Khulls phonology

From FrathWiki
Jump to: navigation, search

Khulls has a very large, unstable "house of cards" phonology that in many respects resembles PIE mixed with modern Chinese. A phonologically maximal analysis would give five vowels, 35 consonants, 7 tones, two distinctions of vowel length, a stress accent, a distinction between pharyngealized and clear vowels, ample consonant clusters, and seven syllabic consonants. However, although the phonology is indeed very large, many of the consonants are marginal, and the tones are interlinked with the stress and length of vowels and therefore all cannot be considered simultaneously phonemic.


Tone

Generally there are considered to be five phonemic tones, with the other tonal realizations being allophones of these. The full range of tones appears only in stressed syllables. It could be said that unstressed syllables have no tone of their own, although there is a distinction between unstressed syllables that resist sandhi and unstressed syllables that are affected by sandhi. A few compound words such as xinănopa "parrot" consist of a word ending in a consonant plus a word beginning in a vowel. The tone here remains ă, not à.

Khulls is not a pitch-accent language because unstressed syllables can have tones, even apart from the tones that result from sandhi. That is to say, there are words like īnčigō "strawberry", which has a tone pattern of H-L-M, where the third syllable is unstressed but nevertheless has a higher tone than the preceding syllable.[1] This pattern can only happen in words that were originally compounds, but this word is no longer perceived as a compound.

Strictly speaking, the pharyngealized "tone" is not really just one tone, because it has the sandhi effects of the tones it grew from. Newly coined words with poharyngerals are treated as coming from á, even though most in fact come from ā. This is because á is a long tone, like â, and does not depress surrounding syllables the way ā, the other long tone, does. So the words that really did arise from ā are pronounced with low tones preceding the pharyngeal.

Since snadhi even affects compiunds, it matters which of the two identical tones is used in a given word. For example ê "fish" and ê "pink" are almost never used as standalone words, but they have different sandhi effects when they combine to make compounds. Also, the "downwards" pushing variety of the â tone is the one that occurs on the very common locative form of a word. Thus:

  • čê "pants" (á-type pharyngeal)
  • čê "in the pants" (ā-type pharyngeal)

sound the same, but have different sandhi effects on surrounding words. That said, even this distinction is not particularly robust. The phrases oé čê "cat pants" and oé čê "cat in the pants" are still identical. It is only when a two syllable word is inflected that the distinction appears. For example porê "mermaid" has tone pattern M-L, but porê "in the mermaid" has L-L. The á-type pharyngeal can be spelled as an alternative.

Allotones

Distinct from sandhi, some tones change based on whether they are in an open or a closed syllable, or on their position in a word. The /à/ tone (high, short, often with a final glottal stop) becomes the [ă] tone (middle, short) when in a closed syllable. Also, all long tones become short in a closed syllable, which means that /ā/ and /á/ both become [à]. As an example, kʷēs "spice", kʷēmpe "spice bottle", kʷēkēs "spice grinds", kʷēko "spice sprout" have different tones. It could be said that this this sandhi after all though because it survives compounds. e.g. pampis + ĭ --> pampĭsi, not pampìsi as would be etymologically expected.

Vowels

If stress, length, and post-glottalization are considered to be properties of the tones rather than the vowels, the Khulls vowel system is very simple: /a e i o u/, with no diphthongs. Khulls is unusual in that its diachronic history has massive tables of consonant changes, but only a few vowel changes, nearly all unconditional. The parent language vowel system was /a i u ə/. In Khulls the /u/ was lowered to /o/ except after a labialized consonant, and /ə/ was either backed to /u/ or reduced to just a non-syllabic lip rounding, creating another series of labialized consonants on top of those inherited from the parent language. (But note that the sequence /əu/ had changed to /ū/ in an earlier shift, so this did not produce ʷu or ʷo/.)

There are also no diphthongs, as these have all been ironed out into monophthongs. In fact /e/ arises solely from diphthongs and is thus the rarest of the five vowels. /a/ is by far the most common, as it is in nearly every language on the entire continent. /i/ is second most common, and /o/ is third most common, unless /u/ is considered to be present in all labialized consonants, which is generally not done as it would lead to sequences of /uu/, thus making /u/ the only vowel in the language that can be doubled.

