Around the year 4178, the last surviving Andanic language went extinct as an everyday spoken language, but its ornate writing system survived, and thus knowledge of the language persisted for thousands of years. This preserved language is called Late Andanese.
- 1 scratch pad
- 2 Phonology
- 3 Phonology and vocabulary
- 3.1 Expressive wordplay
- 3.2 Allophony and sandhi
- 3.2.1 Allophones of vowel sequences
- 3.2.2 Allophones of syllables beginning with /h/
- 3.2.3 Other allophony involving /h/
- 3.2.4 Allophones of syllables beginning with /k/
- 3.2.5 Allophones of syllables beginning with /t/
- 3.2.6 Allophones of syllables beginning with /p/
- 3.2.7 Allophones of syllables beginning with /s/
- 3.2.8 Allophones of syllables beginning with /l/
- 3.2.9 Other allophony involving /l/
- 3.2.10 Allophones of syllables beginning with /m/
- 3.2.11 Other allophony involving /m/
- 3.2.12 Allophones of syllables beginning with /n/
- 3.2.13 Other allophony involving /n/
- 3.2.14 Allophones of syllables beginning with /ŋ/
- 3.2.15 Other allophony involving /ŋ/
- 3.2.16 Rate of speech
- 3.2.17 Voiceless sounds
- 4 Grammar
- 4.1 Classifier prefixes
- 4.2 Private verbs
- 4.3 Use of classifiers to derive new words
- 4.4 Derivation of words
- 4.5 Classifier prefixes and titular words
- 4.6 Dropping of classifiers in compounds
- 4.7 List of common noun classifiers
- 4.8 Consonant-based gender system
- 4.9 Second order classifiers and length of words
- 4.10 Private verbs
- 4.11 Syllable harmony
- 5 Orthography and contact with other languages
- 6 Notes
itanu inuhahuka inuimina inuihupuhu inuihuhatahupuunatata "road to a camp fire wood store"
The above sentence could be loosely glossed as "ROAD STORE WOOD FIRE CAMP", but this does not explain why the words get longer towards the end, and why the increase is much greater on the last word than on the steps in between. Andanese does not need twelve syllables to say "camp"; the last word in the clause must carry the agreement morphemes of all the others, and these are piled on both sides of the root.
Late Andanese has only 12 phonemes: the consonants /l h k m n ŋ s p t/ and the vowels /a i u/. And of these, the consonants /s/ and /ŋ/ are rare because they originated primarily from sequences rather than single phonemes. Vowel sequences are allowed, but final consonants are not. Thus there are only 30 syllables in the language.
The frequency of consonants is roughly the same as their order in the syllabary: /l h k m n p t ŋ s/. However, as detailed below, some consonants are more common before certain vowels, and it is more accurate to think of syllables rather than consonants and vowels as the minimal phonemic unit.
Andanese has three vowels: /a i u/. There are no tones, stress, or length distinctions. Unlike most of the languages of the Gold family, Andanese is vowel-strong: its vowels influence the pronunciation of preceding and following consonants, but the consonants have no influence on the pronunciation of the vowels. Vowels can only change due to the influence of adjacent vowels: an /i/ or /u/ before another vowel (even the same vowel) will contract into a semivowel. Other than this, there are no significant allophones of the three vowels.
Another way that Andanese differs from the Gold languages is that, although /a/ is the commonest vowel, it is noticeably less common by comparison to the other two vowels than it is in most Gold languages. The three vowels are all well represented in Andanese speech.
Phonology and vocabulary
With only 30 syllables in the language, many sentences, even with basic vocabulary items, are highly repetitive. For example hahaha is the word for "hat, cap, headgear" (ha- "shaped like" + haha "hair of the head").
Consonants besides /l h k/ are less frequent. The overrepresentation of the consonants /l h k/ (in roughly that order) is part of the reason why Andanese words are often so long even compared to other languages such as Play that have small phonologies.
Another reason why words in Late Andanese are so long is the use of extensive wordplay.
