From FrathWiki
Jump to: navigation, search

If this page gets too long, see User:Soap/scratchpad/old and User:Soap/history. I may move old content from here to my website, wipe the page here, and start over with a much shorter page.


MRCA consonant stem extensions

17:34, 5 October 2022 (PDT)

Many of these words actually ended in consonants until just before the maturation date of the MRCA, and thus were historically not ambiguous. They remained in use for a short time after the loss of the final consonants, and then disappeared except where fossilized into longer words.

Note that Dreamlandic and PSL retain final nasals as /n/.


The standalone word à can mean:

  1. fish; to swim, to move powerfully forward (earlier am)
  2. cloud; to hide
  3. parent; to carry, nurse (earlier am)

The standalone word can mean:

  1. clitoris
  2. germ, infectious agent
  3. angel
  4. east-to-west; following the sun
  5. (component of the word for sleep)
  6. in front of; the front part of
  7. to smile, befriend; neighbor; to discover
  8. to love; to play with; live in symbiosis; emotions
  9. to run away

Most of the words with /g/ cannot be used as verb extensions however.

The words for "road" and "gold; metal" primordially had two syllables, and so even though they are sometimes reflected as in some daughters, they would not have been used as verb extensions either.


The standalone word ì can mean:

  1. tail
  2. to hide away (from)
  3. deed

It was also once used as a passivizing prefix (eat ---> meat).

The standalone word can mean:

  1. to jump, leap, try to escape
  2. to mess up, defile, make dirty; to contaminate
  3. large expanse of land; country, nation
  4. sea bird


The standalone word ù can mean:

  1. flowerbud
  2. to drink

It can also mean "to expel" but is labeled as "preverbal". The standalone word can mean:

  1. a woman's skirt

other vowels

The standalone word ò can mean:

  1. to blush, turn pink. This was earlier à, and is a true primordial /a/ with no following consonant, but because it was intransitive, it nearly always followed a prefix that colored the vowel to /o/.
  2. part of the morpheme for umbilical cord

The standalone word ə̀ can mean:

  1. child (earlier ən)
  2. (used in words for reaching and pointing motions)

The standalone word can mean:

  1. limestone

The standalone word gə̀ can mean:

  1. thorn
  2. present tense particle in PSL/Dreamlandic; treated as if from bare /ə̀/. PSL has only this, but Dreamlandic also adds /à/

The standalone word can mean:

  1. (component of žaipa-related words)

Number words outside Play

14:58, 5 October 2022 (PDT)

The Play language page covers the number systems of some other languages because it pulls on very old morphemes to build its words. But some words did not survive into Play or became bound morphemes.

  • àhu, a group of something, not necessarily related items. Becomes /bu/ in Play, which is homophonous with the Play plural marker, and therefore came to merge with that. (Originally the Play plural marker was just from /u/ with prothesis.)
  • ŋàhu, a variant of the above consisting of /ŋà + àhu/.
  • hàhu, meaning "fraction, portion" at least by the time of Play and derived from /hà + àhu/. It is possible that this a very early head-final compound rather than a primordial head-initial one.
  • màtə and ndĭḳa-n-, the MRCA words that became Dreamlandic's neuter and epicene plural respectively. These words were added with little thought and have no etymology, so they might be reworked.
  • If assuming that màtə is mà-t- + gə̀ (or with some other weak consonant), it could become a formula, where the first part of the word might change for dual, plural, and perhaps some others, and the second part of the word indicates something that later came to be seen as a gender marker. Therefore perhaps the /ə/ is for neuters and /ḳan/ is for epicenes.
  • , the same morpheme that means to give birth, also means "and" and can stand alone to mean "more than" or "and associated things" in some daughter languages.
  • , to add or increase. This might just be a duplicate of mà-t- "plus one", since it only appears in languages where the final /t/ would have been deleted anyway.


  • , thumb. Always CV even pre-MRCA. Survives in Play as pi but had long since become a fossilized morpheme not even recognized as being the same morpheme in the compounds where it occurred. The survival was because /wi/ was a rare sequence in Gold.
  • tʷò-l-, index finger; to point. Not cognate to tʷò "to show, expose" but quickly confused with it.
  • gà-n-, fingertip. Possibly not the original meaning, since it appears in compounds with meanings more basic than its own.


Use of classifiers on number words may be an areal feature more than a genetic one, thus reappearing in some languages which had lost classifiers and then regained classifiers of another kind. In the MAP branch at least, the classifiers would be there simply because they would be there on any other words modifying the nouns. Play loses this, and when it regains classifiers, it no longer echoes them across phrases the way Andanese does. It is possible, nonetheless, that numbers are an exception to this.

B-B compounds

It is possible, also, that some languages use B-B compounds or an equivalent to Play's B-B compounds, since number words are seen as being the identities of the words they modify. That is, three birds is both "bird (uncountable)" and "three (animals)".

Primordial currency and finance

01:07, 27 September 2022 (PDT)

The MRCA sailors had a single currency, but as the colonists split apart, some adopted different systems and the inherited words changed their meaning. Many societies abandoned the cash economy altogether. The Leapers and the Andanese, who were merchant castes in later years, mostly preserved the original words with their original meanings for thousands of years, one of the few links between the two languages.

Atomic roots and simple compounds

Atomic roots

  • mfù-p-, money; manufactured currency. Reformed in many later languages to mu- or m-.
  • ə̀tu, manufactured currency; a material not inherently valuable.
  • , debt; to owe; to lack; negative number. Primordial CV.
  • àga, gold or another precious metal.
  • pàga, money. Formed by detachment of the /p/ of the prefix and its addition to the àga word above.
  • nə̀hə, a precising morpheme appearing in some languages that pairs with /ə̀tu/.
  • kə̀-n-, a coin.
  • màta, to earn money. This word probably does not contain the /tà/ above, but is merely a coincidence.

Head-initial compounds

These words were mostly formed at the MRCA, but some were formed later in Andanese or other branches that conserved the original head-initial grammar. Head-initial compounds can still appear in Gold and Play for various reasons, but usually are not parsed as compounds because the morphology of Gold and Play is predominantly head-final.

Andanese and PSL arrived in the same area at the same time, and their languages had similar grammar. But they had mostly not been living in the same place on the islands of Laba, so it is possible that head-initial compounds will need to be coined independently in the two language families. It is possible that PSL could loan from Andanese because they were not so resistant to loans as the later Play speakers were, but even at this stage, a calque is more likely.

  • hò-ŋŏgo-n-, a bank. Used at least in Old Andanese, where it had become honŏgo ~ hŏno by 1900 AD.
  • kàma, the currency of the early Andanese society. Made up of kə̀-n- plus gàma, with the expected phonological development. The syncope happened early enough for the word to be loaned simply as /kama/ into various non-tonal languages. Unlike the Play speakers of later centuries, the PSL's and other tribes were happy to loan words from Andanese.
  • An alternative form of the word would be kṅàma, also the expected phonological development but with a coinage date a few hundred years later (after the separation of Dreamlandic).

Head-final compounds

These words are given here with the MRCA roots for comparison, but are head-final and thus can only exist in the Gold branch. They will have parallels in other languages, and in some cases the counterparts will simply swap the order of the morphemes.

  • hà kə̀-n-, a coin used in the Gold branch of the family. Gold reflex would be hāk, which is irregular. The Gold speakers (that is, the Leapers) were not in contact with the Andanese when they created this word; the use of the morpheme kə-n- in both branches is simply because that was the most common MRCA word for coin.
  • tà hà, a statement of owed money. Gold reflex is tas.

Later formations in Gold

  • gauḳan, a bank. This word did not really have an MRCA form because it relies on later Gold formations to exist.
  • miḳànʷa, to set a price. The Play word for price might be mikanuabe or just mikabe.

Areal influence

Although the Dreamlandic and PSL (Play substratum) languages are closely related, they settled in colonies thousands of miles apart, and did not contact each other for many years. Therefore the vocabulary diverged rapidly. For example, the Dreamers had no access to precious metals, and their word for gold came first to be used for a very dark metal, and then for dark objects broadly stated.[1]

Comments on irregular developments

Note that the Gold word for gold should have been *supaʕ, but it was reanalyzed twice just within a thousand years or so because of its great importance. First, the /mup-/ prefix, which was homophonous with the prefix for emotions, was changed to /mu-/, making the word at that time mʷòga instead of /mfupàga/. Then, because /mʷ/ > /m/ in all Gold, the word became mòga. This /ò/ was assumed to be vowel harmony due to the /yo-/ prefix, so it was later reformed to mə̀ga, making the Gold word for gold maʕ. A possible third reanalysis to /màga/ could take place but would have no effects in Gold or Play. Nonetheless, the /māʕ/ form survived into Play even though Play was descended from Gold, meaning that at least two forms of the word existed side by side.

Later currencies and unsorted compounds

  • The KCL currency used by the Matrixes.
  • Possibly a second use of KCL by a different society at a time in which it had come to be used for any currency forbidden to outsiders. These people were the enemies of the Matrixes and so were paying tribute to a defeated enemy only in an ironic sense.
  • STW's meal tokens.
  • STW's general purpose private currency ("asala"). Made of iron and easy to counterfeit because STW's banks tracked the accounts of all of their customers.
  • The indasi coins.
  • The Players' currency.
  • The Lava Handlers' currency.
  • The Dreamers' currency.

Dreamlandic/PSL verb conjugation oddities

01:22, 13 September 2022 (PDT)

By 500 AD, proto-Dreamlandic/PSL had already evolved the same situation that Play had 3,000 years later. Verb conjugations were marked by reduplicating vowels, with that vowel dependent on the stem of the verb, and consonants coming between the vowels in most forms to distinguish the markers from each other.

PSL and proto-DRM have slightly different verb conjugations, because PSL retains /ə/ whereas Dreamlandic merges it to /a/. This means that PSL's /iə/ conjugation is /ia/ in DRM, and DRM spreads it to primordial /a/, for which PSL just uses /i/.

Note that any verb with a 3rd person patient uses a ending, replacing an earlier fluctuation between -ka ~ -kka ~ -Ø for 1p/2p/3p agents acting on 3rd person patients.

Topic markers

Unlike Dreamlandic, PSL distinguishes between the 1>2 and 2>1 prefixes, meaning that they are true person markers, not topic markers. In Dreamlandic it happened that ni ŋi merged as ň, whereas in PSL they remained distinct and then evolved in different ways in the different branches.


The reflexives with /k/ were equivalent to 1p>3p and 2p>3p. In other words, the 3rd person isn't an argument on the verb at all. Thus these passed to /Ø/ when the others uses of /k/ did.

Basic paradigms

1A conjugation

The primary conjugation in PSL, here called 1A, uses the morphemes -l- -y- -ŋ- -p- for 1>2 2>1 3>1 3>2 respectively. The -p- is irregular, as /k/ would be expected (assuming it comes from /h/ and is cognate to Play /-s/), but since another /k/ existed in the paradigm at the time, the secondary reflex of /p/ was generalized.

The vowels in the 1A conjugation can either be /a...a/ or /i...i/. It is possible that some sort of clitic will arise, at least in PSL, to turn the "useless" final vowel into something meaningful. Note that the past tense affix still does not do this.

An alternative analysis can do away with the alphanumerics and speak of just the consonants, which will then be infixes after all, as they will go into morphemes that are mostly of the form (C)V-...-V which carry the grammatical information. This includes the dummy affix, from MRCA gə̀. PSL has only this, but Dreamlandic adds a second dummy verb from the unrelated and etymologically vacuous /gà/.

1B conjugation

1B uses the morphemes -r- -ć- -ŋ- -V:- for 1>2 2>1 3>1 3>2 respectively, where the last is a silent vowel lengthener.

4A conjugation

Note that 2 and 3 are skipped in order to allow 1A and 1B to divide based on their vowels. Therefore the 4A conjugation comes from the primordial verbal suffix -ŋa, indicating an action taken in preparation for another. Thus, for example, this generates "ask" from "know" and "cook" from "eat". It is cognate to Play's desiderative mood marker, but this modifies the verb and is not a mood marker. In bare form it is -ńa.

The 4A morphemes are -ŋilia -ŋiya? -ŋińa -ŋipia for 1>2 2>1 3>1 3>2 respectively. Note that here the full forms are given, but the vowels can still change; that is, this suffix has an /a/ version just as 1A does, and it depends on the final vowel of the preceding verb. However, the final -a is always -a.

There should be a 5A, 6A, etc but perhaps the numbering system is bad.

"Thorn" conjugation

Used in the passive only, and all endings are . Used literally to describe pain inflicted by sharp objects, particularly on the a human body, and with only limited metaphorical use. Possibly best described as a participle or even an adjective, a category which is rarely found in languages of this planet.

Primordial suffixes

The affix -ŋa

12:11, 11 September 2022 (PDT)

The MRCA affix -ŋa "to seek; to act on a desire for" has several developments. In the MRCA, it was used to construct words like "ask" from "know", because one asks when they want to know something. Another example is eat-ŋa = cook. Generally the root describes an action, but roots describing tangible objects can also take this affix.

In the daughter languages, the behavior changes:

In Play, the affix is grammaticalized and conflated with the 1st person patient prefix, which also then becomes reinterpreted as a suffix. It appears in Play's two desiderative moods (if considering the imperative not to be one as well), but cannot occur alone. Nonetheless scholars recognize the unusable bare form as etymologically desiderative.

In Late Andanese, the affix remains with essentially the same meaning, but because Andanese develops a strict head-initial morphology, verbal suffixes can no longer exist, and so it migrates to a carrier verb ha and thus the reflex is hana, which goes before the verb, not after.

In proto-Dreamlandic and the Play substrate languages, there is relatively little change in function or meaning, although the affix becomes less productive in most languages over time as shorter morphemes increasingly tend to become opaque as sound changes reduce the phoneme and syllable inventories. (The same would have happened in Andanese had the morpheme not coupled to a carrier verb.) This appears as -ma in main-branch PSL (elsewhere /na/) but reverts to /ŋ/ in most inflected forms, so it comes to be seen as a conjugation of its own (not all /m/ behaves this way).

Other suffixes

There were other suffixes like this, but they did not survive in Andanese because the language moved to head-initial strictly, and they did not survive in Play because Play grammaticalized almost every possible affix and the phoneme space was used up.

  • mu ~ mbu capable of; prone to. Always used with active verbs (that is, "prone to breaking" means that it can break something, not that it can be broken)
  • nʷo to want to. Often replaced by the simpler /ŋa/.
  • hu meaning "should", but not a true mood marker. Coupled with a carrier verb producing /mbìhu/ in pre-Andanese.
  • pə̆pa-l-, meaning to try or attempt. Despite its length, in PSL this compressed into a shorter sequence and was grammaticalized. With inverted tones, as happened in some inflected forms, this becomes homophonous with the inchoative plus a dummy morpheme /l/, and so it may be reanalyzed as such, meaning that the /l/ could then theoretically stand alone. However, since "try to" and "try to start" might be considered essentially synonymous, the inchoative could also be retained. This happened only in PSL, not in Dreamlandic, because Dreamlandic never elided any /p/.
  • lə̀ ~ lə̀gə, only in PSL. Means both "try to" and "entrap; force to do" and therefore has wide semantic range. The pə̆pa-l- suffix was reanalyzed in PSL as containing this one, but they are in fact unrelated. The "try to" meaning was imported from /pə̆pal/.

In Dreamlandic/PSL, these are verb modifiers, and go inside the person markers, and thus they are not mood markers. It is possible that they evolve into mood markers in Gold, and then disappear in Play, because Play's own mood markers come from a different source. Likewise, they may survive into early Andanese, but then disappear in the main branch of Andanese that becomes exclusively head-initial. Late Andanese at least retains /-ŋa/ and /-hu/ by coupling them to carrier verbs, producing hana and liku respectively, which go before the content verb just like the normal word order would predict them to.

The copula verb

The Dreamlandic/PSL copula was inherited from MRCA nə̆ga-l-, a long verb because it was usually not needed either derivationally or inflectionally. This changed in DRM/PSL and regular sound change brought the verb stem down to just , though this was still longer than the other affixes.

Cross-cultural eponyms

04:52, 11 September 2022 (PDT)

See also Orphaned languages of Teppala.

Rather than derive full definitions through individual morphemes, various languages could derive words for certain things from people's names, as exemplified by English guy, French renard, and so on. It is possible that Play and Andanese would translate foreign names through a cipher rather than attempting to represent the complex sounds of various languages like Leaper.

Possible examples of eponymous concepts could be

  • to approach a loyal customer asking them to buy even more, while ignoring potential new customers in the perception that they will be more difficult to convince
  • to use someone else's property because you have more need for it than they do; often causative ("X assigned Y's Z to W"); a possible quadrivalent verb
  • the "gumption" word set might be derived from a single eponym or several. Alternatively, it could just be a set of words calqued from Andanese or even native to Play, so long as they behave as a set
  • a tourist. even the richest nations produced few tourists, either coming or going, because transportation was both slow and inconvenient.

More can be found in the GUD dictionary, intended for a draft of Andanese in which such word formation is normal although the words are compounds.

Predators and sharp objects

08:59, 6 September 2022 (PDT)

Some languages have a noun class for sharp objects, especially those found in nature, which violates the animacy hierarchy, putting them above even humans. It typically does not include weapons, as these are in the handheld object noun class or in a specialized weapon noun class.

MRCA situation

The MRCA specifically separated sharp objects from all others. Moreover, the MRCA had "private nouns" as well, meaning that there were very short words such as "thorn" under the realm of the /tu-/ classifier prefix. Since the classifier prefix did not need to be repeated in the MRCA, a sentence like ŋatugeḳù would mean "the thorn cuts me" just by itself (the tone shift is part of the grammar). At this stage, the passive marker /ŋa-/ was still required, however; therefore the /-tu-/ affix was not violating animacy hierarchy. This only came about in the languages where the 1st person agent and patient markers merged and the agent role was assumed whenever the other argument was inanimate.

