Zompist Bboard Again

Thanks for the encouraging replies. @Ares Land; I doubt that to be the case, but it'd be interesting to see how pied-piper-inversion can be integrated into an IA language naturalistically.

As for Dr. Anna Dew, in my mind's eye at least, she is a somewhat xanthippean natural blonde, late thirties, built like a cucumber. An academic at heart, she has cut her teeth in the budding field of Conprakrit Studies, working hard at researching, tutoring, supervising theses, publishing articles, meanwhile surpressing a black, phobic dread that one day she'll be professionally unmasked as a fraud. She drives a 2005 Fiat Multipla.

Vātika Prakrit: basic phonology

The basics of Vātika Prakrit phonology have more or less been canonized. The more exciting and intracte bits like accentuation and euphonic change (i.e. sandhi) are still under repeated revision, so that will be covered in a later post.

Before diving in, I want to apologize for the frequent references to Old VP and Sanskrit throughout this discussion; I simply had too much time not to make a mess! The earliest stages of the language, and the relationship to Sanskrit probably would have deserved a dedicated exposition. The earliest stages of the language and its relationship to Sanskrit probably deserve their own dedicated exposition. If there’s interest, I’ll gladly write it up in the near future.

For now, let’s clarify: VP refers to Vātika Prakrit in its golden era, known linguistically as Middle Vātika Prakrit, and belletristically as Dramatic or Scenic Vātika. The language’s evolution is divided into three stages: from the quasi-a posteriori proto-language Old VP, to Middle VP, and finally to Late VP. Unless otherwise noted, VP should be understood as referring to Middle Vātika Prakrit.

Consonants
By and large, the consonant inventory of VP is in line with those of what might be expected of a conprakrit. There are 32 phonemic consonants:

Labials: /m, p pʰ b bʱ, ʋ/; <m, p, ph, b, bh, v>
Alveolars: /n, t tʰ d dʱ, s z, l r/; <n, t, th, d, dh, s z, l r>
Retroflex: /ɳ, ʈ ʈʰ ɖ ɖʱ, ʂ ʐ/; <ṇ, ṭ, ṭh, ḍ, ḍh, ṣ ẓ>
Palatals: /ɲ, tɕ~cç, tɕʰ~cçʰ, j/; <ñ, c, ch, y>
Velars: /ŋ, k kʰ g gʱ, ɰ/; <ŋ, k, kh, g, gh, w>

Like other Middle Indo-Aryan languages, Vātika Prakrit has nasals in all points of articulation, and symmetric pairs voiced/devoiced/aspirated/unaspirated. This inventory closely resembles that of Sanskrit and the earliest stages of Vātika Prakrit were even more similar to it, likely functioning as one of Sanskrit's close relatives. There are some differences though on closer inspection. The palatal affricates have no voiced counterparts like they do in Skt. (*j *jʰ). The reconstruction of Old VP omits these sounds entirely. The Skt. and Old VP palatal fricative *ś manifests as c, s, or ṣ in VP.

To connoisseurs the odd one out in the table above would probably be /ɰ/. As a generalization we can state that this sound maps on Skt *h and visarga *-h̯. Old VP *h was probably more velar in quality than in neighbouring languages. I should probably reconstruct it as a a voiceless velar *ɣ~ɣ̟ rather than the plain glottal fricative.

Vowels
There are 9 vowels. These come in pairings where one maps onto one of the 5 cardinal vowels that all have a "reduced" variant. The mid high vowel /ɨ/ is shared between /i/ and /u/:

/a ɐ/, respresented as <ā a>
/u/ and /i/ both pair with /ɨ/, as <ū i u> respectively.
/e ɛ/ as <ē e>
/o ɔ/ <ō o>

All vowels are nasalized when followed by a coda nasal. The nasal articulation of the vowels is a allophonic function of the presence of a nasal, for example डुण् ḍūṇ /ɖũɳ/ 'meadow, pasture'. That syllable thus cannot have another coda consonant. This behavior can not be completely equated or traced back to that of Skt anusvāra *-ṃ.