Boys' names were often extremely long-winded due to the deliberate repetition of similar-sounding words and syllables. Simple reduplication was avoided, but since the language had many homophones, long sequences of identical syllables piled up even so. By omitting classifier prefixes, the inventory of homophones increased vastly, with dozens of possible interpretations for each morpheme. Thus, even two boys with the same name could claim unique derivations.
The repetitive names were inspired by the Late Andanese process of syllable harmony, in which a whole syllable in a word changes to that of its classifier prefix. But, as with reduplication, Andanese names avoided syllable harmony, since parents considered it unimpressive to coin a name whose internal assonance was simply due to their language's grammar. Instead, every morpheme in each name was a proper word on its own and could be interpreted in a fixed number of ways.
One common name, Kukukukukukuku, was an extreme example, since it consisted of the thematic syllable /ku/ seven times in a row. All speakers recognized kuku as their word for "leg, run", but from there, thousands of possible interpretations of the name opened up, since, with classifier prefixes removed, the same kuku could also mean "canopy; place high up", "night", or "road"; meanwhile, the remaining ku could also mean "forest", "hand", "arrow (weapon)", or "to expose, show publicly". Thus one boy with the name Kukukukukukuku might tell friends his name meant "running down trails at night" (kuku kukuku kuku; trails being roads in a forest) while another would say that it meant "sitting on a treetop with arrows in hand" (kuku kukuku ku ku, treetops being high places in a forest). Stress was placed only on the final syllable, irrespective of morpheme boundaries, just as in the language as a whole.
A more common name template would intersperse a dominant thematic syllable with several minor ones. Kaaaaaaya was a common boy's name, where the thematic syllable is /a/ and the two minor syllables are /ka/ and /i/. Haaaaaaaaaaa was also a very common boy's name. Taaaamaaaaaamaaaaa was less common. Aaaaaaaaaaatataaaa is another common boy's name. Each /a/ is a separate syllable. The longest boy's name in common use was Kakakaaakakatakakakakakakakakaka.
Some names had two thematic syllables; an example of this was Matamataamatatataamaataaaatata. The bare /a/ was common enough in the language as a whole that its repetition here was not considered to create a third thematic syllable.
Some names had abbreviated forms. For example, Lilalaayilalalalalalaa was another very long boy's name, but could be abbreviated to Lalaaalai.
A classroom of children would often sit quietly when a teacher called out their names, as though hearing lottery numbers read off, since a boy with a lengthy name would need to listen for quite a while to know whether the student being called up to the front of the class was him or his friend whose name differed only on the thirteenth syllable.
Girls' names typically did not use repeating syllables, and were thus typically shorter, but parents often chose long names that did not use repetition. The number of girls' names in use was always much greater than the number of boys' names, and many parents coined novel names never used before.
Other proper names
The same pattern of repeating syllables showed up in placenames such as Mipatatatatai (Land of the Ruling Children of Tata) and party names such as Laaatilalatitiaa (Rusted Pearls).
Allophony and sandhi
Allophony in Andanese is best understood as a process that affects whole syllables rather than individually affecting consonants or vowels. With only 30 syllables in its phonology, the syllable has come to be the minimal phonological unit, and each of the 30 syllables behaves uniquely. For example, si behaves sometimes like ti, sometimes like su, and sometimes unlike either of them.
Allophones of vowel sequences
The vowel sequences ia ii iu are pronounced [ya yi yu], and this is reflected in Romanization. Likewise, ua ui uu are pronounced [wa wi wu.] In the case of ii and uu, the on-glide is weak but still distinctly present, and the second vowel is lowered slightly. Note that the Romanization here uses v for IPA /w/ and y for IPA /j/.
Triple vowel sequences are resolved by starting from the right. For example, tiui can only be [tiwi], never [*tyui]. And kiiii is [kyiyi].
The vowel [a] changes very little in any context. The sequence aa is simply [a:], and the sequences ai au raise the /a/ only slightly. It is common to find long sequences of vowels, including /a/, especially in derived words. Here, the same rules are followed, and thus aaa, aaaa, and so on, are simply more lengthened versions of /a/. However, with sequences of three or more /a/'s, the intonation resets on every odd-numbered vowel (counting from either direction). Thus vapaaa is pronounced [wapa:a].