Situation in Late Andanese

Late Andanese preserves an ancient morpheme -tu- as a second-order classifier prefix that specifically indicates a sharp object, whether inanimate or animate, acting upon a (usually human) patient. For example, one can say

Natukuuni tuhupiihu.
The knife cuts me.[2]

This sentence is structurally similar to

Nanulakiputa nuhupiihu.
I cut the pear.[3]

This sentence therefore structurally appears to say "I cut the knife", but the /-tu-/ morpheme overrides animacy and therefore the knife is the agent even though the first morpheme in the sentence is the singular 1st person marker. (Note that in Late Andanese, this prefix is not marked for agent or patient; it simply means 1st person.)

Second-order classifiers are an innovation that appeared in Andanese and several other branches of the family, but were not present in the MRCA. Therefore, at the MRCA stage, the morpheme (already /tu/ even then) stood alone.

Late Andanese has a similar morpheme, -hu-, that behaves in exactly the same way but indicates an animal, usually but not always a predator, attacking a patient that is usually human. Since humans are above animals on the animacy hierarchy, this morpheme also violates the usual order.

The long verbs in the Andanese sentences above are accessible regardless of the noun class or even the animacy of the agent, but Andanese also retains a small class of private verbs bound to their noun class, and the -tu- and -hu- morphemes override any outer prefixes, meaning that one could say "the knife cuts me" in Andanese using a verb that would be ungrammatical if the speaker was trying to say "I cut the pear":

Natukuuni tuu.
The knife cuts me.

Here, the word for cut is just /u/, and is specifically tied to the /tu-/ prefix.

The subject can also be omitted, reducing the sentence to

It cut me!

In Andanese, the only noun classes without a set of private verbs are the inanimate objects and the humans; this is tied to the fact that humans are often manipulating inanimate objects.

The Andanese prefixes are at once verb and noun classifiers, and the short morphemes known as private verbs are a closed class. Metaphorical uses could be employed, but most speakers would simply use the longer open class of free verbs. There are no "private nouns" in Andanese.

Proto-Dreamlandic and Play substratum languages

The reflex of the MRCA word ŋatugeḳù would be nia tu ye kku assuming the original connected word was parsed as four individual words and that the required new morphology can be ignored for the sake of this comparison. (Normally, there would need to be a suffix something like /-a-ŋa/, to mark that this is a verb and then to mark that it is 3p>1p. See below.)

However, it is possible that these languages take an extreme position. First, by retaining the private nouns. Then also compressing the three syllables /tu ye kku/ into kuekku "to be cut by a thorn". And lastly by allowing it to omit the person markers because it is clear from context that the sentence has a 3rd person agent and a 1st person patient.

Thus it would be quicker to say something like "the thorn is cutting me" in Proto-Dreamlandic and Play substratum languages than to say an ordinary sentence like "I am walking". This was in common with Late Andanese, but the structure of the one-word "sentence" is different. Even so, in both languages the presence of /tu/ is what triggers the grammar structure to change to a form that is otherwise unavailable in the language.

Comparison to developments in Play

Play did not inherit private verbs or an animacy-violating classifier morpheme. A faintly similar construction exists, however, in that Play can produce verbless sentences consisting of a noun marked with a suffixal morpheme. For example, suifa means bee, and one can say

The bee got me!

This is functionally equivalent to an ergative, and therefore the same suffixal morpheme appears on the noun even when a verb is present. The difference is that while Andanese drops the noun, Play drops the verb. Thus there is no good way to translate these type of short sentences between the two languages.

For this reason, Play speakers who spoke Late Andanese would often use the Late Andanese interjections when they got hurt, and these became known even to those not fluent in Andanese. Play speakers would also drop the Andanese 1st person marker na-, making the words even shorter. (This /na-/ is cognate to Play's /-m/.)

This practice was not for the sake of efficiency but because Play speakers believed that Andanese was the language spoken by animals.

New color terms

22:18, 20 August 2022 (PDT)

The a ~ i ~ u color symbolism is ancient. It is likely that color terms were intransitive verbs from an early date as well, and that as the morphology changes in each language, the color terms change too rather than remaining as fossils. This means that in Play, they will all need infixes like /-k-/ because in Play intransitive verbs are a marked form. In some other languages, intransitive is unmarked and these words will function like adjectives.

Currently the root stock is very deficient in color terms, because most languages are supposed to get their color terms by affixing a derivational morpheme to a word for something with a characteristic color. For example, Play pamanaa "red" from pamapu "cherry". This is sufficient, but it does not mean that a language needs to rely on it.

The color terms assigned to Poswa have empty etymologies and should be discarded entirely.

Terms for people

It is possible that separate color terms existed for people and for everything else; or for animates and inanimates.

Exaggeration of skin color differences among tribes will exist, but not to the point of using white and black.

Pink and pastels

The PSL people had a term upi (MRCA gùhi) for their own skin color, which was used for pastel colors generally and for the light pink skin color more specifically. It may have originally been a broad term for pastel colors, since it does not have an /a/, and only come to be used for pink because it came to be used for people. It was not the word for white however, nor related to it. The Moonshine people have even lighter skin than the PSL's, as did the aboriginals of the area where the Moonshines settled, but the PSL's had no contact with these people and considered themselves to have the lightest skin of all known human tribes.

If this word survives into Play, it cannot be anything besides pes, which would yield pasaa with the assumed color affix on. This is in line with the association of pink with red and therefore /pasaa/ could be seen as a shade of red. It is possible for this to coexist with the unrelated Play word /pes/ "waterfall; slope to catch running water" since that word will have a classifier suffix on. On the other hand, it is possible that Play simply inherits this in the sense of pastel colors and not for human skin color.

An unrelated word in the MRCA was simply ò and meant to blush. This could be padding for some other color word, perhaps to add a red or pink hue.

Note that the predominance of light skin colors in the tropics is due to the climate of the planet, and in particular the eccentric orbit. For most of the year, the planet is much further from the sun than Earth, and therefore sunlight is very weak even at the Equator. During the short summer, the sun is far more powerful, but the climate is also very hot, so humans in the tropics tend to stay indoors, or in caves, or even in the water in order to protect themselves. Dark skin colors are more common in climates where the summer temperatures are tolerable and people remain outside, but only if a diet sufficient in vitamins can be found to last through the long dark winters that occur even in low latitudes. Put another way, skin color is more closely related to diet than to climate, although near the poles the skin color has always been very light because the cold temperatures require people to dress very warmly even in summer.


The MRCA root meant white, and găgu-n- also meant white. In Gold, these words might be assumed to be related, but they are not. In Play they would come through as ku and pu respectively, both too short to stand alone, but they could pair with something else and come to associate blue with white as both would use the /u/ vowel. As above, this was only used to describe hair color, never skin color, even for those who are extremely pale. It was also used, at least primordially, for flowers and birds.

Dark colors

Meanwhile the MRCA term gàga meant a black object; this is the same as the root word for road, but this may be merely a coincidence as many consonants merged together as /g/. PSL's reflex of this is yaćća. This term was used to describe people's hair but never skin; for dark skin the PSL's may have used terms derived from wood, or may have had atomic roots.

Play is unlikely to retain this term, and will instead create a new term based on the word for night, which is itself probably a new creation since the inherited word for night is also too small to stand alone. Currently Play has natuum "work at night" and paifaep "night" but the latter was created as an etymology for Poswa's word and probably not used in Play. Therefore Play does not have a word for night as of yet.

Yellow and blonde

Blonde hair was common and widely distributed around the world, but for a long time, only among people with light skin. Therefore the term for blonde could be used metonymically for people with light skin, or the other way around. It is possible that this is an atomic root as well, but it could be replaced in many languages.

In many languages, the word for sun provides the word for yellow. It is possible that there is no word that means yellow just by itself. Since there is more than one word for the sun, the word used to mean yellow is ìḳi and not the "fire" root. This is because the /i/ vowel carries the color symbolism.

Play almost certainly uses pipanaa, "color of lemons", as its word for yellow and does not retain an atomic root even if one could be plausibly passed down from the MRCA. The original root for lemon in the MRCA was /wìpa/, a rare example of true /wi/ that is not from consonant lenition. (Thus, the fact that the root word for sun happens to be pip is a coincidence.)

In Andanese, the sun root ìḳi will reflex as just /yi/ in Late Andanese and ì in Galà. This is presumably too short a root to function on its own, but the vowel symbolism means that it could serve as part of a compound.

Units of measure

00:25, 6 August 2022 (PDT)

Play's words might be largely calques from Andanic languages, particularly Late Andanese. However, the Andanese typically did not work with food (either in production or distribution), so farmers' traditional units could be native.

Late Andanese and friendly cultures

  • lita indicates a measure of size, or more rarely distance. It is inherently dimensional, so units measured in lita are simple, square, or cubic.
  • hahugu indicates a one-dimensional measure of distance. The morphemes here are cognate to Play paus and pūm, both still used in measure words, but not together in the same word.

  • litapi, a measure of volume, that is, cubic (feet). Note that the number is parsed as a total measure, therefore as
9 [ft^3], and not as
[9 ft]^3.

Weather, climate, and forecasting

08:31, 9 October 2022 (PDT)

Number scales were not well developed in even the most advanced and math-literate societies such as the Leapers and Andanese. Weather forecasting was mostly the interest of farmers, who often did not learn mathematics at all, and navigators, who did learn math but had little use for it outside of keeping track of their position at sea.

Contact sports

04:59, 4 August 2022 (PDT)

This is not a list of sports, but a description of how they are categorized. Hundreds of different games came and went over the years in all of the various cultures across the planet, and many of them were shared between cultures. But all of these sports can be described in terms of a set of differentiating traits.

Sports may be divided in different ways, and these categories could overlap. Different ways to categorize sports would spread across cultures as the sports themselves were rarely culturebound.

Primary categories

Team sports

Most sports are played by one team competing against another team. However, some sports rely on solo competition, and some are group team sports where players form teams but there are more than two teams competing at the same time.

The Play word for team might be the same as "party (size); for a group". Other cultures might use the word for army. Still others would use a word relating to a team of people at work.


Most contact sports are competitive, but small children and older adults often play noncompetitive sports, as they are the most likely to be of widely differing ability levels such that any game with a score tally would be inherently unfair.

The rare non-competitive contact sports played by children are mostly tests of athletic ability, often races in which the players are allowed to collide with each other and attempt to block others from reaching a goal, and therefore interference is explicitly allowed. A player who is either very large or very fast may join a team and dedicate themselves to slowing down the progress of the other team's players; because these games are not scored, they are not costing their own team any points by doing this.


There are symmetrical sports and asymmetrical sports.

For example, a sport might consist of a team of humans competing against a team of animals such as pigs or dogs. Either team can win, but their winning strategies are very different. Many large animals are naturally tame, so minimal training is required, and animals may even be trained alongside the humans they play with.

Most asymmetrical games are configured such that both sides consider the game to be fair, but there are exceptions. Typically an inherently unfair game will draw both teams to the competition by promising the weaker team a large prize, while taking little or nothing away from them if they lose.

At the extreme are games which are wholly unwinnable for the weaker team. If such a game involves a team of humans against a team of animals, the animals are always set up to win, and the humans merely enjoy the suspense of seeing how long they can hold on. For example, there is a game in which the only legal means of offense is to bite an opponent, but because this is played by a team of humans against a team of large dogs, the humans cannot win, but they can run and climb trees in an attempt to keep the dogs at bay.

Another example of an asymmetrical sport is a match of men competing against women, not simply because of inherent athletic ability but because the two sexes follow very different rules and therefore have different winning strategies. These games' rules are based on overtly sexual metaphors and are played by teenagers or young adults. The male team's goal most often involves pushing through a defending wall of females, which may or may not outnumber the males; if they do not, they will have some ability assigned to them that the males do not, such as using their whole bodies to defend instead of merely attempting to push through like the man is made to.

Objects of play

There are object-centered sports and direct contact sports. Essentially this describes whether the game involves a ball or some similar object (called vasu in Play), or whether it is just humans competing against each other.

Mixed-species play

As many large tame animals exist, many sports involve animals playing alongside humans. Trained animals are teamed up with the human players and therefore this category is different from the asymmetrical battles described above where one species fights another.

Combat sports

Each sport's rules define where unrestrained combat is allowed or not. A sport often makes use of nonviolent mock combat in which a person may feel some pain but is not at serious risk of injury, such as snowball fights, but other sports involve true combat in which the only way to end a fight is to surrender and leave the game. These sports are not simply fights, because the way to win is to accomplish some unrelated goal, often a race, but a losing team can resort to violence to stop the opponents if they realize they cannot beat their opponents to the goal.

Weapons and instruments

Similarly, players in some sports carry weapons onto the playing field. Most often, this is paired with the mock combat scenario above, and the player's "weapon" is not capable of causing serious injury, but yet is capable of disabling the other player's weapon. For example, using a soft sword one may attempt to remove the sword from an opponent's hands, and then win the game by striking the disarmed opponent in an imitation of actual sword combat.

In languages such as Late Andanese, the word for a nonviolent weapon is the same as the word for the traditional weapon it most closely resembles, but with a different classifier prefix. Late Andanese has more than one noun class for weapons, but ki- is common, and for a nonviolent weapon the most common prefix is u-. Thus a spear is kitaguta and a soft spear used in games to disarm an opponent is utaguta.

Other categories

These categories usually do not define a sport.

Division by sex

There are sports where teams are allowed to be mixed-sex, and sports where teams must be all of the same sex. In sports where the teams must be of the same sex, they are further divided by whether men are allowed to compete against women or whether the opposing team must also be of the same sex. This is separate from the symmetry criterion above because it does not define the sport.

In a culture where men and women are nearly equal in athletic ability, a male vs. female sport can still be symmetrical. Most of these sports are rituals played by very large teams, such that the entire able-bodied adult population of a town may show up to play the game. If women outnumber the men, they tend to win, but in some cultures, such as the Dreamers, men are physically strong enough to win even when outnumbered. Even in cultures where women are taller than men, men tend to have more experience with manual labor and muscle-building tasks, and are considered equals of women in the sports most popular in those cultures, but nearly all cultures have more women than men in their population.


Sports differ on whether happenstance encounters with nature must be compensated for or not. For example, if a stray animal bites a player, the entire game might be called off in some sports organizations, whereas others would simply say that avoiding stray animals is part of the game and cannot be ruled as interference.

Some cultures play a game which is no more than a race across a field filled with nonvenomous biting snakes, essentially a dynamically shifting obstacle course. Another game involves rushing through a forest filled with thorny plants towards the center, in which each contestant is allowed to choose where to start, but none of them know which path to the center has the least thorn plants.

Prizes and penalties

Competitive sports can be played for their own sake, where winning the game is the goal by itself, or with the intent to win an external prize. Often the losing team will need to buy that prize or otherwise pay a penalty. This can be asymmetric in both ways: in some matches there will be a prize for the winning team and no penalty for the losing team, and in others there will be a penalty but no prize. The latter case is common when both teams are seen as owing a debt of some sort but are unwilling to split that debt.

As above, the prize and penalty do not define a sport.

Comparison with noncontact sports

In most (perhaps all) cultures, so-called non-contact sports such as sledding and climbing would simply be called games, and not categorized as sports, even if they involve heavy physical activity. These may nonetheless resemble some of the traditional sports with an additional element of safety involved to protect the players, and therefore would have names derived from them.

For example, many contact sports involve racing towards a goal, but players are allowed to interfere with each other. In most cultures, this would be considered the basic form of the race, and the derived form in which interference is forbidden would be given a derived name to show this.

Exercises such as coordinated dancing are in the same category as these non-contact sports. Even academic pursuits such as trivia games may be considered more closely related to these non-contact sports than the contact sports are.

Other terms relating to sports

Late Andanese hamiki and Play numunam referred to a game in which each team was deemed to be undeserving of sympathy from the standpoint of their opponents, meaning that it was legal to use pain and extreme physical violence so long as the game's other rules were strictly followed. (This is the same word used to translate "video game".) NOTE: this may simply be the same as the "unrestrained combat" sports up above.

Players are considered good athletes but the war caused the Play government to ban most sports to conserve energy.

There may have been sports in which whole families played against other families, including all of the women, children, and the elderly. These would by necessity be noncompetitive, and perhaps even go beyond this and become cooperative sports (see Lilahaa#Cybernetic_warfare for a hypothetical modern-world equivalent). If the teams shared a common enemy, the teams would all be on the same side and all would share in the victory when the enemy was defeated. The separation into teams would nonetheless encourage each family to strive harder for the prize, bonded by family ties and seeking the ability to share in the glory of having contributed the most.

Remember also the word gìtʷi.

Examples of individual sports

10:09, 26 October 2022 (PDT)

Fingers, Beaks, and Teeth

This is a three-team version of the dog game above in which a species of small firebirds (ordinarily a very large predatory bird), dogs, and humans compete in simulated combat to see who can bite or strangle the others into submission.

This is an asymmetric game. The firebirds have a clear advantage over the dogs due to their greater speed and mobility. The dogs have a clear advantage over the humans because they can both bite and outrun the humans. But the humans have an advantage over the firebirds, because once the firebird gets close, the human has much greater maneuverability due to the small size of the birds allowed to compete in this game.

All of the animals are tame and are instructed not to actually kill their enemies, but to make it clear that they could go on to kill the opponent in true unrestrained combat. This includes the humans, the only species of the three that cannot bite (they are not allowed weapons). This means that the human must hold the bird in place for a period of time without the bird escaping or severely wounding the human.