There are some additional differences between VP and Skt.
- An absence of diphthongs. OVP *ai and *au have been resolved to /ɛ ɔ/.
- There are no long vowels. The length distinction that existed for some of the OVP cardinal vowels collapsed.
- The syllabic vowels *r̩, *l̩ did not survive the early stages of sound changes from the proto-language, but in VP they are morphophonemically regurgitated and resurface in the form of debris (colouring on vowels, stress shifts, aspiration shifts).

Syllable structure
Initial syllables can begin with a vowel or with one or two consonants, following the pattern (C)(C)V(C). However, all subsequent syllables follow the simpler maximum structure of CVC. Vowel sequences are strictly disallowed and the ways in which to rejigger consonants abck into their allowed clusters is governed by a swath of sandhi rules. Old VP shared most of its sandhi rules with Classical Sanskrit. This is not the case anymore in the middle period of VP.

Romanization
Some notes on romanization are warranted. During the earliest stages of development, I experimented with several transcription options, especially for vowels. My intuition told me to write the cardinal vowels without diacritics and to mark the corresponding "reduced" vowels with a circumflex, i.e.:

/a/ represented as <a>, /ɐ/ as <â>
/e/ as <e>, /ɛ/ as <ê>
/o/ as <o>, /ɔ/ as <ô>
/u/ as <u>, /ɨ/ as <û>
/i/ as <i>.

The system above was generally the most pleasing to me, with one exception that turned out to be a deal-breaker. The /ɐ/ is by far the most common vowel, and I disliked seeing reams upon reams of <â>'s. Another option was using digraphs, with <oa> for /ɔ/ and <ea> for /ɛ/, which looked promising. However, none of the digraph options for the high central vowel /ɨ/ like <ui> or <iu> were satisfactory and I similarly was at a loss on how to transcribe the reduced vowel /ɐ/. This left me with macrons as the final and definitive option.

So the current system indicates the cardinal vowels /a, i, u, e, o/ with the macron diacritic, while "reduced" variants of those vowels are transcribed without diacritic. Clearly, the most systematic option won, but at the expense of ease of understanding without explicit instruction. Usually, macrons indicate long vowels. In VP, macrons do not indicate vowel length, but vowel quality. Admittedly, this is confusing to me too at times.

The consonants were the easy part. The only wildcard is /ɰ/, which is written as <w>, mostly because of the graphemic similarity. For the most part, I just kept with the standard Sanskrit romanization, adding <z> and <ẓ> to parallel <s> and <ṣ>. The engma is written as such, as I don't care much for the dyslexia-inducing Sanskritic convention of a superdot (as opposed to the underdots marking retroflexion). Thats just too many little dots. The marginal or non-phonemic /h, ɽ/ get their own sign <h, ṛ>.

Raholeun wrote: ↑Tue Oct 08, 2024 8:16 amAnother option was using digraphs, with <oa> for /ɔ/ and <ea> for /ɛ/, which looked promising. However, none of the digraph options for the high central vowel /ɨ/ like <ui> or <iu> were satisfactory and I similarly was at a loss on how to transcribe the reduced vowel /ɐ/.

/ɨ/ eu
/ɐ/ eo/ao

Man in Space wrote: ↑Tue Oct 08, 2024 8:20 am

Raholeun wrote: ↑Tue Oct 08, 2024 8:16 amAnother option was using digraphs, with <oa> for /ɔ/ and <ea> for /ɛ/, which looked promising. However, none of the digraph options for the high central vowel /ɨ/ like <ui> or <iu> were satisfactory and I similarly was at a loss on how to transcribe the reduced vowel /ɐ/.

/ɨ/ eu
/ɐ/ eo/ao

Thanks for the feedback and the latinesque perspective. For now, I’m content enough with the direction taken, so I’ll be sticking with my current approach. The macroned non-long vowels will take some getting used to though.

One minor thing I would change is, for the sake of consistency, making /i/ <ī>, to match the other macron'd graphemes.