Allophones of syllables beginning with /h/
As /a/ is the commonest vowel in the language, the allophone of /h/ that occurs before /a/ is considered the primary one. This varies between [x] and [χ] depending on stress and emphasis. Before the vowel /i/, /h/ is fronted to a voiceless palatal fricative [ç], and before /u/, it becomes a sound that varies between a rounded voiceless bilabial approximant [hʷ] and a spread voiceless bilabial fricative [ɸ].
Since the allophones of /h/ are entirely determined by which of Andanese's three vowels follows it, in a word-initial syllable the vowel may be spoken so quickly that the listener does not even hear it, and perceives instead a consonant cluster beginning with one of the above sounds.
Other allophony involving /h/
In rapid speech, syllables beginning with /h/ resemble those beginning with vowels, as the [h] spreads backwards over the vowel of a preceding syllable. This also entails contraction of the VC sequences -ih- -uh- into voiceless glides [ç ɸ]. For example, puha may sound like [pʰwa] or [pʷxʷa], and hihahuhi "womb" will be pronounced [çḁɸi], with the whisper tapering off only towards the end of the final vowel. In careful speech, this process is avoided if the preceding consonant is another fricative (/s/ or /h/), or if the preceding syllable is vowel-initial, and therefore in careful speech hihahuhi would sound like [çixaɸuçi], with four clear vowels, albeit very short ones.
Allophones of syllables beginning with /k/
In isolation, the syllables ka ki ku are pronounced like [ka ći kʷu], with the [ka] also slightly further back than velar. Before a vowel (including /h/+vowel in rapid speech) ki ku contract to [ć kʷ]. The sequences ki ku are still audibly distinct from kii kuu, however, because in the latter the starting point of the double vowel sequence is higher, and thus the glides [j w] are audible.
Before a syllable beginning with /p t s l/, the vowels in ka ki ku are spoken very quickly, and may be inaudible, the syllables thus being distinguished by their consonantal allophony only.
Allophones of syllables beginning with /t/
In isolation, the syllables ta ti tu are pronounced like [ta či ṭu]. Before a vowel (including /h/+vowel in rapid speech) ti tu contract to [č ṭʷ]. The sequences ti tu are still audibly distinct from tii tuu, however, because in the latter the starting point of the double vowel sequence is higher, and thus the glides [y w] are audible.
Allophones of syllables beginning with /p/
Before a vowel (including /h/+vowel in rapid speech) pi pu contract to [pj pʷ]. The sequence pu is still audibly distinct from puu, however, because in the latter the starting point of the double vowel sequence is higher, and thus the glide [w] is audible.
Allophones of syllables beginning with /s/
In rapid speech, the sequences si su are pronounced [š ṣʷ]. This contraction occurs even before consonants, except before another /s/.
Allophones of syllables beginning with /l/
The lateral approximant /l/ is advanced slightly before any /u/, and retracted slightly before any /i/, but these differences are very minor by comparison to the changes that affect /t/.
Other allophony involving /l/
The vowels in the VC sequences -al- -il- -ul- are often very short, or even inaudible, when the preceding consonant is a plosive (p t k). Furthermore, the /l/ itself often then becomes voiceless.
Allophones of syllables beginning with /m/
Other allophony involving /m/
The VC sequence -um- is pronounced like a syllabic [m̩] unless the preceding consonant is another labial.
Allophones of syllables beginning with /n/
Other allophony involving /n/
The VC sequence -un- is pronounced like a syllabic [n̩] unless the preceding consonant is one of t n l.
Allophones of syllables beginning with /ŋ/
Before a vowel, the sequences ŋi ŋu are contracted to [ŋʲ ŋʷ], and in rapid speech, these can become IPA [j̃ w̃].
Other allophony involving /ŋ/
The VC sequence -uŋ- is pronounced as a syllabic [ŋ̍] unless the preceding consonant is one of k ŋ.
Rate of speech
Andanese is spoken so quickly that speakers of neighboring languages perceive it as a harsh, guttural language consisting mostly of consonants, especially /l h k/, the three most common consonants in the language. However, even though /s/ is rare, it is the loudest consonant in the language, and speakers of neighboring languages often perceive Andanese to be rich in /s/ as well.
