Hlʉ̂
This is a new language I’ve been working on recently. Key features include a tiny (yet interestingly tonal) phonology, as well as an alignment system which isn’t quite ‘evilly twisted’ so much as ‘slightly bent’. I haven’t developed too much of it yet, so this will likely end up as a more traditional scratchpad series.
Phonology
There are 8 consonants, as follows:
/m n/
/p t ʔ/
/b d h/
Though a phonetic transcription may be more revealing:
[m n] ⟨m n⟩
[p t~k ʔ] ⟨p t~k ʔ⟩
[b d~ɡ~l~l̥ s~h] ⟨b d~g~l~l s~h⟩
/t d/ are [k ɡ] before /d/, while /d/ becomes [l~l̥] after another consonant. /h/ is [s] before /i e/, [h] otherwise. (Also /b/ may end up with an allophone [β] intervocalically, though I’m not sure.)
There are 8 vowels:
/i ʉ u/ ⟨i ʉ u⟩
/e ɵ o/ ⟨e ɵ o⟩
/a/ ⟨a⟩
These can take on five surface tones/registers: /à á â ǎ a̤/ ⟨a á â ǎ ah⟩. Also creaky voice /a̰/ ⟨aʔ⟩, though its status is ambiguous; see below. Underlyingly, some syllables (mostly clitics) are toneless; these syllables get a low tone if they surface. When it matters, I’ll transcribe those as ⟨ȧ⟩.
Maximal syllable structure is
(C)(l)V(C) [
EDIT: whoops, that should be (C)(d)V(C)], where the final consonant cannot be /h/, and an initial cluster of /nd/ is forbidden. Vowel hiatus is forbidden within the phonological word.
Creaky voice in Hlʉ̂ poses one of those interesting problems of phonological analysis you run into now and again. You could analyse it as a tone, since it doesn’t co-occur with any low tone, and doesn’t interrupt vowel hiatus (so e.g. *[plɵ̰.e] is forbidden). But you could
also analyse it as a final /ʔ/, since it doesn’t co-occur with coda consonants and it doesn’t participate in tonal processes. I’ll use the latter analysis here.
Tonology
This is where it starts to get interesting. Hlʉ̂ tones are best analysed autosegmentally (I highly recommend reading
Tengado’s introduction if you haven’t already). Underlyingly, there are three tone autosegments: H(igh), L(ow) and B(reathy). Each syllable is associated with zero or more tone autosegments (though not all tones need be associated with a syllable). Valid surface tones are the following:
Code: Select all
σ σ σ σ σ σʔ
| | | / \ / \
L H B L H H L
To resolve marked tone sequences, the following tonal processes occur to convert underlying tones into surface tones:
- Contour relinking — the second (and subsequent) tone(s) attached to a syllable move to next syllable in the clitic group, if there is one and if that won’t form an invalid contour
- Tone defloating — a floating tone attaches to the next syllable
- Tone merging — two identical tones merge
- Contour legalisation — the second (and subsequent) tone(s) of an invalid tone contour detach
These rules are applied opportunistically from left to right, e.g.:
Code: Select all
dab taʔ hlʉ=e nɵ ══[TD]══> dab taʔ hlʉ=e nɵ ══[CL]══> dab taʔ hlʉ=e nɵ ══[TD]══> dab taʔ hlʉ=e nɵ ══[CR]══> dab taʔ hlʉ=e nɵ
| / \ | | | / \ | | / \ | | /|\ | | | |\ \
H H L H L H H L H L H H L H L H H L H L H H L H L
Or, as a glossed sentence:
Dáb taʔ hlʉ́ ě nɵ.
[dáb ta̰ hl̥ʉ́ ě nɵ̀]
//dáb ˊ taʔ h⟨dɵ̀⟩í=e nɵ̀//
1 LNK cow ⟨PL⟩go=3 DECL
My cows go.
A somewhat contrived sentence, admittedly, but it shows off the tonal processes well enough.
(I’m not 100% happy with this system; I’d like to do a rigorous Optimality-Theoretic analysis of it at some point, which should be easy enough once I get around to properly learning OT.)
Phonological processes
The following phonological processes convert underlying forms to phonemic representations:
- The forbidden contour //nd// is changed to /d/ when it occurs.
- Vowel hiatus resolution: ɵ {a,e,i,ɵ,ʉ,o,u} → {a,ɵ,ʉ,ɵ,ʉ,o,u} and ʉ {a,e,i,ɵ,ʉ,o,u} → {ɵ,ʉ,ʉ,ʉ,ʉ,u,u}. (Other vowel sequences never occur.)
These apply before tonal processes: e.g. //dáb ˊ taʔ h⟨dɵ̀⟩í=e// → /dáb ˊ taʔ hdʉ̌=e/ → /dáb taʔ hdʉ́ ě/ → [dáb ta̰ hlʉ́ ě].
Basic syntax
Basic clausal word order is topic-prominent and essentially verb-final, as follows:
topic S O [verb complex]=S PCL
Though topic, S and O are optional due to widespread ellipsis:
Deb pa di á nɵ; plɵʔe a di?
//dèb pà dí=a nɵ̀ | plɵ̀ʔè=a dì//
tree TOP see=1/2 DECL; like=1/2 Q?
(I) see a tree; do (you) like (it)?
