09.6.13.23:57: THE LANGUAGES OF HADANUS: DWARANIR CONSONANTS, SYLLABLE STRUCTURE, AND TONES

Here is a list of all consonants I have found in Dwaranir words:

S is from Sellec the Earth Shaker, a name I had forgotten about until recently. This is presumably a stem Selek- whose final consonant only appears before vowel-initial endings:

Sele 'Sellec' (isolated form)

Selek-a 'of Sellec'

Sele-nir 'the Sellec'

Selek-a-nir 'of the Sellec'

Compare the -k- of Selek with the -ch- that only appears in certain forms of Old Irish sailech- 'willow tree' (Stifter 2006: 149):

sail (nominative singular) < pre-Celtic salik-s

sailech (genitive singular) < pre-Celtic salik-os

sailchib (prepositional plural) < pre-Celtic salik-bhis

sailchea (accusative plural) < pre-Celtic salik-ns

If not for the lone voiced stop d, I wouldn't mind proposing a small consonant inventory like

with t (instead of d) and p to fill the first row. However, the name Dwaranir is established after nearly twenty years and I don't want to change it in English. Here are two possible solutions:

1. It's already established that Dwaranir 'of the people' is a Dwaranir word but is not Dwaranir for 'Dwaranir'. The Dwaranir call themselves Dwar 'people' or Dwarnir 'the people'. It seems that the Superiors heard Dwaranir 'of the people' and assumed that was the name of the Dwaranir.

What if the Superiors heard [twaʳniʳ]  with an unaspirated t which sounds like a d? Perhaps Superior t is an aspirated [tʰ]. A real-life example of an unaspirated t romanized as d is Korean 대우 [tɛu], romanized as Daewoo.

If Dwaranir t sounded like Superior d, Dwaranir p and k would sound like Superior b and g, and Dwaranir ch would sound like English j (cf. how the voiceless [tɕ] of [pejtɕiŋ] is romanized as -j- in Beijing).

Like Korean p t ch k, Dwaranir p t ch k could be voiced between vowels (and syllabic consonants that exist in Dwaranir but not Korean): e.g.,

Selek-a > Selega 'of Sellec'

ylt-a > ylda 'of yuld'

2. Dwaranir has a row of voiced consonants including three that have not yet appeared in the literature: b j g:

This solution allows existing words with the attested consonants d ch k to remain unchanged.

I could fill in all but one of the remaining gaps for the bottom rows of either solution:

So far there are only two known Dwaranir clusters:

dw (which would be [tw] according to my first proposal)

hy as in hyakl (kl is not a cluster, as l is a vowel-like syllabic consonant)

It is highly unlikely that these are the only clusters in Dwaranir: cf. how Japanese not only has hy but y after many other consonants: py, ky, etc. Perhaps the Japanese analogy could go further: what if Dwaranir had no ty, dy, sy anymore because they had become ch, j, sh as in Japanese? So cha 'home' would be from Old Dwaranir tya (just as modern Japanese cha is also from earlier Japanese tya). Old Dwaranir might be the tongue of the first Dwaranir on Hadanus and the ch, j, sh shift could have occurred in the speech of later generations. Perhaps ritual storytellers speak of Dwar-Tya instead of Dwar-Cha, the everyday name of the forest.

Dwaranir syllable structure is very simple:

(C)(S)V + tone

All syllables must contain a vowel V. V can be any of the 36 basic vowels, 60 diphthongs, or syllabic l or r. (Although syllabic r is not attested, I assume it exists since I don't know of any language with syllabic l that lacks syllabic r.)

S is a sonorant w, l, r or y. The combinations ty, dy, sy don't exist because they have become ch, j, sh. The old cluster ny is now a single consonant like Spanish ñ though it's still written as a cluster ny to avoid typing a tilde ~ on top of n. Clusters with l and r are needed for forms like hyakl-a 'of hyakl' in which syllabic liquids have become nonsyllabic.