The voiceless consonants /k s p t/ are always voiceless unaspirated, even between vowels in rapid speech.
Andanese uses prefixes for inflection, and bound postbases for derivation. There are no exceptions to this pattern, despite the strong influence of the many Gold-family languages surrounding Andanese, which use suffixes and infixes, but never prefixes. There are still some infixes inherited from Old Andanese, but because they can only infect the prefixes, Late Andanese is best described as a language that has tables of related prefixes, with one row of prefixes for each infix, rather than a system combining prefixes and infixes.
There is no morphological distinction between nouns and verbs in Andanese, but there is a fixed word order of Subject-Object-Verb, showing that verbs and nouns cannot be considered the same part of speech. In this respect Andanese resembles isolating languages such as Chinese and to a lesser extent English.
All words have classifiers, except for a few that are sometimes considered to have a null morpheme as their classifier. For example, the classifier gi- means "humans, human body parts". Its accusative form is na-. However, classifiers for inanimate objects do not have distinct forms for their accusatives.
Repetition of classifier prefixes across nouns and verbs
Note that classifiers stack on top of the verbs. That is to say, any verb in the sentence will take a prefix agreeing with the noun classifier of the subject, unless it happens to already have the same classifier.
However, humans are spread across several classifiers, which mark different genders and ages of humans. These all share the same verbs, so it is not necessary to repeat the classifier before the object if the subject and object of a sentence are both human. For example, hinuhuki means teacher (adult female), tukuu means student (young child of either gender), and tuupi means to kiss. The accusative form of hi- is mi-. Thus one can say
- Tukuu minuhuki tututami.
- The student kicked the teacher.
Stacking of classifiers
Classifiers stack across nouns that are dependent on other nouns. For example, kuha means "(a) piece", and sikupi means "wood", so one can speak of
- Kuha kusikupi.
- A piece of wood.
However, when a two-word phrase such as this is used as an object in a longer sentence, the subject's classifier is repeated only on the first word:
- Hinuma hikuha kusikupi hikigi.
- The woman scratched the piece of wood.
This can be thought of more clearly with an analogy to mathematics. The sentence above can be visualized in three dimensions, with kusikupi as a branch off of hikuha, or as an equation:
- Hinuma hi(kuha ku(sikupi)) hi(kigi).
- Words with no classifiers
A small number of words have no classifier prefix. Many of these are proper nouns or loanwords. These are treated as if their first syllable were a classifier prefix, and therefore behave exactly the same as other words except that they usually seem to be in the wrong noun class.
Andanese preserves the private verbs of Tapilula, which also persisted in the Gold language but were dropped in all of Gold's descendants, along with the classifier system itself. Private verbs are morphemes, usually only one or two syllables long, whose meaning depends on the preceding classifier prefixes. For example, there exists a verb vutami "to gallop, run quickly on all fours". This verb begins with vu-, which is also the first syllable of the word for horse. Thus, if a horse is the subject, there is no need to repeat its classifier prefix before the verb; the classifier prefix is already there. If a man is galloping, one would say
- Kilatu kivutami.
- The man is galloping.
- The man is running like a horse.
Whereas if the subject is a horse or a similar animal, the sentence would instead be
- Vuhapi vutami.
- The horse is galloping.
- The horse is running like a horse.
Without the need for an extra syllable before the verb. Likewise, if the situation were reversed, one could say
- Vuhapi vugitami.
- The horse is running like a human (on its hind legs only).
The persistence of verbs like these is the main reason why Late Andanese has so many noun classes for specific animals such as turtles and rabbits with few nouns but larger numbers of verbs.
Note that the neuter prefix gi-, rather than the masculine ki- or feminine hi-, is used whenever animals are personified, regardless of the semantic gender of the animal.
Use of classifiers to derive new words
New words can be formed by copying a word from one class to another. For example, there exists a word pair of gipihi "sharp tooth; canine" and kipihi "to bite". Thus, a noun became a verb by simply changing the classifier prefix.