No winning strategy

The asymmetric nature of the game makes it highly volatile. Any momentary two-species alliance will be hampered by the realization that once the weakest species is out of the game, one of the two remaining species will have a very easy time defeating the other.

For example, if humans were to ally with dogs against the birds, the humans would quickly eliminate any birds that tried to attack either the humans or the dogs, but once the birds are defeated, even just a small number of dogs could easily bite the humans into submission and thus win the game.

Likewise, if the dogs and birds teamed up against the humans, the birds could let the dogs do most of the work, and then when the humans surrendered the birds would turn against the dogs and bite them into submission just as the dogs had done to the humans.

Lastly, if humans and birds teamed up against dogs, the dogs might take down a few humans because of their speed advantage but would quickly go down themselves from the much more numerous firebirds biting them from behind, without the need for humans to even involve themselves. But then the firebirds would need to approach the humans up close in order to finish the game, which would make them vulnerable as the humans would no longer need to run away from the dogs.


The popular name of the sport is translated as such into various languages rather than using loanwords. The name is intended to intimidate human players and spectators, but like other such games, the playing field is leveled by increasing the number of players of the weakest species, and so the humans nearly always outnumber the dogs, whereas the birds are numerous but physically small.

All-human variants

As trained animals in this sport typically do little else, the sport can only be played properly in areas of high human population density. A simulated version of the sport in which all three teams are human, identified by colorful clothes, substitutes in rural areas and is considerably less dangerous for the players, as even tamed animals can cause injuries.

The bird team in this version of the sport is given the ability to remain outside the playing field since they cannot fly; the dog team is given full representation rather than being less numerous than the others; and the human team moves just as quickly as the others but no longer has an advantage in maneuverability.

Obscene and vulgar language

04:46, 3 August 2022 (PDT)

Speakers of some languages, such as Play, claim that their language has no obscene words at all and that any language that allows obscenities belongs to a depraved culture. The Moonshine speakers believed obscene words existed in their own language, but had very strictly culturally defined ideas of what they were.

Once the Play and Moonshine speakers divided the world into two halves, resisting languages were categorized as consisting entirely of obscenities, particularly by the Moonshines, but still looked down on by the Play speakers.

A comprehensive list of terms will not be posted here, but there are some elements which show up repeatedly.

Male anatomy

The MRCA roots nə̀gi, , nàlə, and hàtu all mean penis and can be used interchangeably. They are not related to each other at even a very deep level. Even at the MRCA stage, the /pì/ word was compounded as ṗə̆nu-p- "pointing penis", which is the source of Play's preferred word, tenu.

Lava Beds with six roles

14:02, 31 July 2022 (PDT)

Gold may have grown the Lava Bed role matrix from 4x4 to 5x5 or even 6x6, but left many holes. Then, later, Play would compress the matrix into a 4x3 rectangle with no holes. This relies on the idea that the case markers of Poswa and Pabappa are descended from morphemes that had similar meaning in Play, though perhaps not true noun cases, and that these in turn go back to Gold, when they were still tied to the Lava Bed system.

Gold would have (values refer to arg1 and arg2 speech act roles):

  1. -Ø- AGT:PAT
  2. -k- IDT:OBS (earlier AGT:OBS). Cognate /p/ used in Play to mark any noun as being the speaker, since it sets the arg1 speech act role to IDT and in Play arg1 always means the speaker if not explicitly named in the sentence. Note Play also having a second /p/ with a nearly opposite meaning, which did not get confused with this because it was used in sentences with a different structure.
    If Gold had already lost all its pronouns and person markers, this /-k/ will need to perform the same function it does in Play.
  3. -g- PAT:AGT. Possibly an accusative case in early MAP if there was a need for one; evolves to a locative case in Gold, but this use disappears in Play and the infix comes to have no meaning whatsoever, merely being used to pick which of two paradigms to use in certain constructions.
  4. -h- OBS:IDT (earlier both OBS:PAT and OBS:OBS). Used in Play to mark arg2=IDT just as /p/ is for arg1.

These four are the same as Play. In Play it could be said that IDT is more like BEN (beneficiary) in its most common contexts, but it still is nonetheless used to say "I am a doctor", etc., and IDT was almost certainly its original sense. Gold also has:

  1. -ḳ-, used to specifically mark the arg2 participant in Play, where it merges with /k/ above as -p. Earlier meant OBS:AGT. This is different from the /-h-/ above and also different from the arg1=IDT above.
    This morpheme likely changed its meaning frequently throughout time, evolving from OBS:AGT (within the verb) to a similar function but on a noun, then to marking the arg2 participant even when it was not the agent, and in post-classical Play then to an accusative case.
  2. -ŋ-, which evolves to a locative in Play, from an earlier meaning that could be described as bundling circumstantial, benefactive, and dative all together. Earlier meant PAT:OBS.
  3. -l-, which has a cognate in Play, but in a different part of the verbal system. Earlier meant AGT:IDT, and futhe r\back, AGT:PAT. Possibly an escapee of the verbal system, as in Play, where it appears outside the Lava Bed paradigm.
  4. -y-, likely obsolete already once it became a semivowel. Earlier meant IDT:AGT, and further back, PAT:AGT. Thus this was the original passive morpheme. At the time, /l/ had been the morpheme for AGT:PAT. Thw two meaning remained the opposite of the /l/ morpheme in Play even as those meanings changed.

For comparison, the proto-MAP system was

          AGT     PAT     IDT     OBS
 AGT      ---      Ø      əl      ək
 PAT      əg      ---     ---     əŋ
 IDT      ey      ---     ---     ---
 OBS      əḳ      əh      ---     əh

Note the two /h/'s, a flaw in the system as it existed at the time.

Tense/Aspect/Mood outside Play

07:44, 20 July 2022 (PDT)


02:23, 7 November 2022 (PST)

Most likely an aspect rather than a mood, because it extends to inanimate agents. Old Andanese ṅti, from MRCA ŋə̆ndə-n-, describes an agent that cannot complain.

It may have meant to volunteer in MRCA, as it does in PSL, and thus in Andanese evolved into a near-opposite of its earlier meaning.

It is the same root as Galà honŏdo "underwear" with the vowel shift being due to vowel harmony with the classifier prefix.


It is likely the only tense marker that goes back to the MRCA is the past tense infix -i-. Dreamlandic's /ndà/ morpheme originally signified perfect aspect.

The MRCA had a morpheme hṅ "at a specific point in time" which could be used in the manner that Poswa and perhaps Play use their serial verb marker -s.

Late Andanese

The Late Andanese past tense particle is probably hi, based on the /g/ > /h/ rule applied to the MRCA particle ə̀ ~ gə̀ that also appears in the Dreamlandic copula. It is not clear whether it would appear before the verb or after it. If before, the logic is that it is the main verb of the sentence, and the content verb is merely a modification of it. If after, the logic is also that it is the main verb, but that it governs the content verb, as if producing an S(SOV)V sentence with the SS redundantly compressing to a single S.

The infix form of the morpheme was available in the MRCA, and the copula is merely a carrier for it in Late Andanese, as if I saying "I did brush" when "I brushed" would suffice. This is the only reason that Andanese and Dreamlandic can share a copula; it is not present in most of the other languages that are "in the middle" of the family tree.


MRCA had the following aspect particles, which were most likely treated as verbs that put the main verb into a "serial" mode:

  • àḳi "to repeat"
  • ănu resumptive aspect; "to resume". The /a-/ might not be real, as I believe I added this at a time when I was expecting to form a pattern.


At least some of Play's moods do trace back to the MRCA, though much of the system was elaborated.

  • tə̀, used extensively in Play compounds, but originally a free morpheme in MRCA. Most likely simply indicates an irrealis mood.
  • ŋà, used in three different moods in Play. It is possible that this is a Play innovation though as it is identical to the 1st person patient marker and could have been seen as a way to indicate the speaker's reaction to something. Possibly identical to the ŋà morpheme used in MRCA to indicate a goal-seeking verb.
  • , a syllabic /n/ indicating the imperative mode. Very old.
  • , the interrogative mood, and identical to one word for "or". This might not go back to MRCA with the same variety of senses, and therefore might not be a true mood marker, but the morpheme did exist in somse sense.
  • ŋì, the negative. Possibly a Play innovation because the Dreamlandic negative comes from /man/ and is not a mood marker. Possibly contracts to just the consonant when followed by a vowel.

Verbal derivations resembling TAM

These are grouped here for convenience. They are better considered derivations rather than inflections, but might interplay with inflections in some languages, such as Late Andanese where the inflections and derivations were treated in much the same way.

  • gṁpă and gàha, both meaning "to attempt, try" in the MRCA. Essentially equivalent to proto-Dreamlandic pifia ~ pafia, but languages that have both words will put the /pifia/ word after the main verb, treating it as inflectionary, whereas the words beginning with /g/ will be standalone verbs.
  • , to want or desire. This is a true primordial /o/, not just from schwa.

Clothing terms

15:49, 17 July 2022 (PDT)

The MRCA speakers lived in a tropical climate and allowed nudism. The only atomic root words for clothes were for clothes covering the genital area, and those words were few. This trait continued in Dreamlandic, whose speakers remained nudists for over 5,000 years, and held on to some extent in other languages of the tropics. The Play speakers once stated that they lived in the world's coldest tropical nation, aware of their heritage but also their strong identification with the rugged mountainous terrain of their northern regions.

MRCA words

Protective from moisture

MRCA had a word for diaper: . It was a very short word, and could even be used with no classifier prefix, although it was more common to attach one, which also specified the person wearing it, if there was one. With a different classifier prefix, it was once the word for cradle, but this usage died out.

The word for diaper was also used to mean a hat in some languages, as they both kept away moisture. A possible cognate is MRCA àki "hair of the head", but for the most part these double meanings did not become established in the lexicon.

Other clothing terms

The MRCA had several words for skirts, which later evolved into ethnonyms in some languages to describe the particular skirt that the particular tribe favored. These words were: hàgə ~ yàta ~ lŏpa(-l) ~ pà ~ gù. The last two, as short as the word for diaper, were unusually short words for this language and belonged to very old inherited vocabulary.

The MRCA word gùta could mean "to defend, support; to comfort" and "diaper, underwear". (It is probably cognate to /gù/, but note also unrelated /ĭgu/). A related word, gəgùta, means a cloth fiber from which the garments were once made.

One more word, hìŋə, denoted close-cropped underwear for women only, preferred for comfort but difficult to make. This word was sometimes augmented as hiŋènde.

Other MRCA clothing words

MRCA had words for clothing intended for specific purposes, such as protection during battle or for people working with sharp objects.

The clothing classifier prefix m(f)ʷo- could precede any body part to describe clothing to cover it. But words for clothes that did not follow this formula still needed the same classifier prefix, so the language could be analyzed as having redundant terms for body parts such as mbùhə, another word for leg, normally pʷŏti or even just , because of the existence of a term /mfʷombùho/ describing thick leg coverings used to safely walk around areas with many thorn plants.

There was never a classifier prefix for body parts because MRCA used the classifiers for human referents (men, women, etc) and also had one for humans generically.

NOTE: earlier wrote that soap was considered a type of clothing, but this is an error caused by confusion of two different morphemes that merged in Gold but not in Dreamlandic or certain other languages.

Alcohol and other stimulants

14:04, 17 July 2022 (PDT)

This section excludes Play, except for the sake of comparison. See Play_language#Shopkeeping_and_retail_terminology_in_Play instead.


Wine is the predominant form of alcohol, but most of it comes from the Play-speaking area of the world, whose various governments periodically shut down wine cultivation, particularly during famines, to focus on other fruits (grain agriculture was unknown) and to grow forests where game animals would dwell (livestock was also unknown because the animals were too tame and thus too self-aware to be tricked).

The Dreamers would have had access to palm wine but not grape wine, and the PSL's likely the other way around. But from trade, the PSL's would likely have known of palm wine's existence.

Sleep flowers

The Gold word for the sleep flower is ʕul. It grew in the northern highlands of AlphaLeap, hence a short native name was created. The structure of the word, with a final low tone in the MRCA, requires derivation from a proto-form such as BɜlɨBɨ, where the first B but not the second could also have been /h/. This form is so far back in history, however, that it most likely never existed, and rather the word was compounded after the pieces had become too short to stand on their own.

A good word for the sleep flower in PSL is ukuuria, which literally means "back flower" because it was grown out of sight. Although PSL is functionally identical to proto-Dreamlandic, the two tribes made separate migrations and therefore came upon the sleep flower independently. Therefore Dreamlandic still needs its own name.

In proto-Dreamlandic, the word could be ukuŋuuriān, which is cognate to the word for dream, but not cognate to the PSL word despite its similar shape. It is likely that this word did not exist in proto-Dreamlandic, but the form here is valid in the sense that the daughter languages could independently put their reflexes of these morphemes together to make the word.

Play simply uses meviya, "sleep flower", with an appropriate classifier suffix. The Late Andanese name would likely be a calque of this, as the hypothetical cognate to the Gold word above would just be lu, and even with a classifier prefix the word would only be tilu. The word for sleep is naa, which is only slightly longer, but this could be sufficient; alternatively, a more poetic word could be used, since Andanese has other words for sleep.

More Dreamlandic ideas

Equative compounds

It is possible that Dreamlandic uses Ø for equative compounds, unlike Play, because this is the OBS:OBS role marker. Alternatively, a copula may be needed even for equative compounds, with the zero-marked form being a genitive, which unlike the inherited genitive, is head-final.


14:15, 15 July 2022 (PDT)

The plural prefix u- could be up- before a noun beginning in /i/ (possibly also including /j/), to prevent the otherwise forbidden sequence /wi/ from appearing. It is possible that the /u-/ was in fact once /up-/, since a /u/ is often followed by a labial, but this is still possible to create from analogy only.

What is the oblique, anyway?

08:41, 8 July 2022 (PDT)

The oblique morpheme -n disappears in MRCA and causes the tone of the final syllable to be high, which makes the preceding syllable low. It has no apparent meaning in MRCA but may be needed to turn nouns into verbs. This is why Play always reflects a high-toned final syllable in its verbs, even when no person marker is added.

However, Dreamlandic does not use the oblique to turn nouns into verbs, and it would actually greatly complicate the system if it did, since in Dreamlandic the reflex is consonant mutations instead of tone shifts. It is possible that the oblique is embedded within the MRCA locative suffix /-ni/, which Dreamlandic retains as -nii, and which would still cause the same consonant shifts, but might be easier to analogize away since it would be perceived as a morpheme on its own and not a bound subsyllabic inflection. Similarly, the verbal suffix was analogized away.

The oblique might also be a genitive, assuming that DPR and Baywatch got their genitives from later formations as they lost the inherited noun cases.

Initial consonant fluctuations

02:13, 6 July 2022 (PDT)

If yan "hand", which is also the instrumental prefix, retains its /-n/, in theory any word beginning with a fricative could change it to a stop, although perhaps it would only occur for inanimate objects. A new rule will have to be written to handle /nr nl/, which would have almost never come up in the previous writeup for Dreamlandic. It is most likely that these would be treated the same as /nd/, and that there would also be /mb nġ/ since some /-n/ comes from the other nasals.

The word could easily lose its /-n/ due to analogy, because Dreamlandic also retains the oblique marker /-n/, and the speakers could assume that the instrumental prefix was not simply the word for hand but the oblique case of some otherwise lost word /ya/. It depends on exactly what the role of the oblique case marker is. The inherited accusative marker is -ii, even after a vowel, although it causes mutation if after a vowel instead of appearing as /yi/. Therefore the oblique is not part of the accusative. In the MAP branch, the oblique evolves into a tone shifting morpheme with no clear meaning; it makes the last syllable of any word high-tone, and therefore the previous syllable low tone. It is not the source of the case marking tone shifts in MAP, which do the same thing, because those come from infixes.

Liaison in Dreamlandic

It is possible that Dreamlandic could set up French-like liaison, since the disappearing final consonants of bare-stem words often preceded vowel-initial prefixes, such as the instrumental prefix yan "hand". Spread of liaison onto following content words is extremely unlikely, however.

Dreamlandic maturation date

Proto-Dreamlandic may have matured as early as 300 AD, essentially making it a sister language of the supposed MRCA, but 700 AD is perhaps a more accurate date. This means that Baywatch was mature by around 2668 AD. This date is of no particular consequence in Dreamlandic history but aligns with the Oyster War being fought to the east. Nonetheless, both Baywatch and Dolphin Rider will need to have another thousand years of history added to their sound changes and therefore the languages as they stand now are for intermediate periods.

If proto-Dreamlandic split off around 1000 BC, this means that they took 2,700 years just to find the mainland, while the MAP sailors did it in a historically negligible period of time, such that it can be considered instantaneous. This is viable if assuming that the Dreamers were for some reason bound to use the much longer eastern route, all the way around the planet, while the MAP's sailed to Fox Island and then from there to the mainland. What is needed is a good reason why the Dreamers were not able to use this same route. If it is because the MAP's already controlled the seas, they would need to have controlled the seas from at least as far back as 1000 BC, which means that there are 1,000 years of unwritten history and that it may have in fact taken the MAP's about a thousand years to reach the mainland as well.

Words for parents and family relations

05:04, 3 July 2022 (PDT)

Pre-MRCA had /am/ "parent; to carry, nurse, shoulder" which was already down to just a single à by the time of the MRCA, meaning that only Dreamlandic has the final consonant.

Words with precise meanings in the parent language

People by age and sex

  • tə̀ means man, not father, and therefore will not appear in ordinary content words, but might be used in baby talk (replaces the earlier idea of using /tàta/).
  • means woman, not mother, but as above may appear in baby talk. This same morpheme also means breast and milk, with the proper classifier prefixes.
  • means child (young person), not child (offspring), and probably will not appear in any words for family relations.
  • ṗò, meaning a teenager of either sex. Shifts to meaning adult males in Dreamlandic and disappears into the grammar in other branches. It is the source of Play's /ta ~ a/ "career" morpheme that indicates the habitual agent of a verb.

