Icecap

Icecap Moonshine is a highly divergent language spoken around the year 6843 in cold climates famous for its small root vocabulary, compact morphology, and wide gap between male and female speech registers. When men are allowed to speak at all, they use a much more difficult speech register than women do, and when women speak to men, they use a speech register that omits crucial information, so men have to listen closely and think quickly whenever a woman gives them a new chore to do.

The first Moonshine speakers arose in the year 3948, and committed the Great Conspiracy, forever abolishing all male social power structures and spreading their revolution to foreign nations as well. The Moonshines prospered in their radical new society for about 150 years, whereupon a traditional male army invaded and crushed the Moonshine empire. Nevertheless, the winners of the war were unable to occupy Moonshine territory, and the Moonshines became even more feministic as they retracted into supreme isolation for the next three thousand years.

Moonshine women are much taller than their men, and it soon became unnecessary to apply social pressure to force men down to the bottom of the society; female superiority was seen as the only possible natural order, and few men even contemplated fighting back.

Although there have been other societies in which female power was even more unfairly stacked against men, the Icecap Moonshine language is notable even by comparison to these other societies for the great extent to which the social way of life has become entrenched in their language.

Scratchpad

See ht tp://w ww.frat hwiki.co m/inde x.p hp?tit le=Ice cap&old id=14 7218#Scra tchp ad for removed ideas.

Noun classifier suffixes

14:04, 18 August 2022 (PDT)

Because the noun classifier suffixes were formed already in Gold, it is likely that Leaper and Moonshine inherited them, and they may have a system similar to Play's setup, or perhaps something resembling Play's but also working the animate person/gender markers (which have a separate origin) into the system by making them classifiers as well.

Inherited noun classifier suffixes (Gold/Leaper forms given)

Here, the vowels are assumed to all be low-tone and unstressed, regardless of their origin. Many noun classifiers begin with /g/, which is then dropped after any consonant stem except a nasal.

• -ḳa, for objects in buildings. If even Play had a problem with this affix being two syllables instead of one, and thus not becoming a classifier, in Leaper/Moonshine the /g/ would simply be taken as an error for /ʕ/ and dleeted. Therefore this classifier is in fact valid after all.
• -Ϙ, a silent classifier indicating a human whose gender and other information is made clear by the word stem, and therefore needs no additional classifier suffix.
• -ra, the otherwise expected epicene human classifier suffix, cognate to Play -(t)a, from earlier Gold -da. Not originally a classifier, and still not one in Play.

• There may be a homophonous -ra, spelled differently in Moonshine only, that means a young boy. However this requires that Moonshine scholars were aware of the dual origins of this suffix even after the two had merged for thousands of years.

• -pɜ, possibly denoting adult women or females in general.

Gender will probably be taken over by classifiers as well, but there may be two sets: one set for people, and another for handheld objects (or perhaps objects in general) that they possess. This second set of endings replaces the classifier of the free form rather than attaching to it; this is unlike the behavior in other languages such as Play, and probably also unlike the parent language Gold.

Use of symbols

It is possible to represent the supersufficient early Moonshine alphabet in Roman letters.

The symbol ♈ can be used to mark feminine property. It is silent but by convention changes the pronunciation of certain following consonants, making /l s n/ into syllabics, making "the weakest x" still be pronounced, and turning /ʔ/ into /p/. The symbol is used here in part because of its resemblance to a uterus (which was used as an ideogram) and in part because of its resemblance to the gamma symbol /ɣ/. There may still be a use for ♀ but it shows up weak in some fonts.

The symbol ♃ represents the inherited "boy" gender which is almost certainly a generic human gender for both sexes and all ages by the time of proto-Moonshine. See above, as this is just homophonous with /ra/ and may not exist.

The symbol ¤ represents a former /x/ phoneme that is usually silent, and most often marks the locative case. Two symbols may be in use in Moonshine as well.

Masculine symbols

• 💮 used to mark masculine gender and several other things. Alternate forms: ⁕ ⚘

• Γ A reminder to men to pronounce any adjacent /x/ as /g/ (that is, IPA /ɣ/).
• Π A reminder to men to pronounce any adjacent /f/ as /p/.
• Ω A reminder to men to pronounce any preceding /a/ as /u/.
• ϵ A reminder to men to pronounce a short schwa-like vowel instead of silence. Possibly also still converts /i/ > /a/.

Other symbols

• ¤¥ marks essive case. Pronunciation depends on the word, and the actual pronunciation might be added too.
• ¤₩ marks instrumental case. see above for pronuncation notes

Question marker

05:18, 31 July 2022 (PDT)

Early Proto-MS likely has at least tīs, and perhaps also kṡ, in common with Play as question markers.

Goals

These are plans for a situation in which I never actually create the Moonshine language, but still write about it from the perspective of speakers of other languages.

Homophonous consonants

11:57, 18 August 2022 (PDT)

One early design goal of Moonshine that never got realized even in the 1994 draft was that it be a language that "works in written but not its spoken form". This could be created by constructing a large inventory of "virtual consonants" that in fact resolve to only about 15 to 20 distinct phonemes, but which in the written language count for much more. The speakers would simply accept this, as they would know nothing else. It would mean that written language could be considerably more compact than spoken language, but even in spoken language, certain homophones would be in common use. These would be at the phonemic level rather than the word level; that is, there could be four different m letters, and not merely four homophonous words all containing /m/.

Some of these consonants may even just be arbitrary semantic splits, where the scribes made up a new glyph and used it only in words with a certain semantic scope.

The vowel inventory would be much more difficult to handle if it had virtual sounds as well, so it is perhaps best to keep it at three vowels in both the real and virtual configurations.

Gendered possession markers

06:44, 5 July 2022 (PDT)

Inherited from MRCA ĭkə, Leaper/proto-Moonshine had a morpheme that indicated property belonging to a woman, and this particle usually did not add any syllables to its base word because of the sound changes that had taken place. Thus in a sense there were two stems for every word: a free form and a "her" form. This was likely mistaken for a classifier suffix at some point after the split of Play and Leaper. If Play had retained the affix, it would have essentially just become a vowel lengthener except after a schwa in which case it would have done nothing.

