Nashalq (conlang for hyenafolk)

[aporaporimos]

Context

Nashalq is the language of the Shikazi, a race of nocturnal hyena-like creatures who live in tribes as hunter-scavengers on the outskirts of human civilization. (This setting is not fleshed out at all, but assume the humans are on the technological level of late antiquity.) The Shikazi and humans mutually fear and distrust each other but rarely come into violent conflict.

Notable features

My idea when I started making this language was to have it reflect animal cognition. At first I thought it would only be a partial system of communication limited in what it could express; but it's gotten more complex as I've worked on it and this point it's clearly becoming a "Turing-complete" language capable of detailed and precise expression. However, it still has two prominent features based on the non-human psychology of its speakers:

1. Imperfect recognition of other minds existing equally but separately and independently of the self; and
2. Emphasis on expressing direct perceptual and emotional experiences, rather than events or ideas considered abstractly.

These are reflected mainly in two corresponding grammatical features:

1. Rather than a 3-way distinction between 1st, 2nd, and 3rd person, there is only a two-way distinction, between Ego (prototypically 1st person) and Alter (prototypically 3rd person). Depending on the context, the speaker may include the listener within Ego, or within Alter; the former is typical in friendly, cooperative conversation, the latter in hostile interchanges. (In glosses I use "one" for Ego and "it" for Alter; in translation into English I use whatever pronouns are contextually appropriate, most often 1st person for Ego and 3rd person for Alter.)
2. For a large number of common words referring to perception, emotion, and the like, the semantic role of the experiencer is implicitly and immutably assigned to Ego. For example, the normal word for seeing is n'tuh. It is intransitive, and its subject is the thing seen. The sentence tfig n'tuh may mean "I see a lizard," "we see a lizard," or even "you see a lizard," depending on context, but there is simply no way to use this word to say "he sees a lizard."

Phonology

Nothing about the phonology is intended to be unhumanlike; as far as I know it's a plausible phonology for a natural human language. For now I'll just talk about the structure of roots; the process that occur in morphologically complex words can be put off till later.

Vowels

Plain	a	i	u	əi̯ <ai>	əu̯ <ou>
Mid	ɤ <ǎ>	e	o	əe̯ <ae>	əo̯ <oo>

A single root contains either only plain vowels or only mid vowels, according to rules that will be explained below. (In fact the plain and mid variants of the diphthongs are probably too similar to distinguish audibly.) Accordingly, in normal transcription I use <a> for both /a/ and /ɤ/; it will not cause ambiguity outside of corner cases.

Consonants

	Labial	Alveolar	Palatal	Velar	Uvular
Plosive/Affr.	p	t	tʃ <ch>	k	q
Fricative	(ɸ <f>)			x <h>	χ <x>
Sibilant		s	ʃ <sh>		(sʁ <š>)
Nasal	m	(n)		ŋ <g>	(ɴ <ň>)
Approx.	w	l	j <y>		ʁ <r>

Where a transcription isn't given, it's the same as the IPA. The digraph <sh> will cause no ambiguity, as /s/ only occurs before vowels.

There are a few things that need explaining here:

1. <f> is an allophone of <h>. Either realization is possible in most positions, but most often <f> is used in the onset of a syllable and <h> in the coda, and this is the rule I follow in transcription. I spell them differently for aesthetic reasons (and to remind readers that both realizations are common).
2. The contrast between <n> and <l> is very limited; strictly speaking [n] and [l] never contrast within a word. At the beginning of a word only [n] occurs (and I write <n>), and at the end of a word, or before another consonant, only [l] occurs (and I write <l>); in other contexts there is free variation (and I write <l>). The clusters [ɲtʃ ŋk ɴq] are written <nch nk nq>; the first of these is difficult to analyze as anything other than /ntʃ/, and contrasts with <lch>.
3. <ň š> don't contrast with <g s>; the former appear next to mid vowels, the latter in other positions. Accordingly I spell both sets as <g s> in normal transcription. But <ň š> are phonologically relevant allophones and it's useful to have a transcription for them.
4. <s> is historically a reflex of <k> before <i>. However, in the modern language there is sometimes a contrast between them: sit means "short," -kit is a suffix form of kait meaning "track."

A few other notable cases of allophony:

1. The nasals <m g ň> are sometimes realized as voiced oral stops: optionally in between vowels or at the end of a word, and obligatorily in certain clusters. (Note that, while I spell /ŋ/ as <g> out of typographical convenience, it's not any more prone to being realized as a stop than <m> is.)
2. On the other hand, the plosives (and affricate) <p t ch k q> are sometimes voiced, especially when adjacent to a nasal. <t ch> are usually voiced between vowels.
3. <y> can be realized as a nasal [ɲ]. Naturally, this happens most frequently next to another nasal.
4. <r> varies a bit; the typical realization, as shown, is a uvular approximant or voiced fricative [ʁ], but it may be trilled [ʀ], and some speakers substitute a pharyngeal [ʕ].

Uvular assimilation and mid vowel harmony
As already noted, the vowels and diphthongs are grouped into pairs of plain and mid vowels. The velar and uvular consonants can be similarly paired up:

K	k	g	f/h	s
Q	q	ň	x	š

I'll call these K-consonants and Q-consonants, since they don't line up perfectly with phonetic categories (<s> participates in these processes but is not velar, while <r> is uvular but does not participate in them at all). There are four rules concerning these classes of consonants and vowels:

1. A K-consonant cannot be adjacent to a Q-consonant.
2. A K-consonant cannot be adjacent to a mid vowel.
3. A plain vowel cannot be adjacent to a Q-consonant.
4. A single root cannot contain both plain vowels and mid vowels.

Violations of these rules are repaired by assimilating K-consonants to Q-consonants and plain vowels to mid vowels; this will be more important when we start forming compound words.