The sequences /ye/, /yi/, and /yu/ occur ("y" is IPA /j/, not IPA /y/), but /ye/ has changed to /e/ except after another vowel, making the /y/ purely allophonic, and both /yi/ and /yu/ are quite rare. /yi/ occurs mostly in noun plurals. Thus /y/ occurs as a phoneme only before the high vowels /i/ and /u/. In earlier stages of Khulls, /y/ could occur before all five vowels and could also occur after labialized consonants. In modern Khulls, the labialization has swallowed the /y/.

Allophony

Khulls vowels are "weak" in that they don't affect the pronunciation of surrounding consonants in any way. Instead, it is the consonants that infect the vowels. This is still fairly minor, howeever.

Normally, all five vowels are close to their cardinal IPA values. However, the front vowels /e/ and /i/ are pronounced slightly further back when following an ejective consonant or when pharyngealized. When both conditions are true, the vowel is still pronounced the same as if only one were true; it is not doubly-backed (this does happen in some descendants of Khulls however). Likewise, the vowels /o/ and /u/ do not labialize the consonants they surround; there is still a distinction between, for example, "doll" and xʷū "seed". It is the vowel that changes, being somewhat more rounded after a labialized consonant. Vowels do not change due to consonants that follow, however. The very rare minimal pairs such as làpʷ "carrot" and làp "dentist, teacher of mouths" have identical vowel allophones. It is not generally considered that Khulls has ten vowels instead of five because there is only a limited set of labialized consonants, rather than a Poswa-like setup with one labialized consonant for every plain consonant, plus a pure /w/ to match silence. Thus, the post-labialized allophones of vowels cannot occur in vowel-initial syllables. Moreover, labialized consonants can occur with no following vowel.

Consonants are never palatalized, even allophonically, by a following /i/. The sequence /ji/ (spelled yi) does palatalize a preceding consonant, but in Khulls the /y/ behaves as a consonant rather than a vowel. This occurs mostly with the plural suffix -yi. There was a historical process of palatalization, but it was defeated by any labialization and also undone in certain other environments, so most plurals of nouns that end in consonants do not change the final consonant of the stem. However, there are many nouns which today end in vowels, but in the past ended with -k or -ḳ; these all have plurals in -či. Note that unlike many neighboring languages, palatalization of the ejective was allowed. (Most other languages, including Khulls itself in an earlier stage, shielded all ejectives from palatalization.)

distribution

Most words that begin with vowels have pharyngitis there. /e/ and /u/ are rare on some tones, but allotones make up for this. Phars are omitted cuter y. Closed syllable s g pad vowels and no rounding on O.

Consonants

Khulls has many labialized consonants in its bilabial and velar columns. Since there is no true /w/ in the language, they could be analyzed as clusters, but this is not done because then the /w/ would only exist after certain other consonants. However, the very common /ʕʷ/ sound is pronounced [w] in unstressed syllables because pharyngealization always disappears in unstressed syllables. Thus an analysis of, for example, ḳʷ being /k/ + /ʕʷ/ could work, but would still be unusual in being the only type of cluster permitted in certain positions such as before other consonants. Moreover these consonants are not pharyngealized; ʕʷ is simply used because its voiceless counterpart, hʷ, would be inappropriate for non-aspirated stops.

Another reason for not analyzing /kʷ/ etc as /k/ + /w/ is that in older stages of the language, it was possible for labialized consonants to precede /j/ (spelled "y"). e.g. kʷyâ "on the sea ice",[2] </ref> hʷyas "war" etc. (These words have become kʷê and hʷes in classical Khulls, but *kʷê is no longer widely used.) This does not happen anymore, because most /j/ was deleted in a late sound change and what remained was deleted after all labialized consonants. However, these sound changes did not happen in the proto-Moonshine dialect, and therefore Moonshine began its history allowing such clusters after all. However, except for /hʷj/ they were rare because of the dilution of vocabulary with Bābākiam loans. Khulls has very short words, and there are 20 consonants that can stand alone as words: /p ṗ b m n ŋ l s š z pʷ ṗʷ bʷ kʷ ḳʷ ġʷ xʷ gʷ hʷ ʕʷ/. There are ten each of the plain and labialized consonants, but this is a coincidence because all of them, even the p ṗ b / pʷ ṗʷ bʷ pairings, have different origins.

