As opposed to my more mainstream way of describing my languages by fleshing a grammar page out as I go through the process of making them, I'll be doing something more unconventional with Classical Marpanni: I'll document the process of me making the language in a linear, chronological manner, in hopes of giving others an insight into how I work.

Initial DesignEdit

Before I begin working on a language, I tend to contextualise it either by figuring out key ideas or positioning it within some conworld context. For Marpanni, it's primarily going to be the second. I then go on to rough out the general morphosyntactical and typological traits of the language.

Compared to modern contemporary Mestian, Classical Marpanni (from which contemporary Marpanni, a neighbour of Mestian, is descended) is a language no longer spoken, but is still used by a fairly large amount of educated people as a literate lingua franca, a position it is losing to the influence of Mestian. It owed its influence to the literate merchant classes that spoke it and used it as a language of literature. As such, a vast body of native literature in it is preserved.
Classical Marpanni, much like its descendants, is a very inflecting language. Uncharacteristically of languages on the Dragonforge, it is morphologically split-tripartite; its contemporary vernaculars have, in various ways and to various degrees, resolved this oddity.

I don't tend to put much weight on typological classifications like "fusional" and "agglutinating": many agglutinating languages have the occasional fusional morpheme, and the same also applies the other way around.

At this point I usually decide what orthography I will use for the language:

Classical Marpanni is written in its own native alphabet, even when typeset within texts written in the Imperial script.

This just means I'll have to make a new script for it eventually.

At this stage I'm still working in the roughest of drafts and vaguest of ideas. I decide the syllables of the language will probably be (C)(C)V(C)(C), with a strong preference for a maximum internal -CC- clusters. I'll also employ phonation distinctions beyond just plain voicing (or absence of it). This is where I get into phonology.

Preliminary PhonologyEdit

Even though the end result is usually fairly intricate, I usually start work on the phonology by taking some general design principles and broad strokes, and then whittling it down until it resembles something satisfactory. I never really finish work on the phonology in one go: the preliminary design is always a rough indication of the phonemes and occasionally also phonotactics and some morphophonological processes. I don't actually work out the phonotactics of my languages all too often: they're sometimes implied by the few sample words of the language I begin working with — but more on that later.

I usually stary working on a phonology with a change of venue: I carry my sketchbook with me to my lectures and internships; I usually work on my projects during my breaks. I "doodle" the things I plan on implementing and occasionally etch out a few preliminary phoneme lists. I find it helps get my creativity flowing.

The ideas I wanted to look into were the seldom-used phonation features such as breathy and creaky voicing.

The first thing I do is think about whether I want to conjure some special look or inspired feel. Due to how long I've been doing this, I have a lot of mental reference to draw from without looking things up much. I usually have my languages only vaguely resemble natural ones, if at all; I tend to form language zones in my conworlds so that they vaguely resemble each other in ways that suggest areal groupings. I've done enough conlanging to be able to make fairly naturalistic phonologies that are typologically not incredible. The more you do this, the easier it becomes.

A primary freature I want to include is prominent glotallisation, and the phonation features employed will be creaky~stiff, modal and breathy~slack.

I often use a different colour in my notes for intercalated phonetic text and orthographic representations of my conlangs, to set them apart. As I use fountain pens, my main text is usually written in blue and I use red for emphasis: here, I'll use blue instead of red.

I usually begin with consonants, most frequently plosives, and slowly work out the phoneme distribution according to places of articulation. The preliminary plosive list:

Plosives: ɓ tʰ tʼ ɗ ʈʰ ʈʼ ᶑ kʰ kʼ ɠ qʰ qʼ ʛ ʔ

This first, rudimentary plosive list is usually way too cluttered: I tend to drop some plosives quite quickly:

Plosives: ɓ tʰ tʼ ɗ ʈʰ ʈʼ kʰ kʼ ɠ qʰ qʼ ʔ

This more assymetric system follows a typological tendency about implosives — fewer than egressives — and is generally more pleasing to me. Still, it has a bunch of plosives. Here I introduce the primary phoneme feature set:

The phonemes from the set /tʼ ʈʼ kʼ qʼ ʔ/ are treated as fortis glottalised consonants, and those from /ɓ ɗ ɠ/ as their lenis counterparts. The tenuis set /tʰ ʈʰ kʰ qʰ/ doesn't have a fortis/lenis distinction. Their phonetic realisation depends on the phonation of nearby segments.

Ha, that showed them!

I usually leave the plosive (and, generally, consonant) features implicit in the main lists, instead preferring to describe them outside tables and lists, inside a body of text.
The places of articulation I'm going to use are numerous and diverse: the plosives span the labial, alveolar, retroflex, velar and uvular POAs, and I'll introduce a palatal one too, for other segments.