People by relation

  • , meaning a parent of either sex. Disappears in Gold because it merges with the word for hand, which also happened in Dreamlandic but with a different word. Survives in Andanese as just yi although as with other nouns still reuiring a classifier prefix.
  • , meaning a parent of either sex. There is likely some difference in meaning between these two words even in the MRCA.
  • , meaning mother. Found only as a bound morpheme because of homophony with grammar words.
  • ŋà, meaning father. Found only as a bound morpheme because of homophony with grammar words.
  • , meaning father, but also appearing in other words such as /wùṁə/ "father of the womb; husband". Synonymous with /ŋà/ above but used in different ways.
  • , meaning son. Always used with a classifier prefix which indicates the child's age; therefore this word also indicates adults. In all languages other than Dreamlandic, this word is a homophone of the 1st person verb marker, but never occurs in the same context (these languages have no pronouns and the word for son cannot be a classifier prefix on its own).
  • , meaning daughter. Reflects an ancient word shape like /čɨw/ or /čɨrʷɨ/, but the final consonant never appears even in Dreamlandic and even when vowel-initial suffixes are attached.
  • Either this word or a homophone of it appears in a Play word for teenager, implying that it may have shifted meaning from child (offspring) to child (young person).
  • ndăhi-n- meaning a male relative. Apparently comes to mean husband in Play, even though Play also has /tus/ meaning husband. Despite the /d/ this could be cognate to /nà/ above.
  • ḳùka, meaning a sibling. No obvious shorter morphemes and /ḳ/ always comes from an earlier cluster with an elided vowel, so this word was once at least three syllables long. Because of the classifier prefixes, it is always clear whether this word is referring to brothers, sisters, or both, and also to some extent whether it is singular or plural and whether the referents are adults of children
  • MRCA had ka- as the epicene agent prefix, but did not allow other words to stack before it, so if there is a relation here it must be quite old.

There may have been a lexical gap in that there was no gender-neutral word for child (offspring).


  • ŋə̀, meaning family in general. Possibly cognate to /hàŋə/ which is used in polygamous marriages.
  • òki, meaning marriage, and likely used only in the Gold/Play branch where it becomes monosyllabic. The Gold cognate is žĭ, where /ž/ is the rarest sound in the language, and this may help later associations between this sound and acts of love and intimate relations. Play shifts /g/ > /ž/ but retains this emotional association. In Play, the word for love itself could even be reanalyzed as beginning with this morpheme even though it is wholly unrelated.
    This word may be cognate to ə̀ku-p- "couple". If so it needs no further explanation and may even be the older of the two words. The initial /ò/, however, would need to come from an adjacent word or a classifier prefix.
    It is also possible that this word is related to the word for diaper, meaning that the word for diaper would be older than both the word for marriage and the word for couple. However it is difficult to make both connections at once, even with classifier prefixes, as it would need to be something like diaper > cradle > childbirth > marriage > couple all in a short time.
  • , meaning wife. This is a true primordial /o/, not due to ablaut, and therefore always appears as /a/ in Play and never as /ə/, even when adjacent vowels are schwas.


  • à, the word meaning to parent, carry, or nurse. It may be cognate to a word for shoulder and for hip, the parts of a parent's body that a child is prone to embrace. However this would be a late creation in MAP because the medial /m/ is not there.
  • , meaning diaper, a very ancient CV primitive that might be used in more than one branch of the family to create new words for babies and for parents. Languages with classifiers might also retain the meaning of diaper.
  • This word is thousands of years older than the classifier system and cannot be traced to an earlier longer form such as /kin/; it was CV back to the stage when /i/ became phonemic. It is possible, however, that the original meaning was cradle, and that the split of meaning into "diaper" and "marriage" was the result of two different classifier prefixes being added, if /o-/ started out still with the cradle meaning and then later shifted. But the word still ended up meaning diaper even with no classifier prefix by the time of the MRCA.


  • ṁə̀, meaning womb. Almost certainly a compound of /ṁ/ above and some other ancient word that ended up as just /ə̀/. In Play, this word was conflated with /ma/ and therefore mother and womb shared the same word.
  • hə̀, also meaning womb, and with a primordial /-n/ that was lost in the MRCA. The word above may actually contain this word although the development is irregular if so.
  • ùŋi, meaning fetus, but probably originally a plural /u-/ attached to an ancient /ŋì/.
  • ùŋu, meaning pregnant. Not necessarily cognate to the word above; if it is, the /ù/ is probably a plural in both words.
  • tùŋa, to give birth. Possibly explains /tu/ and /tus/ above and below, particularly given the abundance of /ŋa/.

Tribe and extended family

The following words all mean simply "tribe" in at least one descendant language, and all trace back to the MRCA, but they did not all have the same meaning in the original language.

  • pìli
  • hège
  • găhə-l-
  • ìmfa-nʷìha Reflected in Late Andanese as ihagika. It is most likely that this compound was formed at a late date and only in the Andanese branch.
  • ŋaŋe perhaps being simply the reversed form of the word for adolescent below (but note that there were two /ŋa/'s). Therefore it could be either "father family" or "father of family" depending on the age of the compound, more likely the first one.
  • žibu

Other words

  • tàlə, meaning orphan. Direct ancestor of Play tāu with the same meaning. A contraction of earlier tà gàlə, "girl who searches with her hands", though it may be that /tà/ was not always female, or that the /tà/ used here was unrelated to the word that means girl.
  • nʷò, meaning to touch. Used in terms for close relations of all kinds; originally a verb but could correspond well with English "beloved", "dear", and so on.
  • mbàgə, meaning a younger person related through one side of the family: a nephew or niece, including adopted children. Thus the word simultaneously indicates age and relation. It is not related to /hàmba/ "mother", whose /mb/ crosses a morpheme boundary. This word first comes to mean "boy" in Lava Bed languages, and then becomes grammaticalized, taking out of the lexicon entirely.
  • lì-n-, to adopt a child. Can be used as a suffix on other words to indicate both adoptive parents and adopted children. This word has broader application, and can be used to mean "bring into the household", not just with children.
  • hàŋa, an adoptive family. Appears to be a compound of the words for mother and father but is in fact "mother" plus a suffix indicating a goal seeking verb.
  • nə̀mi, indicating a child who closely resembles or is meant to closely follow one of their parents.
  • ŋə̀ŋa, appears in a Play word for adolescents, perhaps originally indicating a young woman seeking to start a new family.
  • găḳa-n-, a young boy. Originally a term of address, and therefore possibly ultimately from a word unrelated to all others on this list.
  • ə̀gə, a man of marriageable age. Appears in Old Andanese kège and wògo, each with a prefix. Likely also cognate to hège "tribe".

Words with unclear meanings in the parent language

  • , earlier /Bà/, appearing in the word for mother in at least Dreamlandic. The consonant here appears as /b/ because it follows an /m/, and therefore may have originally had a different value. No matter what the initial consonant was, it would merge into the reflex of the word for hand unless this too had a lost final nasal. The most likely choice is /B/ because the alternatives /g/ and /h/ could have been mistaken for pieces of grammar. This word may also appear outside Dreamlandic, whether in the same position or a different one. The MRCA form would have just been /hàmba/ and would have been CVCCV for many years. It still needs a classifier prefix, however.
  • , appearing in one word for wife. Possibly related to Play /tus/ "husband", which itself probably cannot stand alone and might have had some other meaning. May mean to give birth, since /-ŋà/ could mean any of several things.
  • hʷò, appearing in a word for a polygamous wife.
  • ĭkə, meaning "for women only" in Gold, Middlesex, and likely others. It could have been a simple collocation with a zero-marked genitive, however, which would make this simply another word for woman. This meaning could then be preserved in Dreamlandic and perhaps FAM (Fern-Andanese-Middlesex) but not in Gold. The tone may be wrong: ìkə.
  • ŋùni, the male version of above. Certainly a compound, as /u/ always comes from a sequence, and the second element could be from /nɨ/ which also occurs below.
  • ə̀ndi, which is the same as above but refers to boys. This word simply became an ordinary word for boy in most languages but sometimes with extra affixes. The morpheme structure is suspiciously similar to that of a classifier prefix for children, but the vowels do not line up. Rather than connect the two words for boy, it is perhaps more likely that this word was once ɜn-nɨ, with the second element being the same as in the word for adult male up above.
  • , apparently meaning males, but perhaps extinct as a standalone morpheme long before the MRCA. This is the word that would have meant "his" if the Lava Bed languages hadnt bundled age and number into their person markers.
  • This may be cognate to morphemes such as Play ni "penis", but not in the obvious way, since the Play word comes from MRCA /nə̀gi/. Nonetheless, the /nə̀/ in this word could still be a direct cognate of the standalone /nì/ that means males.
  • ə̀-n-, from primordial /ɜn/ meaning child. This word may have been used in a diminutive function in the MRCA, though also used literally, such that suffixing it to any noun would make a word that means "___ of a child" or "____ fit for a child".
    • This could survive marginally into Gold, though nonproductively because it would rely on knowledge of lost consonants. That is, MRCA gàn ~ gănən would become Gold gà ~ gănə, with no indication from the bare form of the word that the lost consonant is an /n/. This then dies out in Play, stereotypically because "in Play culture they respect children and don't use cute names." The Andanese, who lived only where Play speakers lived, also lacked diminutives.
  • ŋù, could mean "adult" or "adult male" just by itself, but may also be spurious, as /u/ can come from sequences and it may not have been a word of its own.
  • , possibly meaning girl or female. There is no word for girl in the later languages that can be traced to the MRCA; they all formed new words through semantic shifts. Even this word only appears as the second element of a compound, ndʷotà, where the first element is the same as the "reversed" word for boy up above. Nonetheless Dreamlandic would not have made such a compound and so this word is likely to be real, but may not have meant girl by itself.
  • , meaning girl in Dreamlandic. From primordial /pup/, presumably from an earlier word like /kʷɨkʷ/ or perhaps an unusual formation based on a compound. Meaning unclear outside DRM. The same word also occurs in the number system and again only in Dreamlandic.

Other isolated words

These words belong up above but are placed here for convenience.

  • nò-p-, any male. Can be preceded by the classifier prefix ka-, ordinarily epicene.
  • mfà-m-, any female. Can be preceded by the classifier prefix ka-, ordinarily epicene.

The nʷ words

There are a few words added to the lexicon very early on that are difficult to use:

  • nʷìgi, a very young girl.
  • yìnʷo, a very young boy.
  • nʷìnʷo, a very young child.

If real, these reflect a very different morpheme structure than the rest of the language. They belong to a very early series of words from Fojy in which such exotic consonants were common.

The last word, at least, seems to be the source of Play nina "toddler; to smear, mash; to have an idea to improve something". Play also has a word for orphan that could be traced to /nʷìgi/, although it is not sex-specific. No Play descendant of /yìnʷo/ (the expected Play form would be /yinua/) is in the dictionary at this time.

Standalone words



Play words are mamšaa or maaa (not possible to reduce to two /a/'s) depending on the age of the compound.

Play could also simply create a new compound meaning "milk nurse", which would appear as mamba or mama or ma, or be variable between these forms. The other two words above would still remain in the language as synonyms perhaps in more formal speech, particularly if /šaa/ continues to be recognized as a word on its own. /mamšaa/ would then mean "milk mother".


Dreamlandic would simply continue the primordial word /hàmba/ (ha-a(m)-ga), whose reflex would be yampia, but because of classifier prefixes, is likely to be ampia since the applicable classifier prefixes would all swallow the /y/.

Child (offspring)

Languages will need new words for offspring.


A phrase that could produce a new word is such as lìŋa ḳà, which would become either vīmpa or viŋap in Play depending on when the compound became opaque. The second word means barren when describing humans and pregnant when describing egg-laying animals; this is due to a longstanding cultural trait relating children with birds and other non-mammals.


Possibly a relation with a word that evolves into the word for boy in the Lava Bed langauges,

Comments on different languages


The phrase "in the shadow of the womb" meaning someone very young was probably coined very early and would develop into just mata by the time of Play, with more historical morphemes than phonemes. It may be reanalyzed as matam to keep in step with the contemporary grammar of Play, where /-m/ came to be the new locative marker.

Play inherits none of the words for husband or wife from the MRCA, except possibly fapa "polygamous wife" though this would already be colliding with many other words.

The word for toddler is used directly to mean both "to have a plan for the future" and "to smear, mash, make everything the same", both describing typical toddlers' behaviors. This word was originally supposed to be nina, from MRCA nu- (as above) followed by an /ina/ of unclear derivation, but this no longer seems viable for a language like Play. A new word such as čitana or just tana could be a better choice. (Although či in its bare form did not mean diaper in Play, the morpheme was still in use and still had that meaning with the clothing classifier suffix on.)

Play also has a reflex of ə̀ndi "boy". That reflex is ti ~ i, mostly seeing /i/, and it means "little" and can apply to inanimate objects. However, /tai/ is not used to mean a child where /ta/ would denote an adult; instead, taā takes its place.

Possible survival into Poswa

If Poswa inherits the Play paradigm with verbal embedding, it could have a conjugating paradigm with nappa ~ napa ~ naba as the words for "my son, your son, [the] son" respectively. This could survive despite the ability of the language to handle kinship terms just like any other nouns. The normal Poswa noun paradigm would instead give nabo ~ nabe ~ naba.

One good reason for the retention of this system is that /nappa/ could be seen as a contraction of /nabosa/, with the obedience morphemes, and avoid the uncomfortable problem of putting obedience morphemes on the words for close relatives.

A disadvantage is that the archaic Play paradigm leaves no obvious place to put expressives like Poswa's /-ambo/ "that I love", since the morphology is opaque. It is possible that they could be assumed to be bare stems instead of being 3rd person nouns, but if this is the case, the "good reason" above might disappear.

Late Andanese

Although it has the smallest phonology, Late Andanese may preserve the original kinship terms better than most other languages because it continues to use classifier prefixes and compound terms. If words are inherited naively, it may happen that nearly all of the valid syllables in the language attain meanings as independent words relating to families, and that baby-talk could take over. This is not a problem because the classifier prefixes effectively keep the kinship terms in a private lexicon of their own in which they will not be easily confused with other words, even if those words are exact homophones apart from the classifier.

Very young children would use the words without the classifier prefixes, and it is most likely that as terms of address they would also be without classifier prefixes since proper names also were. But in ordinary speech, they would need classifier prefixes just like other words, as there was no "null" word class.

Play substratum language

PSL (or at least Shining Wave) gets a new word for boy, ňunzun, from "eager; brave". The other words probably disappeared altogether since PSL did not have classifiers and also did not have Lava Bed morphology.


If keeping with the original draft for Moonshine, the words for child (offspring) and child (young person) are the same, and are the passive form of the verb "to have", whatever that may be. In the original draft the word for have was a and therefore the word for child was its passive form, à.

Lava Bed 3rd person gender marking

03:26, 20 June 2022 (PDT)

It may be that at least in Galà, the /g/ that marks so many nouns is "B" after all. A new morpheme "T" could be created, echoing the thematic consonant of the topic, thereby making most inanimate nouns agree with some animate argument of the sentence, since most topics are animate. (But not all, and inanimates can have their own thematic consonants.)

Early Andanic gluons

16:04, 25 May 2022 (PDT)

This is a near-duplicate of the Gala language seciton below but is here for convenience.

Stage 1

The MAP clade branched off after the IDT fields had reappeared, and therefore the minimal matrix is

          AGT     PAT     IDT     OBS
 AGT      ---      Ø      əl      ək
 PAT      əg      ---     ---     əŋ
 IDT      ey      ---     ---     ---
 OBS      əḳ      əh      ---     əh

The zero morph had already moved up to the AGT:PAT position, meaning that there were now two /h/'s. This was because /l/ had moved to the new AGT:IDT slot, freeing it up.

The vowels were entirely schwas, which is important because all MAP languages have at least conditional schwa loss.

A new morpheme must be found to fill the OBS:PAT slot, and it must come from /ə/ followed by a primordial CV sequence. For example, if the verb were "to see", the sequence would be /-əb-/ in early Andanic, from an earlier /-mb-/. This morpheme arrived too late to conflict with the earlier unrelated /-mbə-/, so this is no problem. And since Andanic handles its schwa loss differently, there is no reason why a consonant such as /b/ cannot participate in this operation.

Stage 2

Therefore a list is:

          AGT     PAT     IDT     OBS
 AGT      ---      Ø       l      ək
 PAT       g      ---     ---      ṅ
 IDT       ē      ---     ---     ---
 OBS      əq       ō      ---      h

The schwa is still the only vowel. /əb/ regularly proceeds to /ō/ even before a vowel. However, a new problem has appeared, in that /g/ cannot be relied on to contrast with /h/ or with /Ø/ because the vowel is not always present. Therefore this too might need replacement. Alternatively, the /e~o~i/ alternation freezes out and it becomes a vowel-only infix just as IDT:AGT is.

Stage 3

A possible further development from this state is

          AGT     PAT     IDT     OBS
 AGT      ---      ō       l       Ø
 PAT       ē      ---     ---      ṅt
 IDT      nē      ---     ---     ---
 OBS       k      hō      ---      h

There are no new morphemes, just semantic shifts. Note the /k/, now without its schwa: it was contracted from /həq/, and therefore when a /k/ cannot appear, the sequence /heq/ takes its place.

Stage 3a

If the development below is not followed through, a stopping point along the way may be found:

          AGT     PAT     IDT     OBS
 AGT      ---      ō      ---      Ø
 PAT    (n)ē      ---     ---      ṅt
 IDT      ---     ---     ---      l 
 OBS       k      hō      ---      h

Finding a new OBS:IDT morpheme is very important, however, as this is not only the 2nd person identity marker but also the 2nd person possessive whenever it is used with in animate object.