For some classes, the entire vocabulary can be assumed to be copiable. As above, gi- nouns denoting body parts pair with ki- verbs denoting the action of striking or hitting something with that body part, and with mi- denoting that body part served as food.
Likewise, the word for a young student, tukuu, is actually derived from the word for book, ikuu, by changing the prefix from i-, which denotes handheld objects, to tu-, which marks the "human children" noun class.
Derivation of words
Andanese is a head-initial language. That is, within a word, a classifier prefix will come first, giving the general broad meaning of the word, and each additional morpheme will narrow the meaning of the word further. This is the opposite pattern of most surrounding languages. Thus Andanese could be described as having a taxonomic vocabulary. Someone hearing the first part of a word will not know its precise meaning, but will have a general idea of what it might be. But someone hearing only the end of a word will have no idea which of the many possible categories of the vocabulary the word belongs to.
An exception to the rule that compounds are head-initial (i.e. "ears of corn") is that if one party is animate, the animate partner goes first; thus from vukia "horse" comes vukiakiki "horse ears".
The genitive prefix si- is never used within a word; this appears only heading up a full, independent word describing a definite object. Thus, all compounds are single words, and some can be very long. In general, Andanese words are often highly precise, and may seem long even given the small phonology of the language. Many words have been lengthened by adding additional morphemes over time, even when such lengthening was not necessary to prevent collision of the word with a homophone. For example, the word for heart, as inherited from Old Andanese, is vi. But this is padded with the body part classifier prefix li- on one side and the precising morpheme tu "blood" on the other; thus, the resulting word livitu can be analyzed as "body part heart of blood".
Reassignment of classifier prefixes
Words have moved from one class to another over time. For example, the inherited root word for snow, reflected in Late Andanese as gina, is now indifferent to the form of precipitation and only pairs with classifiers: gagina "snow"; vugina "rain".
Classifier prefixes and titular words
Classifier prefixes cannot be used as words of their own. Every classifier prefix has at least one titular stem, whose meaning simply repeats one of the meanings of the classifier prefix so that it can be used as an independent word. For example, the classifier prefix sa- means "love", but the proper verb for "to love" is sanala. Thus one would say
- Kikuhigi nanuma kitasanala.
- The soldier loved the nurse.
Likewise, ka- means "tree", but the full form of the word for tree is kakupi.
Classifier prefixes with more than one meaning will have more than one titular word. Since ka- also means "insect", the word kakui, meaning "insect, arthropod" is also considered a titular word.
Dropping of classifiers in compounds
Compounds of two nouns generally drop the classifier from the second noun, using semantics to disambiguate the possible meanings. For example, hikala "seashell" and lakala "bear" share the same two-syllable root, and have different noun classifier prefixes. But the compound word pugikala, which adds pugi "claw", can only be used to mean "bear's claw" because seashells do not have claws. If a speaker did want to specifically say "seashell claw", then they would use pugi puhikala.
Dropping the second classifier is mandatory when it is the same as the first classifier or of the same syntactic field. In fact, some teachers posit null morphemes before each element of a compound, each of which agrees with the outermost classifier. Thus, for example, they analyze pugikala "bear claw" as pugi pukala, and claim this proves that there exists a secondary Andanese word pukala "bear" alongside lakala, which is never used in bare form but is called up when speakers create compound words such as the above pugikala. This theory is one way of explaining the limits of which compounds are allowed to drop the classifiers and which are not. However, a dictionary based on this theory would list over a million words, most of which would be unusable duplicates of others.
- Note, the morphemes here may be in the wrong order, since animacy overrides head-first construction.
Locative and thematic classifiers
Assignment of newly coined words to classifiers usually corresponds either to the place ("locative classifiers") or the purpose of the object ("thematic classifiers"). For example, a spoon is not an edible object, but it is frequently found with them, and therefore the word for spoon, miguha is in the food class rather than the handheld object class; iguha instead means "shovel".
Inanimate objects have classifiers that do not change for syntactical active or passive roles. Since the subject of a sentence is always animate, however, their classifier is always buried underneath another classifier that repeats the subject's classifier. Note that there are no sound changes; if two vowels come together, they are still pronounced separately. Since this happens also to the verb, often a sentence will consist of three alliterative words. Thus one can say
- Tulata tuinuhu tuyula.