Next up go the fricatives:

Plosives: ɓ tʰ tʼ ɗ ʈʰ ʈʼ kʰ kʼ ɠ qʰ qʼ ʔ
Fricatives: sʰ s ʂʰ ʂ ɕʰ ɕ xʰ x χ
Lat. Fricatives: ɬʰ ɬ ꞎʰ ꞎ ʎ̥ʰ ʎ̥ ʟ̥ʰ ʟ̥

You can see I've introduced lateral fricatives into this, in almost all the places I have central fricatives (only lacking uvulars), but I believe I have too many. Fricatives now come in aspirated and tenuis/fortis and lenis pairs, no glottalisation involved. A more refined list of fricatives:

Plosives: ɓ tʰ tʼ ɗ ʈʰ ʈʼ kʰ kʼ ɠ qʰ qʼ ʔ
Fricatives: sʰ s ʂʰ ʂ ɕʰ ɕ xʰ x χ
Lat. Fricatives: ɬ ꞎ ʟ̥

I gutted five fricatives, all lateral. What bugs me hard is the lack of dedicated palatal and velar lateral fricative symbols, as well as a voiced retroflex lateral fricative. The fricative inventory completely lacks labials. This is a design choice: Marpanni will have a limited repertoire of labials, as is already obvious.

Next up are the nasals:

Plosives: ɓ tʰ tʼ ɗ ʈʰ ʈʼ kʰ kʼ ɠ qʰ qʼ ʔ
Fricatives: sʰ s ʂʰ ʂ ɕʰ ɕ xʰ x χ
Lat. Fricatives: ɬ ꞎ ʟ̥
Nasals: m n n̥ ɲ ɲ̊ ŋ ɴ

"Not a lot" of nasals at first, but they also have a dimension of voicing. I tried toying with phonation and nasals:

. . . .
Slack Nasals: m̤ n̤ ɲ̈ ŋ̈ ɴ̈
Stiff Nasals: m̰ n̰ ɲ̃ ŋ̃ ɴ̃
Modal Nasals: m n ɲ ŋ ɴ
Voiceless Nasals: n̥ ɲ̊

Aside from crappy font support for combining diacritics in italics, this would make for a whooping 17 phonemic nasals! This is, obviously, way too many, thus I go back to what I did, cutting out the extra nasals and re-delegating them to allophonic status. Furthermore, I think I'll gut the uvular nasal (making it, too, only an allophone next to uvulars), and restrict voiceless nasals to a prevocalic, initial position. The final nasal inventory is thus:

Plosives: ɓ tʰ tʼ ɗ ʈʰ ʈʼ kʰ kʼ ɠ qʰ qʼ ʔ
Fricatives: sʰ s ʂʰ ʂ ɕʰ ɕ xʰ x χ
Lat. Fricatives: ɬ ꞎ ʟ̥
Nasals: m n *n̥ ɲ *ɲ̊ ŋ

Next up are the affricates:

Plosives: ɓ tʰ tʼ ɗ ʈʰ ʈʼ kʰ kʼ ɠ qʰ qʼ ʔ
Fricatives: sʰ s ʂʰ ʂ ɕʰ ɕ xʰ x χ
Lat. Fricatives: ɬ ꞎ ʟ̥
Nasals: m n *n̥ ɲ *ɲ̊ ŋ
Affricates: tsʰ tsʼ ʈʂʰ ʈʂʼ tɕʰ tɕʼ
Lat. Affricates: tɬʰ ʈꞎʰ kʟ̥ʰ

Nothing too fancy here. Lateral affricates have no ejective counterpart, there is no lateral palatal and no central velar affricate. Uvulars and labials are fully lacking, as is the tendency established by now. Furthermore, I'd note below:

The lateral affricates /tɬʰ ʈꞎʰ kʟ̥ʰ/ can be realised as the aspirated fricatives [ɬʰ ꞎʰ ʟ̥ʰ]; these realisations are in free variation.

Thus, I reintroduce aspirated lateral fricatives, though only as occasional variants of their affricated counterparts.