Possible Galà developments

Sound changes would cause the /ē ō/ above to shift to /i u/ since they only occurred before a vowel. Likewise, clusters such as /pk tk/ could become /pp tt/ (essentially another way of spelling `p `t) if the /heq/ allomorph is lost for some reason. It is also possible that /heq/ will take on an independent existence as the missing OBS:IDT morpheme.

The OBS:OBS morpheme remains as h, which means that an /h/ is infixed into every noun that does not involve the speech act participants, even though these nouns must also be marked for 3rd person identity using vowel length in the final syllable. Thus for example hulintū "moon". The combination of these two morphemes might come to be seen as an interfix h...: or as a gradation that adds an /h/ to the final syllable and makes its vowel long.

It is also possible that lh > s, contradicting the sound change list on Andanic languages.

Note that IDT:IDT is a gap, as in other languages, and that 1st person dual and plural are expressed in other ways. This applies to all combinations, not just to 1:2.

The 3rd person IDT marker appears in both the word for an object itself and the word for "his/her/their [object]", because classifiers are still present. With the words for the 1st and 2nd person possessive, however, the 3rd person IDT marker is omitted because the 1st or 2nd person marker(which are both IDT) makes it logically impossible for there to be one.

Dreamlandic gluons

16:04, 25 May 2022 (PDT)

Dreamlandic probably branched off about a thousand years earlier than the MAP clade, so it could have had a gluon system that was underdeveloped.

Past tense

21:08, 25 June 2022 (PDT)

Use the suffix version of the gluons below. Use either a specific past tense verb or the infix -ig-, as in Play, and choose the system that more closely aligns the two sets of suffixes with the two tenses.

ndà system

If the past tense morpheme is from MRCA ndà, then Dreamlandic would treat it as nč-...-ia because even the suffixes below are originally infixes which were reanalyzed because the two vowels on either side of the salient consonant were the same. Here, the vowels would not always be the same, and the final /-ia/ is a fixed choice. This means that the tense marker is outside the person marker, as a "hatelang" might do; this is the system that Galà escaped above. One advantage of this system is that it would allow the speakers to use the "B" morphemes below and not worry about the two separate verb conjugations.

ìg system

If DRM instead uses Play's /-ib-/ morpheme, this morpheme would be an infix into the dummy verb rather than replacing it. Thus the past tense marker would be /-ìgə-/ in the same sense that the present tense marker is /-ə̀(g)-/. It would also be affected by the two verb conjugations since it doesn't have a consonant to protect it from the verb stem.

Naively the person markers would then be infixed into the /ì/, essentially ruining the system and making it worse in all ways than the /ndà/ system. But whatever system DRM inherited would be quite old, and would not have lasted long were it so inconvenient. Since these are ultimately the same morphemes that Play uses, and were not originally person markers, it could be that the tense marker is in fact outside the person markers, just as Play uses opaque alternations like /-ap/ > /-aša/ instead of /-ap/ > *-ibap.

If the Play setup is the original, DRM would infix /-ìg-/ (that is, -(y)ić(ć)-) into the outermost morphemes listed below, and this might bind them to only using the A list. Therefore the tense marker would be further from the root than the person markers are, but would also seem quite heavy, as it would appear to carry the final vowel in addition (which would no longer obey vowel harmony with the root). That is, the past tense markers could appear to be something so long as /-a/ > /-yićći/.

NOTE: this might be a mistake, as it seems that the consonant should be /p(p)/ instead of /ć(ć)/. In either case it will probably be generalized to the singleton.

Mixsed system

Both morphemes could be used if /ndà/ becomes an aspect marker. It would not work the other way around.

proto-Dreamlandic stage (infix version)

12:16, 6 June 2022 (PDT)

By proto-Dreamlandic (1700 AD) there was no longer any way to summarize the required morphemes in a chart, as below. They had developed forms dependent on the words they were in, and appeared throughout the word (as in Lava Beds) rather than in a single compact morpheme. However, it is not a Lava Bed system either as it is missing two key features of the Lava Bed paradigm.

  • 3p > 3p tuppu-a
  • 1p > 2p nia-tupp-ili-u-n-
  • 2p > 1p i-tupp-ī-u
  • 1p > 3p nia-tupp-ik-u-n
  • 3p > 1p nia-tupp-iŋ-u
  • 2p > 3p i-tup(p)-ikk-u
  • 3p > 2p i-tuppu-: (that is, the vowel is lengthened. Historically, the morpheme causing this was an infix, just like in the words above)

It is possible that /ili/ above is just /il/, making it look more like the others, but the two major Dreamlandic languages both shift /lʲ/ > /l/ anyway.

The above are just one set of morphemes, since they depend on the shape of the word. For example, 3p>2p can be i-...-ip- instead.

It is also possible that the final vowel that above is /-a/ (which also changes) will be generalized to the other forms, even though it provides no useful meaning.

proto-Dreamlandic stage (suffix version)

This assumes that Dreamlandic essentially generalized all verbs into the single MRCA morpheme gə̀ and that it became a carrier for the infixes. Since it was itself a suffix, these infixes would appear to be suffixes.

For reasons not explained above (but due to tones), the /-n/'s have been removed as they no longer serve their intended purpose.

  • 3p > 3p tuppu-a
  • 1p > 2p nia-tuppu-a-la (this assumes deletion of the other /i/)
  • 2p > 1p i-tuppu-a-ya (this assumes DRM allowed /ʷe/ for a period)
  • 1p > 3p nia-tuppu-a-ka
  • 3p > 1p nia-tuppu-a-ŋa
  • 2p > 3p i-tup(p)u-a-kka
  • 3p > 2p i-tuppu-a-pa

These are still only half of the morphemes, because a second set is required for verbs whose final syllables had high tones in the MRCA. This is true even if vowel sequences like /ua/ are taken as having been /uga/, and so on. The "B" forms of -la -ya -ŋa -pa might be -ra -ća -ŋa -:, the last being a vowel lengthener. The /k/ morphemes will not change because they are moved out of their context.

Thus it could be said that the morphemes nia and i are 1st and 2nd person topic morphemes, as they give no information about agent or patient. Something similar happened in Andanese, but the Andanese speakers repaired the system by stacking the topic markers (which also served as classifiers) together. Dreamlandic would have less motivation to do this, though it would still be possible.

Dreamlandic is less likely to move morphemes from word to word than languages like Andanese, except for the topic markers.

Possible future evolutions

See Lenian languages for diachronics.

Note also that the final vowels must agree with the preceding vowels, making them useless, unless they take over the job of tense marking, but it would be awkward for tense markers to appear further from the root than person markers.

Perhaps DRM will create a past tense verb that then takes person markers.

It could happen that -ka and -kka erge into an emphatic active voice marker, bringing it to the 3p>3p position as well, while also allowing omission in the original positions. Thus there would be a system of only four morphemes, -la -ya -ŋa -pa, for 1>2 2>1 3>1 3>2, with the others usually unmarked. Again remember that the vowels are tied to those occurring earlier in the word, so the morphemes really just consist of single conosnants, /l y n p/, but there is no way to get rid of the final vowel.

The system above faintly resembles Play's /-av- -ay-/ and /-ām -ās/ (that is, the "dirty feet" morphemes). The optional /ka/'s are the cognate to Play's /-p/. If the Play system is stable, the reduced form of it could appear here, even though the Dreamers had no contact with the ancestors of the Players at this stage of their language. Proto-Dreamlandic was spoken about 2,300 years before Play, and it could be that the system above does not actually survive much longer.


NOTE: the below are the proto-forms, and are from different stages.

Maximal development

          AGT     PAT     OBS
 AGT      ---      əl      ək
 PAT       ey      ---     əŋ
 OBS       əḳ      əh       Ø

Minimal development

          AGT     PAT     OBS
 AGT      ---     lin     nan
 PAT      yi      ---     yi
 OBS      gə      li       Ø

Compare the older system below:

         AGT     PAT     IDT     OBS
AGT      ---     ln       nn      n
PAT       re     ---     rit      ri
IDT      mbə     lm      ---      m 
OBS       gə      l        t      Ø

A good incentive for Dreamlandic using -əli-...-n is that it would preserve consonant qualities. The MAP languages would not need this. At some point, the inserted -ə- needs to be justified; presumably it comes from a /gə/ which lost its meaning, which implies that it cannot have remained in an important slot such as OBS:AGT even in a far-off branch such as Dreamlandic.

The IDT fields disappeared and then resurfaced in the MAP clade, but the morphemes were totally lost and therefore the 4x4 matrix is for convenience. Nonetheless the functions were similar at both stages and it may be that the basic system works better with a 4x4 matrix.

Lava Bed 1P/2P matrix

10:58, 25 May 2022 (PDT)

There will need to be thirteen forms for every content morpheme to handle the 1P and 2P morphemes. This is 4x4 minus three, since AGT:AGT, PAT:PAT, and IDT:IDT cannot exist, but OBS:OBS does exist. (The "we" morphemes would be considered 1st person and marked in an entirely separate way.)

Each Lava Bed language will have its set of thirteen mutations. Since AGT and IDT often overlap, this list could shrink to nine, removing PAT:IDT, OBS:IDT, and their inverses. But AGT:IDT and IDT:AGT are still valid. This results in a graph with a stranded OBS:OBS off on its own. The others are AGT:PAT, AGT:IDT, AGT:OBS, PAT:OBS, and their inverses.

The system below assumes that the MRCA had the ancestors of Play's Ø k ʕ h gluons, plus another four corresponding to l ḳ y ŋ, for a total of eight. Note that /y/ was once /r/.

Possible MRCA matrixes

Minimal efficient MRCA system

Here there are nine gluons, but /h/ occurs twice in the assumption that context will disambiguate. It is possible that one of the /h/ sounds comes from an earlier sequence like /ti/ since at the very least /ḳ/ was also originally a sequence.

           AGT     PAT     IDT     OBS
  AGT      ---      Ø       l       k
  PAT       g      ---     ---      ŋ
  IDT       y      ---     ---     ---
  OBS       ḳ       h      ---     (h)

However, it is not likely that the system was ever this clean. The distinction between /Ø/ and /g/, for example, is mostly an elaboration of the MAP clade, and so there would be a collision between AGT:PAT and PAT:AGT, which are important to keep distinct. This indicates that perhaps /g/ was not the original value of PAT:AGT, or that the MRCA did not use morphemes in this slot at all.

Minimal defective MRCA system

           AGT     PAT     IDT     OBS
  AGT      ---     ---      l       k
  PAT      ---     ---     ---      ŋ
  IDT       y      ---     ---     ---
  OBS       ḳ       h      ---     ---

Here there are six morphemes, but AGT:PAT, PAT:AGT, and OBS:OBS are all missing, meaning that they cannot be indicated in any way by an infix. This stage of the language is too early to have vowel harmony and therefore cannot use e~o~ə for that (although such an elaboration would make perfect sense since there are exactly three empty slots).

It is possible that OBS:OBS was in fact filled, but with a morpheme that was later replaced by /h/.

A smaller matrix, assuming that a distinct IDT form was a Lava Bed innovation, and that OBS:OBS was the original zero-marked form, could be as such:

           AGT     PAT     OBS
  AGT      ---      l       k
  PAT       y      ---      ŋ
  OBS       ḳ       h       Ø

Because /ḳ/ is historically compound, this system cannot be very old. Since /ḳ/ often comes from clusters like /tr/ and /y/ comes from /r/, it may be that an even older system had a /t/ in the bottom left corner, which was lost because of sound changes that made dorsal phonemes more convenient. Therefore a still older system suggests itself:

           AGT     PAT     OBS
  AGT      ---      l       k
  PAT       r      ---      ŋ
  OBS       t       $       Ø

Here the dollar sign indicates a dummy phoneme, probably an /s/, that had become /h/ by the maturation date of the MRCA.

It happens that /t/ appears in the pre-primordial 2nd person pronoun, and since the infixed vowel is schwa, it could be that the pronoun itself was the infix, although this would not explain any of the rest of the table except perhaps the /ŋ/ for PAT:OBS (since /ŋ/ was the first consonant of the 1P passive morpheme). These consonants are in different places in their respective morphemes, however.

Further breakdown is unlikely, because while in theory there could have been /ti/ > /$/ > /h/, /ni/ > /ň/ > /ŋ/, and the like, the system itself may not have existed at such an early date.

Applesauce system

This assumes that the least appealing system was the original, based on the idea that /t/ was originally /gət/ and that therefore the full pronouns are being infixed at least for one set of morphemes. There is no way to get from /nam/ to /k/, of course, so this requires that another set of pronouns must have existed.

           AGT     PAT     OBS
  AGT      ---     l?m     nam
  PAT      rit     ---     ri
  OBS      Gət     l?       Ø

Here, question marks indicate unknown phonemes, but the required slots may not have even existed.

Note that this system only explains, at best, the /l y ḳ Ø g/ morphemes, requiring that /k h ŋ/ all be innovations and that the distinction between Ø~g be a separate (later) innovation.

This system is projected so far back that the syllable structure was different, and it could be that they were not infixes at the time, but freestanding words that came to be seen as infixes when CVC roots began taking them before vowel-initial suffixes that later came to be seen as part of the roots. That is, /CVC V/ alternated with /CVC-VC V/ and later it came to be seen as /CVCV/ and /CVCVCV/.

An /-ɨ/ suffix, for the accusative, could have supplied this paradigm just by itself, since it begins with a vowel and could conceivably appear further from the root than the morphemes that later produced the Lava Beds. e.g. yam nam ɨ would be the accusative form of yam nam "my fish".

Supersufficient MRCA systems

Contrapositively, the MRCA may have had more morphemes than it needed, including a robust distinction between AGT and IDT. This would make most sense if the morphemes were originally very short and that some of them arose from compounds such as /ḳ/ being originally /tr/. Here, the only blanks are AGT:AGT, PAT:PAT, and IDT:IDT. The rest need to be filled in, possibly with clusters of the others.

"Half Lava Bed"

          AGT     PAT     IDT     OBS
 AGT      ---      l       tm      m
 PAT       r      ---      ŋt      ŋ
 IDT       nt(r)   tn     ---      n 
 OBS       t(r)    ti      G       Ø

Here, /g/ and /Ø/ are distinguished even though the language may not have been able to do this phonetically. This is based on the assumption that the grammar was fairly complex and allowed for irregularities that might patch up where a lost phoneme once was.

/l r/ now look out of place and it may be that /G/ was not distinct from /Ø/ in the first place.

/tm tn/ were probably pronounced as /mm nn/. /nt/ and /ŋt/ would be distinct, however, and thus the (r) is unnecessary.

N Rim system

A much more likely alternative:

          AGT     PAT     IDT     OBS
 AGT      ---     ln       nn      n
 PAT       re     ---     rit      ri
 IDT      mbə     lm      ---      m 
 OBS       gə      l        t      Ø

There are three morpheme slots at this stage, and they are for AGT, PAT, and IDT rather than for 1st and 2nd person. Thus this is nothing like the Lava Bed system and a lot more like a traditional person-marking language.

Thus the morphemes are:

1st person
n agent, ri patient, m identity/possessor
2nd person
agent, l patient, t identity/possessor

Any fusion is due to regular sound change.

It is most likely that the -i accusative was added anaologically, and that the patient morphemes are really just r and l, and that these may have been both /l/ at some point, which could have been a pronomimal accusative affix that attached to the agent form. Thus, for example, /nl/ > /nr/ > /r/ and while /gəl/ > /gəl/, the /gə/ drops off by analogy or repetition. Thus there is no missing vowel, and the morphemes are:

1st person
n agent, ::m identity/possessor
2nd person
agent, t identity/possessor
Accusative suffix
l 1st or 2nd person pronominal accusative suffix (common nouns use )

Galà matrix


Tentative Galà system

           AGT     PAT     IDT     OBS
  AGT      ---      Ø      ol      ok
  PAT      e(n)    ---     ---     ṅt
  IDT      ē       ---     ---     ---
  OBS      ēh      è(k)    ---     əh

AGT:OBS could also be Ø in context, as in Play.

There is no schwa in Galà, so the /ə/ vowel here indicates a harmonizing vowel that can be any of /e o i/, with the last being very rare, or not show up at all (depending on the preceding consonant).

Likewise, the capitalized letters are harmonizing consonants that contrast with Ø and with each other. Σ is not B, but might come to be analogized to each word's B-value without necessarily becoming B.

Underlines and empty spaces are undetermined values.

ək for AGT:OBS can only have come from an earlier system in which it was IDT:OBS or PAT:OBS. Galà got rid of the distinction.

The /ol ok/ morphemes might actually be /il ik/ if some analogy is taken from prefixes. è for OBS:PAT is difficult but could have arisen at a time when there were still reliable classifier prefixes. Note that the distinction here between /e/ and /o/ is not etymological, but is still sound because it comes from the variants of /ə/.

There is no reflex of /q/ because it seemingly can only occur where /h/ occurs, and /qh/ > /k/.

2P agent can have a [B] value of /d/ or /k/, while 2P patient has a [B] value of /s/. This may allow the gluons to share homophonous values, especially if the single-consonant morpheme gets duplicated. This cannot be reflected back to the MRCA, however. The MRCA may have relied on vowels to show the 2P morphemes, but note that because /-u-/ is the plural, this would cause problems.

1P=IDT + 2P=AGT means such as "you use my soap to ____" and therefore a 3rd person morpheme is mandatory. The speaker is the owner of the soap, but not the agent.

More Lava Bed ideas

11:20, 17 May 2022 (PDT)

The locative construction only exists for objects in the "place" classes, which include o- for natural places and probably at least one for buildings and one for political entities. It is probably a bare locative. Therefore for example "under the sky" as in the Ring Poem would need to be two words, one for "sky-LOC" and one for the verb that means "under". However, this verb would be part of the antecedent, so e.g. "three-rings-for-the-elven-kings-under" would all be one word.