Root structure
Roots are subject to a few constraints on their form. They can have one syllable or two syllables; two-syllable roots always end in a vowel. A root can contain at most three consonants; a side effect of this is that a root can contain at most one consonant cluster. Consecutive vowels are forbidden. Diphthongs only occur in monosyllabic roots, and are never followed by a cluster. Additionally, each root is either a plain root, which uses the plain vowels <a i u ai ou> and may not contain <q> or <x>; or a uvular root, which use the mid vowels <a e o ae oo>, must contain at least one occurrence of <q> or <x>, and may not contain <s>.

Edit: one more constraint: <w> and <y> can't occur in a syllable coda.
Edit2: <s> occurs only before <i>, whereas <k> is forbidden in that context. As a result, <k> and <s> do not contrast in roots. (The cluster <pk> is an exception--more on that later.)

Here are some examples of roots: ni, xa, uk, aih, ooq, rat, xeg, sit, kait, urk, wulp, ugu, piku, qoyo, shri, kta, qro, txash, pkil, tkuh, tkala, ishmu, yuglu, eqshe, ishpla.

OK, I think this post has gone on long enough; next time I'll talk about what consonant clusters are permitted in roots, and then we can get into syntax.

[Pabappa]

Very nice. I cant remember the last time I saw a language designed for speakers with a non-human psychology. I hope to see more soon.

[aporaporimos]

Very nice. I cant remember the last time I saw a language designed for speakers with a non-human psychology. I hope to see more soon.

Thanks!

[aporaporimos]

Phonology (cont.)

Initial clusters
The initial clusters all can easily be classified and enumerated; here, have a table:

p

t

ch

k

s

q

m

g

f/h

x

sh

y

shC

shp

sht

shk

shq

shm

shg

shf

shx

shy

Csh

psh

ksh

qsh

msh

gsh

fsh

xsh

tC

tp

tk

ts

tq

tm

tg

tf

tx

chy

Ct

pt

kt

qt

mt

gt

ft

xt

Cr

pr

tr

chr

kr

qr

mr

gr

fr

xr

shr

Cl

pl

tl

chl

kl

ql

ml

gl

fl

xl

shl

Cw

**

tw

chw

kw

qw

**

gw

hw

xw

shw

Notes:

1. In place of *<mw>, there is a cluster /mŋ/. Sometimes the <w> reappears in derived forms. I write /mŋ/ as <mn> because *<mg> is ugly and I hate logic. Similarly, in place of *<pw>, there is <pk>, or <ph> if <k> occurs later in the word, or <pq> if a mid vowel follows, or <px> if a mid vowel follows and <q> occurs later in the word. Thus we have roots like pkil, phika, pqex, and pxoqe. As pkil demonstrates, <pk> is an exception to the normal rule that <s> occurs before <i> instead of <k>.
2. <tk> occurs before <a u> and <ts> before <i> (the usual distribution of <k> and <s>). <ts> is a consonant cluster, not an affricate. Likewise <tl> is a cluster /tl/, not any sort of lateral affricate.
3. <m g> are realized as [b g] at the beginning of a cluster. In particular, <mt tm msh> are pronounced [bd db bʒ], and likewise <gt tg gsh> are [gd dg gʒ]. But these clusters often get devoiced in speech.

Final clusters
The final clusters are mostly the initial clusters, but reversed, subject to the constraint that <w> and <y> don't occur in codas. There's also a series of nasal + stop finals. Another table:

	p	t	ch	k	q	m	g	f/h	x	sh
shC	shp	sht		shk	shq	shm	shg	shf	shx
Csh	psh			ksh	qsh	msh	gsh	fsh	xsh
tC	tp			tk	tq	tm	tg	th	tx
Ct	pt			kt	qt	mt	gt	ht	xt
rC	rp	rt	rch	rk	rq	rm	rg	rh	rx	rsh
lC	lp	lt	lch	lk	lq	lm	lg	lh	lx	lsh
NC	mp		nch	nk	nq

As with initial clusters, <m g> are realized as [b g] in many final clusters: in careful speech <tm> is [db] and <shg> is [ʒg]. In normal speech they're partially or fully devoiced.

Medial clusters
Any legal initial or final cluster can also occur between two vowels. Additionally there are sometimes three-consonant clusters, which have the form <sh> + stop/nasal + <r, l, w>, as in ishpla "urinate." But three-consonant clusters are uncommon, and speakers tend to be lazy and drop the middle consonant.

The prefix n-
One last bit of phonology (I promise). There is an extremely common grammatical prefix <n>; I separate it from the stem with an apostrophe, like so: n'tuh, n'shoqo, n'ugu. It is pronounced:

1. As a homorganic nasal before stops <p t k ch q>, and the stop is voiced.
2. As an identical nasal before nasals <m g>, making a single lengthened consonant. Note that <mr> is [bʁ] but <n'mr> is [m:ʁ].
3. As [ɲ] before <y>, so <n'y> is also a long nasal, [ɲ:].
4. As [nə] before fricatives <f s sh h x>. The schwa is epenthetic, and quite brief.
5. As [n] before vowels and other consonants. <n'n> has two interchangeable realizations: [n:] or [nl].

OK, now we're ready for basic syntax... coming soon!

[Raphael]

I'm sorry that I don't have any more detailed feedback to provide, but the idea certainly sounds fascinating.

[aporaporimos]

Some Syntax

Nouns 'n' verbs
Words (with the exception of a few particles) can be classified as nouns or verbs. And most roots are also words. Some examples of roots that are nouns: qoyo "eye," tkala "breath/life," ouh "nose," qeme "full moon." And some roots that are verbs: ich "take," shak "stand (on)," gik "bite, sting (of an insect)." Verbs can be either transitive (taking a subject and an object) or intransitive (taking just a subject); most verbs are transitive, except for the so-called internal verbs (introduced below).