- The student threw the torch.
Here, while a clever student might expect the sequence /tui/ to contract to /pi/, as it historically did in many similar constructions, the morphemes remain separate and the sequence is thus pronounced [twi].
List of common noun classifiers
All noun classifiers are one syllable long, but some second-order classifiers can follow other classifiers and create what are effectively two-syllable prefixes. Even so, the outermost classifier in such a word is the one that echoes throughout the sentence, not the pair.
a : roads, streets
la : large land animals
i : handheld objects
ha : to worship; needle
ka : trees; some flying insects
u : water, liquid; land features (e.g. "beach", "hill")
ma : some grasses
ga : winter, things encountered in winter
na : accusative of gi- (all senses)
li : rung, plank, flat surface(?); water, liquid (alternates with u-)
sa : love
pa : some words for loose-fitting clothes
pi : pregnant women, babies, and couples
hi : tree (bark); worm; the ocean(?); bowl, cup, dish; adult women (nominative); men, boys (accusative)
ti : foot, motion
si : genitive prefix (etymologically hi-i-; often seen in contracted form as s- before vowel-initial words)
gi : protective objects; sharp, firm, protective; humans, human body parts; birds
mi : food; breast, nourishment; adult women (accusative)
ni : young girls; place of X, generic placenames; snakes
pu : succulent fruit (doublet of tu-); sexual reproduction, obscene body parts (only when following another classifier such as li-)
hu : fire; celestial objects; insects; tight clothes, "shaped like"; hair of the head, back of body;
tu : small plants; children; blood, bodily humors
su : oceanography (hi- + -u-)
lu : some body parts (a 2nd-order classifier that often comes between the gender marker and the root word)
ku : arrow, sharp weapon
gu : breast, nurture, nutrition (secondary use only)
nu : fruit; buildings
ki : verbs of motion (corresponds to li- body parts); weapon, claw; men, boys; darkness, night, sleep; some buildings
The sequences /ja ji ju va vi vu/ have taken on the role of pseudo-classifiers, since in initial position they are monosyllabic:
ya : some grasses; pineapple, large fruit; meat, food
vi : eye, vision, knowledge
vu : rain, water; horse, rideable animal
If the root begins with a single vowel, the syllables /yi yu vi vu/ reverse to [iy iv uy uv]. If the root word begins with /i/ or /u/ followed by another vowel, however, the normal pronunciation is restored.
However, sequences such as /sia/, /nua/, and so on have not been reanalyzed as single syllables, and cannot serve as classifier prefixes.
Note that there are many monosyllabic morphemes that are not classifiers. For example i means "shoulder", but is only used with a body part classifier li-.
Consonant-based gender system
Andanese inherited the consonant-based gender system from Tapilula. The genders are:
li: Babies; also used for humans of indeterminate age and gender (accusative is na-)
pi: Pregnant; adult males and adult females together; parents, childbirth
ki: Men and boys (contracts to k- before a vowel; accusative is variable, alternating between hi- or s- when acted on by "weak" agents and a contracted form of the prefix of the agent itself for "strong" agents)
tu: Young children (accusative is ti-)
ni: Young girls
hi: Adult women (accusative is mi-)
There is no neuter gender and no epicene gender. Note that females have two genders (three if "pregnant" is considered to be exclusively feminine) but males have one. This is a trait common to many languages of the area, and was present in their common parent language of Tapilula.
As in Gold, all objects found in bodies of water are grammatically feminine even if syntactically masculine.
A very small number of words show relics of the older internal mutations that marked gender in the Tapilula language. In Tapilula, classifier prefixes were dropped in more situations than they are in Late Andanese, and therefore the existence of separate stems for different genders was necessary. In Late Andanese, only a few such words survive, such as the word pair kilitu "king" and hiihu "queen". Even though the stems are different, the use of the prefixes is mandatory in Late Andanese.