The rhotic and lateral consonants I tend to group into a general class of liquid consonants. They'll be fairly simple in Marpanni, possessing only a single phoneme with a lot of phonetic wiggle room:

Plosives: ɓ tʰ tʼ ɗ ʈʰ ʈʼ kʰ kʼ ɠ qʰ qʼ ʔ
Fricatives: sʰ s ʂʰ ʂ ɕʰ ɕ xʰ x χ
Lat. Fricatives: ɬ ꞎ ʟ̥
Nasals: m n *n̥ ɲ *ɲ̊ ŋ
Affricates: tsʰ tsʼ ʈʂʰ ʈʂʼ tɕʰ tɕʼ
Lat. Affricates: tɬʰ ʈꞎʰ kʟ̥ʰ
Liquids: r
The liquid /r/ possesses a wide spectrum of allophonic realisations, either retroflex or alveolar, with a tendency to surface as a lateral flap; it is generally found somewhere in the set: [r ɾ ɻ ɺ ɽˡ].

I took a two days' break to sort out the approximant ideas I had. I generally find myself working on languages either almost too much, or not at all for weeks on end — this all helps me work better, refreshing my mind and keeping my work from going stale.

I'll have a very generic set of approximants that are underspecified — their pronunciation will depend a lot on what vowel they precede:

Plosives: ɓ tʰ tʼ ɗ ʈʰ ʈʼ kʰ kʼ ɠ qʰ qʼ ʔ
Fricatives: sʰ s ʂʰ ʂ ɕʰ ɕ xʰ x χ
Lat. Fricatives: ɬ ꞎ ʟ̥
Nasals: m n *n̥ ɲ *ɲ̊ ŋ
Affricates: tsʰ tsʼ ʈʂʰ ʈʂʼ tɕʰ tɕʼ
Lat. Affricates: tɬʰ ʈꞎʰ kʟ̥ʰ
Liquids: r
Approximants: j ɰ ɦ

I will have to sort their pronunciations out once I delve more seriously into rudimentary allophony, but with this addition the consonant list is done.

The keystone of the phonology of this language will most definitely be its vowels. They carry phonation and are the only allowed syllable nuclei. I usually start with an inventory roughly like what I want the end result to be, and then adjust it as I see fit. The first draft was:

Preliminary inventory 1
High Vowels: i ɨ ʉ
Mid Vowels: ɛ ɜ ʌ ɔ
Low Vowels: ɑ

I tend to like central vowels — this is just a peculiar preference of mine — so that's what tends to go into my inventories by large. A few more I toyed with:

Preliminary inventory 2
High Vowels: i ɨ ʉ
Mid Vowels: ɛ ɜ ɔ
Low Vowels: ɑ

.... and:

Preliminary inventory 3
High Vowels: i ɨ ʉ
Mid/Low Vowels: ɛ ɑ ɔ

Working through these I finally settled on the following phonemic vowel inventory:

High Vowels: ɨ
Mid Vowels: ɛ ʌ ɔ
Low Vowels: ɑ

What a fairly asymmetrical, yet reasonably balanced five-vowel system! The crucial element, the phonation, also gets introduced here:

All Marpanni syllables may have one of three different phonations: they may be modally, slackly or stiffly voiced. The carrier of phonation is the vocalic nucleus. These phonations are, respectivelly, marked with diacritics as: /ʌ ʌ̬ ʌ̰/.

Furthermore, I decide to implement tone and vowel length:

All Marpanni syllables carry a tone regardless of phonation: they can either be high or low: /ʌ́ ʌ̀/; high syllables are generally left unmarked. On short vowels, tone is level. As Marpanni long vowels are actually made up of two segments (and are always written as such), their segments may carry different tonemes; in such a situation, the low toneme acts as a tonal downstep that lowers the tone of the the vowel. This gives five possible length-tone combinations:
Short: ‹ʌ ʌ̀› — high and low
Long: ‹ʌʌ ʌ̀ʌ ʌʌ̀› — level high, level low, and contour falling
The sixth possible combination ‹ʌ̀ʌ́› — a theoretical contour rising tone — always merges with the long level low tone due to the behaviour of the low toneme.

Fairly elaborate (for a sketch) interactions between tones, but this doesn't solve the problem of the underspecification of semivocalic approximants that I mentioned and left for later. To this end, I introduce a new concept of semiphonemes:

Theoretically speaking, Marpanni has three semiphonemes that do not have an independent surface realisation of their own, but instead facilitate the interaction between vowels, approximants and their surface realisations. Most vowels come with a semiphoneme. They are: /(ʌ) wʌ jʌ ɥʌ/ — the labialising, palatalising and labiopalatalising semiphonemes, respectively. Their presence adds a very opaque layer of uncertainty to the language.

