʔ-ay=mah

push-FACT-snap-throw-send-enter-TEL-INCH=REP

baʔtɨ̌bʼ-ɨ́h

evil.spirit-DECL

Then, to that which she had caused to be tied up (i.e., the door), (he) pushed it until it snapped, threw it out of the way, and entered, they say, (did) the evil spirit!

(Sometimes people call these ‘verb compounds’, though multi-word SVCs occasionally get called that too.)

Oh lol, Yeah, I forgot about those.