Marking the accusative case
Late Andanese marks the accusative case by changing the noun classifier prefix, if there is one. For example, the noun classifier for adult women is hi- in the nominative (agentive) case, but this changes to mi- when a woman is the patient (direct object) of a verb. This case is still padded with the additional case marker of the agent itself, meaning that there will be two noun classifiers stacked together. This is the normal case in Andanese when an noun is the direct object of a verb. Note that inanimate nouns and most non-human animate nouns do not have separate forms of their noun class prefix for showing the accusative.
The masculine noun class, marked by ki- in an agent role, is the most changeable of all noun class prefixes, and the only one whose accusative form depends on the noun class of the agent acting on them. That is, a man stung by a bee will have a different noun class than a man kissed by a woman, in addition to the noun class of the agent which is stacked before the noun class for the man. Furthermore, the agent and patient forms merge before certain vowel-initial words, and the agent form also takes a special form when the patient is male.
The various interrelated patterns are all due to etymology. The original form of the agent prefix was tə- in pre-Andanese, and the patient was hə-. In early Old Andanese, this vowel either disappeared or changed to /i/, depending on environment. The /t/ soon changed to /k/, part of a general shift. Then, the sequences ki hi merged as s- before a vowel-initial word. But this change did not spread to words that became vowel-initial later on, meaning that only a few vowel-initial Late Andanese words trigger this assimilation, one of which is /atu/ "soldier". Further, although the nominative forms for male and female soldiers are both /satu/, in the accusative they diverge, with the masculine form being still /satu/ while the feminine form changes to /miatu/.
Any sequence of /hihi/ resulting from these shifts reduced to /hi/ by a general process of analogy. This occurred when a verb had a female agent and a male patient, but only when one of the other rules above did not force the inner /hi/ to contract into /s/. Lastly, the /hi/ disappears before many consonant-initial roots, as it evolved early on into a simple /h/ in these positions, which was then dropped.
Sample sentences below illustrate the different possibilities:
- Male soldier.
- Kuinau kusatu kutakiu.
- The bee stung the soldier.
- Hiuma hisatu hitakiihumi.
- The woman hit the soldier.
- Male farmer.
- Nianu nikapi nipihi.
- The snake bit the male farmer.
- Female farmer.
- Nianu nimikapi nipihi.
- The snake bit the female farmer.
Second order classifiers and length of words
Because of the noun classifier prefixes, most word bases are at least three syllables long. Exceptions are of two kinds: some very common words are used with no classifiers, and some words have classifiers but stems that are only one syllable. Generally these result from recent sound contractions.
Often, a classifier prefix is not sufficient to precisely define a word, and the word will take a second-order classifier between its "exposed" classifier that interacts with the grammar and the rest of the root word. For example, many body parts are classified under the prefix lu-, such as gilulali "head", which breaks down as gi- + -lu- + lali.
Most words of this type acquired their second-order classifiers only fairly recently, when drastic sound changes led to phonological collision of words even within the same noun class.
Some syllables can be used either as primary or secondary classifiers. For example, the -lu- above can only appear after a human gender classifier, but ku- "arrow, sharp object" often follows i- "handheld object" to create words for handheld weapons such as ikukivuni "slicing knife" and ikukuhigi "one-handed sword".
Andanese preserves the private verbs of its parent language Tapilula, whereas in the Gold branch of the family they disappeared early on.
Private verbs are verbs whose meaning is entirely dependent on the noun classes of the subject and object that precede it. For example, if the subject is "boy" and the object is "orange", a verb spelled lua could mean "to eat". If the subject is "boy" and the object is "girl", the same verb would mean "to kiss". The only commonality between the two verbs is that they both involve the mouth, and indeed, lua is derived from the word for mouth. The opposite meanings can be indicated by stacking additional classifier prefixes. For example, nulua means unambiguously "eat" in any context, so one can say
- Kupu kihipu kitanulua .
- The boy ate the girl.
This could be translated narrowly as "The boy mouthed the girl like she was a fruit."
Through cultural osmosis, similar phenomena later reappeared in some of the Gold languages, particularly those most closely culturally linked to the Andanese.