Note that these same 20 consonants are also the only consonants that can occur at the end of a word, if the few exceptions like word-final /nt/ are read as /np/. The fact that 14 of the 20 consonants are labialized or bilabial or labialized bilabials leads to the language sounding in some ways like Classical Poswa. However, the so-called "syllabic" /p ṗ b/ are rare because they only occur as the reflex of sequences like /ŭk/, meaning that they can only come from vowel-initial syllables. (The sequence was uk > ukp> up > ʷp > p.)

Single-consonant words

Some single consonant words are: p "teacher", "eye", b "pine tree", "breast, inedible body part", "one, singulative", ŋ̇ "clam", "sleep", ṣ̌ "bomb", ż "to injure, hurt", "wet, soaked with water", ḳʷ "God", "insect", ġʷ "to believe; to stand up", "to bite, grasp firmly", "human, soldier", ʕʷ "fire". These single consonant words can combine with other words readily, even though the single consonant is often difficult to hear despite greatly changing the meaning. However, apart from a few suffix-only forms such as the agentive suffix -hʷ, most of these compounds were created before the missing vowel dropped out. Thus pàpo "student" was /həkpàpo/, etc.

Because these single-consonant words were originally stressed syllables, one might expect them to lower the tones of the rest of the word; however, the schwa never carried stress in early Khulls in any word in which a non-schwa vowel was present. Likewise, since labialized consonants can also arise from primordial /u/, one might expect tone sandhi here, but this also did not happen because the only /u/ that collapsed into consonant labialization was that which was unstressed.

Stops

Aspirated Stops

  • These are by far the most common series of stops in the language. There are five: /pʷ p t k kʷ/, and they are for the most part inherited directly from the parent language without any changes. Sometimes they have arisen from unaspirated stops coming into contact with an /h/ or other voiceless fricative. They are always strongly aspirated, even when unstressed, and Khulls has no counterpart of Thaoa's Grassman-like shift that deleted aspiration when two aspirated stops occurred in consecutive syllables. Of the five aspirated stops, three can stand alone as words without an additional vowel: /p pʷ kʷ/. /pʷ/ and /kʷ/ generally come from the parent language /pu/ and /ku/, but can have other sources, while the plain non-labialized /p/ arises from the sequence /ək/, which changed as follows: ək > ŭk > ŭkp > ŭp > p. Thus Khulls historically shares with Poswa a preponderance of bilabial consonants, especially the series /p pʷ ṗ ṗʷ/.

The voiceless aspirated labialized velar stop /kʷ/ is also very common, far more so than Khulls' parent language (see Gold language), because of the many sound changes that lead to it. Some seem impossible, such as /səkū/ > kʷō "bed", but are explained by the fact that labialization always dominates in a disharmonious cluster. The sequence was sək > sʷk > sʷkʷ > hʷkʷ > kʷ. Note that this happened too late for the labialization to rescue the /ū/ from slipping down to /ō/.

Ejective Stops

  • Ejective stops, also known as glottalized stops, are also inherited from the parent language. There are five: /ṗʷ ṗ ṭ ḳ ḳʷ/, assuming the marginal /ʔ/ is not itself considered glottalized. Sequences of /ʔ/ plus an aspirated stop have never produced a glottalized stop, so there are few diachronic paths besides direct inheritance to produce them, and they are thus somewhat rarer than aspirated stops. They lose their glottalization when unstressed, which gives Khulls the unusual trait of having plain voiceless stops only as an allophone of a somewhat rare series, while strongly aspirated stops predominate even in unstressed syllables. This is a similar system to that of the parent language, Diʕìləs; but Diʕìlas had only one glottalized stop: /ḳ/, which means that /ḳ/ is the most common glottalized stop in Khulls and that for the most part the other four are derived from what were once mere allophones of /ḳ/. There are no paths leading from /ḳ/ to /ṭ/, however, so /ṭ/ is very rare outside of loanwords.
  • The commonest native source of /ṭ/ is from sequences of /ʔd/ in Gold; that is, /d/ after a high tone. However, most words in which one would expect to see /ṭ/ from this particular sound change have since been analogized and thus have /r/ or /l/, the otherwise expected reflex of /d/. Although /ʔb/ similarly changed to /ṗ/, this did not happen often because in proto-Khulls /b/ could only occur word-initially or after a falling tone.
  • Of the five glottalized stops, three can stand alone as words without an additional vowel: /ṗ ṗʷ ḳʷ/. The above sound changes applying to /p/ also appear here with /ṗ/.