The Leaper language may not have even had 1P/2P possession markers; if it did, they are presumably the same as Play's "self" and "non-self" markers, /p/ and /s/. Play does not have a generic 3rd person possession marker, but this may be a loss from Gold, as Play evolved into its classical state by constantly distancing itself from its Lava Bed ancestry.

Male property

There was originally a similar suffix denoting male property, from MRCA ŋùni, and this contracted as well, but always added at least one syllable and sometimes two to the roots it was applied to. It may be that the Moonshines simply do not use this affix, or if they do, they see it as no longer part of the grammar and therefore require an additional morpheme to indicate something owned by a man. This would likely have existed even at the Leaper stage and the speakers of both languages would have simply accepted it as an asymmetric feature of their language just as Spanish speakers accept del:de la.

It is also possible that /ŋùni/ comes to be seen as a pejorative, such that e.g. chair-ŋùni means toilet. This morpheme would still be required when used to indicate the belongings of a man, meaning that "his chair" and "his toilet" would be the same word. It is not clear whether the /ìkə/ morpheme would also evolve in the opposite direction, or if it would simply stay the same by continuing to indicate both objects that are "fit for a woman" and an object that is owned by an individual woman named in the sentence. An earlier draft of Moonshine did have such a difference, but the proto-language has taken a firmer shape now and it is difficult to imagine how the distinction could arise in the way it did.

It is very difficult to see how Leaper would evolve IE-style gender after losing its classifier prefixes, and therefore the original goal of having inherently feminine words is likely unattainable in Moonshine. The ikə/ŋuni system is the only gender that Leaper could have had, and therefore even over thousands of years also the only gender that Moonshine can have.

Pronunciation

Even in the egalitarian Leaper society, the grammar was biased towards females in some ways. There were also strict sex differences in speech habits where it could be argued that men and women were about equal after all. Moonshine may change this towards a situation in which women have the advantage in every single instance where there is a difference.

In Middlesex, men and women had different pronunciations for both vowels and consonants and the system was very complicated, such that even adults made frequent mistakes. They wrote with a script in which the correct pronunciation for both sexes was always clear, and therefore the people who made mistakes were the people who did not know how to spell, rather than people who had mislearned the rules.

Leaper did not have this feature, and Moonshine only acquired it when they absorbed the Middlesex speakers as an underclass, whom they quickly elevated to full citizenship even as they moved towards speaking Moonshine only. They borrowed the twin pronunciations directly from Middlesex, treating the Moonshine original pronunciations as feminine and therefore creating whole new phonemes just for men.

The gendered differences in pronunciation had helped keep Middlesex's phonology stable, as any change would need to be picked up simultaneously by both sexes. The consonants in which both genders had the same pronunciation all along were the ones most likely to change. When Moonshine takes over, the women will be in control, and therefore the rate of change will probably not be greatly affected by the two speech registers.

Consonant inventories from older writeups

Go-carts

                      VOICELESS              VOICELESS      LIQUIDS AND
                        STOPS         NASALS   FRICS         LATERALS
Rounded labials:      ¦     pʷ ¦ bʷ ¦     mʷ ¦ fʷ     ¦     ¦     w  ¦    ¦    ¦    ¦
Neutral labials:      ¦ p      ¦ b  ¦ m   ṁ  ¦ f   ṗ  ¦     ¦        ¦    ¦    ¦    ¦
Spread labials:       ¦ pʲ     ¦    ¦ mʲ     ¦        ¦     ¦        ¦    ¦    ¦    ¦
Lower dentals:        ¦     ṭʷ ¦    ¦     ṇʷ ¦        ¦     ¦        ¦    ¦    ¦    ¦
Upper dentals:        ¦     ṭ  ¦    ¦     ṇ  ¦     ṣ  ¦     ¦     ḷ  ¦    ¦    ¦    ¦
Alveolars:            ¦ t      ¦ d  ¦ n      ¦ s      ¦  z  ¦ l      ¦ r  ¦ c  ¦ ʒ  ¦
Postalveolars:        ¦ č      ¦ ǯ  ¦ ň      ¦ š      ¦  ž  ¦ ʎ      ¦ ř  ¦    ¦    ¦
Palatals:             ¦ ć      ¦    ¦ ń      ¦ ś      ¦     ¦ y      ¦    ¦    ¦    ¦
Neutral velars:       ¦     k  ¦ ġ  ¦     ŋ  ¦        ¦     ¦ r̄      ¦    ¦    ¦    ¦
Labiovelars:          ¦ kʷ     ¦ ġʷ ¦ ŋʷ     ¦ hʷ     ¦  gʷ ¦        ¦    ¦    ¦    ¦

Helicopter

Rounded labials:        p   b   m   ḟ   w
Spread labials:         pʲ  bʲ  mʲ  f              (ṗ   ṁ)
Dentals:                            ṣ       ḷ      (ṭ   ṇ)
Alveolars:              t   d   n   s   z   l   r  (c   ʒ)
Postalveolars:          č   ǯ   ň   š   ž   ʎ   ř
Palatals:               kʲ  ġʲ  ń   ś   y
Velars:                 k   ġ   ŋ   h   g   r̄

Lack of contact with Poswa

Since the maturation date of Icecap Moonshine is now 6800 AD, it cannot have participated in Poswa's later shifts, and perhaps the old consonants ṗ ṃ ṭ ṇ need to be thrown out altogether.

Comparison to Play

The proto-Moonshine language (long before Icecap) and Leaper could both be described as a defective version of Play in that morphological processes that work perfectly in Play have many exceptions, have often split into separate paradigms with no clear semantic differences, and apply only to closed classes of words. And yet, Moonshine/Leaper has very few innovations of its own. Even its elaborate case system, for example, is just the same as Play's with the addition of a dative case.

Therefore in a sense Moonshine is just as difficult to learn as Play, but much less impressive, and also much less stable.