C is any Dwaranir consonant. Sonorant-sonorant combinations other than lw rw ly ry are not possible before vowels: e.g., ww, wl, wr, wy, ll, lr, rr, rl, yy, yl, yr. Note that sonorant-syllabic consonant combinations are possible if no vowel follows: e.g., wl may exist, but wla cannot exist. 

high: dwár 'people'

low: chà 'home'

rising: chǎ 'of home' < chà + -á  (low + high contraction)

falling: châ 'of ?' < chá + -à (high + low contraction)

high followed by low: hyákl̀ (note how syllabic l has a tone)

chá is a word I made up which would be distinct from chà 'home' to Dwaranir ears.

Simple vowels with complex tones may also originate from Old Dwaranir VCV sequences in which V was identical: e.g.,

châ '?' < Old Dwaranir cháCà (C = unknown lost consonant)

Diphthongs are written with two or more vowels but can have only one tone in a syllable (kai is an arbitrarily chosen example)::

high (ˊ): kái < Old Dwaranir kái

low (ˋ): kài < Old Dwaranir kài

rising (ˇ): kǎi < Old Dwaranir kàCí (C = unknown lost consonant)

falling (ˆ): kâi < Old Dwaranir káCì (C = unknown lost consonant)

hyákl̀-a > hyákl-à 'of hyakl'

Diphthongs shift to vowel-glide sequences before vowel-initial suffixes:

kái + -a > káyà 'of a kái'

pàu + -a > pàwá 'of a pàu'

The toneless suffix -a (and presumably others) bears the second half of the rising or falling melody that may be absent from monosyllables.

The complex tone of a diphthong is spread out over a toneless vowel-initial suffix:

kǎi + -a > kàyá 'of a kǎi'

pâu + -a > páwà 'of a pâu'

Tones are unwritten except in technical linguistic works like this blog entry.

Next: How to create your own Dwaranir names and words and type them in English.

09.6.12.23:59: THE LANGUAGES OF HADANUS: DWARANIR DIPHTHONGS

I said I would cover Dwaranir consonants, but I'd like to address some vowel suggestions from David Boxenhorn:

Vowel length: In "Melodies of Dwaranir", I created the form chà-á 'of home' from the established forms cha 'home' and -a 'of'. I thought of chàá as having a long vowel, but later I thought that if Dwaranir were sung, there would be no way to distinguish between chà 'home' and chàá'of home' except through context if they were stretched out: "chaaa ..." So I thought that the two words might be distinguished only through melodies:

Changes in tone signify stem-suffix boundaries. So if a monosyllable has a high or low tone, it has no suffix, but if it has a falling or rising tone, that indicates a vowel suffix: e.g.,

Falling melody root + suffix

é- + -u > éù > êu (diphthong with falling tone)

Rising melody root + suffix (chǎ 'of home' above is another example):

ò + -i > òí > ǒi (diphthong with rising tone)

Identical vowels are merged: e.g., à + -á > ǎ in chǎ 'of home'

Diphthongs: I realized that Dwaranir might need an eu after all since that letter combination appeared in the name Aireund (for a non-Dwaranir living among the Dwaranir, so it may not be a Dwaranir or Dwaranirized name). I would now interpret that name as Airẽu in isolation and as a stem Airẽud- before vowel-initial suffixes: Airẽud-a 'of Aireund'. ẽu is a nasalized diphthong combining ẽ with ũ (though I only write the tilde for nasalization once in ẽu).

To keep the diphthong inventory manageable and to avoid combinations that are difficult to spell and pronounce like aoui [ɒy], I propose that all diphthongs can end only in -i or -u:

These 20 diphthongs can also be nasalized or retroflex: e.g.,

ẽi, õu, etc.

eiʳ, ouʳ, etc.

Nasalization is written with etymological nasals: e.g., Airẽu is from earlier Aireund, so the isolation form could be spelled as Aireun without diacritics. The name Aireund in Globe-Hurler was based on the stem form Airẽud- before vowel-initial suffixes.

Retroflexion is written with -r after diphthongs: e.g., oiʳ would be written as oir.

36 (= 12 x 3) simple vowels plus 60 (= 20 x 3) diphthongs equals 96 possible rhymes in Dwaranir, a number comparable to the 105 rhymes of Tangut, another language without final consonants.

Next: Dwaranir consonants (at last!)