Xeniades' Indica

All this medieval India stuff may seem quite fanciful at first glance. One might even suspect themselves at the mercy of the fevered fabrications of this author. Fortunately, the situation is not exactly as such. We have more to build on than speculative fantasies when it comes to the historical veracity of Vātika Prakrit and its speakers. So let us instead turn to something far more tangible. Through the literature, we gain access to the accounts of those who traveled, observed, and recorded these distant lands. In this case, no account is more relevant to our purposes than that of the Byzantine traveller Xeniades. In his work Indica (Ἰνδικά), Xeniades offers a brief but valuable geographical account of the peoples, customs, and cities he encountered on his travels to the Subcontinent. Among many other places, he describes the the city-state of Pūravādha, situated in the northwestern reaches of India. We have reasons to believe that this area corresponds to the region where Vātika Prakrit was spoken. He writes of the place:

Indica wrote:“Pūravādha lies to the north and west, not far from the river-fed plains. Though it is on the edge of the great kingdoms that rule this land, it has remained independent for generations. The polis is governed by a peculiar form of kingship, unlike those seen in the empires to the east or west. Here, the king rules in name only, having no real power in the governance of the city. Upon the death of the reigning monarch, the oligarchs, who truly hold power, choose one of their number to ascend to the throne though it is often the most disfavored among them who is selected. This man is then sequestered in the royal compound, burdened with but ceremonial duties, leaving the governance in the hands of the council."

Xeniades seems to view this arrangement with a degree of incredulity, writing:

Indica wrote:"Thus, the oligarchs have contrived to rid themselves of their most despised members while at the same time maintaining the facade of kingship. The king is treated with all the respect his title commands, but in truth, he is isolated in the confines of the palace. It is a curious and ingenious means by which they avoid conflict among themselves, banishing troublesome individuals to the throne while they, the ruling few, continue to steer the course of the realm. [...] The people of Pūravādha trace their ancestry to the mountainous regions far to the north, whence they are said to have migrated before the campaigns of Alexander the Macedonian reached these lands. Their traditional garb still reflects the climes of their ancestral homeland. They favor their attire elaborate, colorful and not seldomly lined with pelts of deer and fox. Their women drape themselves in silks and fine linens, richly embroidered with intricate patterns and adorned with gold thread, and they take great pride in the dyes used to give their clothes hues of crimson, saffron, and deep indigo. Jewelry too is abundant; most pleasing to eye and ear are the anklets that chime gaily as they walk."

Xeniades does not merely confine his observations to the political and sartorial:

Indica wrote: “Much in the manner of the Thracians and the Corycians, for the Pūravādhans place great faith in the oracles and soothsayers that dwell in the wilderness. Men and women irrespective of caste travel into the countryside to seek the words of these pythonic maidens, who inhabit caves and forest groves of repute and where they prophesy about the seasons of the year, or of any impending public calamity, or even on private matters to individuals, or because it is not undignified for them to trouble about such things."

Of particular interest to the topic of this thread is the language of the Pūravādhans. As for this language, which is often equated with the Vātika Prakrit of classical plays Xeniades observes:

Indica wrote:"The Pūravādhans speak a corrupted form of the ancient Indian tongue, much debased by colloquialisms. However, they are not ignorant of grammar and recognize three sources of their speech. These are as follows: Tadbhava: words derived from older roots, Tatsama: borrowed terms, chiefly from Sanskrit, though they may also originate from other neighbouring languages, Pali, Gandhari, and even Hellenic, Deśya: vernacular words and expressions common among the populace."

Raholeun wrote: ↑Fri Oct 11, 2024 6:28 am Xeniades' Indica

Fascinating!