Possible inaccessibility

06:02, 5 May 2022 (PDT)

It is possible that Moonshine will simply never be realized, as it requires a great deal of work, and even a well-ordered draft of Moonshine can be easily upset if something needs to be changed to harmonize the language with the more important languages, especially Play and the shared MRCA, Gold.

At the very least, I will be working on the tropical languages (Middlesex, Andanese, and Play), and then on Gold (and Leaper), before committing to anything in Moonshine.

It is even possible that Middlesex will effectively be what Moonshine was meant to be, and that the Moonshines gave up their own language instead of teaching it to the Middlesex speakers, but using Middlesex as "Moonshine" would require very rapid changes.

Size of phoneme inventory

If created, the phonology will likely be much smaller than in the early drafts of Moonshine, where I believed I could increase efficiency by having a very large phonology and loose syllable structure. A CVC language with the original Moonshine phoneme inventory would have had about 24,000 syllables, but using the inventory this way would require that nearly all syllables be closed, which was never my goal even then. Thus C₂ in Moonshine cannot be much more salient than C₂ in a language such as Play; even saying that Moonshine might have twice as many CVC syllables in running speech compared to Play might be already too much. Therefore the relevant comparison is not CVC but just CV, where the original Moonshine would have about 600 syllables and Play has about 50. Taking the natural log of both numbers shows that Play only needs to be spoken about 50% more quickly than Moonshine to match its efficiency, and therefore that the original model is mathematically unsound. Andanese would need to be spoken as twice the speed of Moonshine if assuming that /sia/ etc are counted as two syllables in Andanese and that Moonshine is allowed a CVC structure for every third syllable.

Stereotypes of foreign languages

See User:Soap/scratchpad#Cultural_divides_in_the_tropics and the latter half of Play language/history for now.

Giri tile and toy block scripts

14:52, 4 April 2022 (PDT)

The canonical consonant inventory of Gĭri, the Middlesex and early Moonshine children's speech register, was:

Bilabials:                   m   b
Coronals:                t   n   l
Palatals:                        y
Velars:                  kʷ  ŋ  (Ø)

And the vowels were /a i u/. Thus there were 27 syllables, representable in various means based on the number three. Giri had no writing systems of its own because the speakers never lived independently; if they wrote at all, they used their parents' alphabets. Nonetheless, when the proto-Moonshine people met the Play people, they were amazed at how young Play children adeptly wrote their language with only two letters, stacked in two or even three dimensions to form shapes of arbitrary height, width, and depth. (This was an elaboration of a Late Andanese toy block script.) For political reasons, the Moonshines actually preferred the Play language to their own, but they soon set about creating toy block scripts for their children to play with as well.

Pronouns

All pronouns in early Moonshine are derived from verbal embedding, of either type. Thus, they are an open class, and the person marker is that of the highest social status participant in the sentence. Thus "your servant", etc for 1st person when the listener is superior, and a simple pronoun like "I, speaking" for 1st person superior. The rules are complicated and it could thus be still said that early Moonshine, like its ancestor, has no fixed set of pronouns.

Later, diachronics and semantic shifts create true pronouns in Moonshine, but person is marked indirectly, as an emergent result of the gender markers and other things. These pronouns may then help fill the role of speech discourse markers (that is, gender marking for the speaker and listener) alongside other morphemes. This would mean that every sentence would need to have a pronoun in it even if it's a third person event.

Volition and obedience

05:50, 2 April 2022 (PDT)

Volition should probably be folded into the obedience morphemes, rather than using the old system where obedience morphemes were for males only and were marked by suffixes, while volition applied to all situations.

The involuntary obedience morpheme requires a 3rd person argument, as the Moonshines consider it impolite to accuse the listener of forcefulness and 1st person situations could be handled as they come up, possibly even through possession. Thus Moonshine is not like Poswa, which allows involuntary obedience markers associated with forcing agents of all three persons.

The voluntary obedience morpheme can take either a 2nd or 3rd person argument (possibly 1st person in some situations). The 2nd person feminine external agent is a zero morph if the speaker is male, and possibly also if the speaker is female. Third-person marking is the same as it is for the involuntary paradigm. Remember that there are also morphemes that mark the gender of the speaker and of the listener.

It is possible that semantic shift will turn the volition distinction into a person marker, creating the need for a new volition distinction, but this hypothetical later stage cannot be built without first creating the original stage.

Doctorate Moonshine

05:50, 2 April 2022 (PDT)

Rather than evolving Doctorate Moonshine as a highly artificial IAL, it could be that this is already the state of the language by 6800 AD and that it came about naturally. This would require a number of astonishing coincidences, however, even if assuming that some of Doctorate's IAL-like features were not actually very useful.

New genitive infix

16:19, 2 March 2022 (PST)

The new sound change list opens up the possibility for a genitive infix appearing around 4800 AD, such that e.g. ŋàsiḳa "door" and its genitive ŋasiḳas could appear as ŋàsʲk and ŋàsʲikˠ. There may need to be a way to stop the palatalization from staining the following consonant in the resulting cluster. The stabilization of the accent on the root is not a problem however.

Possible use of Play block scripts

16:19, 2 March 2022 (PST)

The Police in Play territory wanted to write Moonshine using the Play block script or an adaptation of it, at least for pleasure. They admired the small children they saw writing messages in a script that their own adults could scarcely make sense of, let alone read, but which the Play children all seemingly understood without a moment's pause. Moonshine had too many phonemes to use the Play script directly, they realized, but the Police hoped that they could make a new version of it, whether it be more complex to handle the larger phonology of Moonshine or less complex to show their admiration for the superiority of the Play language's adaptability to artistic means of writing.

Diachronics and other information

Feminist Compact Imperial (3948) to Icecap Moonshine (~6800)

The expansive inherited phonology simplified quickly during the settlement period as the proto-Moonshine speakers passed through territory inhabited by speakers of Play and other languages with similarly small inventories.

Early shifts (Feminist Compact Imperial to Police)

The term Police is used instead of proto-Moonshine here, as it could be unclear whether the proto-Moonshine language was spoken in 3958 (when it was identical with Leaper) or around 4300. Properly the proto-Moonshine language should be considered to begin in 3958 because even then they considered it a separate language from Leaper for political reasons.