The noun phrase
I haven’t totally figured out noun phrases yet, but I’m leaning towards the following linear structure:
possessor/relative LNK noun adjectives numeral demonstrative classifier
Possessors and relative clauses (which are mutually exclusive) go before the noun. To use
WALS’s terminology, Hlʉ̂ genitives/adjectives/relative clauses are ‘moderately differentiated, with genitives and relative clauses collapsed’:
dáb ʔlí ‘my dog’ vs
tʉ́p ʔlí ‘sitting dog’. The genitive/relative clause is joined to the noun via the ‘linker’, a floating high tone. (So
ʔlí in the previous sentence is really //ˊ=ʔdi̤//; the breathy voice has been pushed off and is now floating after the noun, hence
tʉ́p ʔlí dih á nɵ ‘I see the sitting dog’ vs
ʔlih dí a nɵ ‘I see the dog’.)
The noun itself is pretty boring. As you may have noticed, Hlʉ̂ is pretty isolating; the noun is thus non-inflected. However, this is probably a good time to present the pronoun inventory. The personal pronouns are pretty minimal, lacking number distinctions and a third person (though distinguishing possessive forms):
| Person | Personal pronoun | Possessive pronoun |
| 1 | ná | dáb |
| 2 | ba | bab |
For third person pronouns, use the demonstratives: ‘this/here’
bap and ‘that/there’
dahn.
The rest of the noun phrase is pretty much as outlined above: adjectives immediately following the noun, followed by numerals and demonstratives. (No numerals yet, though… sorry, Janko.) Numerals and demonstratives require the use of a classifier. The classifier system is likely to be very interesting, though I haven’t worked it out yet. A postposition can occur after the noun phrase;
beh is optional after S/A (see below), while
pa obligatorily marks topics.
The verb complex
The verb complex is even less well-developed than the noun phrase. It seems likely to contain a verb, followed by some auxiliaries. After the auxiliaries comes the negative particle
plé.
Following the verb complex are the subject-agreement clitics, which make a minimal distinction of SAP vs non-SAP:
| Person(s) | Clitic |
| 1/2 | =ȧ |
| 3 | =ė |
(Not that this is set in stone; at some point I might change this to conjunct/disjunct marking or something more interesting.)
Finally, at the end of the sentence is an obligatory ‘sentence-final particle’, mostly evidentials (and some other modals):
| Meaning | Particle |
| Declarative | nɵ |
| Visual | ehb |
| Hearsay | bap |
| Inference | dá |
| Assumption | hleʔ |
| Question | di |
Alignment
If you know me, you should be entirely unsurprised to learn that morphosyntactic alignment is the best-worked-out part of the grammar so far (not that that’s saying much). Let’s get started.
Noun phrases are marked using a vaguely ergative pattern, with the A argument getting the ergative postposition
beh:
Taʔ dahn blep beh ná dí a nɵ.
//taʔ da̤n blèp be̤ ná dí=a nɵ̀//
cow that CL ERG 1 see=1/2 DECL
That cow sees me.
Of course, there are many subtleties here. Most obviously, the ergative case-marker is optional. It tends to get used most often with inanimate nouns, but is rarely attested with pronouns. It’s obligatory when A is focused, as well as when O is topicalised; on the other hand, it is forbidden with topicalised A.
But
beh isn’t truly ‘ergative’: it can mark some S arguments as well. It’s used only with agentive verbs, most commonly when S has contrastive focus; on the other hand, it’s forbidden with pronominal S.
Now on to verbs! Verbs agree with their ‘subject’ and ‘object’ (using the terms loosely). I’ve already mentioned the pronominal clitics
=ȧ and
=ė; these agree in person with the A argument. More interesting, however, is the object inflection, where verbs agree with O in number. This alternation seems to be irregular, almost suppletive, a combination of prefixation, infixation, consonant and vowel mutation and tone shift:
[EDIT: whoops, forgot to account for vowel mutation in some of these]
(SG → PL)
na →
da dɵ
ma →
mla mɵ
dába →
glabâ glɵbâ
dí →
glʉ̌
dí →
dɵdí
But in fact, this is entirely regular, with plural inflection regularly being an infix ⟨dɵ⟩ after the first consonant. Phonological processes, allophony and variant underlying representations account for the rest:
//nà// → //n⟨dɵ̀⟩à// →
/dà/ → [dà] /dɵ̀/ → [dɵ̀]
//mà// → //m⟨dɵ̀⟩à// →
/mdà/ → [mlà] /mdɵ̀/ → [mlɵ̀]
//dábà// → //d⟨dɵ̀⟩ábà// →
/ddàbâ/ → [ɡlàbâ] /ddɵ̀bâ/ → [ɡlɵ̀bâ]
//dí// → //d⟨dɵ̀⟩í// → /ddʉ̌/ → [ɡlʉ̌]
//ndí// → //n⟨dɵ̀⟩dí// → /dɵ̀dí/ → [dɵ̀dí]
All the above is for transitive verbs. Intransitive verbs, however, have the same inflections — only now, both markers agree with S! So clitics agree with S/A, and the infix agrees with S/O, with the agreement ‘doubling up’ for S arguments.
Exercise: Here’s a little exercise, to see how many of you were following along. (Also, to see how good my explanation was.) Translate the following sentences:
Ná bab taʔ klɵ́ glʉ á nɵ. Klɵ é ehb. Dáb ʔlí dih é di? If you’re feeling really confident, write down a gloss and the underlying representation. (New words:
klɵ = black, adjective;
té = ‘sleep’, intransitive verb.)
EDIT: Some corrections to sentences in exercise
I don't know if you based some of the "bent" bits on a natlang, but this definitely feels natural. The interaction with topicality is a really nice touch, in particular.
I've looked at Tibetan & Basque, but I haven't gotten so deep that I saw any behavior like that. Neat.