The real meanings of third and fourth person

08:48, 25 May 2022 (PDT)

Consider that the 3rd and 4th person arguments might be renamed as proximal and distal or some such thing. Traditionally in Earth languages (and even languages like Poswa) a 4th person is always a patient, whereas in a Lava Bed language the patient is usually (not always) closer to the verb.


It is possible that sentences like "The student read the book that the teacher assigned him" need to have identical markings on the word for book for the student (3P=AGENT), teacher (4P=AGENT) because it is the patient of two different verbs. These verbs would also both mark the book as the patient (4P=PATIENT), but the verb for "assign" is trivalent, so there seemingly needs to be yet another argument.

"i will give for him the Jedi Knights the justice they so truly deserve" \

Test in Galà

Leipzig glosses in Lava Bed languages may lie outside the limits of human comprehension; a layered system may make more sense. Or perhaps a table where each word is on its own line and columns show the mandatory morpheme slots.

hikìqa hunasàqa
hu-ì-h-Ø-ʔ-h-a hu-nal-h-a-ʔ-h-a
The sun loves me.

The Leipzig gloss is (with each word on its own line)


Note that the only difference between the two words is that "sun" (ì-...-Ø) changes to "love" (năl-...-a). The phrase thus can appear as just hinasàqa since all morphemes enclosed by like circumfixes can meld to each other. However, while saving time, such compressed forms are typically used for set phrases; here, it would be most appropriate if "sun-love" (with the sun as the agent) was a common concept.

Note also that syllables are split CVC-V because the infix always goes after the consonant.

A table-based gloss could be such as

   TOPIC      ROOT       1P       2P       3P       4P
   hu         ì          PAT      OBS      AGT      OBS
   hu         năla       PAT      OBS      AGT      OBS

Although AGENT and IDENTITY are separate grammatical concepts, it is most likely that they are merged in most of their forms, especially for inanimate objects.

Remaining issues

There are problems with the above use example, one being that -h- already means "3rd person is patient, and is male". In most languages this might not be such a problem, but here, animates and inanimates share the same morpheme slots and a sentence can make sense either way. It is not clear what the listener would think with a sentence where the arguments were PAT-OBS-PAT-OBS, but it might be that the actual table of morphemes will automatically shift one of the OBS arguments into an AGT/IDE argument and therefore change the meaning of the sentence.

The topic prefix hu- partially helps disambiguate, but it is not required that the topic be the agent or even that it be 3rd person. Even so, it could be said that the /h/ here is not a literal /h/ but a repetition of the first consonant of the word. Therefore it would still collide with /hi-/ and some other prefixes but not with wider sets.

Test in Galà (2)

This time the third person is not the topic.

   TOPIC      ROOT       1P       2P       3P       4P                       meaning
   hu         onàku      OBS      OBS      AGT      OBS                      planet
   hu         nat       (                              )                     to visit, be located at
The boy is on the planet.

It is possible that the /kā/, here placed in the third person slot, should in fact be in the fourth person slot with the third person slot occupied by the morpheme for celestial objects. This morpheme is just an /h/, but in the example higher up it was assumed that it acquired a CV shape over time.

It is also possible that nàto above will need to become natòko, in the thinking that if one /ə/ can be inflated into an /o/, all of them must be, and therefore there are two /o/'s.

Split-up spelling

For clarity, the morphemes could be spelled out as separate words, so the above would be h onaku nàto kā.

But this will cause issues with morphemes that overlap syllables; already there is a stranded /h/ and the morpheme meaning planet has lost its stress due to the following "word". It will still nonetheless make morphemes easier to recognize, and the "stuck" morphemes could be considered clitics. For example, the 1P and 2P morphemes would go to the last content word in the phrase; here nàto is thus visit.1POBS.2POBS.

If the proper word for "boy" were added into the sentence, it might require the topic to change. This could mean that the topic must always be in contact with the first content morpheme, and that they could be considered to be a single word even in the "split up" spelling style. This would not affect verbs however.

Some Andanese homophones

The Andanese word kupu means:

  1. pine tree (from MRCA kòpo, earlier /kaipə/)
  2. pine tree (from MRCA tʷòpo "treetop, canopy")
  3. pine tree (from MRCA tŏmbo "tree")
  4. pine sap (from MRCA kawòndʷu)
  5. young boy (from MRCA ndʷowòndʷu)
  6. barrier to soldiers (from MRCA ndʷòtʷo, cognate to DRM word for peace)
  7. nettle (from MRCA tʷŏpʷo)

The first three words were entirely unrelated in the MRCA but all fell together due to sound changes and therefore converged on the meaning of pine tree, the most common type of tree in Andanese territory. The word for pine sap was also unrelated, and had a root homophonous with that for young boys even in the MRCA, but in this case it was the classifier prefixes that coalesced. The two remaining meanings are also unrelated to the above and to each other.

Many of these words have cognates in Play, which lost its classifier prefixes and then added suffixes; the words here are presented without those suffixes. The first word, kòpo, is cognate to Play te and still means pine tree. The meanings "treetop, canopy" and "nettle" merged in Play as tapa, and are distinguished by classifier suffixes (tapafa and tapaa respectively), or by use in compounds. The word that originally meant a generic tree appears in Play as taa, with a meaning specifically narrowed to maple trees; it has also merged with an unrelated word for cactus. Play has pau for both "pine sap" and "young boy"; this word has also merged with some other unrelated words, and it is not the most common word for boy in Play, that being taā which looks related to the above words but is not. The remaining word, meaning a barrier to soldiers in Andanese, is not found in Play.

Play-speaking scholars recognized that many of the Play words on this list began with the voiceless stop t, and that the two words that did not also happened to be homophones, just as they were in Andanese. The Players knew of some other words that began with t in Play for which corresponding words began with k in Andanese. Like other scholars, the Players did not understand the details of sound change processes, but did understand pattern recognition, and therefore the Players realized that their language was related to Andanese.

Pine cones and fishing boats

09:16, 12 May 2022 (PDT)

The MRCA was a head-initial language, meaning that nouns preceded the object they belonged to. Thus "house of cards", not "card house", and so on. The exception was that animate nouns always came first, so that "horse ears" was correct, and "ears of (a) horse" was always wrong. This is important because it means that the derived nouns are themselves animate.

Primordially, the head-final construction was also head-initial, but the second morpheme was a verb. The classifier then served as a nominalizer. Thus, "horse ears" was really "that by which the horse hears", and so on.

With objects such as trees, the decision was more fluid, but because pine cones are parts of a pine tree, the animal model is followed, at least in languages where trees are considered animate. Though trees were not commonly the agents of verbs, they nonetheless still had some, and pine cones could be "that by which the pine reproduces", or the verb could simply be unique to the construction as though in English the word "cone" could be a verb specifically meaning to reproduce like a pine tree does.

Because a fishing boat is not made of fish, it is an inanimate object despite the fish morpheme, and therefore the order is head-initial, "boat of fish". Nonetheless, it does not simply pattern like the essive or partitive case. For example, describing furniture, such that the tree is no longer visible or attached, the head-first order is more common.

/ə ~ əni/

02:01, 12 May 2022 (PDT)

It is possible that /ni/ alone will not be the locative prefix in Andanese. Dreamlandic uses it as a suffix, but there is no imperative to suppose that the Dreamlandic morpheme order would be preserved in Andanese or Lava Bed languages generally. Rather, the inflected word /ə̀ni/, "in a place", would shift to just /nì/ after schwa loss, and this would become a standalone morpheme that more likely would behave as a prefix than as a suffix.

Note that the primordial form of the /ni/ morpheme was with a true /i/ vowel, not the more common /ɨ/. This means that if the construction were switched to /nì-ə̀/ even in pre-Andanese, it would have evolved into /ŋə/, and that a standalone prefix /ŋ-/ would take the place of /ni-/ whenever a vowel followed. This would be more efficient than any other system, but note that both Andanese and Dreamlandic separately lost the distinction between /ŋ/ and /n/ at least partially.

The ŋə form of the morpheme would evolve to ŋ̇ in Old Andanese, potentially leaving its effects in Late Andanese through sound changes but not surviving into the open-syllable era. In Galà it would simply become ŋ, filling the optional coda slot and thus meaning that no other morpheme could appear there.

Various related Dreamlandic ideas


If the Dreamlandic locative /nii/ is from earlier /-n-i/, it is more difficult to explain how it was retained in Andanese.

Phonemic gaps

05:52, 17 April 2022 (PDT)

Dreamlandic has a lot of gaps in its root stock because it derived new consonants from tones, and because it went through a stage early on in which all words had to have alternating tone sequences (either HL or LH), except that classifier prefixes and perhaps some suffixes were always low-toned. This meant that, for example, there were no words such as /pitu/, /sisu/, and so on, because fricatives and stops were forced to take turns heading the syllables, again with a few exceptions such as geminates and classifiers. Also, nasals never occurred together either, because /mimu/ > /mpimu/, and so on. This last rule was ignored if compounding two CV monosyllables, but even this process could not get stops or fricatives to stack.

Dreamlandic's m/t pronouns

16:37, 10 April 2022 (PDT)

Dreamlandic preserves the ancient suffixes -m- "my" and -t- "your" when padded by a following vowel of any origin. These were drawn from a tub of bath toys and thus the similarity to IE is a coincidence; even subconscious influence is not possible. In all other languages, the affixes were unnecessary because the early non-Dreamlandic languages used classifier prefixes such that "i" and "my" were the same, and so on. Tiny relics survive, such as Late Andanese /nu/ for 1p>2p (na-m-hə > nambə > nabə > nab > nō > no > nu).

MAP-associated morphology

00:05, 30 April 2022 (PDT)

The so called MAF languages are the best examples of Lava Bed morphology: Middlesex, Andanic languages, and the Fern languages that originated in the core tropical area but were pushed through Play territory to the east. Later, Play pushed further west and separated the Andanic languages from Middlesex as well. Thus, Play appeared at the center of this discontinuous distribution, and came to be associated with the languages even though it was not a Lava Bed language itself. This was helped by the fact that Play was notoriously difficult to learn even early on. The paraphyletic grouping MAP can be used for the cultural association between the MAF languages and Play. Here, Fern drops out because of its weak cultural prominence, the other languages being spoken primarily in the tropics. But, remembering the Ferns' early contributions, the grouping could also be labeled FAP.


Remember that Old Andanese and perhaps Middle Andanese (no longer listed here) had a stage intermediate between Lava Beds and the Late Andanese syllable harmony wherein words were built of portmanteaus with a single root and a single "person marker" which was itself a full content word. Canonically, this required at least a CV sequence in common between the two roots, out of a 75-syllable inventory (tones were ignored).


remember Yurok.

sappa sappi

and remember the type of word formation where only one element changes, meaning the information content is " 2nd syllable /a/ > /i/" ... this is slightly more nutritive than just saying /a/ > /i/ because there are four possible places for the change. in the original intent of this change, /sap/ was probably functioning as a pseudo-classifier, but it would not need to be so in a new language.

Lava bed third person markers

In some languages mostly spoken near Play, all words in the sentence are marked for their relation to all the nouns in the sentence. This includes the nouns themselves.

One such language is Galà, only distantly related to Play but with some remarkable coincidences in its grammar. Play and Gala were not in close contact; it is merely that both were very conservative languages and preserved traits of the MRCA that had in other languages fallen away. For example, the verbal conjugation paradigm was similar in both languages even though Gala was head-initial and Play was head-final.

Third person markers (additional information)

12:57, 28 April 2022 (PDT)

In Galà, the roots for boy, "teenager" (see above), and woman all have the null consonant /Ø/, just as they all have /b/ in Middlesex. This could lead to a setup where new consonants are inserted to signify biological sex and the vowels are left to signify age. Perhaps this is unlikely, though, and rather the vowels could be used to signify the 4th person marker while consonants signify the 3rd person marker. Then, the words for boy, teenager, and woman would shift in meaning to male, epicene, and female, meaning that the 4th person would not have marking for the age of the participant. Note that it is still possible to distinguish agent from patient in this morpheme even with no consonants, so the 4th person marker would not simply become a patient marker.

Yet another word for boys is găḳa, which survives in Play as a term of address (/žakas/ "hey, boy" and /žakap/ "I, the boy" (needed because of Play's lack of pronouns)). In Galà this would evolve into àa, which constraints with the existing /ā/ and which would also evolve to have the same form whether it stood for the agent or the patient. It might be seen as more "distant" and therefore default to a patientive meaning, and then later evolve into an atomic morpheme that indicates both agent and patient. This would not happen in the other Lava Bed languages and the term might not even participate in LB morphology outside the Andanic family. Thus it would be part of an isogloss with Play, meaning that the term was geographically bound rather than following linguistic family boundaries.

Gold has the word dʷàta "girl, woman", which is cognate to Andanese puta "child". The first syllable is the same as that in the word for boys below. This could replace /pùgu/. It might also be present in Andanic despite not having a /b/.

Middlesex 3rd person markers

The ā/ă alternation is early enough to appear in Middlesex and other Tropical Rim languages but they would not have the word for boy evolving to just a single vowel. Therefore this might be confined to Galà or at least to Andanic, and would need a name of its own, as it is not crucial for Lava Bed morphology and arguably does not "erupt" in the same way that the core Lava Bed infixes do.

Middlesex would be able to get to a stage where bā ~ bă meant "3P is agent/identity ~ 3P is patient" and that the 3P was a boy; with the /b/ remaining in place, it is most likely that the meaning stays close to the original instead of becoming generalized to humans as in Galà. Perhaps suffixes could be added to disambiguate meaning, resulting in the perhaps odd situation of the affixes for adults being derived from those for children. The existence of this word is mere happenstance; it seems as though Galà had it planned all along, but in fact, the root word was just one of many words for people. Thus, its existence in Middlesex is no surprise.

Words for humans at large

If Play's ta "human" has a cognate with a similar meaning in Middlesex, that word would be shorter than the word for boy; this is no guarantee, however, as the Play word (earlier /dà/) almost certainly underwent semantic shift from some narrower meaning and thus might not mean "human" in Middlesex or even in Andanic. Play's ta word is probably MRCA ṗò "teenager; adolescent", however, as it was gender-neutral from the beginning and had an easy path to supplant any preexisting word for adults because it was monosyllabic. Moreover, it came to also signify adults in Moonshine.


Middlesex could perhaps use the unrelated word ḷbă "human", which is their cognate of the ethnonym Lephal. This word also means "skirt", but this sense is most likely secondary, as it could have been used to set the people apart from tribes who did not wear skirts, but would have had no such meaning when they were in isolation. The PATIENT form of this would probably be ḷbà. Thus, a chain could set up, where ā <-> ă <-> à, in both Middlesex and Andanic, and would define the perception of tones. The syllabic /l/ would become plain except when preceded by a consonant. Then, this would pair with the preceding syllable and take the stress; this could be a problem since it would take the tone away. It is possible that /l/ will not do this in Middlesex, even though patterns suggest that it should.

Thus, for example, Middlesex /gìma/ "to trace" produces gimpabā "the boy traces" and gimpabă (or /gimpàba/ by stress-shift) "the boy is traced", but with the generic human affix, the resulting word gimpalba can either mean "the human traces" or "the human is traced".

Note that the dictionary for Middlesex currently has nh > (in other words, no change), but md > mb, which is unlikely to coexist. It is perhaps more likely that Middlesex will do /mh nh ŋh/ > /mp nt ŋk/ like its relatives in the Andanic and Fern families.

There is ANOTHER word for skirt which, with a different classifier, can mean "people" in Dreamlandic and Play, but means "woman; woman's skirt" in the Lava Bed languages, despite the fact that Play emerged from within Lava Beds and Dreamlandic was out of contact for 1,500 years. This is essentially a coincidence but can be explained by the Dreamers and pre-Players having their men wear skirts while the people in between did not.

Sex-based speech registers

Note that Middlesex has b as one of the consonants which men and women are made to pronounce differently, with women using [f] or [v] while men must only use [b].

Other Lava Bed ideas

Derivational cards

11:51, 27 April 2022 (PDT)

Also remember that "cards" were originally used for words like Altotta, which was built of the thematic consonants /l t/ and perhaps the vowel /u/. However, this word in particular could have also been syllable harmony if assuming was originally something like /alatautata/ with a thematic syllable /ta/ and the rest being "proper" syllables.

Like Play, the compounding is between two open classes, and there can be more than two root words in each compound word, highlighting extreme examples such as Late Andanese inuihuhatahupuunatata "road to a camp fire wood store" (derived from ihahukahaha "road to a store") bundling five inanimates together.

This system is derived from the gĭri speech register, which in Middlesex came to be seen as a children's speech register, but was merely seen as playful in other branches such as Andanic. It would also have survived in the Gold branch since it mimicked genitives, but then died out early on in Play.

This is derivational, not inflectional. Thus, the difficult word formation process was only encountered when the speaker created a new compound word, either as a proper name or as a new word for a complex concept uniting two or more basic concepts. Thus, unlike Play's noun-verb couplings, the listener did not have to immediately understand the new words in running speech, and the use of essentially opaque forms was commonplace.


dùhai dìta dìta dù shows that /ai/ behaved as one syllable. This phrase was actually pre-tonal. In fact, primitively even the vowel /ù/ is composite, so the thematic syllable is /dì/ and the /dù/ at the end is also composite. Nonetheless this situation would not have lasted into the recorded history of Andanese.

Still, one could respect a dissection of this such as

-w-ha-ɨ -ta -ta -w.

Late Andanese would preserve the morpheme boundaries instead of using both /di/ and /du/ as classifier prefixes. This would result in hiatus, so the resulting sentence would be such as

Suku tika tika su.