1. All high rising tones became ordinary long tones.
2. All pharyngealized vowels became ordinary low (mid) tones. The stress became weak.
3. After a high tone, the voiceless fricatives x xʷ shifted to k kʷ. Note that this shift is subtly different from Leaper's because they did not also become fortis.
4. The rare labialized glottal fricative hʷ (sometimes spelled ħʷ for distinctness) shifted to a voiceless bilabial fricative f.
5. Labialization was lost in the syllable coda; pʷ mʷ kʷ ŋʷ xʷ gʷ became p m k ŋ x g. All of these codas occurred only after the short low tone. Thus, these new codas joined the existing codas /s l n/ in not occurring after high tones (but /s l n/ could also occur after long tones). Since the codas could not occur after high tones, the high tone came to be seen as if it were a coda by itself, /ʔ/.

Middlesex influence

The lower classes of Moonshine society at first had their own language, which had different phonologies for men and women. In Middlesex society, the genders were equal, but in Moonshine society, women quickly took control. Thus, if any sound changes were taken from Middlesex speech registers, they would either affect men (and perhaps women pronouncing "men's words"), or affect women and then pass on to men, since men could not preserve a distinction their women (who spoke the "correct" form) did not have.

These changes might be able to violate traditional diachronic sequencing principles because the Middlesex language persisted for a few generations, perhaps for hundreds of years, and most sound changes would be conscious on the part of the speakers. And yet, at the same time, the sound changes were almost all simultaneous, because they all came from the same source and applied unconditionally. Therefore, the separation here is artificial. Also, female stubbornness could create situations in which different groups of women had different pronunciations for the same sounds, both for themselves and for their men.

Also note that the sound change list for Middlesex ends at the year 3370, allowing another thousand years for additional changes to occur between the two speech registers. However, it is likely that few new differences appeared because the living standards of the Crystals, the primary speakers of Middlesex, had changed.

Possible sound changes are:

MSX Vowel changes

1. Women came to spell any ʷa sequence as o, but did not change their pronunciation. Men also came to spell /ʷa/ as /o/ but came to pronounce it ʷu. (Middlesex has a plain /u/, but this vowel would likely have been rounded in Middlesex but unrounded in proto-Moonshine, so /ʷu/ is what the listeners would hear. Thus, the rare inherited plain /u/ did not shift to /ʷu/.)

   It is possible that Moonshine teachers would spell this with a sequence that could be represented here as aΩ or aω, but because this new symbol would only occur after /a/, the sequence would be still seen as a single vowel.

2. Women and men both shifted any inherited plain i to ʲi, except after a labialized consonant. That is, the Middlesex plain /i/ was heard as rounded and lowered, whether or not it bordered a labialized consonant.
3. The remaining plain i, which occurred only after labialized consonants, now came to be spelled e by both men and women. Women continued to pronounce this as a somewhat centralized IPA [i], but men shifted it to a. (Not /ʲa/, because even though proto-Moonshine allowed simultaneous labialization and palatalization, Middlesex did not.) It did not merge with the Moonshine schwa, which was further back because it corresponded to Leaper's /o/ vowel, and was rounded when occurring after a labialized consonant. Nonetheless, this new vowel spelled /e/ occurred only after labialized consonants, where the inherited schwa was rare.

Thus Moonshine retained its four-vowel inventory, but spelled it with six vowels. The pronunciations were:

 SCRIPT       a    e    i    o    u    ɜ
 WOMEN        a   ʷi   ʲi   ʷa   ʷu    ɜ
 MEN          a   ʷa   ʲi   ʷu   ʷu    ɜ

Note that the labializations in the /e/ column are present because this vowel occurred only after labialized consonants in inherited vocabulary. In loans, this may not have applied, but such loans would have been irregular and more likely borrowed with either [ʲi] or [a] instead of using a rare plain [i].

The /u/ column above implies that Moonshine could not use this vowel except after a labialized consonant; this may be a matter of analysis. But, importantly, Moonshine did not do Leaper's shifts of /ɜu/ > /ū/ and of /ĭʕʷ/ > /û~ŭ/.

This system was complicated, but nonetheless, simpler than Middlesex in important ways:

1. Both men and women always agreed on consonant coarticulations. By contrast, in Middlesex, palatalization and labialization could be undone by certain consonants, and men had a palatalized sequence [ʲa] where women had a plain [i].
2. Women and men only had different vowel pronunciations when these vowels occurred after a labialized consonant. (Even so, a tiny number of exceptions could be pulled from the Middlesex substrate if /e/ was borrowed as [ʲi ~ a] instead of homogenizing on [ʲi] or [a].)
3. Furthermore, in Middlesex, sometimes sequences such as [ʷa] appeared that were not spelled as "o" and thus were pronounced identically by both sexes. Since the Moonshines were analogizing their inherited system to the borrowed one, they spelled all female [ʷa] as "o" and therefore did not create new exceptions to the rule. And likewise for the other situations.

Because the three script vowels e o u are all restricted to occurring after labialized consonants, labialization can be omitted in Romanization. In the native Moonshine scripts, it was nonetheless retained, as it did not greatly affect the appearance of the words.

1. Then, the vowel sequence əi əu shifted to ī ū, without staining a preceding consonant. These sequences never occurred in a closed syllable, even stressed, so all such syllables were open. (The collapse of syllabic consonants had not happened yet.) This change was not directly forced by Middlesex, but was influenced by the fact that Middlesex did not have an independent schwa and also did not have many closing diphthongs. This change may have lagged the other changes by a few generations.

Sporadic and reversed shifts

1. Because Middlesex had a rare /ʷi/ sequence of its own, which men pronounced [i], the /ʷi~ʷa/ paradigm could be disrupted in some words. This bare [i] was also the sound that Middlesex women used for what men pronounced as [ʲa]. Thus, just as Moonshine had its [ʷi > ʷa > ʷu] masculinization chain, Middlesex had [ʷi > i > ʲa]. But note that in Moonshine, neither men nor women had a bare [i] and therefore that this would have been heard as either a schwa [ɜ] or the sequence /ʲi/.