Due to historical sound changes such as /pua > pʷo > po > pu/, where a syllable beginning in a vowel was completely swallowed up by the syllable preceding it, Late Andanese evolved syllable harmony where most related languages evolved consonant harmony or nothing at all. Any a in the second element of a compound word could change to match the syllable of the classifier prefix. Often, a further classifier prefix will then appear before the original one since the meaning of the word has changed. Alternatively, the second word will agree with the first syllable of the first word in the compound rather than its classifier, since the grammar allows for the reinterpretation of originally independent words as sequences of classifier and short word.
This process is not used in people's names, as scholars consider the use of such vacuous alliteration to be a sign of an unintelligent mind, but it is used in common nouns and some place names. For example, a road to a store may be called ipunapupu, the compound of i- "road, open place", puna "store", and aa "road". Note also that the two main morphemes are in the reverse order from what one might expect, because the reinterpretation of the word for store as a one-off classifier makes the compound head-final instead of head-initial as in most compounds.
Orthography and contact with other languages
Andanese has had several scripts. Commonly, the Andanese wrote their language with one of a series of artistic syllabaries, each with 30 glyphs, which were based on square tiles. Of these, the commonest one was based on squares with 90° and 45° angles inside them. In multi-line texts, the boundaries of the squares would often be omitted, resulting in a shape that resembles a Tangram puzzle.
The primary script was derived from the 100-glyph Tapilula syllabary in a highly irregular manner. As the phonology contracted and evolved, many glyphs were dropped while others were repurposed. The actual original glyph derivations are:
[see red notebook]
Because most of the glyphs were taken from the top half of the syllabary, many of the original shapes did not appear, and the glyphs were smoothed into simpler designs over time.
During the last few decades when Late Andanese still had tones, they were spelled with the inherited glyphs for /va vi vu ya yi yu/. When Andanese lost its tones, these glyphs did not recover their original values, nor were new glyphs created, and therefore these sequences continued to be spelled with digraphs /ua ui uu ia ii iu/.
The syllabary had only thirty glyphs because Andanese teachers considered their language to have only thirty syllables. At first, these thirty syllables included two sets of symbols for /a i u/ depending on whether the hiatus was primordial or of recent origin; this had important effects on grammar. They had no glyphs for /sa si su/ because they considered these syllables to be underlyingly /hia hii hiu/. But the final stage of the language shifted the spelling to a phonetic one incorporating the earlier hiatus glyphs as /s/ glyphs. (That is, the glyphs themselves were repurposed, meaning that /a₂ i₂ u₂/ came to spell /sa si su/.)
The teachers never considered the sequences /wa wi wu/ or /ya yi yu/ to be proper syllables, even though they behaved as such in the grammar. Neither were sequences such as /mia mii miu/, whose surface pronunciation was monosyllabic, incorporated into the syllabary with single glyphs. The earlier practice of not writing /s/ made sense from this standpoint, as the distribution of /sa si su/ is very similar to that of sequences like /mia mii miu/ and unlike that of more common syllables like /ka ki ku/; the change happened when the script came to be widely used to write foreign languages in which /s/ was a common sound and could not be analyzed as an allophone of an underlying CV sequence.
The oddest looking script, "batam" (an exonym), was not a script at all but a means of drawing objects with the angular shapes of one of the many 30-letter syllabaries.
Persistence of Andanese words in other languages
Late Andanese loanwords in other languages are mostly pronounced according to the closest match in the recipient language's spelling, since knowledge of the original language disappeared shortly after the war in 4175. Thus the pronunciation of these words varies from language to language even as the spelling has remained the same for thousands of years. However, no tradition has introduced phonemes not directly descended from one of the 30 syllables in the Late Andanese language.
- "book" in dict, though
- this probably isn't the reason why gold switched, though, because in goldnthe anime pRtenr oftem.goes last
- NOTE. "humans, human body parts; birds" was originally listed under li but this is an error. Both come from Tapilula nʷə-, meaning that humans and birds were considered as one even then. There is alternate form of the prefix, nu-, which is the basis of the accusative.
- "Large enough to be eaten one at a time, but small enough to hold in one hand."
- from təlin "penis"