Voiced Stops

  • The voiced stops are also five: /bʷ b d ġ ġʷ/. The labialized voiced stops /bʷ and ġʷ/ are common because they arose from labialized nasal consonants. The plain voiced stops are rare, occuring mostly after nasals, and could very nearly be also considered allophones of the nasals were it not for a few words with geminate nasals. Alternatively, they could be considered allophones respectively of /ʕʷ r g/ (note that /g/ is a fricative) after nasals, even though there are also a small number of words with clusters of nasals plus these consonants, because such clusters only occur across morpheme boundaries. Nevertheless, voiced stops, mostly /b/, do occasionally occur in places other than after a nasal. E.g. Lobexon, a placename, where the /b/ arose from an earlier /mj/. The earlier name for the territory was Lomyoxon. A late sound change affecting only central Khulls also dropped the /l/ in the clusters /ml nl ŋl/, which had for a long time been allophonically [bl dl ġl], leading to a few more voiced stops. These clusters were rare, so the new voiced stops were rare, but nevertheless occurred in some common words, such as the name of the country Kʷoxʷudas ("land given up to God"; cognate to Babakiam Ketufumuas). This sound change did not apply to syllabic nasals, however, so the cluster /ŋl/ still occurred almost entirely in loanwords, and /ġ/ was thus still the most marginal of all of the marginal consonants even in late central Khulls.

The Glottal Stop

  • The glottal stop /ʔ/ is not normally considered phonemic, because it occurs only after high-tone vowels when not before certain consonants. It occasionally arises from /ʕ/ or even the cluster /ʕʕ/, because in the parent language an /ʕ/ after a vowel only occurred when that vowel's tone was high, and because /ʕ/ is deleted in unstressed syllables (which includes anything immediately after a stressed syllable) these /ʕ/'s would be deleted, leaving a hiatus, allowing the allophonic /ʔ/ that accompanies high vowels to reappear. As said, though, there is no letter for /ʔ/ in the Khulls alphabet because it is considered to be part of the preceding vowel; /ài/ implies [àʔi].

Affricates

  • The only affricates in the language are /č ǯ/, and they are fairly rare, occurring primarily in place of velars before the vowel /e/. However, it was not the /e/ that caused this shift; generally it is the /e/ itself that results from the environment, having come from a sequence such as /kia/. These would normally be grouped with the aspirated and voiced stops, respectively, if not for their many shared characteristics. However, the coronal fricatives /s z š ž/ are pronounced as affricates by most speakers when occuring after an allophonic glottal stop; and /s š/ are affricates in this environment for all speakers.

Fricatives

The system of fricatives closely parallels that of the stops, but there are two series (voiced and voiceless) instead of three; there are no labial fricatives, and there is a glottal series corresponding to no stop series. It is tempting to align the glottal fricatives with the labial stops, since each covers a gap in the other, but there is no special relationship involved between these two groups of consonants. Note that the glottals, even labialized ones , do not cause preceding final /n/ to become /m/.

Voiceless fricatives

  • The full list of voiceless fricatives is /s š x xʷ h hʷ/. /h hʷ/ are much rarer than /x xʷ/, but are definitely not marginal phonemes as there exist many minimal pairs in many different environments between the velars and glottals. /s/ and /š/ can explicitly be syllabic, and even stand alone: s means "sleep", and š means "bomb". /xʷ/ and /hʷ/ often occur in syllabic-like positions, but when joined by vowels in compounds they do not remain syllabic, so this is not considered a phonemic contrast. There are no significant allophones of any of these fricatives, either in voicing or point-of-articulation assimilation. However, when preceded by high tone vowels, /s/ and /š/ become /ts/ and /tš/, although this is properly a property of the preceding vowel rather than the fricative, because this is the glottal stop.
  • Note that in the Gold language, the affricate pronunciation extended to closed syllables such as those found in the very common genitive/possessive form of a word, but that in Khulls, these syllables have changed to mid-tones, and therefore now pronounce the /s/ as a true [s] rather than the Gold language's [ts].