Verbs agree with their subject in person. Recall from the intro post that there are only two persons: Ego (~1st person) and Alter (~3rd person). So each verb has two forms: Ego-subject, which is unmarked, and Alter subject, which is marked with the prefix n-. We can draw up a little paradigm with shak:

Subject	Form	Gloss
Ego	shak	one stands
Alter	n'shak	it stands

That's the entirety of the verb conjugation; there is no inflection for such exotic features as tense, aspect, mood, etc. There's also no independent personal pronouns, so you can just as well think of n- as a prefixed pronoun.

Similarly, any noun can be verbalized by prefixing it with n-. Tfig means "lizard"; n'tfig is an intransitive verbal form meaning "it is a lizard." If there is no explicit subject, n'tfig may also mean "there is a lizard." (What if you want to say "I am a lizard"? The verbal form can't have an Ego subject; instead you'd say ai tfig, literally "behold, a lizard.")

Meanings we associate with adjectives are covered by nouns: qom "big thing," fshi "sharp thing," kak "excrement; bad, worthless thing." The corresponding verbal forms are n'qom "be big," n'fshi "be sharp," n'kak "be bad." Some of these words are used almost exclusively in the verbal form, in which case they're more like intransitive verbs than nouns. It doesn't make much of a difference either way.

Internal verbs
"Internal verb" is my not-very-clear term for a verb with the following properties:

1. It is intransitive;
2. It refers to a perceptual or emotional sensation, or a reflexive action tied to such a sensation;
3. Its subject is the source of the sensation; and
4. The experiencer of the sensation is implicitly and immutably Ego.

As a result of point 3, these verbs always have an Alter subject and so are always prefixed with n- (when used as a verb). Examples of such verbs are n'tuh "one sees it," n'sip "one hears it," n'xa "one laughs at it," n'ooq "it hurts one," n'qloga "one is hungry for it." (The idea behind the term "internal verb" is that the experiencer role is "internal" to the lexeme rather than being expressed separately. I'm open for suggestions if anyone has a better term.)

Basic sentences
The order of constituents in a clause is SVO. The subject may be omitted, in which case it is inferred and could be anything, whether specific or vague. If the verb subject is Ego, no explicit subject can be supplied. Similarly the object of a transitive verb may be omitted as well, in which case it could be anything at all, whether Ego or Alter. A clause may contain multiple verbs, and is usually tagged with one or more internal verbs at the end. So now we can put together a sentence:

mirp n'gik xeg n'tuh.

fly it-bites leg it-one.sees

A fly bit him on his leg.

In this case I left the internal verb n'tuh untranslated, but that doesn't mean it doesn't contribute anything to the meaning of the sentence. Quite the contrary:

mirp n'gik xeg n'ooq.

fly it-bites leg it-one.hurts

A fly bit my leg!

The internal verb specifies the perceptual character of the thought and very often, as in this case, clarifies whether the object of a verb or the possessor of a noun is Ego or Alter. There is no direct means of marking an Ego object or possessor, so internal verbs are the main means of expressing this information. While internal verbs are not grammatically required as such, a declarative sentence without one often feels complete, difficult to understand, or possibly even meaningless.

[Vilike]

I especially like the system of indexing on verbs.
Is the speakers' culture influenced by the Hyenas in Digger, by any chance? With Ego refered to as Alter if its name were ever "eaten" (read: banished)?

[aporaporimos]

I especially like the system of indexing on verbs.
Is the speakers' culture influenced by the Hyenas in Digger, by any chance? With Ego refered to as Alter if its name were ever "eaten" (read: banished)?

I am indeed inspired by Digger! (Was it the avi that gave it away?) But the inspiration is pretty loose, and I don't plan to reproduce all the distinctive parts of the Digger hyena culture (mine aren't cannibalistic, for instance).

Yes, if an individual were banished from the tribe, they would be unable to included in the sphere of Ego. (Note that non-present individuals are always referred to as Alter, so it would only come up if someone were speaking directly to the banished individual, which they may not be supposed to do anyways.) On the other hand, while Digger has Ed referring to himself as "it," I don't think self-reference as Alter is possible for a speaker of Nashalq; it would suggest psychological pathology.

On the subject of names, I plan to talk about them when I get to compound words, since that's what most names are: Qoyolni "Blue-eyes," Qrolshi "Sharp-claw," Shyogshol "Stripes" (lit. "rivers-in-fur"). Shikazi don't really distinguish stable personal names from nicknames the way we do; any individual can have multiple names in use, and can acquire a new one at any time; and it's common for one Shikazi to make up a nickname or pet name for another one and use it exclusively, even if no one else picks it up. So the notion of "eating a name" might not mean much to them as a metaphor for obliteration of social identity; but still it's likely that someone banished from a tribe would be referred to by some appropriately imprecatory title rather than any names they held formerly.

[aporaporimos]

Syntax: noun phrases

A phrase can be made of two nouns in sequence, in which case the second modifies the first:

qoto maksh

horn deer

Antlers

You can chain together multiple nouns like this, although it's not so common:

fith sisi fi

wing songbird tree

The wings of the bird in the tree.

The first noun can be prefixed with n- in its copular use:

qeme n'nakma mir.

full.moon it-mother night

The moon is the mother of the night.

An alternative construction, especially used for expressing possession, consists of the possessor, followed by the possessed head noun, which is prefixed with n-:

maksh n'qoto

deer it-horn

The deer's antlers.

I haven't fully worked out when the distinction between the qoto maksh and maksh n'qoto constructions; the latter generally requires the possessor to have a definite specific referent. If I found a shed antler on the ground I could call it a qoto maksh but not *maksh n'qoto.