MSX consonant changes

1. Men came to pronounce f as b unconditionally.
2. Both men and women shifted š ž to s z unconditionally.
3. Women came to pronounce g gʷ as x xʷ unconditionally. They used the symbols that Leaper was using for its own /x xʷ/, which was cognate to Moonshine's /h hʷ/. (Leaper's /h hʷ/ were mostly cognate to Moonshine's /p f/, but this was more complicated.) The unified symbol can be represented as ɣ.

Sporadic and reversed shifts

1. A sporadic shift of kʷ to p, or perhaps to pʷ, could occur in a few words, for both men and women.
2. Women may have shifted inherited ṭ to kʷ, but note that the Middlesex dental "ṭ" sound is not particularly close to the pre-proto-Moonshine alveolar ejective "ṭ". This shift depends on the speakers hearing two unrelated "ṭ" sounds as identical. The same applies to a possible opposite shift for men.

Idiolectal variation

1. Because the only voiced:voiceless fricative pair was /s z/, both men and women experimented with new fashions in pronouncing fricatives. Some Middlesex women had been pronouncing /f/ as [v] all along, and this change came back into fashion. Meanwhile, Middlesex men actually had two phonemes, /b v/, to the women's /f/, and the choice of /b/ was assigned by teachers; however, some teachers instead chose /v/ and therefore [v] was a valid realization of this phoneme for men as well.
2. Likewise the glottal fricative /h/ could become [ʕ] or [Ø], and even /s/ could become [z]. These latter two changes created phonological collisions, so they were more stigmatized and less widespread than the others.
3. Likewise, a second /x/ existed in Middlesex, which was pronounced [x] by both sexes, and therefore in some loans from Middlesex, a "flexible" [g] appeared where the source language had had the "hard" [x]. This had not happened in Middlesex itself because the two /x/'s were always distinct in spelling.

Cultural reflections

There were a few syllables containing /ʷi/ and /ʲu/ in the Moonshine branch, whereas Leaper only had /ʷi/.

Importantly, the vowel and diphthong inventory was very similar to that of contemporary Play, which also had /a i u ə/ and the diphthongs /ai au əi əu/. The differences were that Moonshine also had a long schwa /ə̄/ but lacked Play's distinction between long vowels and vowel sequences. That is, Play had both /aa/ and /ā/, but Moonshine allowed only /ā/.

Prenasals existed in word-initial position, also unlike Khulls. e.g. /mpʷà/ "house" vs Khulls pà.

Contact with Giri

The Middlesex children's speech register, Gĭri, had a consonant inventory of /m b t n l y kʷ ŋ Ø/ and a vowel inventory of /a i u/. Importantly, only two vowels per word could appear, and at most three consonants. All syllables were CV, for a total of 27 possible syllables, just shy of the 30 of Late Andanese, which also influenced Moonshine's children.

Imitating the slow speech rhythm of young children, Moonshine teachers may have used separate consonant symbols for eight of the nine the Giri consonants (all but /y/), or perhaps for all but /y/ and /b/ since there was not a symbol for /b/ in Moonshine itself at the time.

Development of feminine speech style

1. The script vowel e became delabialized, both for women and for men. Thus men now merged /e/ and /a/ into one sound, but women kept them distinct.
2. The script vowel u also became delabialized, from a phonological point of view, as it was phonetically equivalent to /ʷu/.
3. Lastly, the script vowel o (perhaps already spelled aΩ) became the last one to drop its labialization. Because women now had control of the education system, the female speech register was considered the only correct one, and the recognition of /o/ as a separate phoneme disappeared, while /e/ remained.
4. The sequence ʷɜ shifted to ɜ, removing labialization from the language altogether. (This usually corresponded to Leaper's /ʷo/.)
5. The above changes also shifted the children's speech register's kʷ phoneme into k (not /p/).
6. Then b > p.

Thus the vowel inventory now was

SCRIPT       a    e    i   (o)   u    ɜ
WOMEN        a    i   ʲi   (a)   u    ɜ
MEN          a    a₂  ʲi   (u₂)  u    ɜ

Words that had been spelled with /o/ were now spelled with /a/, and pronounced [a] by both men and women, with a few rare exceptions confined mostly to words specifically relating to masculinity and used mostly by men with other men. Men may have also had a rare surface [i] phoneme not dependent on palatalization, but it would not be spelled as /e/ because the rule about the separate pronunciations was still in effect.

Consonant table

The language was spelled with a highly redundant consonant inventory at this time, because they indicated gender, age (because of the children's register), and relevant etymological information that would help learners inflect words. The full inventory was

VOLATILE CONSONANTS
SCRIPT       f   ɣ   ṭ       
WOMEN        f   x   k       
MEN          p   g   t       

STABLE CONSONANTS
SCRIPT       k   s   z   š   l   r   m   n   ŋ   x̣   ḥ   p
WOMEN        k   s   z   s   l   r   m   n   ŋ   x   h   p
MEN          k   s   z   s   l   r   m   n   ŋ   x   h   p

Therefore the voiceless bilabial stop /p/ could be substituted for any /f/ to show masculinity, but the sound was rare. The voiced /g/ sound was weakly articulated and not distinguished from hiatus or silence. Therefore, in this case, an /x/ spelled as ɣ could be removed rather than inserted to mark masculinity in a given word.

It is possible that Moonshine teachers used a digraph like xΓ for what is here spelled as ɣ, again in the belief that male speakers were inserting an extra sound into their words rather than pronouncing a single sound differently. For convenience, this can also be spelled here as xg, but the letter "g" would be foreign to women entirely since they did not have that as a phoneme. Both men and women would still pronounce a plain x as /x/.

Likewise the men's /p/ could be spelled as fΠ. The /b/ > /p/, which was not native to Moonshine, could be p₂ since it was not gendered.

Mixing speech registers to show gender of words

At this point, women began to use the male speech register to indicate that an object was male, either literally or figuratively. Men could also use the women's speech register to indicate feminine objects, but only in certain social situations such as with their loved ones or when a woman asked them to repeat speech.