Voiced fricatives

  • The full list of voiced fricatives is /z ž g gʷ ʕ ʕʷ/. The voiced alveolar fricative /z/ occurs in only one word, z "to injure, hurt" and its derivatives, but there are a few unrelated words which developed a /z/ and were reanalyzed as compounds containing the morpheme z. Although this was originally a syllabic /z/, the lack of contrast with any other /z/ led to a loss of syllabicity. As for /ž/, there was never a syllabic /ž/ to begin with. Thus, there are no voiced counterparts of the syllabic voiceless fricatives /s/ and /š/. Because pharyngealization is pronounced only in stressed syllables, /ʕʷ/ becomes [w] in unstressed syllables and is Khulls' first choice for spelling any foreign word with /w/. /ʕ/ simply becomes silent in unstressed syllables. The voiced fricatives /g gʷ/ are pronounced as [ġ ġʷ] (that is, they become stops) after a high tone, which means that almost all [ġ] in Khulls is actually /g/ rather than the marginal independent /ġ/ phoneme. The other fricatives have no significant allophones, and do not even assimilate in voicing to neighboring voiceless sounds.
  • Note that for most of Khulls' history, /ġʷ/ was rare, and it is only the recent change of /ŋʷ/ to /ġʷ/ that made it become common. Thus, for most of Khulls' history, the fact that the fairly common fricative /gʷ/ had [ġʷ] as an allophone did not cause any ambiguity; it is only in the last few hundred years that the two phonemes came to partially merge. Also, this shift did not happen in the Moonshine dialect.

Nasals

Nasals are very common, but there are only three of them: /m n ŋ/. Labialized nasals became voiced stops unconditionally and those are no longer perceived by the speakers as simple allophones of the nasals. Palatalized nasals also become stops in all dialects but Moonshine, and this palatalization itself was usually discarded. Hence "bottle" becomes "in a bottle; bottled" (but proto-Moonshine myâ). Indeed nasals have no significant allophones and can occur in any position within a word, and even be syllabic. If a syllabic nasal touches an open vowel than an epenthetic simple nasal is added: e.g. lŏṁ "womb, uterus" is pronounced /lŏmṁ/.

Liquids

There are only two liquids in Khulls, /l/ and /r/. The /r/ in this case is a flap, not a trill, and cannot be doubled because it arises only from an earlier /d/ which itself could not be doubled. /l/ comes from both the original inherited /l/ and from /d/ in certain positions. /l/ can be syllabic and is much more common than /r/. In descendants of Khulls, /r/ is often eliminated entirely and the sound spelled "r" is usually a uvular approximant deriving from Khulls /ʕʷ/.

The sound /j/, usually spelled /y/, is arguably phonemic in Khulls, although it has traditionally been grouped with the vowels, since unlike all other consonants, it can only occur before the high vowels /i/ and /u/ (but note the persistence of [j] allophonically in sequences like /ae/, /oe/, etc).

Sources of marginal phonemes

Marginal phonemes have arisen from smoothing of consonant clusters. For example, the rare voiced alveolar stop d can come from earlier sequences of -bn- or -nl- (though the syllabic nasals did not trigger this change).

NOTES on phonology

Some consonants are far rarer than others historically, but they have gained at the expense of the "major" consonants due to the extremely short average wordlength of Khulls and its love of extreme compounding. For example ṗĕhʷ "writer, author", composed of two rare consonants and one fairly rare vowel, with each phoneme being an independent morpheme. The etymology is "eye, front" + ĕ "to lick" + "soldier". Someone who licks things frequently makes a lot of marks, and someone who makes a lot of marks is a writer or artist; meanwhile the word that originally meant soldier came to signify an adult male and then just an adult, hence becoming a career marker.

It is rare for vowel-initial words to begin with the à or ā tones, since both of those would normally have changed into â due to the automatic insertion of ʕ several thousand years earlier. Exceptions, like àpo, are generally due to restoration of nouns from previous verb-only stems.