A verb or verb+object may follow a noun and modify it attributively, like a relative clause:

tfig n'kshah moqse

lizard it-exhales fire

A fire-breathing lizard

This is how adjectives are used:

tfig n'qom

lizard it-big

A big lizard

Edit: There doesn't actually have to be an explicit head noun:

n'kshah moqse

it-exhales fire

The fire-breather

At this point you may have noticed some ambiguity in the syntax. First of all, the construction X n-Y can be either possessive (maksh n'qoto = "the deer's antlers") or copulative (maksh n'qom = "the deer that is big"). This turns out not to be a big deal in practice; it's normally obvious from context and the meanings of the words which of the options is appropriate. A more serious problem is that a verbal form can be either attributive or predicative: tfig n'qom can be either "the big lizard" (attributive) or "the lizard is big" (predicative). This ambiguity can be resolved using a particle u, which is sort of like a verbal comma; it marks the end of a noun phrase. So we can distinguish:

tfig n'qom u n'tuh.

lizard it-big U it-one.sees

I saw the big lizard.

tfig u n'qom n'tuh.

lizard U it-big it-one.sees

The lizard is big!

This is quite useful when sentences get more complex:

tfig n'qom u n'kshah moqse n'tuh.

lizard it-big U it-exhales fire it-one.sees

The big lizard breathes fire.

The particle u can maybe be analyzed as a determiner or pronoun. As you can see I just gloss it as "U". It's enclitic (pronounced as part of the preceding word) and has a couple allomorphs. If the preceding word ends in a Q-consonant (one of <q x ň>) it has the form o. If the preceding word ends in a vowel it's lu (it's common for <l> to be inserted to avoid consecutive vowels). However, if the preceding word is multi-syllabic and ends in <a>, that final vowel is elided before u, which takes the form o if the elided vowel was mid <ǎ>. Examples of all the allomorphs:

1. n'qom > n'qom u
2. n'shxeshoq > n'shxeshoq o
3. qoyo > qoyo lu
4. tkala > tkal' u
5. pxaqa > pxaq' o

OK... next planned post will be on compound words (ie taking the phrases and smushing them together).

[aporaporimos]

Compounding

Compounding is the most prolific form of word derivation in Nashalq and is extremely productive. The phonological rules governing compounds are for the most part regular, so it is normally straightforward to identify the component morphemes of a compound; the meaning, however, is often idiomatic. I cover the phonological rules at the end of this post; they may be of interest but they aren't necessary to follow the rest of the exposition.

VO compounds
In many cases, you can easily imagine a compound as a phrase that got reanalyzed into a single word. It's a good starting point for verb+object compounds:

Qrolshi n'shak metq Shyogshol n'tuh.
Qrolshi it-stands neck Shyogshol it-one.sees
Qrolshi stood on Shyogshol's neck.

Qrolshi n'shakmetq Shyogshol n'tuh.
Qrolshi it-stands-neck Shyogshol it-one.sees
Qrolshi overpowered Shyogshol.

This example is typical in that the meaning of the compound is idiomatic, and the compound as a whole is a transitive verb. Sometimes VO compounds are nouns (usually referring to the subject of the verb): n'kshah xshol "it blows on fur" > kshaxshol "wind." Nashalq makes a great number of VO compounds from body parts and a few basic verbs. Here's a sampling:

Phrase	Gloss	Compound	Gloss
shak metq	stand-on neck	shakmetq	overpower, subdue
rat metq	grab neck	ratmetq	command, compel
rat ouh	grab nose	ratouh	lead, guide
qwo ashp	open paw	qwolashp	let go of
kta ashp	close paw	ktalashp	grab hold of
qwo paig	open mouth	qwopaig	begin
kta paig	close mouth	ktapaig	finish
gaik mouk	bite tail	gaikmouk	backstab
rat mouk	grab tail	ratmouk	pester
taq mouk	eat tail	taqmouk	waste time, be stupid

Gaikmouk, ratmouk, and taqmouk are all bad things to do, but the first is the only one you'll get killed for.

SV compounds
Subject-verb compounds can often be seen as reanalyzed phrases as well:

Shyogshol n'piku n'malk n'ish.
Shyogshol it-ear it-dull it-one.senses
Shyogshol doesn't seem to have the greatest hearing.

Shyogshol n'pikulmalk n'ish.
Shyogshol it-ear-it-dull it-one.senses
Shyogshol doesn't seem very bright.

Observe that the prefix n- gets absorbed right into the compound, taking the form <l>. My category of SV-compounds includes those where the second element is a noun verbalized by n-. (Many adjective-type words are used almost exclusively in the n- form so it's hard to say if they're actually nouns or verbs anyways.) The compound itself can be either a verb (often intransitive) or a noun. We can find lots of examples of the noun+adjective form:

Phrase	Gloss	Compound	Gloss
piku n'malk	ear is-dull	pikulmalk	be slow-witted
qoyo n'fshi	eye is-sharp	qoyolshi	be clever
qoyo n'roq	eye is-heavy	qoyolroq	be sleepy
ouh n'qom	nose is-big	ooxqom	be greedy
qro n'tul	claw is-long	qroltul	be old
metq n'tul	neck is-long	meqtul	strain one's self (to do something)
ig n'qom	sound is-big	enqom	loud noise, boom, crack
ig n'pitk	sound is-small	igpitk	soft noise, rustle, whisper
shoxe n'qom	shadow is-big	shxenqom	something frightening

Shoxe is one of the minority of roots that has a dedicated combining form used in compounds. In fact, it has two: shxe- at the beginning, and -shox at the end.

By the way, the personal name Qrolshi is a compound of this type too: it derives from qro n'fshi "claw is-sharp."

NN compounds
Noun+noun compounds are the last major category of two-element compounds. Unfortunately I don't have very many examples of them yet! But the name Shyogshol is one: it's the ordinary word for "stripe," derived from shyug xshol "stream (in) fur."

Compounding: Phonological rules

When two roots or words are joined in a compound, the junction between them may produce a cluster that can't be pronounced. The phonotactic rules for clusters are a little freer than those for roots—most two-consonant sequences are allowed in a compound, aside from geminates—but many compounds still need to be modified somewhat to be pronounceable, most often by deleting some consonants. I'll give the relevant rules according to the phonological shape of the junction between the elements.

