Well, it's been freaking YEARS since I've posted anything on a conlang I've worked on, so here goes. Tl'acho is a vaguely North American language, with bits taken from Penutian, Algonquian, Muskogean, and others. I've been working on a grammatical sketch forever, but I think it might be more interesting and perhaps clearer if I introduced Tl'acho through examples. So, here goes (NB: I'll probably fuck up the glossing tags at first. Bear with!):

/d-3.Sg.O(e)nitl=sleepane/near.visual
[de.nee.tla.ne]
"S/he is sleeping (I see it)"

Morphosyntax:
- 'sleep' /(e)nitl/ is a stative root, so it takes a single, O-marked agreement (in this case 3rd singular /d-/). Tl'acho is Split-S (but with some degree of fluidity), and distinguishes O-marking (stative subjects, active objects) prefixes from A-marking (active subjects) suffixes.
- the 'near.visual' /ane/ is a tense/evidentiality suffix (i.e., a suffix that marks both tense and evidentiality), which is obligatory on verbs. The "near" tense can be recent past, recent future, or present; with a stative root, the default reading is present imperfective. "visual" evidentiality contrasts with "auditory", "other senses" and "reported" evidentiality for near tense.

Morphophonology: /d-(e)nitl=ane/ -> [de.nee.tla.ne]
- The parenthesized (e) vowel in /(e)nitl/ only appears when needed to satisfy disyllabic minimality on the Stem. Since the only other Stem morpheme is /d-/, the (e) vowel appears.
- Tl'acho parses the Stem into iambic feet, and always lengthens stressed vowels in open syllables. Lengthened /i/ gets regularly lowered to [e:]

/d-3.Sg.O(e)nitl-sleepha=causativeno-2.Sg.Aet/near.reported
[dineetlaano'te]
"You put her/him to bed (so you said)"

Morphosyntax:
- The causative /ha=/ adds an A-marked argument to the stative root 'sleep' /(e)nitl/. A-marking suffixes (here, the 2nd singular /-no-/) are outside the Stem.
- The causative also makes the Stem telic, so that the "near" tense is a recent past perfective.

Morphophonology: /d-(e)nitl-ha=no-et/ -> [dineetlaano'te]
- Because the causative suffix /-ha=/ is within the Stem, the (e) vowel of the root does not surface. Instead, the front vowel [ i ] is inserted to break up the illegal /d-n/ cluster.
- The /tl-h/ cluster is also illegal, but is resolved by coalescence, not vowel epenthesis. A coalesced /h/ typically aspirates an obstruent, but /tl/ is already aspirated.
- The /o-e/ hiatus is fixed with glottal stop epenthesis (/no-et/ -> [no'et])
- This word is in isolation, so it counts as a phrase. Obstruents are illegal in phrase-final position, so a vowel is epenthesized after /t/. Because /t/ is alveolar, the epenthetic vowel is [e] (/-et/ -> [ete])
- These two changes create the environment for syncope of the underlying /e/ vowel: /no-et/ -> [no'ete] -> [no'te].

Here's where it gets even interestinger!
/wón-bedd-3.Sg.O(e)nitl-sleepha=causativela-potentialn,non.pastgi1.Sg.As/near.direct
wúúnineetlaalangise
"I might be sleeping in my bed soon (I know this)" OR "I might put someone in my bed soon"

Morphosyntax:
- The incorporated object /wón/ 'bed' messes with the syntax and creates an ambiguity here. The bed in this case is definite (interpreted as "my bed" because of the 1st singular subject), so it requires O-marking agreement. In order to express the subject of 'sleep', the causative /ha=/ is needed to add an another argument. This causative may be here only to allow A-marking agreement, and be semantically vacuous; this is the "I might be sleeping in my bed" reading." Alternatively, the causative may be "real", and the causee is left out due to syntactic reasons (only two arguments can be marked).
- The combination of the potential (aspect/mood), the non.past (mood/tense), and the near.direct (tense/evidentiality) give the meaning "might do something soon". The non.past plus near tense plus telic Stem gives the recent future reading "will do soon", while the potential gives the existential modal (here with an epistemic reading). Direct evidentiality is used when an event is neither reported nor directly sensed through vision or hearing, in this case since it's a plan in the speaker's head.

Morphophonology: /wón-d-(e)nitl-ha=la-n,gi-s/ ->  wúúnineetlaalangise
- The acute accent in /wón/ is high tone. Since /wón/ is a prefix, the tone is fixed in position, and surfaces on the stressed vowel of the foot that the tonal vowel is in (here, it's the same vowel /o/).
- The /n-d/ cluster coalesces to [n], and also nasalizes the preceding vowel to [ u ] (the sole nasal vowel in Tl'acho).
- The relative position of the suffixes /n/ 'non.past' and /gi/ '1.Sg.A' is determined by sonority and phonotactics. With a different selection of mood/tense and A-agreement suffixes, you could get the opposite order.

Ok, enjoy this. I will hopefully post more in the coming days ... weeks ... months.

Interesting. I especially like the detailed explanations. Are you  continuing this?

in a word: YES. caveat: my dissertation needs to be finished in one month. so maybe in just over a month?