Here is where it starts getting complex, just because I want to see the system utilised to its fullest potential:

These subphonemes essentially behave phonotactically much like consonants: they disappear after a consonant cluster of two or more consonants. Word-initially and after non-semivowels, they transform their host vowels as such:
Phonemic Phonetic
ʷɨ ʲɨ ᶣɨ [ʉ ɨ ʉ]
ʷe ʲe ᶣe [ø e ø]
ʷʌ ʲʌ ᶣʌ [o e ø]
ʷo ʲo ᶣo [o ø ø]
ʷa ʲa ᶣa [o e ø]
With the palatal approximant, they interact as such:
Phonemic Phonetic
jʷɨ jʲɨ jᶣɨ [jʉ jɨ jʉ]
jʷe jʲe jᶣe [jø je jø]
jʷʌ jʲʌ jᶣʌ [jo je jø]
jʷo jʲo jᶣo [ɥo jø ɥø]
jʷa jʲa jᶣa [jo je jø]
With the velar approximant, they interact as such:
Phonemic Phonetic
ɰʷɨ ɰʲɨ ɰᶣɨ [wɨ jɨ jʉ]
ɰʷe ɰʲe ɰᶣe [we je jø]
ɰʷʌ ɰʲʌ ɰᶣʌ [ɰo jʌ jo]
ɰʷo ɰʲo ɰᶣo [wo jo jo]
ɰʷa ɰʲa ɰᶣa [wa ja jo]
With the radical consonant, they interact as such:
Phonemic Phonetic
ɦʷɨ ɦʲɨ ɦᶣɨ [ʕʉ ɰɨ ɰʉ]
ɦʷe ɦʲe ɦᶣe [ʕø ɰe ɰø]
ɦʷʌ ɦʲʌ ɦᶣʌ [ʕo ɰʌ ɰo]
ɦʷo ɦʲo ɦᶣo [ʕo ɰo ɰo]
ɦʷa ɦʲa ɦᶣa [ʕo ɰa ɰo]
Vowels without subphonemic features don't alter the approximants in any significant way. Due to a large amount of homophonic mergers, some phonemic confusion occurs, where speakers analyse a certain sequence to be a different one altogether; e.g. they could see /ɦʲo/ as /ɰo/ since they both surface as [ɰo]. This massive ambiguity caused a slew of reanalyses in the vernaculars.

What's left after this cacophony of shifts is some tidying up in the form of allophony:

Phonetically, the series of aspirates is always realised with aspiration, regardless of phonation. Ejectives merge with their aspirated equivalents in word-final position. Unaspirated fricatives are voiceless regardless of phonation. Voiced nasals and approximants absorb the phonation of their syllable and reflect it fully.
In stiff syllables, the lenis glottalised plosives /ɓ ɗ ɠ/ are realised as implosive [ɓ̰ ɗ̰ ɠ̃], and their fortis counterparts as ejective [tʼ ʈʼ kʼ qʼ tsʼ ʈʂʼ tɕʼ].
In modal syllables, the lenis glottalised plosives /ɓ ɗ ɠ/ are realised as varying between implosive and glottalised realisations of voiced plosives [ɓ ɗ ɠ] ~ [bˀ dˀ gˀ], and their fortis counterparts as ejective [tʼ ʈʼ kʼ qʼ tsʼ ʈʂʼ tɕʼ].
In slack syllables, the lenis glottalised plosives /ɓ ɗ ɠ/ are realised as slack voiced [b̬ d̬ ĝ], and their fortis counterparts as tenuis [t ʈ k q ts ʈʂ tɕ].

With this, I generally consider the work on preliminary phonology done. What's left to do is phonotactics and prosody, and I tend to go look at those once I've established the general morphosyntactic traits that the language will be based upon. The only thing I add, which will be fairly significant later on, is the rough syllable structure outline:

Marpanni allows syllables that are maximally (C)(C)V(C)(C), with a strong preference for a maximum internal -CC- clusters.

The final phoneme inventory (simplified per recommendation of the IPA) of Marpanni is thus:

Plosives: ɓ tʰ tʼ ɗ ʈʰ ʈʼ kʰ kʼ ɠ qʰ qʼ ʔ
Fricatives: sʰ s ʂʰ ʂ ɕʰ ɕ xʰ x χ
Lat. Fricatives: ɬ ꞎ ʟ
Nasals: m n n̥ ɲ ɲ̊ ŋ
Affricates: tsʰ tsʼ ʈʂʰ ʈʂʼ tɕʰ tɕʼ
Lat. Affricates: tɬʰ ʈꞎʰ kʟʰ
Liquids: r
Approximants: j ɰ ɦ
High Vowels: ɨ [ʉ]
Mid Vowels: e [ø] ʌ o
Low Vowels: a

Ad blocker interference detected!

Wikia is a free-to-use site that makes money from advertising. We have a modified experience for viewers using ad blockers

Wikia is not accessible if you’ve made further modifications. Remove the custom ad blocker rule(s) and the page will load as expected.