These genders were originally of the property type, that is "a chair for women", and not of the traditional identity type, but the logic was soon extended, because animate nouns could not be possessed. Therefore the first stage had nouns like "chair for women", the second stage had nouns like "body for women", and the third stage had nouns like "teacher for women", which meant not someone who teaches female students, but rather a female teacher.

The possession system was still symmetric, unlike the alignment detailed below where men are grouped with handheld objects and therefore women can own men, but men cannot own either women or other men.

A new glyph was invented to modify the pronunciation of the women's words containing /i a/ into /a₂ u₂/ to show this change.

First deletion of vowels

This is around 4300, when the speakers will still in contact with Play.

1. Accented schwas (ɜ, not /ə/) took on the quality of the vowel in the "real" stressed syllable (usually the following one). Even men did this, so they had a proper [i] again. This did NOT entail labialization of any preceding consonants.

   It is not clear whether this new unstressed men's [i] will then immediately shift to the low vowel, or if it will be retained as [i]. This is important for the shift below.

2. Presumably, the unaccented ɜ also shifted.
3. The high vowels ʲi i u (in women's speech) were deleted to ʲ Ø ʷ in unstressed position, assuming labialization had hung on as a phonetic detail. If not it was just /ʲ Ø Ø/. Men's corresponding shift was from ʲi aₓ u to ʲ ϵ ʷ, however, where the symbol ϵ here spells a very short vowel rather than a consonant; the use of yet another symbol here is to prevent confusion with /e ɜ ə/ all of which were different. So now words had different syllable counts for men and for women. The true /a/ sound did not shift.

   This may mean that unaccented /ɜ/ > /Ø/ if it can be analogized from words in which there had been a plain /i/, but note that this plain /i/ was rare.

   The new symbol may be partly or entirely changed out for ʕ later on, but note that /ʕ/ was originally used to mark syllabic consonants, and so if it does get changed, it's a spelling reform rather than a sound change.

4. Therefore the only unstressed vowels were /a/ and the much rarer /ɜ/ vowel, which was lower and further back than that of Play, but which now moved towards a true schwa. There were no unstressed high vowels at all.

Note also that [u] had existed only after historically labialized consonants, so this shift did not create any new labialized consonants and the labialization may have been allophonically present all along. If there had been Middlesex loans with bare [u], they would have been handled in various ways as by this time the Middlesex language was primarily in writing.

The men's extra [a] sounds were no longer spelled by females, so the new letter ʕ was created to keep track of them and they were considered to be consonants. It is possible that labialization survived long enough for a distinct pronunciation to have been retained, so that they were not truly [a]. They may have even become schwas.

1. It is possible that the two-vowel rule becomes applied to the whole language here, with the inherited /ɜ/ vowel always being an echo vowel of the tonic syllable, perhaps even if it is /a/.
2. The earlier deletions likely also cause ai to become a followed by palatalization.

The retained /ʕ/ sound may have been identified as an allophone of /g/, of /ɜ/, or of both at once, meaning that the schwa vowel /ɜ/ would be considered a consonant, although one with a continuant quality like nasals.

Discourse marker

The above sound change creates a discord marker for gender, where men have an extra morpheme, spelled as ʕ, found unpredictably in words of all types indicating that the speaker is male. Its use is most likely also influenced by the gender of the listener, as men would not be eager to produce the extra phonemes among their own kind. On the other hand, because they would be required to pronounce the extra sounds whenever they were merely in the presence of women (not only when speaking to one), the inflated words would be the unmarked forms and therefore they might pronounce the sounds at all times, and perhaps would speed up their speech tempo somewhat when women were not listening in.

It is possible, perhaps with influence from Lava Bed languages and perhaps Play, that the discord marker turns into a "poly-syncretic" morpheme (that is, fusional and encording more than one underlying morpheme at a time) which encodes the gender of the speaker and the listener, which would mean that the morpheme would need four forms instead of two, and perhaps that men would be made to use it among themselves after all. This would eliminate the need for gendered verb marking whenever the participants in the verb were 1P/2P, and perhaps even the need to mark person on the verbs at all (this would also affect Leaper, however).

If Lava Bed influence is stronger still, it could come about that men use the discord morphemes not just to mark the gender of 1P and 2P, but also for the gender of the words themselves (which could be seen as a 3P marker), and of other participants in the sentence, which could be seen as 4P and even beyond if inanimate objects acquire gender.

This is quite different from the typical Lava Bed system, and its reliance on deleted phonemes makes it unlikely that it could evolve smoothly from the inherited grammar. But in common with Lava Beds is that the inserted morpheme can appear anywhere within a word, and can appear within any type of word.

Note also that consonants were probably still distinct for men and for women, but that this was at parity, with no "extra" consonants, so would not really be a discourse marker in itself.

Possible new orthography

09:50, 9 October 2022 (PDT)

It is possible that the Moonshine teachers introduced glyphs that could be represented as ʔ ʕ around this time, standing for tone markers that came to be seen as consonants. The glottal stop marked any preceding vowel as high tone, and therefore could not occur word-initially, but it may be that it could occur after a consonant, as an orthographic innovation marking a place where a vowel could be inserted in certain grammatical alternations.

Possession markers

Poswob Rare (talk) 11:07, 25 October 2022 (PDT)

There will be some irregular nouns, but it is possible that a generic possession marker table can be created:

                              SPEAKER
                          Female     Male
  
  Free                    A          A
  1st person                   (B-₁) 
  2nd person                    B-s
  3rd person Female (I)   A-ɣʷ       A-ɣʷ
  3rd person Female (II)  A-r        A-rɛ
  3rd person Male         A-ra       A-ra

The -₁ and -s above have the same forms for male and female speakers, but -ɣʷ and -ra do not always, even though the affixes themselves are identical. This is because of Moonshine still retaining some relics of the earlier fusional grammar. Put another way, speech registers by this time will be sufficient to mark the gender of the speaker and listener, and there is no need for a grammatically overt means of doing it.