NEW WORD: [gich'aatse] "hello"

I know you don't have time, but words are more fun with glosses or etymologies. ;-)

i made this one up on the spot for a tv interview. therefore no gloss or etymology... yet. i'll back-form one, i promise!

quoting kodé, Deacon, this fucking hole we call LA:

for a tv interview

Tell us more!

well ... so i made an appearance on Jeopardy! (the famed US game show with the erstwhile mustachioed canadian host) last week, and alex trebek asked about my hobby of language creation, so i told him about tl'acho, and made up the word [gich'aatse]. the episode airs june 16th, though i think it's only aired in the US and maybe Canada.

Great! Did you win anything? (I've heard about the show, but never watched the original version. In my experience, game shows tend to be copied or franchised with local hosts and candidates, but not to be aired with the original hosts and candidates outside their home markets. There seems to be a German version, I think I came across it once or twice when channel-hopping)

i can't disclose any winnings, but i *can* say i had a whole bunch of fun! i think you're right about game shows outside of home markets, so you'll have to wait for a pirated version. not that i'm providing one, since i don't want to jeopardize any winnings, if there be any

jeopardize any winnings
But that would be very fitting. ;-)

When does it air?

june 16th, on your local ABC affiliate, almost certainly at 7 pm.

I've put it on my calendar. I expect to be dazzled.

Here's a couple more examples. Despite the intriguingly named "ram-ship", I really haven't done anything interesting on the conworlding side. But hopefully that'll change, soon enough...

/kalh-ram-ship dzo=make e^,past gi-1Sg.A s/near.direct
[kaatsóódegse]
"I built a ram-ship"

Morphosyntax:
- /kalh/ "ram-ship" is an incorporated object; with /kalh/, the verb root /dzo/ "make" takes on the more specific meaning "build". With different objects, /dzo/ could mean "cook", "sew", or something else; its root meaning is something along the lines of "make a physical artifact". Incorporating a noun with /dzo/ is the default means of talking about making something; there are more specific roots for some processes of making, like /ch'aang/ for "build a reed-house" (see, I'm getting to the conworlding!).
- the post-Stem suffixes are pretty boring in this example: "past" + "1st.Sg.Agent" + "near tense/direct evidential" = "I did it (relatively recently)". I'm not sure when one could use 1st person subject marking without a direct evidential; maybe being under a spell or a god's influence, or doing something in a dream, or while intoxicated from /shvxwe/, a fermented barley drink?

Morphophonology:
- the cluster /lh-dz/ simplifies to [ts] with compensatory lengthening of the previous vowel. This is basically what happens when a coronal fricative meets another coronal fricative or affricate: the second fricative is preserved with its place of articulation but the glottal state (voicing, aspiration or ejectivization) of the first fricative. The first fricative deletes but the weight of the syllable is preserved through vowel lengthening. Typically, the second fricative is affricativized (if it's not already an affricate).
- the rule of phrase-final vowel insertion after an obstruent feeds syncope of the preceding vowel /i/. Thus, /e-gi-s/ -> [egise] -> [egse].
- the tone of the post-Stem suffix /e^/ "past" gets realized on the last stressed vowel of the Stem, here [óó].

/kalh-ram-ship (o)gongw=sail e^,past gi-1Sg.A ta^ remote.direct w-1Sg.P (o)gongw-sail k=adversative.reflexive o^l-terminative e^-past ta^/remote.direct
[kalhgóóngwektá wogúúgóltá]
"I sailed a ram ship but it crashed (it happened long ago)"

Morphosyntax:
- this is an example of a paratactic construction, where two fully inflected verbs are placed one after another without any subordination. The reading here is that the first verb (really a clause, but there's no other (overt) words) happens first, and the second verb second. The sequential reading is pretty much the only one available here, since both verbs have identical tense marking. I'll have to see what happens when, say, the first verb has future or near tense marking, but the second verb has a more past reading.
- the verb root /(o)gongw/ is an activity, but with the adversative reflexive /-k/ suffix, the subject of the verb gets demoted from agent to patient. This means any previous theme argument must now be marked obliquely; in this case, no theme incorporation is possible in the second verb (cf. the first verb).
- the terminative suffix /-o^l/ indicates that the action stopped, though there is no necessary completion implicature. The combination of the terminative and the adversative reflexive yields an "it ended badly" reading, which with the root "sail" implies a crash.

Morphophonology:
- this example is a fun illustration of the changes that happen to underlying forms: /(o)gongw/ is realized as [góóngw] in the first verb but [ogúú] in the second verb. In the first verb, there are no problematic consonant clusters, so the root only undergoes vowel lengthening to form a stress foot; because the Stem is already disyllabic, the optional vowel (o) does not surface. In the second verb, the Stem is not disyllabic without the /(o)/ vowel surfacing, so it does. The end of the root gets badly mangled: the coda /ngw/ gets dropped, but nasalizes and lengthens the preceding vowel and voices the following obstruent, so /ongw-k/ -> [uug] (in X-Sampa this would be /ON_wk/ -> [V~:g]).
- the example also shows what happens when you get multiple post-Stem suffixes with tone. The first tone is always attracted to the nearest foot, which is the Stem-final vowel. Any following tones stay on their suffixes, like /ta^/ -> [tá]; however, syncope of the vowel /e^/ of the "past" suffix forces its tone onto the nearest preceding vowel. By the way, every successive high tone is downstepped, so the three tones in [wogúúgóltá] go from highest to less high to somewhat low.

Whoo! Let's do more of this.

Inventory dump: http://anthologi.ca/?id=260833

Hopefully this'll make translating easier.