Although it is unlikely Andanese would preserve such short morphemes.

Compounds and word trains

15:37, 20 April 2022 (PDT)

If a word break occurs, the 1P and 2P inflections need to occur again, but not the others. The outer morphemes (topic/class and 3P/4P/etc) are only used once in any clause.

Therefore, compounds of arbitrary length can form provided all the words within have the same arguments for 1st person, 2nd person, and any other nouns present in the sentence. This is mostly seen with adjectives and serial verbs (e.g. "runs and jumps and skips" would be a single verb). This avoids the notorious "Angoram problem" where nearly every word rhymes.

Comparison to Play

11:54, 20 April 2022 (PDT)

Note that it is not the case, as one might expect, that each noun in a Lava Bed language is marked for its agent, patient, identity, and observer. Rather, they are marked for their relation to the speaker, listener, third party, and fourth party; the latter two of these are the two nouns in the sentence. In each of these four morpheme slots, one of four morphemes must be placed, marking whether the relationship is one of AGENT, PATIENT, IDENTITY, or OBSERVER.

Old ideas

The "Laban" language

08:04, 7 February 2022 (PST)

It is written in the red notebook that ALL of the Sea Turtles (Bombadiers (sic)) could speak Laban, and this language is the same language that arrived in Play territory in 4186. They knew that they were learning the language of their historical enemies, but did not consider it to be Dreamlandic. They considered Laba a distinct cultural entity even though, at the time, it was indeed part of Dreamland. (It broke away in 4186.)

NOTE: The red notebook ideas are extremely old, but I believe I had at least separated Late Andanese from Laba at the time, since Late Andanese was never the language of Laba, only "borrowing" from it in a plot hole that I later eliminated.

It is possible that this so-called Laban language could be one of the Lava Bed languages. This is not a subconscious association between the name of Laba and the English word "lava", because my teenage conception of the Laban language was precisely one that was so difficult to comprehend that outsiders could not even find the word boundaries. However, this would make it difficult to explain politically. So, another possibility is that the language the Turtles learned was simply Play, and that this was significant because at the time, there was still a sizable fraction of monolingual Andanese speakers in their population. Thus, rather than learning a minority language, they spoke exclusively the majority language. Still, this has problems of its own, since Šasuasa also speaks this language, and it would be unusual for her to speak Play in a Play-speaking nation and yet be known for standing out from the rest.

It is essentially impossible for "Laban" to be Middlesex, as Laba would have no incentive to let the Middlesex-speaking Crystals in; note that this was really Laba, not a rebel faction such as Lohi.

Possible survival in Gold

This system dies out in Play. It may have been reduced to a closed class already in Gold, but perhaps these words with rotating consonants could be seen as their own part of speech, maybe participating as verb endings for a time before freezing out into a closed class. For example from gās "feces" one could form tās "he (did|does) [VERB] disrespectfully" just by changing the neuter /g/ into a masculine /t/, thereby marking it as having a masculine agent, making it transitive. This has no tense marker, and it uses the bare stem of the noun, not an instrumental or some other case. There would also be some means of marking the patient on this, possibly Gold-specific, since it would likely rely on analogy such as schwa-dropping in the Gold branch and same-vowel dropping in the Andanic branch.

Doubly open fusional compounds

07:13, 6 April 2022 (PDT)

One very difficult aspect of Play for outsiders was the idea that two open classes, nouns and verbs, could fuse together to create new words that seemed atomic because they did not have an audible morpheme boundary and because both often underwent stem changes in addition. For example, pupa means book and vāu means to read, but neither of these words is audible in the compound word

The book of yours that you read to me.

Which functions as a noun just like any other, since it begins with a root and ends with a classifier suffix, just like /pupa/ itself.

This is the verbal embedding written of below. Although Play did use a certain small set of a few dozen verbs much more commonly than all others, the fact that a phonetic formula existed for the embedding meant that the class of embeddable verbs was in fact the class of all verbs, and therefore was open.

Comparison to other languages

In most other languages, even highly complex ones, when fusion of two content words occurred, one of them belonged to a closed class.

Survival in Poswa

Poswa, a daughter language of Play, allowed fusion of the full set of nouns and verbs in its inalienables, such as pupabo "the field I plowed", and like Play this involves stem-changing, but the mutations are confined to the middle morpheme and are more severe than in Play; therefore Poswa speakers did not coin new words such as this quite so freely as did the Players, and teachers taught these words as augmented possessives. Poswa's augmented possessives are directly descended from Play's verbal embedding paradigm, but despite both morphemes theoretically covering the full open classes of nouns and verbs, in practice the middle morpheme was constrained both semantically and phonetically. This is because of sound changes in Poswa.

Moreover, these words are always inalienable nouns, because in Poswa there is no way to embed the verb without immediately attaching a person marker to that verb. In Play, there were no person markers, and the final morpheme in a verbal embedding construction was a classifier suffix, making such nouns behave like ordinary nouns.

Moonshine circumfixes

06:18, 1 April 2022 (PDT)

Moonshine's circumfixes (e.g. s-[woman]-č "woman's bathroom") can arise from zero-marked accusatives and locatives, but rapid change and analogy is required. These can be thought of as a rearrangement of Play's verbal embeddings, perhaps taking them inside-out, since the middle morpheme is always a noun and is bookended by morphemes that could be thought of as verbal; nonetheless, the result of the Moonshine construction is still a noun.


11:40, 25 March 2022 (PDT)

Poswa has no demonstratives, and must use verbs with person markers on. (That is, "by me", "by you", etc.) It is possible that Play was the same way; but Play has a very complex noun structure already and adding a demonstrative suffix to it all would not overburden the speakers since it would only appear where certain other suffixes could not appear. For example, it seems logical that the demonstrative suffixes would never appear on nouns that were also tagged with the question particle tīs ~ tes (see below for derivation); if the question particle is also part of the same series of affixes as the "belief" mood markers, then it stands to reason the demonstratives can not cooccur with them either.

It would appear that Play does not have demonstratives, since verbal embedding and locatives are sufficient to cover both demonstratives of place and of deixis. Gold might have some inherited demonstratives, however, since verbal embedding was only just beginning to form.

Classical particles

Other particles

15:52, 3 May 2022 (PDT)

Dreamlandic requires the existence of a particle nə̀, with a meaning like "(if....) then", and which is a prefix to the second clause. Dreamlandic does not even preserve this itself, but this is because Dreamlandic essentially loses the entire system and replaces the particles with verbs and possibly nouns. This would probably appear in Play as na, either because of early /ə/ > /o/ > /a/ or because of the later vowel harmony rule. In Old Andanese it would be a syllabic , changing to a normal /n/ in Galà but with an unclear future in Late Andanese. If it survives it might be tied to a "rescue" morpheme. Normally, schwa can be rescued as /i/ or /u/ through the /e~ə~o/ alternation, but in this case /ni-/ and /nu-/ are important prefixes already.

Dreamlandic uses this particle in generating its word for "or", nimia(-ni), meaning that there was a longer form, nə̆ma-n-, either in the MRCA or in the very early history of Dreamlandic.

This also requires the existence of a particle mà-n- meaning "not", though this need not survive into Play.

Dreamlandic/PSL might be the only branch with a reflex of pre-MRCA , which is probably related to MRCA "person, human". It survives because the /e/ vowel is unusual and therefore the word did not collide with other words. Even so, it was not used alone, but only in conjunction with other particles. At least PSL also evolves MRCA hàli > a because of the pine tree rule (expected *yayi).

Question markers

06:46, 25 March 2022 (PDT)

Remember that Play's question markers tīs and tes both include fossilized 2nd person patient markers, meaning that the speaker is addressing someone, essentially saying "I ask you", and that this needs to be taken into account when deriving question words for languages such as Dreamlandic, which never had that set of person markers. Even Andanese would not have been able to use this.

Question markers in Andanese

Whatever solution is devised for Andanese may also serve as the general purpose solution for all languages not part of the Dreamlandic or Gold clades (Gold includes Play but not Trout). Andanese preserves the irrealis mood marker ki, which is the same morpheme that appears as te in Play (and its many variant forms in which both the consonant and the vowel can be changed). It is not cognate to Andanese ki "if" but the similarity of meaning may have led to them being merged early on.

Question markers in Dreamlandic

Dreamlandic will probably also have the irrealis mood marker, which would appear as si. (Not /ši/, which would be the cognate of the word for "ask".) This, however, is likely to have no role in forming questions in Dreamlandic.

Possible non-Play IE-style setup

As detailed at Play_language#Question_particles_and_suffixes, Play does not use case markers to form question words as does IE, even though the Play case system would serve such a purpose very well as the Play noun cases NOM/NOM/GEN/LOC/LOC correspond well enough to IE "who/what/when/where/why" and the duplicates could be rescued by additional morphemes. The accusative would also appear when the "who/what" words were the object of a verb. It is possible that somewhere else in the family, a system like IE's does exist.

The construction would most likely set up by taking one or a few bare CV stems, meaning "object", "place", "person", and so on, and attaching case markers to them, followed by the /ti/ question marker, which would later decay in some manner or other to leave either bare case markers or a slightly different form of the case markers. Even in such a language, it is possible that the Play strategy would be reflected in some manner.

Here is a setup for a toy language based on Trout:

"who/what" (subject or intransitive)
bəḳ "who/what" (accusative)
bəh "when; whose" (two meanings that merged)
bəl "for whose benefit" (dative); probably also covers "why"
bən "by what" (circumstantial); note that this noun case is rare on Earth but common in these languages because of the lack of pronouns. Essentially it presents in a sentence with an object and a verb, and asks who the subject is. It might not be necessary, however, if the nominative covers this.
bəg "where". This would probably have the same pronunciation as the nominative above.

There is no *bək. Also, note that /bə/ is the cognate of the numeral for 1, and so this form would be singular, and there would be corresponding duals, plurals, etc and most likely other forms indicating animacy and other aspects so that the auditory distinction between the various words would not fall on just the final consonant. This is particularly important since the system needs to have arisen at a time when it would be padded with a question suffix, which would mean that the consonants in the words above would not even be word-final.

This system could not have existed in Trout. Rather, these are the forms that would be constructed if the system had been in place at the time, and their reflections in a languagse such as Thaoa would follow the ordinary rules of sound cvhange.

tìhu doesnt exist

Since it would be odd to have a disyllabic question particle in a CV(C) language (that is, pre-Tapilula), the word pair tìhə ~ tìhu is likely spurious, and instead the two words could have been tìhə ~ tə̀hə, which would have given the same results in Gold and Play (at least for freestanding forms), and allow a cleaner analysis in the proto-language since /tì/, /tə̀/, and /hə̀/ already exist and have etymologies going back thousands of years. Play could still assume that its /tes/ particle had come from *tìhu by passing through a stage in which that particle was never inflected and was later adapted to the /-iCu/ paradigm.

Play/Gold innovations

The hə̀ morpheme is the primordial 2nd person patient marker. Therefore /tìhə/ would have meant something like "[I] ask you" and then evolved into a question marker as the patient markers fell out of use. Strictly speaking, the patient markers were placed after nouns, not verbs, so the /tì/ morpheme would need to be capable of functioning as a noun even if it were primarily a verb meaning "ask" or some such thing. Since only Gold would do this, Andanese *kiku cannot exist, and neither can there be disyllabic cognates in any of the other branches. This means that /tì/ and /tə̀/ were freestanding particles at some point, and it is possible that neither of them were able to function as a question marker on their own. If there is a question marker, it might need to rely on a morpheme somewhere else in the sentence to give it meaning since the particle itself would probably just be /tì/ and /tə̀/ would be a mood marker.

It is even possible that Play continued to see the true etymologies .... tìhə ~ tə̀hə .... and that the creation of *tifu happened post-classically. This would make sense if the scattering of the /ə/ in the mood markers to /a/ and /i/ messed up the original paradigm such that /t[V]/ was seen as the original question particle. Even so, it is unlikely that Play would be able to use the /tə~ti/ word without the 2p patient suffix.

Mood markers

06:41, 17 March 2022 (PDT)

See Play_language#Verbal_mood_and_associated_morphemes; these morphemes are primarily CV and the class traces back to Tapilula (even if the morphemes were slowly created and discarded). It is possible that Play's -pa, expressing a strong desiderative, is actually the same morpheme that means "or" and that it already had such a double use in Tapilula.

Late Andanese mood markers

Play's te, pa, and na particles are likely to have direct cognates in Late Andanese, and /pa/ is the same /pa/ below. Play uses ŋi for the negative, and this too is likely found in Andanese, but perhaps not as a mood marker. It would appear as /ni/.

Late Andanese does not use suffixes, so these morphemes could only be preserved if they were suffixed to an auxiliary verb that later lost its meaning. Since /tə/ > /t/ > /Ø/, the stem of the auxiliary verb would take over for the lost /tə/ morpheme. In the other Andanic languages such as Galà, suffixes were rare but still in use, and at least some preserved closed syllables, so the structure in Galà would be much as it was in Play, despite Galà and Play being otherwise almost opposite in morpheme order.

Late Andanese uses ki as its question particle. This is not directly derived from /tə/, even though that would be the expected form (when stressed), but rather derived from Tapilula /tìhə/. It might help preserve the /tə/ mood marker, though, so that it could occur freely instead of needing an auxiliary verb.

Basic four free particles

The "and/if/or/but" particles that are ka/ki/pa/pi in Late Andanese are directly traceable to the proto-language Tapilula, and are all high-tone CV sequences. Therefore they would all be expected to lenite their fricatives to stops in proto-Dreamlandic; however it is possible that they irregularly behave as though they were low tones because of their usually unstressed nature. Remember that /i/ > /ii/, and these particles are not the source of proto-Dreamlandic pi "of", which actually comes from Tapilula and shows the change of /h/ > /p/ because the PREVIOUS word imparted its stress pattern to the particle.

It is most likely that the particles ki pi had /i/ in Tapilula, even though Andanese could have derived its /i/ from an earlier /ə/. If the original vowel was /ə/, it would have exhibited the triple reflex /e~ə~o/ and then evolved to /i~Ø~u/ in Late Andanese instead of just holding /i/. In Gold and Play, the reflexes would be /a~ə~a/. The Play reflexes či pi therefore assume original /i/.

Dreamlandic particles

10:35, 3 May 2022 (PDT)

Bound particles

It could be said that there is no such thing as a bound particle, and that these are all classifier prefixes with very abstract meanings, the most common of which is "of".

archaic morphology in Dreamlandic

07:23, 11 March 2022 (PST)

Dreamlandic morphology, particularly the acc suffix -i, is very sick (as Moonshine teachers would describe it) by ~500 AD. It is sick in the sense that:

  1. it demands lost knowledge (the final consonant that didnt appear in the bare form),
  2. it ruins retained knowledge (it merges final /-i/ and final /-ia/ into just /-ia/),
  3. it even merges with a form it is supposed to contrast with (because /-ia/ > /-ia/). This is because primordial -a -ai -ə -əi > proto-Dreamlandic -ia -ia -i -ia.

It is possible, however, that /-ai/ > /a/ in this instance, using the "pine tree rule" where primordial /ai/ was read as though it had come from /aɨ/ which always shifted to /o/ and then to /a/ in Dreamlandic. Primordial /ɜɨ/ would most likely also shift this way although it is not listed in the sound changes. In this case, the /ai/ > /aɨ/ substitution is actually legitimate.

cardinal directions

See also Play_language#cardinal_directions.

left and right

The words for left and right are nouns in the Gold-branch languages, meaning "the right side", etc, which means that compounds and inflections are required to express concepts that would be atomic in English, and therefore that the morphemes can collapse to CV monosyllables and still be intelligible. Tapilula had ŭdə-k- for "right side" which turns into dʷə in Gold, a rare use of the sequence /ʷə/.

The word for right may disappear in Play, or be padded with additional morphemes. Even though Gold preserves /ʷə/ here, instead of the much more common /ʷa/, Play would change the vowel to /a/ because the classifier suffixes -ba -ya would both trigger the /ə/ > /a/ rule. The labialization would also drop out, potentially leaving just /a/ as the root.

One word for left in Tapilula was muhŭŋi-k-, which is cognate to a word that means treasure chest, and is not an atomic morpheme (but the word for treasure chest also has one additional morpheme, so neither of them contains the other).

Dreamlandic might preserve the /k/'s by taking both words as being verbs and therefore padding them with an affix. The nouns would then be derived from these verbs, meaning that two extra morphemes essentially cancel each other out, but the derivation would have long since become opaque by this time.

north, south, east, and west

Words used in navigation, from the Tapilula stage:

  • boat; navigable water (classifier prefixes differentiate the two meanings). It is possible the two meanings are untethered later on because Play has a word /pipi/ meaning "line, straight path", but which could have come from a meaning such as "boat in water", and this word could have even been used in Gold.
  • (g)à fish. Possibly the same as the next word.
  • (g)à to move, navigate, push.
  • ìḳi the sun. Gold reflexes point to wìḳi, which may be a remnant of a classifier prefix that hung on, or generalization of the plural /u-/ to uncountable nouns (there is only one sun, so it is uncountable).
  • wò ~ ùga water; ocean. Different from /pì/ above. Both forms of the word were in use even in the MRCA.
  • ndăku-kʷ- migratory; moving in all directions. this requires preceding /gà/, so it is actually gandăku, just as gòḳi below.
  • ndò-kʷ- a possible variant of above, assuming that /pʷŏndo-kʷ/ "fishing boat" is just /pì + gà + ndò-kʷ/.
  • gòḳi moving along an east-west axis (gà + ìḳi)
  • pʷò boat in motion (pì + gà). Possibly also the source of the Gold verbal mood marker -pa-, but largely irrelevant to navigation.
  • pʷòḳi boat in motion along the east-west axis; boat following the sun (pì + gà + ìḳi)
  • hʷò-kʷ- directional location (from earlier /$uBakʷ/, thus not part of the /mʷ/ > /mfʷ/ > /hʷ/ shift). importantly, /hə̀n/ + /hʷò/ produces hòtʷo, not *hòhʷo, because n$ʷ > ntʷ while nmfʷ > mfʷ > hʷ. In Gold and Andanese, this was almost certainly conflated with the unrelated /hʷò/ below, which led to different accent patterns.
  • hʷò cloud; to hide. Again, likely not involved at all, but listed here because of homophony in Gold & Andanese with the other hʷò. Dreamlandic preserved the distinction.
  • hàla north
  • hə̀n south
  • mà-t- east. possibly the same root as "plus one" in the numerics, because the new day's sun comes from the east
  • hàmʷu west

Gold developments

In Gold, pʷòḳi > paiḳ, and it stays as an atomic unit, even if the morpheme order of the other morphemes changes. This would pass on to Play paip, and Play scholars would still recognize that the /-ip/ part was related to their word for sun, pip, appearing just as if there had been lenition in a compound. Likewise pʷŏndo-kʷ and pwondăku-kʷ would appear in Gold as păda and padăku respectively; the first is already the Gold word for boat, as /pì/ was too polysemic. Thus /păda/ could take over.