The prefix n-
...when it begins the second element of a compound, becomes a homorganic nasal before <p ch k q>, and <l> otherwise. Then continue with the appropriate section below.

VV junction
Two consecutive vowels aren't allowed. If the first element is a single syllable, just insert <l> as a separator: qwo + ashp = qwolashp. If the first element is multi-syllabic, its final vowel is deleted: ishpla + uk = ishpluk. When two elements are joined in this way, the two roots at the juncture count as one for the purpose of mid vowel harmony, so mid vowels may spread from one to the other: ishpla + oqt = eshploqt.

VCV junction
Iff the C is <n> it becomes <l>. If a K-consonant comes into contact with a mid-vowel, it becomes the corresponding Q-consonant, and if a plain vowel comes into contact with a Q-consonant, it shifts to a mid vowel; both of these may trigger harmony within a root or further assimilation. For example: qwo + sih = qwosex.

VCCV junction
First, K-consonants and plain vowels assimilate to Q-consonants and mid vowels, possibly triggering harmony or further assimilation in one of the elements. Most two-consonant clusters are allowed in compounds, even ones forbidden in roots: ich + ma = ichma, ak + piku = akpiku. But the same consonant can't occur twice in a row; if it does, one of them is dropped: ak + kusi = akusi. Accounting for assimilation, this means ak + qoyo = aqoyo. A few other pairs of consonants have special outcomes:

• tsh, tch, chsh > ch
• chs > ts
• shs > s
• ty > chy

VCCCV junction
First, apply the rules from the last section to both pairs of adjacent consonants. If you're lucky, you'll be left with only two consonants, or with one of the few permissible 3-consonant clusters. (There are no 3-consonant clusters permitted in compounds beyond those which are permitted in roots.) If the cluster still needs to be simplified, follow these rules until there are only two consonants left:

1. If the sequence has the form C₁shC or C₁chC, delete C₁.
2. If the sequence has the form PNC, where P is one of <p k q> and N is one of <m g ň>, delete N and replace P with the nasal at the same place of articulation.
3. If the sequence has the form ClC, delete the <l>.
4. If the sequence has the form tCC, delete the <t>.
5. If none of the other rules apply, delete the middle consonant.

If the second element began in <mn> and the <m> was deleted, or if it began in one of <pk, ph, pq, px> and the <p> was deleted, the original following consonant is replaced with <w>. As a result of this rule, ak + pxaqa = akwaqa. Combined with rule 2 above, ak + mni = agwi.

Now, apply the rules from the last section again to the remaining two consonants. Thus ak + tkala = akala: the <t> is dropped and then the two <k>'s, now adjacent, coalesce into one.

More than three consonants
Not going to go into detail here; the rules in the last section apply analogously to longer clusters. <t> is preferentially deleted, <sh> and <ch> are preferentially preserved, and otherwise consonants are dropped from the middle.

[aporaporimos]

Derivation

Noun -> Verb: ak
There is a prefix ak- that derives a verb from a noun. (Actually, ak does occur as an independent word, albeit only in a specialized meaning.) It's enormously productive:

Noun	Gloss	Verb	Gloss
piku	ear	akpiku	prick one's ears; listen
qoyo	eye	aqoyo	look (at)
ouh	nose	(a)kouh	sniff (at, for); seek
ashp	paw	(a)kashp	touch; grasp; walk
num	tooth	aklum	chew (on); think (about)
tkala	breath; life	(a)kala	breathe (in); be alive
sisi	songbird	asisi	sing
troqo	earth	(a)kroqo	bury

What's up with the ones where the initial <a> is in parentheses? Well, for many verbs in ak-, that <a> is dropped in the unmarked (Ego-subject) form. It reappears when the Alter-subject prefix n- is added. When such a word is used to form a VO compound, the <a> is dropped entirely from all forms, so (a)kouh + mouk = kouhmouk. We can distinguish three patterns, plus that of (a)kala, which is irregular:

Lexeme	aklum	(a)kouh	kouhmouk	(a)kala
Gloss	chew	sniff	follow	breathe
Ego S	aklum	kouh	kouhmouk	kala
Alter S	n'aklum	n'akouh	n'kouhmouk	n'akla

Some notes:

1. Which pattern a verb follows isn't always predictable: *klum would be perfectly pronounceable, but it's not a valid form.
2. The <a> isn't dropped if doing so would make the verb form identical to the noun; this is what protects aqoyo and asisi.
3. The irregular form n'akla is a victim of Nashalq's distaste for words with 3 open syllables in a row; the middle vowel of *n'akala is dropped to yield a more-preferred structure.
4. The form kala, while regular, is odd in that it has the structure Ø + ak + tkala, and ends up equivalent to the stem with an initial consonant deleted. It is a victim of Nashalq's refusal to allow geminate consonants in a word, even across a morpheme boundary.

Abstraction: ak again
I mentioned earlier that ak has a limited use as an independent verb. That use is, basically, to quote a single word. It's like the English word "go" in phrases like "go ow!." Many of Nashalq's internal verbs, like n'ooq "it hurts one" or n'xa "it makes one laugh," are similar to English emotive interjections like "ow!" or "ha!" (and indeed the emotive verbs are used as interjections as well). With ak they can be abstracted, that is, applied to someone other than Ego:

Qrolshi n'ak xa n'sip.
Qrolshi it-goes ha! it-one.hears.
Qrolshi laughed out loud.

Shyogshol n'ak n'ooq n'sip.
Shyogshol it-goes ow! it-one.hears.
Shyogshol cried out in pain.

(The reason n'ooq keeps its n- but xa drops it is simply that the former begins in a vowel.)

Naturally, these two word phrases beginning in ak get combined into compounds: aqxa "laugh," (a)klooq "cry in pain." Note, however, that there are limits on which internal verbs can get abstracted this way. It only works for those that entail some kind of visible external reaction. From n'tuh "one sees" there is no abstracted form *aktuh, since sight is a purely private and subjective experience.