The r is cognate to Play's -Ø- and the ɣʷ died out in Play comes from and would have been retained as p if Play had not evolved three other affixes with /p/. The reason for there being two female rows is because in certain environments the original /p/ was changed to /r/ (p > bʲ > d > r), which caused the vowels to also have different reflexes.

PRELIMINARY CHANGES

This section may be little used, as many of the changes feed and bleed each other. For example, there is no need to construct fortis consonants if they simply remain as clusters, and since this would allow the language to keep its closed syllables, vowel shifts that had depended on fortis consonants can simply be reassigned to those involving closed syllables.

Police (c. 4300) to Ice Cream in a Bowl (c. 5547)

1. The clusters mm nn ŋŋ shifted to mˠ nˠ ŋˠ and lengthened the preceding vowel. Since there were as of yet no other fortis nasals, this was nondistinctive but still contrasted at least in some contexts with simple nasals after a long vowel (bcase of ān, etc).
2. The syllabic consonants ṁ ṅ ŋ̇ ḷ , when bordered by vowels in either direction, became fortis mˠ nˠ ŋˠ lˠ as well. THIS IMPLIES A DIFFERENCE FROM LEAPER. The syllabic ṡ then became a normal s, not fortis.

   This means that the vowel + g + consonant declension simply becomes vowel + consonant.

3. Final CVC syllables lost their stress as the words became toneless.
4. Unstressed syllable-final s shifted to h. This has repercussions later on because it leads some vowels to disappear before /s/ and not when in an open syllable.
5. All remaining syllabic consonants (bounded by consonants) became normal.
6. The consonant w in the coda (corresponding to Leaper /ʕʷ/) here became an ordinary u, forming a diphthong. Note that this consonant could occur after diphthongs itself, but only in newly formed compounds, and these words may have been "rescued" by changing the preceding diphthong elements from /j w/ to /žʲ gʷ/.

Loss of unstressed vowels

All four vowels /a i u ɜ/ disappeared in unstressed position, except when propped up by certain conditions explained below. The list speaks of two (perhaps three) dummy vowels.

1. In unstressed position, the vowels a ɜ disappeared to Ø when bordering one of /g h gʷ hʷ/ in either direction. This created clusters.

   This shift may have happened simultaneously on both sides of the consonant, so that for example, /-taga/ > /tg/, or it may have preserved the final vowel, which nonetheless disappears later. The distinction will only be meaningful if words appear in which there is in fact a difference in final outcome.

2. Where the voiced velar fricative g collided with any consonant from either side, it made that consonant into a fortis consonant, marked by Cˠ (perhaps /Cʰ/ for easier typing and display). At the immediately following stage, the entire inventory of consonants could be fortified this way, and thus they were not considered new phonemes but still as clusters. The same thing happened with gʷ except that the consonant was labialized in addition.

   Note that this shift even applies to sequences like /ng/, which became /nˠ/.

3. Where the voiceless velar fricative h collided, it made the preceding consonant voiceless and aspirated. This meant nothing for stops; other consonants are dealt with below. These consonants also became fortis, and can be spelled Cˠ or as Cʰ.
4. All LABIALIZED consonants became FORTIS in addition, while retaining labialization.
5. Any unaccented ɜ shifted to Ø .
6. The diphthongs ai au (of any origin) now became āʲ āʷ, with the coarticulations cleaving onto a following consonant, if any. In open syllables they continued to be spelled with the symbols for inherited /ai au/.

   This may be a problem, however, as it suggests that even the plain vowels /i u/ should stain following consonants.

   It is possible that proto-Moonshine retains a distinction between Leaper's /ē/ and /é/, and likewise, instead of merging both into the same long tone. It would need to be that PMS allowed superheavy syllables like /paiʔ/ or that it allowed tone contrasts with semivowel coda.

7. In stressed position, due to Play influence, the vowels ɜ̆ ɜ̀ ɜ̄ shifted to the color of the vowel in the next syllable, and to ă à ā in monosyllables. They did NOT acquire preceding glides /ʲ ʷ/.

   This shift began under Play influence around 4100 AD but did not become phonemic right away. It is likely that the shift didn't take place all at once, and that allophones of /ɜ/ that can here be spelled as ɪ ʊ existed. It is also possible that this shift can be discarded entirely if it comes about that the coarticulations of a following consonant account for it.

8. The diphthongs ɜi ɜu became āʲ āʷ when an /a/ followed in the next syllable, and ī ū otherwise (even in monosyllables). Any iu (that is, /i/ + /ʕʷ/) also became ū.
9. Unaccented WORD-FINAL a ʲi ʷu (not just unstressed) shifted to Ø ʲ ʷ unconditionally, creating more clusters. This happened even after fortis consonants.

   Now, every word ended with a consonant unless the final syllable was stressed. These shifts meant that for example, /-kas -ka/ had become /-kˠ -k/.

10. Unstressed ʲi u (and /i/ if it existed) now shifted to ʲ ʷ unless they occurred in one of Cˠ_ _Cˠ C_Cʲ where clusters and immediately posttonic consonants also count as fortis consonants. (But they are not actually fortis, so they don't behave as part of the fortis class later on.)
11. Then, unaccented a (and any other vowels) became Ø (sometimes spelled ʕ), unless it occurred in one of the following environments: Cˠ_ _Cˠ C_Cʲ where clusters & IPC's also count as fortis consonants. The dummy vowels ɛ ɔ can be used here to spell /aCʲ aCʷ/, and ɐ to spell /CʷaC/ and /CCa/. These were mere allophones of /a/, however.

At this stage the language had a problem with unstressed closed syllables, and deletion of /ž žʲ/ (which were distinct) would not solve this problem because Play requires a distinction that would have forced Leaper and Moonshine to have the same sound for the instances in which it would be most distinct. That is, Moonshine *-az is a mistake, as Moonshine only ever had /žʲ/ in that position (a separate, much rarer plain /ž/ existed in other morphemes). Nonetheless, deletion of /ž žʲ/ could happen for its own sake, without solving the closed syllable problem.