The scarcity of /j/

The palatal approximant /j/ (always spelled y) became increasingly rare as Khulls evolved. By the year 4700, in the standard dialect of Khulls, /y/ could only occur after /p/, /ṗ/, /l/, or a hiatus (or in word-initial position). This came about due to a sound change affecting all Khulls dialects in which /y/ after any non-labial consonant created a new palatal consonant, and these soon became postalveolar. Thus, for example, /hy xy sy šy/ all became /š/. Later, in all dialects other than Moonshine, /my by ñ/ became /b b ǯ/. Then, in all dialects other than Moonshine, /y/ was deleted after a labialized consonant: that is, kʷyê became kʷê. Yet a further change also affecting all non-Moonshine dialects then eliminated /y/ before all non-high vowels. Thus, in standard Khulls, /y/ can only occur after /p ṗ l 0/ and before /i/ or /u/. These are generally not considered to be palatalized consonants because of the vowel dependency and because historically, it was possible for /y/ to occur after a labialized consonant (and still was possible in contemporary Moonshine) and speakers perceived the labialization as being a property of the consonant but the /y/ as occurring after that consonant.

Since early /i/ collapsed to /y/ before another vowel, some pseudo-prefixes exist such as (again using the standard dialect) b which can mean either "coast, beach, shore" or "human hair". But these can only appear before words beginning with vowels, which are not very common in standard Khulls. Still, examples of such compounds abound, such as čê "pants", a compound of "foot" and ê "tunnel, tube". (Since this word originally contained two accented vowels in a row, the first was "suppressed" and therefore it became possible later on for it to collapse into a /y/.)

Sound rules

Consonant clusters are resolved by a series of rules.

  1. Aspirates dominate ejectives: thus p "teacher" + ḳā "building" combine into "school". On the other hand, "eye" + "rainfall, runoff" combine into "tears, overflow" which also begins with /p/, not /ṗ/. Thus ejectives, a lready rare, have become rarer as time has passed; any word beginning in /k/ or /ḳ/ will change it to /p/, not /ṗ/.
  2. Labials dominate dorsals, and in general, consonants dominate any consonants further back than them. So labials dominate coronals, coronals dominate velars, and velars dominate glottals. There are many exceptions to this rule, such as kʷt > kʷ, but these are due to subsequent shifts. In this case, tʷ > kʷ.
    Words beginning with any of /k ḳ kʷ ḳʷ ġʷ gʷ g ŋ x xʷ h hʷ ʕʷ ∅/ have spawned variants beginning with /p/. Since this /p/ means "teacher", it is commonly found that the new words have a more specific meaning relating to education, but semantic shift has made this not always so.
    Labials and coronals that meet up go with the consonant on the right. e.g. + tăni = tăni "nose". That is, the ṗ completely disappears. However this is actually related to a shift that happened when the ṗ was still a morpheme ending in ḳ, thus the rule goes back to an earlier coronal vs dorsal fight. Similarly, b + = "pine bottom". However this rule does not apply in words like yṅṭ "camel's eye" or ŋ̇k "teacher of clams"; i.e. it doesnt apply after syllabic nasals.
  3. Labialization dominates nonlabialization.

Patterns of phoneme distribution

Khulls has a large phonology, but many of its phonemes are little used or are used in restricted ways. For example, Khulls learners being taught a word such as ṭébu would recognize immediately that the word is likely of foreign origin or a nonce creation of the teacher, as it looks strikingly different from any words in the known native vocabulary. Although all four of its phonemes are well-known in Khulls, in this word they are used abnormally: /ṭ/ does not occur word-initially except in loanwords; /é/ generally appears word-finally, and never appears in loanwords; /b/ is mostly restricted to the edges of words; and unstressed /u/ is very rare in any position other than immediately following a labialized consonant.

Sandhi

A fex examples of sandhi on consonants do exist. For examp,e, the syllabic labials assimilate to the PoA of a precedding nasal. e.g. lṅ "wing" (nominative) becomes lṅṭ in the accusative, not *lṅṗ. However, in the Khulls script, it is still spelled with a , because the sequence /ṅṗ/ cannot exist. Note that this /ṗ/ originally was /ḳ/, but disappeared after vowels, thus being pronounced in the present day language only after consonants. The reason Khulls picks /ṗ/ and not /ṭ/ or /ḳ/ as its true form is because before it disappeared, it had changed to /ṗ/ after /u/, and that was retained as a bare form in words such as "eye" and its derivatives, whereas no bare forms with /ṭ/ or /ḳ/ survive.