By the way, aqxa is pronounced [ɤqqʰɤ]. Word-internal sequences of stop + <h> are pronounced as geminate aspirates as well. (The prohibition on gemination applies only to phonemic doubled consonants, not to geminate realizations of underlying clusters. Not only is <qx> realized [qqʰ], but <mp>, <nk> are commonly [bb], [gg].)

Verb -> Noun: n-

Much like phrases can be reanalyzed as compound nouns, inflected verbs can get reanalyzed as nouns:

n'aqxa n'sip.
it-laughs it-one.hears
I hear someone laughing.

naqxa n'sip.
NOM-laugh it-one.hears
I hear laughter.

The proof that this has occurred is that you can re-prefix n- to the nominalized form: n'naqxa "there is laughter." It only works if the verb stem begins in a vowel. Conveniently, the verbalizer ak- does begin in a vowel, so many words in ak- have a further derived form in nak-. In the case of naqxa, the nominalized form refers to a process. In other cases it refers to the agent, as in nakma "mother," which comes from akma "give suck," which in turn comes from ma "teat." One more table, this time with 3-word families:

Root	Gloss	ak-	Gloss	nak-	Gloss
xa	ha!	aqxa	laugh	naqxa	laughter
n'ooq	ow!	(a)klooq	cry in pain	naklooq	pain
num	tooth	aklum	chew; think	naklum	day-dreamer
tkala	breath; life	(a)kala	breathe; live	nakla	life
ma	teat	akma	give suck	nakma	mother
tsi	seed	aksi	beget	naksi	father

Of note: the meaning may drift significantly across derivations, as from num "tooth" to naklum "day-dreamer"; but on the other hand, tkala and nakla are near-synonyms, both meaning "life." As if that weren't bad enough, there are popular pleonastic phrases like tkala nakla "the breath of life," or, etymologically, "the breath of breathing."

The Shikazi don't usually refer to their own parents by the rather bristly terms nakma and naksi; they use the familiar forms mama and titi. Both the formal and familiar terms are used for any adult female/male of the tribe in one's parents' generation.

[Vilike]

So, can consonant-initial verbs ever get nominalized? What is, if any, the alternate strategy?

[aporaporimos]

So, can consonant-initial verbs ever get nominalized? What is, if any, the alternate strategy?

They can! Consonant-initial verbs are nominalized by phonetic simplification of word. Right now I have three forms of simplifcation used for derivation, all of which apply to the last syllable of the word:

1. Reduction of a diphthong to a pure vowel: gaikmouk "backstab" > gaikmuk "traitor."
2. Deletion of <t> from a final cluster: shakmetq "overpower" > shakmeq "pack-leader."
3. Deletion of the other consonant from a cluster with <sh>: kroqapsh "hide" > kroqash "cache."

I may add more in the future. It's really more accurate to say that, for example, gaikmouk is a verb 80% of the time and a noun 20% of the time, and vice versa for gaikmuk. That is, this form of derivation isn't as cut-and-dried.

These simplifications belong to a larger class of processes that happen to morphologically complex words: kroqapsh itself derives from ak-troqo-kapsh, "bury skull." Nashalq has a kind of principle of conservation of effort where adding morphemes doesn't cause words to get much longer than 2 syllables (ending in a consonant) or 3 syllables (ending in a vowel).

(This is where I start improvising) Of course, many words neither begin in a vowel nor have a final syllable that can be simplified. Some can be either a noun or a verb, like chacha "to babble" or "babbling." Or you can just use a phrase or compound.

[aporaporimos]

Sensory words

OK, let's take a break from derivational morphology for a bit. (There's more but it pretty much continues the same themes.) Let's talk more about those little "internal verbs" tagging every sentence:

txash n'urk kashproq n'sip. n'chacha n'sip.
human on-back path it-one.hears. it-babbles it-one.hears.
I hear humans on the path. They're talking.

txash n'urk kashproq n'tuh. n'patpat n'shoqo n'tuh.
human on-back path it-one.sees. it-cub it-is.like it-one.sees.
I see humans on the path. They look like children.

txash n'urk kashproq n'ish. n'agshir n'ish.
human on-back path it-one.smells. it-shivers it-one.smells.
I smell humans on the path. They're afraid.

(Urk "back" along with several other body part terms has a prepositional meaning when verbalized.)

The sensory verbs n'sip, n'tuh, n'ish tag each thought with the sense by which it was perceived; hearing, sight, and smell in these examples. I'm don't have a comprehensive list of sensory verbs yet, but the classical 5 senses are rounded out by n'muk "one tastes" and n'chwa "one touches." The full list will be pretty long, because there's a lot of specialized ones like mlisi "it is painfully bright." One more important one is n'qa, which covers things you did yourself:

kroqapsh n'ashp fi n'qa.
bury on-foot tree it-one.does.
I buried it beside a tree.

They look kind of like evidentiality markers, don't they? But they're not really. The point of n'tuh isn't that you know something is true because you saw it; it's that the perceptual character of the thought is visual (as opposed to auditory, olfactory, etc). This helps pin down a lot of descriptive words that by themselves are rather vague:

n'ni n'tuh / n'sip / n'qa.
it-light it-one.sees / it.one-hears / it.one.does
It's pale-colored / high-pitched / light-weight.

n'chlufshoq n'tuh / n'ish / n'muk.
it-blood-like it-one.sees / it-one.smells / it-one.tastes.
It is blood-red / smells like blood / tastes like blood.

(You find out how much something weighs by lifting it, or trying to do so, so Nashalq puts weight under the domain of proprioception covered by n'qa.)

The vagueness of words like ni and chlufshoq isn't remarkable—English has similar ambiguity in terms like "light" (color or weight) and "soft" (texture or loudness). What's worth noting is the role the sensory verb plays in pinning down the meaning. Any of these sentences could also be a question; for example,

i n'ni n'qa?
Y it-light -it.one-does
Is it light (in weight)?