Vātika Prakrit: Less basic phonology

Before diving in, I have to admit that writing this in an engaging way was more challenging than I expected. Sandhi seems simple enough on the surface, but as I tried to lay it all out, I found myself grappling with a few tricky issues:

• I found it difficult to demarcate historical sound changes from those currently active in the language. What morphophonological changes are inherited from Old Indo-Aryan, which ones are the result of language contact and which ones are idiosyncratic, etc. The question of which changes are still active and which belong to the past is a bit blurry, and it complicates the decision of what to cover here and what to save for a post on diachronics.
• Secondly, the term "sandhi" should not become more of a catch-all term that it already is. Its exact scope can vary from language to language and it seems to encompass a multitude of processes, including assimilation, dissimilation, reduction and deletion, while also interacting with stuff like metathesis and floating segments. So, finding neat, generalized explanations that work for most cases was harder than expected.
• Personally, I don't really care to read a bunch of paragraphs filled with computationally precise lists of changes. It just doesn't make for a nice read. It's not as bad when the paragraphs are full of examples, but even with examples I don't know if any of this stuff makes for material that is both instructive and interesting.

In Vātika Prakrit, there are two primary applications of sandhi rules, which vary depending on the context in which the sound changes occur. Some rules apply in both cases. Internal sandhi occurs at the juncture of morphemes, members of compounds or between meaningful parts of words. These changes should be considered more historically significant. External sandhi on the other hand applies at word boundaries. These sound alterations frequently affect conjugation and declension paradigms, thereby adding complexity and increasing the potential for ungrammatical constructions. The orthography of Vātika Prakrit reflects changes caused both by internal and external sandhi.

There is another additional category of sound changes that fall outside of the definition of sandhi, as in they are purely allophonic in nature (in fact, they would have best been stated in the first post on VP phonology). These sound alterations are not represented in the orthography. They include: the nasalization of vowels under influence of following coda nasals, affricates /tɕ, tɕʰ/ (rendered as <c, ch>) become [cç, cçʰ] in coda position, the approximant /ʋ/ (written as <v>) shifts to [w] when it follows a labial in initial consonant clusters.

Internal vowel Sandhi
In an effort to partition the sprawling¹ topic of sandhi and other sound changes into comprehensible chunks, it might be best to start with the type of euphonic processes that are triggered by vowel-vowel sequences (consonantal sandhi is for a later post). In Vātika Prakrit it can be generally stated that two vowels should not come into direct contact. That is to say, vowel hiatus is avoided. This is true without exceptions when it is word internally. To resolve hiatus word internally, Vātika Prakrit employs one of four strategies: elision, coalescence, glide formation and consonant insertion. Note that some of the possible vowel combinations are quite theoretical, and won't really occur in practice.

1. Elision
When two vowels clash, one of them is often sacrificed. This typically happens when one of the vowels is /ɨ/ (written as u). For example, a + u simplifies to a, and u + ē results in ē. The same holds true for mid-tense vowels like ē and ō, which tend to overpower other vowels in the cluster. So, ē + o becomes ē, and i + ō resolves into ō. But, as with everything, there are exceptions: when ē or ō is followed by a lax a, the result is a lax mid-vowel, yielding e or o. In more complex clusters like ēā, ēō, or ōē, a consonant gets inserted (see point 4 below).

2. Coalescence
A common resolution for vowel hiatus is coalescence, where two adjacent vowels merge into one. The simplest coalescence, or at any rate the easiest to remember, occurs between like vowels, particularly the non-high tense-lax pairs. So, when two vowels of the same type (a, ā, e, ē, or o, ō) come together, they typically form their tense counterpart: ā, ē, or ō. For example, amrā + aḍavi = amrāḍavi 'mango forest'. Unlike in Sanskrit, Vātika Prakrit doesn’t have neat high-vowel pairs. VP /ɨ/ has two sources, first the reduced OVP vowel *ə̯, but also combinations of high vowels and as a consequence, combinations like i + ū, ū + i, and u + u all resolve into u, rather than producing ī or ū. However, combinations of the same high vowels, like i + i or ū + ū, stay intact as i and ū. When o or ō meets a or ā, the result is usually o, except in cases like oa or oā, which may coalesce into ō. For sequences beginning with e or ē, the result is typically ē.