The only word-final open syllables are those that are stressed, and it is most likely that even here, only the grave and long tones survived, since the others would have been indistinct from unstressed syllables. Tone may need to be taken into account in the rules above. Even if stress shifts back to the root in these vowel-final words, it would still mean that all such words would have low tones on the stressed syllable. Also note that the grave tone is effectively vowel + /ʔ/ and can be considered to end in a consonant, and even the long tone is a heavy syllable.

Monosyllabic stage

Now, the Moonshine language had only three vowels, but an abundance of consonants, including a lenis:fortis distinction and consonants that were palatalized, labialized, or both. All labialized consonants were fortis, but not all fortis consonants were labialized, so there were five possible modes of consonant: plain, palatalized, fortis, fortis labialized, and palatalized labialized. The fortis articulation was velarization for the frontal consonants but not for the velars.

1. Before a vowel, the labial approximant w shifted to a uvular approximant r̄.
2. Doubled consonants (and fortis) simplified to singles and caused the tone of the preceding vowel to become high (à or á).

   This is the best (and perhaps only) opportunity for Icecap Moonshine to gain the ability to use the tones freely instead of having only one tonic syllable per word.If it doesnt happen here, the only viable path is that whole phrases were compressed in speech and came to behave like single words.

3. Any consonant before a nasal disappeared and lengthened the preceding vowel. If the sound had been voiceless, it caused the tone of the preceding vowel to become high rising (á). If it had been voiced, it caused the tone of the preceding vowel to become high falling (ā).

   It might be best to have these going to short high tones (grave accents) instead, and to only lenite fricatives and stops. This would eliminate the high rising tone entirely and limit the use of the long falling tone, while also creating a huge new supply of CV words from old CVC's.

   Possibly /ntn/ > /nn/. Or ʔnn.

4. The clusters nlh nlk shifted to ŋh ŋk.
5. The palatalized labials pʲ mʲ fʲ became the labiodentals fˠ ṃˠ f in all positions. Then, a shift from ṃˠ to a plain bilabial mˠ spread throughout the territory, slowly driving out the new phoneme.
6. The alveolar flap r became l before any stop.

Ice Cream in a Bowl (c. 5547) to Classical Cartwheel (~6800)

1. The palatalized velars kʲ hʲ ŋʲ merged with the palatalized coronals tʲ sʲ nʲ into č š ň. This also includes all such consonants before front vowels, since these were always preceded by /ʲ/. If /gʲ/ existed, it became /ž/.
2. A single, non-fortis labial, when following any posttonic consonant and before a vowel, generated /ʷ/ and then disappeared.
3. The labialized alveolars tʷ sʷ nʷ became the dentals ṭ ṣ ṇ in all positions. This shift is distinct from a similar one that happens later when new labialized alveolars are created.
4. All high rising tones (á etc) merged into either ā or à when not before a nasal.
5. Then lʲ rʲ became y r.

   This may mean that rž and rʲ have distinct reflexes. Consider whether the cluster /rž/ would behave uniquely here or whether it would simply behave like its constituent parts.

   Consider also that there will briefly exist a /lʲʷ/.

6. Unstressed a in a closed syllable became ʲi (or possibly just /i/) before a palatal consonant. That is, /ɛ/ > /ʲi/. There was no corresponding change before a labialized coda.
7. All coda palatalization was lost (and they became alveolars, not velars.)

New palatalization

1. The labialized postalveolar consonants čʷ ǯʷ šʷ žʷ ňʷ delabialized.
2. The palatal consonants č ǯ š ž ň became c ʒ s z n in all positions.
3. The labialized alveolar approximant lʷ shifted to w. Possibly /rʷ/ (the "dark" one) also shifted.
4. Nasals disappeared before a fricative.
5. All nasal+stop combinations become voiced; prenasals mp nt nc ňč ŋk shifted to plain voiced stops b d ʒ ǯ ġ. Here, the coarticulations are omitted for conciseness. ALL VOICED STOPS WERE FORTIS.
6. Posttonic low short a became Ø if an adjacent syllable had a long vowel (even if that long vowel was not the tonic vowel).
7. The velar stops k ġ became labialized to kʷ ġʷ before any labial consonant. Possibly a few other clusters like /ks/ also survived. It is possible also that kl > kʷl > kw, distinct from kʷ.
8. The velars k ġ h g ŋ were fronted to postalveolars č ǯ š ž ň. These behaved as palatalized consonants, cʲ ʒʲ sʲ zʲ nʲ, and any remaining fortification was lost. It is possible that the fortis velars had all become labiovelars by this point.

Second unstressed vowel deletion

Assuming the distinction of lenis/fortis in absolute final position holds on from the beginning, it becomes important here as it determines whether an unstressed vowel is lost or not. Thus for example /nk/ could alternate with /nak/ in some noun to form the genitive.

Also note that new prenasal sequences are created here, e.g. /map/ > /mp/, and these do not become voiced stops or even voiced prenasals. It is possible that the prenasals in the earlier shift never lost their nasal element, and thus there would be /mb mp/ but no free /b/, but this would be unusual.

1. All unstressed vowels weakened: a i u all disappeared to Ø, unless protected by a fortis consonant. Unlike the similar shift about 2,000 years earlier, there was no protection afforded by bordering another collapsing vowel. As before, most unstressed /i/ was really /ʲi/ and thus this created new palatalized consonants. For the most part, the consonants did not stain each other.
2. Nevertheless, palatalization sometimes skipped across a syllable boundary. For example, /hip/ > /hʲp/ > /pʲ/ is possible. It may be that /h/ was the only consonant in the entire language that violated the rule against consonants staining each other.
3. All unstressed syllables with "tone" now became plain. These had arisen from secondary stress in compounds. Thus unstressed ā ī ū à ì ù became a i i a i i.
4. Any remaining fortis consonants not touching the tonic syllable became plain.

Sequence changes

1. The sequences ph th shifted to pw tw.
2. Coronals followed by the labial glide /w/, including those generated above, became dental. Thus for example the verb ending nw became ṇw.