Remember that Ego can include the 2nd person; this is especially common in questions. The "listener inclusion" particle i suggests, but doesn't require, this interpretation; the question itself is marked only by intonation. (I gloss i as Y because it's easier to read than capital I; y- is a pre-vocalic clitic allomorph.)

If you're telling a story—whether a true story about yourself, or a story someone else told you, or a traditional story—you will constantly switch sensory verbs based on what kind of sensation you're describing: n'tuh (sight) for visual descriptions, n'sip (hearing) for dialogue, and so on. Using n'tuh for something you didn't actually see in person is by no means dishonest or misleading; you can use it for imaginary images, as long as they're imaginary images and not imaginary sounds or smells.

Of course, it won't always be clear-cut what sensory verb is most appropriate. The defaults are n'tuh (sight) and n'ish (smell); the latter is used generally for impressions, intuitions, guesses, etc. (You can think of it as corresponding to contemporary English "I feel like...") The Shikazi have a more refined sense of smell than we do, and rely on it more; I want to work this into their language, but I haven't really figured out how to do that yet. (Suggestions?)

[Vilike]

You can put more smell-related metaphors in the metaphorical framework of the language (as you already did with intuitions and guesses), and organize the smell-related lexicon within a well-defined structure, as seen in the human language Maniq (here and here).

[aporaporimos]

You can put more smell-related metaphors in the metaphorical framework of the language (as you already did with intuitions and guesses), and organize the smell-related lexicon within a well-defined structure, as seen in the human language Maniq (here and here).

Ooh, Maniq looks interesting—thanks! I'm planning to lean heavily on synesthesia for descriptive terms, so ordinary (to us) words for sights/sounds/textures etc can double as scent terms; hopefully that will make my work a little easier.

[aporaporimos]

A Short Dialogue

Characters:

• Qoyolni: A Shikazi who has been leading nighttime raids on farms and rural villages. Unusual in his intelligence and in his hostility towards humankind.
• Shyogshol: A kouhmouk (follower) of Qoyolni.
• Yarangil: A (human) trapper who was at the wrong place at the wrong time and got caught by Qoyolni's band. He speaks Nashalq, a rare ability among humans and one that marks you as an outsider—but a useful ability too, as someone who speaks the Shikazi's language language can often extract a favor from them. Not tonight, though...

Dialogue

1. Y: ai numyaka.
behold tooth-none
I'm unarmed!

2. Q: shka txash n’chaku!—u n’numyaka?
startles.one human it-speaks.—U it-tooth-none?
You can speak! — Is he unarmed?

3. S: n’qwolashp tgar n’tuh.
it-open-paw talon it-one.sees
I saw him drop his dagger.

4. Y: ai numyak, n’qwomouk, nipmouk, txash n’yaka. i taq?
behold tooth-less, it-opens-tail, turns-tail, human it-is.none. Y eats?
Look, I'm not a threat. Why don't you just let me go, I'll leave, and you won't have any humans to deal with. OK?

5. Q: xa! u txash n’chaku nisha n’sip, a n’wulp tlung wikra n’ish.
ha! U human it-speaks sweet it-one.hears, but on-underside tongue bitter it-one.smells.
Don't make me laugh, human. You can talk nice, but I know you have a trick up your sleeve.

6. Y: shpak! wikra n’yaka.
spits! bitter it-is.none.
No! No tricks.

7. Q: txash n’igyaka! n’gashm!
human it-sound-is.none. it-pathetic.
Silence, human! You're pathetic.

8. S: y’oqo Qoyolni n’chaku n’sip, oqlo txash n’igyaka!
Y WHEN Qoyolni it-speaks it-one.hears, THEN human it-sound-is.none.
Yeah! When Qoyolni talks, you stay quiet!

9. Y: Qoyolni… qlosh n’sip… n’aq nichkala.
Qoyolni... distinctive.smell it-one.hears... it-smells.of NOM-take-life.
Qoyolni... I've heard about you... you're a murderer.

10. Q: u n’sip.—shkisi.
U it.one-hears.—cut.
Indeed, it seems you have heard of me.—Kill him.

Notes
1. numyaka. The word n'yaka means "there is none," and it's often used in SV compounds, like numyaka here. In this case "teeth" stand for any means of doing harm. N'yaka also has a bound suffix form -yak (sometimes -yaq). There isn't a clear difference in general between -yaka and -yak in compounds. They occasionally occur with the same root in different shades of meaning, and that's the case with num: numyaka is "unarmed," but numyak is "harmless."

2. shka. An expression of surprise; emotive words usually go at the beginning of the sentence (as in English). Like most such words, this one can also be used as a verb, n'shka "it startles one."

txash. The human is addressed in the 3rd person (so to speak). This is normal even in amicable encounters with Nashalq-speaking humans.

n'chaku. Simply means "to speak"; in this case context it means "speak Nashalq." Other languages fall into the category of chacha or "babbling." The root of both words is cha-; this is a case where there are two basic words from the root, one which is diminutive or derogatory and is formed by reduplication, and one which is neutral and formed by adding a meaningless suffix -ku.

u. A discourse particle, always sentence initial. (Nothing to do with the other u that marks the end of a noun phrase). It has several different meanings. In a question, it asks for confirmation of a claim. In this case Qoyolni is asking Shyogshol to confirm the human's claim that he lacks weapons.

3. tgar. Used for talons of birds and as an elevated term for "claw," and also as a term for daggers, knives, etc.

4. n'qwomouk, nipmouk. To "turn tail" is a familiar metaphor. (The Shikazi have actual tails, so it's not even a metaphor for them.) The basic meaning of qwo is "open," but here it essentially stands for qwolashp "open hand > let go of," so qwomouk means "let go of someone's tail" ie "let them go free."

i taq? While taq literally means "eat," it's also a discourse particle meaning "ok, sure, uh-huh." (The metaphor here is that eating something = accepting it; just like how in English to "swallow" something can mean to believe it.) The "listener inclusion" particle i, commonly used in questions, takes on a pleading tone when used from a subordinate position as here.