Some combinations just have to be memorized, though. Notably, ai/āi and aū/āū typically become e and o, respectively.

3. Glide Formation
Another common strategy to resolve vowel hiatus is glide formation, in which the first vowel in the sequence is converted into a glide. This is particularly frequent when the initial vowel is high, such as i or ū. In such cases, i becomes /y/ and ū becomes /v/. For example, 'thus' iti + 'indeed' ē becomes ityēva (and after further sound changes to icēva). The lax vowels e, o show similar behavior, but not as consistently. Also, remember that combinations like i + ū or ū + i resolve directly into u, rather than an onglide.

4. Insertion
Not seldom, a consonant is inserted to avoid hiatus between two vowels. This inserted consonant can be one of the approximants (v, l, r, y, w) or an alveolar (n, t, d, s, z), depending on the phonological context or just plain convention. Once inserted, this consonant can undergo further phonological processes, making the range of actual sounds even greater.

External vowel sandhi
Fortunately, when it comes to vowel combinations at word boundaries, Vātika Prakrit is more forgiving, whereas Sanskrit or other Prakrits tend to be more restrictive. For now, the main principle governing external sandhi is simple: a word-final lax vowel can be dropped (apocoped?) if the following word starts with a tense vowel. We see this demonstrated in the example drūbkh-imi ē-ti → drūbkhim ēti 's/he goes in the fog'. In all other cases, a consonant is inserted to break the hiatus. This prothetic consonant is usually an approximant or an alveolar (see the above paragraph), though the specific consonant can vary and is not always predictable to me yet.

___________
¹) By this point, the vowel sandhi system in VP is evidently quite a bit more straightforward than the natlangs I’m drawing from. Part of the reason is that ē and ō aren’t considered composite vowels like Sanskrit’s /ai/ and /au/. Sanskrit textbooks often discuss sandhi that affects the syllabic liquids under the header of Vowels. By the time of Old Vātika Prakrit, these had gained an inherent "supershort", yer-like vowel that was more centralised and shorter than other vowels. By the time of Scenic Vātika Prakrit these will drive some interesting apophonic mutations, but it's probably best to detail these with the other grades later on.

Since I was digitally accosted today by the illustriously dependable mister Gorenc, my notes about the numbers had to be typed up. I might just share them here, although they are in no way spectacular. The reconstructions are from Old Vātika Prakrit:

1. yek < *ēk
2. dva < *dva
3. dre < *trai; the voicing is irregular, probably influenced by other Northern Indo-Aryan languages, possibly Kashmiri or some Dardic language if I'm not mistaken.
4. catār < *catār
5. panuc < *pañca; here we find that the root *pañca was reanalyzed as ending with a case suffix -a. In the accusative, this -a is dropped, and the stem is used, which was then thought to be *pañc, with insertion of the epenthetic vowel becomes panuc.
6. ṣat < *ṣaṭ; loss of retroflexion on the final stop is irregular.
7. satha < *sapta
8. aṣṭa <*aṣṭa
9. nan < *navan
10. daṣa < *daśa

Moreover, numerals are declined like nouns, i.e. for gender, for case and numbers above one also for plurality. The word for 1 "yek" is frequently used as an indefinite article ‘a, an’. Also, an eroded form is found as the reflexive verb prefix.

I'm really enjoying reading along with this!

Somehow I ended up taking 4+ years of Sanskrit as an undergrad, without ever being seriously tempted to bring that into my conlanging stuff. I don't really know why. But it is very satisfying to see you doing it

Thank you, fusijui. It’s wonderful to connect with another enthusiast. I have to admit to feeling more than a bit envious of your undergraduate experience with Sanskrit. A decade ago, I managed to take a single semester and that was mainly to fill a gap of 6 ECs. It was a refreshing diversion at the time from all the Slavic stuff. Unfortunately, we didn’t get very far in those 6 months; just some basics like learning Devanagari and a brief overview of morphophonology and the essentials of grammar. Now though, it is great fun to continue exploring Sanskrit through the lens of conlanging.