Aspiration spreading

Not true sandhi, but a few very old sound changes are still respected. For example, ḳēs "mud" + ḳî "spill, puddle, accident" becomes ḳēki "mud puddle", with the s and combining into just a simple k. The sound change that caused this happened several thousand years ago, when the s was [h], but is still respected even though [h] is no longer an allophone of /s/ even before a dorsal consonant. The resulting aspirated /k/ is not palatalized to /š/ because that particular sound change happened before the sound change that created the new /k/.

Likewise tēs "spice, fire" + "red pepper" = tēnta, from an ancient change of [hn] > [nh] > [ntʰ]. (The /t/ is still aspirated today in Khulls, like all other unmarked voiceless stops.) Note that in ḳēki, the accented vowel is long, but despite being spelled ē, it is short in tēnta because it is in a closed syllable.

Likewise, previously existing final consonants sometimes reappear in compounds. ḳò "knife" + "hole" becomes ḳòḳi "knife hole, stab wound" because the word for knife previously ended in a ḳ and there was a historical sound change of ḳg > ḳ. (kg became ḳ also, not k.) However, this process has not spread by analogy to words ending in this tone that did not originally end in ḳ.

Spelling

Khulls adopts few loans, but it faces a lot of languages, so among the loans that it does take are a very diverse set of words. Khulls natively uses an alphabet, not a syllabary. It uses its letters for /h/, /ʔ/, and /ʕ/ to spell syllabic consonants in other languages. /h/ spells aspirates and fricatives (that is, /hs/ implies /ṡ/, /hp/ implies syllabic /p/), /ʔ/ spells ejectives (that is, /ʔṭ/ implies syllabic ejective /ṭ), and /ʕ/ spells voiced sounds (that is, /ʕr/ implies syllabic /r/).

Labialization

There is no regular process of labialization active in the language lately, but it occurs in derivatives of some words. For example, the instrumental case of hìni[3] is hĭġʷ. In the Gold language from which Khulls developed, it was possible to form plurals of nouns by making the initial consonant voiced and then labializing it. In Gold, there was less confusion because labialized consonants such as /lʷ nʷ ŋʷ/ etc still existed. This process is not really active in Khulls, and the suppletive plurals that have survived have mostly taken on the sense of collective nouns such as gʷan "vineyard; a place where grapes grow" from tan "grape". Note that words beginning in /k/ or /ḳ/ will change to /gʷ/, not /ġʷ/, because Khulls has a sound change /ġʷ/ > /gʷ/, with the new /ġʷ/ arising from a different cluster. Thus kăḳo "young tree" becomes gʷăḳo "tree nursery".

p  ṗ  m  b  t  n  d  l  r  s  š  z  ž  č  ǯ  y  k  ḳ  ŋ  ġ  x  g  h  ʕ   
pʷ ṗʷ bʷ bʷ kʷ gʷ gʷ ʕʷ ġʷ hʷ hʷ ʕʷ ʕʷ kʷ gʷ ʕʷ kʷ ḳʷ ġʷ ġʷ xʷ gʷ hʷ ʕʷ

/č/ may sometimes become ḳʷ instead of kʷ, but this is in fossilized words (/č/ can come from either ti- or ḳi-.)

Dialects

Some northern outlying dialects of Khulls changed the coarticulated labial-velar consonants /k͡p ḳ͡ṗ gb/ into plain velars /k ḳ ġ/ instead of the normal Khulls result of /p ṗ b/. For these speakers, the alphabet thus begins with /k/ rather than /p/. Standard Khulls has borrowed words from these dialects, but these loans do not include the single-consonant morphemes /k ḳ ġ/, which would have the same meanings as the native /p ṗ b/ since the dialects have not been separate long enough to evolve much of a gap in semantics. Since this shift is the source of almost all /ṗ/, these dialects have no /ṗ/ except for loans from mainstream Khulls and as an "allophone" of ḳ after /m/. Although almost all words with /ṗ/, and most with /b/, in mainstream Khulls in fact have that one morpheme or b in the word, it is not always seen as a compound, and novel coinages like ġī "beaver" do exist.