5. u. Another use of initial u is to mark a sentence as sarcastic, or (as in this case) just cynical.

a. A clause-initial particle that acts as a generic connector. It can usually be translated as "and," "but," or "then."

n'wulp. Wulp refers to the surface of the belly; in its use as a prepositional noun it means "the underside." (Some of the prepositional terms only seem to make sense if you assume a quadripedal posture.)

wikra. A scent/taste term meaning "bitter, acrid." Associated with poison, hence its use here to mean "trickery" or planned betrayal.

6. shpak! Literally spit out, but also means "no," as a rejection of another's words. The counterpart to taq.

7. txash n'igyaka. Commands aren't distinguished syntactically from statements, but if it were a statement we would expect it to be tagged with n'sip "one hears it."

n'gashm. To be gashm is to fall pitifully short of an ideal analogous to what we'd call "manliness," except among the Shikazi, who have little sexual dimorphism, it applies equally to both sexes.

8. y'. This is form i takes before a vowel. In this case it signals agreement with the last speaker (not, of course, with Yarangil, who is the addressee).

oqo... oqlo. A pair of correlatives, whose meaning is broad, but they generally are used for conditionals of the "if... then" or "when... then" type. Before a vowel, oqlo takes the form oqol'. Note that the oqo clause is tagged with a sensory verb n'sip while the oqlo clause is not, suggesting that the latter is a command rather than an observation.

9. qlosh n'sip. Qlosh literally refers to one's distinctive personal scent. (Literally literally, it refers to certain odoriferous secretions.) It's used generally to mean one's personal character and reputation. The sensory verb of hearing n'sip makes it clear that the meaning here is "reputation."

n'aq. N'aq is a sensory verb, but a transitive one. The sentence could end with n'ish, but in this case it's optional (because it's redundant). While n'aq literally means "smell of" it often simply means "be associated with"; here it's used in a negative sense, "be guilty of."

nichkala. Ich "to take" is another verb often used in compounds that conveniently begins in a vowel, allowing easy nominal derivations. Strictly speaking ichkala "take breath/life" simply means to kill, not necessarily to murder; it's the term used for killing intelligent beings. It's not normally used of killing animals for food.

10. u n'sip: It's common to repeat just the final sensory verb of someone's utterance to stand for the whole thing, even in contexts (such as here) where you wouldn't otherwise include that person in an Ego-reference. There are several meanings u can have in this context, and all of them seem to be at play here. First, with a repeated sensory verb it often is just a confirmation: "indeed, that is so." Second, in any declarative context it can indicate surprise or being impressed, possibly mixed with skepticism, which can often be glossed as "apparently..." And finally as mentioned above it can add a touch of irony.

shkisi: Properly to cut, sever, or rend, but, well, you see what it means here. This time the command has an Ego subject, as it's spoken to a friend.

[Bob]

...

Good job on this one. This one is great for concepts and ideas about animal thought.

[dendana]

Wow! This was really fun to read through. I love how you’ve encoded so much of the species’ psychology into the language. The ego/alter distinction is neat, as are all the etymologies and expressions you’ve worked out. I may have to borrow the idea of having those evidential-like verbs everywhere. Morphology is intriguing too, and I like the way a small number of affixes/patterns chain and lexicalize.

Things I’m curious about:

How does predicative possession work in this language? or maybe context alone would distinguish from the copula (or maybe the concept of possession is quite different in their culture)

I’m also curious why in the dialogue we have

S: y’oqo Qoyolni n’chaku n’sip

- I’d have expected ego referent (chaku instead of n’chaku) to describe Qoyolni’s speech as the speaker is on the same side as Qoyolni.

I know I’m a couple months late, but maybe since smell is so primary for them, there could be suppletive internal verbs for smelling different kinds of things? I’m also curious if any of the internal verbs can ever be used as non-internal ones or vice versa.

[aporaporimos]

Good job on this one. This one is great for concepts and ideas about animal thought.

Thanks!

How does predicative possession work in this language? or maybe context alone would distinguish from the copula (or maybe the concept of possession is quite different in their culture)

I don't have an answer to this yet. A couple possibilities are specialized verbs (ktalashp "grasp" could perhaps be appropriate for transient possession) or a locative construction (I don't think I've covered how these work yet, but various body part terms can be used like prepositions, both attributively and predicatively).

I’m also curious why in the dialogue we have

S: y’oqo Qoyolni n’chaku n’sip

- I’d have expected ego referent (chaku instead of n’chaku) to describe Qoyolni’s speech as the speaker is on the same side as Qoyolni.

In this case the point is the authority of Qoyolni, which Shyogshol (the speaker) does not share. A verb with an ego referent can't have an explicit subject, so it would go like this:

y'oqo

Y WHEN

chaku

speak

n'qa,

it-one.does

oqlo

THEN

txash

human

n'igyaka!

it-is.silent

But that would mean "when we speak, you stay quiet!" with no indication of Qoyolni's special authority, which is not his intent. While he is on the same side as Qoyolni, he isn't in the same position as him when it comes to speaking and expecting others to stay silent. So for this thought Qoyolni is an alter referent. In general it's usual to refer to a comrade as alter when speaking to a third party; ego reference normally includes either the 1st or 2nd person.

I know I’m a couple months late, but maybe since smell is so primary for them, there could be suppletive internal verbs for smelling different kinds of things? I’m also curious if any of the internal verbs can ever be used as non-internal ones or vice versa.

I haven't worked on this language much since my last post, so suggestions remain timely The boundary between internal and non-internal verbs probably isn't absolutely fixed but I don't currently have any examples of verbs that act as both. But since so many internal verbs are also used as interjections, a non-internal verb used as an interjection has a likely pathway to gain an internal use.