Play vs Leaper

Play uses šavafa "north", šatua "south", sata "east", and šasuša "west" alongside similarly formed words for things like "north across a sea" and non-cardinal directions. In Leaper, these would be xalăxa, xàla, săta, and xasŏxa through direct inheritance if no analogy happened. Leaper would also have just as many accessory words as Play.

The fact that three of the four compass words begin with ša (Play) and xa (Leaper), while the fourth begins with the very similar-sounding sa- is a strong incentive for the words to retain the anomalous Gold prefixing morphemes like /paiḳ/ even though ordinarily the morpheme order would be reversed. /ḳh ḳs/ would lead to /k ks/ in Leaper and /p ps/ in Play (because the proto-Players would have been using the morpheme independently).

In Play, the initial consonants would all delete when occurring as the second element of a compound where the first is a locative. Leaper cannot do this, but could perhaps use nonsyllabic morphemes such as /kʷ/ to get better use out of the syllables.

Late Andanese

In Late Andanese, all four cardinal directions begin with /h/. They are halahu / hupu / hapu / hahuhu, and thus even more repetitive than those in Play and Leaper. In this case, retention of the /paiḳ/ morpeheme would do little good, since Andanese never did the Gold vowel collapse.


The four cardinal directions are unlikely to be handed down to proto-Dreamlandic. Though Dreamlandic is usually conservative, in this case the four morphemes are essentially ruined by the sound changes, turning into yaria / a / mia / yamuu for north/south/east/west, assuming no analogy and a free word order (that is, they were not fused with the suffix as in Tapilula, even though Tapilula was the parent language). The word for the directional is wapa.

International tile and block scripts

07:00, 2 March 2022 (PST)

The Clover kids used a floor tile script, but did not fully understand it. Essentially they used "Batam", believing that the floor tiles represented the shapes of objects rather than letters. One reason for the children's failure to learn the script was that STW did not teach them Late Andanese, and the Players had not yet come up with ornate scripts (neither tiles nor blocks) for Play. Thus the Clovers did not understand the other artistic scripts either. Yet, some of the kids understood the script better than others, and were able to notice patterns.

wipe attitudes toward the other side


14:50, 9 February 2022 (PST)

Acrobatic semaphore

Early on, sailors imitated the shapes of the letters in their syllabary by using their arms and legs. They used only static positions, not motions; however, some signallers moved so quickly that they were able to use incomplete motions, and sometimes jumping was used in place of certain leg motions.

The semaphore positions varied considerably from one culture to another. On top of this, the syllabary by nature had a fixed number of syllables (that is, there were only (22 * 22 = 484) possible glyphs), and different language communities that continued to use the syllabary adopted different spelling conventions. For example, when Late Andanese evolved a vowel inventory of just /a i u/, they used the glyphs for A E O to spell them, because the E and O symbols were simpler than the I and U symbols. By contrast, during Dreamlandic's earlier /a i u/ stage, the Dreamers used the glyphs that had earlier been used to spell O A U respectively, since that choice meant that all of their vowel glyph subparts faced the left. Thus, a Dreamlandic /a/ was an Andanese /u/, a Dreamlandic /i/ was an Andanese /a/, and the other vowels did not overlap. This made bilingual communication difficult both in the traditional script and with coded forms such as semaphore. This helped create the tradition of using a single maritime language, even when that language was culturally hostile to the signallers. The language that served this role towards the end of the "Gold" era was Late Andanese, primarily because of its simplicity and not cultural relevance.

Arms and legs only

The letters on which the semaphore code was based resembled human arms and legs, and always had a horizontal line through the middle, resembling a belt. (This was the only way to ensure the tops and bottoms of the letters always touched; note that Andanese "loosened the belt" by making it contrastive whether or not the belt was drawn in, and yet all its letter strokes were still connected because its total inventory was much smaller.)

The motions requiring only one arm are signed by leaving the other arm down by the waist. Two of the 22 consonants' arm motions are little used and 6 are undefined; none of these eight was considered a single consonant at the time of Tapilula. Only the least common consonants involve having two arms in different non-limp positions.

Tapilula had six vowels and there were seven leg motions defined in the semaphore guide. The unused seventh vowel, here symbolized as , stood for syllabic consonants and for any consonant found in isolation, as in a foreign word. Originally, Tapilula's few clusters had their own symbols because they were analyzed as units, but some descendant languages continued to use semaphore and had evolved many more clusters than their arms could handle. These languages thus came to use the ∀ motion to indicate a lone consonant.


The script was also tonal; the tones are marked by replacing the 6 leg movements with more difficult ones. Since there is only one tonic syllable per word, this does not cause great difficulty for the signaller, and helps the viewer identify the rhythm of the word.

High pitch was indicated by kneeling positions and low pitch by sitting down. Only the tonic syllable was marked, and by tradition, only the high tone was indicated by the signaller. Tapilula did not have low tones on monosyllabic words either, so the tone was not marked at all on a monosyllabic word. Since only the kneeling motion was required, the signallers often substituted a simple bow-like motion, bending their legs but also bending their upper body to create the impression of a greater leg motion. The Ǝ vowel could not be easily executed in a kneeling motion without injuring the signaller, so they substituted the otherwise unused (or "upside down A") vowel. This, in turn, was no problem, since there was no such thing as a high-tone vowelless syllable.

The high tone could also be executed by having the signaller jump in the air. Typically they jumped only very slightly, since a high jump would make it difficult to keep their arms and legs in the right positions. At first, the kneeling variant was the standard, since it mimicked the appearance of the high tone letters in the script.

But signallers felt it was more reasonable to indicate a high pitch by having the signaler jump higher, saying that the bar at the bottom represented the signaller being lifted up, and then languages arose in which low tones could have stress too, and therefore there needed to be more than one way of indicating a stressed syllable. At this point, the semaphore system lost its connection to the inherited script, since the inherited script only had a high/unmarked contrast, and therefore the signallers created their own signs. Jumping was used by some signallers and kneeling by others, meaning that the contrast between the two could not arise, and the low tone had to be indicated by sitting down. Still later, a three-way tone contrast appeared in some languages, but these typically did not use semaphore.

Variants with other body parts and motions

Because the standard semaphore script used only the arms and legs, there was no obvious use for the rest of the body, or for motion within a single sign. Early on, some signallers used hip motions instead of turning their bodies to indicate the I and U vowels, but still turned their bodies to indicate the E and O vowels. This made these signs more distinct. I and U were the only vowels in which the signallers' two legs were in different positions, again because of the script letters, and therefore were more difficult to sign than the less common /e o/. (This is why the Andanese selected the mid vowels e o to spell their /i u/ when their vowel inventory shrank.)

Preservation elsewhere

It is possible that a maritime culture that had a very conservative language could have preserved the body signals fairly well, though even here it would be impossible for the exact original values of the consonants to be preserved, since the speakers would be unlikely to remember the assignments for consonants that dropped out of the language and then later reappeared.

AlphaLeap could have preserved the signals even though their language quickly grew beyond 22 consonants, making sign language impractical unless they were to innovate even more arm signals, perhaps relying on using hands separately.

Remember the dream about Play culture preserving the original 22x22 syllabary despite its inapplicability to the developing Play language.

Andanese semaphore codes

The inherited semaphore signals disappear early on in Andanese, and Andanese cultural influence may keep them out of Play as well because even though both languages had much use among sailors, the Andanese speakers were more literate for most of their coexistence. This is because the Andanese reordered their syllabary to make common syllables easier to write, so that, for example, the /wa/ syllable came to be /u/. This would make it difficult for the signaller to keep things together in their mind. Play may nonetheless borrow the script from Gold or even rediscover it in its original form after the year 4175. AlphaLeap, also a maritime power, may have kept semaphore alive but this does not mean that the Players would imitate it.

The Andanese were not typically a seagoing people, but by this time they lived only among the Play speakers (the Pubumaus people), and therefore sailors learned to speak rudimentary Andanese even when their daily language, the language they spoke on land, was Play.

Stick semaphore

Late in their history, the Andanese nonetheless reintroduced a type of signalling similar to semaphore. They did not imitate the shapes of the letters, however. Instead, the Andanese had associated the letters of their syllabary with a specific body part, typically one whose first or last syllable was the syllable in question. Then, the signallers indicated each syllable by pointing with a stick to the proper body part, either with the end or the middle of the stick (that is, the stick could either point to or cross over the body part being indicated). This was done at the 36-syllable stage, not the 30-syllable stage. However this was not an efficient means of signaling, and did not replace semaphore.

Note that the signallers carried a long, rigid rod, and not one rod in each hand. This is why so many of the signs crossed over the body part in the middle and not at the end. Notably, left and right sides of the body were considered different signs; even though the language did not have inherently different words for e.g. left arms and right arms, the signs were different because they mostly had originated from a distinction between front and back or between outside and inside.

Visual impressions

Five of the thirty body parts used in the sign language could be considered obscene: hip, buttock, vagina, urine, anus. These were differentiated by holding the signaling rod in five different positions as it crossed over the signaller's genitals; this is why the hip sign was grouped with the others. Since Andanese words were quite long, this meant that nearly every sentence included one of these signs, and many sentences included quite a few, sometimes more than one per word.

The militant Helper faction of the Play party admired the Andanese body part sign language, even as they ruled that the Dreamers' semaphore code based on body motions was obscene and therefore illegal in Play territory. This was due to political bias: the Players claimed the Andanese script as part of their culture, and sought to cast it in the brightest light possible. They said that it would be highly inconvenient for a script based on body parts to ignore a region of the body that was so large, important, and conveniently located near the hands' resting position. By contrast, the Helpers and even the non-Helper Play factions criticized the Dreamers for what they called gratuitous obscenity, adding obscene body motions to a semaphore code that had clearly been fully sufficient without such signs, as it had been initially based primarily on arm movements with minor involvement of the legs.

Two-stick semaphore

Traditional semaphore was reintroduced to Andanese shortly afterwards. Here, they still did not imitate the shapes of the letters, but rather used a system where the signaller carried a stick in both hands, brightly painted and easily visible. The leg were not used.

There were two arms involved in all 36 signs, with six positions each, 0 1 2 3 4 5, of which the zero was the resting position. This would mean that there would be no way within the system to indicate word boundaries, but the signaller could simply turn around, bend their legs, or wave their arms, so this is no problem.

These numbers were counted as a two-digit base-6 number. Therefore, when the syllable inventory reduced to thirty, the signs with "5" as the big-end digit were removed.


06:43, 7 February 2022 (PST)

Cannot use the 1994MS pronouns *ā ē ō because they are used to generate the vowels for Tapilula. Nonetheless it suggests that there could have once been a generic third person pronoun instead of using gender markers.

oct 14, 2021

Vanamaa_Fana is a ridge, not a swamp

re-read "close to original writeups" document ... it has many unused pieces of history, e.g. "tinks on Nama" (sic)

Oct 1, 2021

"leaving just Ezra" indicates that at a very late stage of the war, STW rebelled and came to support Amade, and in particular the Firestone party. However they were likely moribund by that time.

Sep 23, 2021

Search all documents for sarabism.

"close to original writeups" document

There are details about the Rapala stage of the government which could be projected backwards to the Anchor Empire generally or else attributed to a revival of Thunder-era government policies. For example, note that one person was able to outvote the entire Parliament on issues relating to the military, but not on the other issues. This could also be projected forwards into Fayuvas. In this same system, Emon (who may not be mentioned at all in the current writeup, but is canonically the same person as the Red Sun) managed to have a total lock on power within a specific geographic region of the country, in all domains — meaning we could overrule his friend the Golden Sun on military questions, even though the Golden Sun had more than half of the Parliament's power on military matters by himself.

An old writeup called "close to original writeups" describes Taboo as "a lukewarm Crystal". These events seem to describe the war in 3958, not the later wars involving the Players and Raspara. Thus, this is the stage of government that preceded the Anchor Empire. Nonetheless, it is possible that it could have been revived later on.

The "Jecaja" city that the Womb Justice party moved into (whose name was Mirebane in at least one Dreamer language) was also mentioned in this writeup.

This CTOW document also states that the Tinks considered themselves a wing of the Crystal party after their treaty with the Crystals, meaning that they would no longer have been able to wield authority over the Players. At this time, the Players in Paba were bound by an agreement that the Play party was subsumed to Tinker authority, but the old writeup ignores the Players and may not have addressed the situation from their perspective; it could be that the Tinks had already pulled away from the Players within just the first few months of their reign.

CTOW also says that a Dreamer politician named Paetal (Nettanetu?) had been promoting another Dreamer invasion of the Anchor Empire and that the Flower Bee invasion was in response to this, rather than being unprovoked. It also describes the Ik army (in the Yoy language) as wanting to live in Tata, which may mean that they considered themselves Players, but because CTOW does not mention the Players by name, this is not certain. Tata would not have been thought of as a "middle ground" territory at the time because the Players in Tata were the ones occupying Dreamland, and were thus more anti-Dreamland than the Tinks.


Remember that the Dolphin Riders are the same group as the Neamaki, who were known for their contacts with Moonshine. CTOW even describes them as defending the Cold Men and the Crystals, while opposing Wax and AlphaLeap, and supporting both pacifist movements and dissent from within their own party. CTOW has the Neamaki victory in the year 4112, four years off of what it is in the current writeup. This document also puts the renaming of the DPR party to Gold in this year; thus, they would have been practicing a Gold-style international government for nearly a hundred years before they took over the Crystals.


Camia under the Theyape government was a democracy - meaning all citizens are in the government, and are at once executive, judicial, and military. (Since the country was technically controlled by the military, it was necessary to grant all citizens membership in the armed forces, whose actions were voted on by its members.) There were no offices, only quotients for each citizen showing the amount of voting power that person would have on a particular question. Everyone had a different quotient for each situation, and these quotients were constantly changing to reflect changes in the person and his environment. Superficially, this was essentially the same system as the Cold government. All actions had to be presented as questions and voted on by the entire population, which by 4425 numbered about 18 million, though it was growing very fast.

Dreamland's STW clone

Dreamland had over 400 years earlier come to make an alliance with Adabawa to fight against Camia. When the war ended, the war-era emergency government (called "Wamia Major") refused to pull out, and became even more repressive on formerly democratic Dreamland. Even when Adabawa fell from power in 3992, Wamia Major held on, moving toward a government independent of Adabawa. In 4014 they sealed off their territory, trapping the rest of the Dreamers inside.

Afterward they refused all change unless absolutely necessary; they felt their original order was correct and wished to preserve it as well as possible. Just as STW had grown inside the "Camian" government, a new STW-like corporation called Kapa (in full, Nobolē Kapa) began to form inside Dreamland. The Kapa corporation was entirely controlled by the stupendously rich Yukiese family, which kept the population weak and poor by monopolizing all wealth and refusing to sell goods to anyone who carried weapons. By 4544,placeholder date[4] the Gold-style Dolphin Rider government was vestigial; the real power lay in the Yukiese quasi-cephalist system that placed all military power in their hands, so that they would be able to survive and keep their wealth even if the entire country revolted against them.

The Yukiese enterprise was a tangled mass of red tape that was, in fact, a very good imitation of STW. The difference was that the Yukiese were concerned only with staying in power and keeping money for themselves.

Note that the Kapa name of the corporation is the basic form, but it could have appeared as any of kapa ~ opa ~ papa ~ pepi given different coinage dates (it was a compound) and analogy. The names beginning with /p/ would be possible reflexes only if the word had been created thousands of years earlier.

Other early developments

This may be moved toCosmopolitan Age.

Note that the "Camians" were planning on war against Dreamland, but canceled the war because their own allies were also arming themselves and Camia had more to fear from Bèdom than from Dreamland. Thus, Dreamland was never invaded.

In 4150, the Sepu Resinio party formed in Dreamland. Their name could be translated as Combs, Cover, or (pejoratively) as Underwear or even Diaper, as the party had purposefully chosen to use a term (resi)[5] that without its classifier prefix had a variety of possible interpretations. The Comb party that formed later on in Play territory was not named after this party.

Feb 15, 2021

User:Soap/Playful planets

this is partly on wikipedia now.
  1. This assumes that the original word is not simply the word for road, however, in which case both branches would have diverged.
  2. Sorry I still dont have the past tense morpheme placement figured out.
  3. Note though that /nanu-/ is normally replaced by /pu-/.
  4. from an unrelated event; essentially saying, "by the time this happened, in Dreamland they...."
  5. possibly cognate to a verb meaning to remove dirt rather than to a noun