The Sound Change Champions

  • šôn "bathroom", from siʕiʕaʕun
  • p "teacher", from hək

Sound changes' influence on vocabulary

Because of the sound changes, and the fact that many words even in the parent language were very short, many homophones exist. Many words for basic concepts are thus compounds. Nearly all speakers retain knowledge of the meanings of the components of the compounds, however, so the pieces can still be combined with other morphemes to form other compounds. For example, in the standard dialect, the commonest word for water in general is , a compound of "lake, pond" and ō "water". Thus a word for "lake water, fresh water" came to mean water in general. But although ō is not normally used by itself, it still appears in other compounds such as ṗō "eye water, tears, crying". Note that in the Moonshine dialect, the sound change /jo/ > /e/ did not happen, so early Moonshine still preserved the more transparent original form *pyō.

Because many basic words are compounds, and there are many possible formulations to choose from, some words for basic concepts have more than one standard form. Although all speakers understand , in some dialects, particularly further north, the commonest word for water in general came to be pʷū instead, a compound of pṡ "water" and ō "water"; in other words, it was originally an emphatic form meaning "water water" or "true water".

However, even here there is a difference in meaning between the two components; although both pṡ and ō mean "water", pṡ is primarily used for the sense of water in motion, particularly a running stream or an artificial water slide, whereas ō was used primarily for still water, particularly captured water in an artificial container such as a cup or bowl.

Collapse of initial consonant clusters

Although Khulls allows clusters in initial position, many clusters that arose in the parent language were smoothed out into single consonants (often labialized). The history is often complex; for example, Gold səs dŭdi "sleep flower" (similar to the opium poppy) becomes Khulls kʷŏri, as the initial syllable of the parent language form changed using the path səsd > sədh > sət > sʷt > hʷt > tʷ > kʷ. A disappearing schwa nearly always leads to a labialized consonant; the only exceptions are the small number of words in which it instead created a syllabic consonant. Note that in this word, metathesis occurs twice, first with /h/ (which causes the -d- to become -t-), and later with /hʷ/ (which causes the -t- to become kʷ-). These sound changes were separated by several thousand years, but are considered to be part of the same process. The reason that the word did not immediately change to kʷŏri in the first stage of the shift is that at the time, the second step of the sound change was being blocked by the intervening schwa.

Kʷŏri is cognate to Babakiam sevi, since Babakiam kept the schwa vowel intact but lost all three internal consonants. Note that the labiovelar stop in this word arose much later than the shift of /u/ > /o/ that occurred after all nonlabialized consonants; most sequences of a labialized consonant followed by a short /o/ are due to sound changes like this.

The word for a flower in general is lŏri; thus, Khulls speakers think of kʷŏri as an English speaker might think of learning that a "pable" is a kind of table.

Any word prefixed with "sleep" behaves as if it were prefixed with "human; soldier", and thus these two morphemes have come to share meanings when used in word-initial position. They do not merge in other positions.

Consonant-based gender system

Khulls inherited the consonant-based gender system of the Gold language. The genders are:

Conso	Applies to
-----	----
 p	Pregnant women; couples
 ʕʷ	Babies
 t	Adult men and sometimes boys
 r	Boys
 l	Young children 
 n	Girls and young women
 m	Adult women
 s	Epicene (groups of humans of all genders and ages)
 ʕ	Neuter (nonliving things)

Note that unlike its contemporary neighbors Andanese and Babakiam (and their descendants Pabappa and Poswa), Khulls usually distinguishes between men and boys. In the other languages, men and boys are grouped into the same gender as if there were no important difference between them, whereas women and girls are always carefully distinguished. Khulls has also revived the distinct gender for babies, which is unisex, despite having merged it early in history with the /p/ gender for pregnant women and couples.

Additionally, it is not common in Khulls to casually refer to an adult male with one of the pronouns for young boys, even in a friendly manner. Instead, it is more common to hear boys referred to as men, particularly in terms of praise. Khulls speakers historically believed that this showed that their culture appreciated men's natural masculine powers whereas the peoples around them seemed to be ashamed of their male population and wished to keep them forever as boys.

Orthography

See Khulls script.

Khulls in herited the Gold language alphabet.

Notes

  1. I had this word before I knew Japanese ichigo, so even subconscious borrowing is not possible here.
  2. Note that kʷyâ is from kʷī with an infix of -(a)ʕ- which both changes the vowel and pharyngealizes it.
  3. sic; not *xìni