09.6.11.23:32: THE LANGUAGES OF HADANUS: DWARANIR VOWELS

Three Dwaranir names have vowel sequences that I don't remember seeing in Superior or Urdreh names:

Coen

Haen

Meana

For nearly twenty years, I've pronounced Coen as [koen]. However, I recently realized that Dwaranir could have a rich vowel system absent from the two other major tongues:

IPA symbols are in brackets. Typable spellings are in italics.

I made up ui, eo, oa, ao to fill out this system which is similar to French, though the spellings are different: e.g., my Dwaranir ui is like a French u, Dwaranir u is like French ou, etc.

a e i o u are as in Spanish.

ea is halfway between the e of bed and the a of bad; Meana is similar to mana(tee).

ae is the a of bad, so Haen is like hand but without a -d (or an -n; the vowel is nasalized)

ui is an i pronounced with the lip-rounding of an u.

oe is an e pronounced with the lip-rounding of an o.

eo is ea (see above) pronounced with the lip-rounding of an o.

ao is an a pronounced with the lip-rounding of an o or a back a [ɑ].

- low rounded [ɶ] is a very rare vowel; I don't know of any language with it and can't find any language with it in UPSID. I find low rounded vowels to be very hard to pronounce because lip rounding conflicts with a mouth that's wide open.

- I don't know of any language that has full sets of both front rounded vowels and central unrounded vowels, perhaps because they sound similar, at least to me. Before I learned IPA, I used to use the German letter ö for mid front round vowels to write [ə].

- I couldn't think of easily understandable ways to write central unrounded [ɨ ə ɐ] using only two simple vowel letters each. iu for [ɨ] looks like 'eew'. I thought of using eu for [ɨ] (used to romanize the similar Korean vowel [ɯ]) but there is nothing e-like about [ɨ].

Each of the twelve basic vowels can be combined with

- nasalization: ã, ẽ, ĩ, etc.

- retroflexion: aʳ, eʳ, iʳ, etc.

resulting in a total of 36 (= 12 x 3) vowels (compared to only 5 vowels in Superior).

Nasalization can be written with etymological nasals: e.g.,

[olũ] as olum (since the word once ended in -m which is still preserved before vowel-initial suffixes: olum-a)

[kø̃] as Coen (since the word once ended in -n which is still preserved before vowel-initial suffixes: koen-a)

Retroflexion can be written with -r: [aʳ] = ar, etc. R before a vowel or glide (w, y) is always a real [r].

Nasalized retroflex vowels are extremely rare. They may have existed in Tangut, but the only language that has them for certain may be Kalasha (see p. 28 of Jan Heegård Petersen's PhD dissertation) which only has four (not twelve) nasalized retroflex vowels: ãʳ, ẽʳ, õʳ, ũʳ.

09.6.10.23:57: THE LANGUAGES OF HADANUS: DWARANIR STEMS

To make Dwaranir "melodic", I'd like it to have no final non-syllabic consonants. Yet obviously Dwaranir ends in an -r. Doesn't that directly contradict my wish? No. The spellings in the existing Hadanus literature are English-influenced and are not in phonetic notation. As I wrote last night,

The -r in dwar 'people' and other Dwaranir words could be interpreted as retroflexion rather than a final -r.

This retroflexion would be indicated with a superscript [ʳ] in phonetic notation, but could simply be typed as a regular -r.

-ld: yuld

-m: olum

-n: Coen, Haen

I can account for them by regarding them as parts of stems:

Stems are abstract forms that can be used to generate all other forms. They are not actually pronounced by Dwaranir given my rule that no Dwaranir words can end in non-syllabic consonants. Dwaranir names in English are based on stems, not isolated forms, with minor adjustments: yl > yul, Ko > Co.

If a stem ends in a nonsyllabic consonant

- final stops are dropped: yld > yl

- final nasals are dropped and the preceding vowels nasalize: olum > olũ, Koen > Koẽ

Note that there is no way to tell whether a stem has final -m or -n based on the isolated form.

- final r is dropped and the preceding vowel becomes retroflex: Dwar > Dwaʳ

Vowel-initial suffixes are added directly to the stems.

If a stem ends in a syllabic liquid, that liquid becomes nonsyllabic before a vowel:

hya-kl (two syllables) > hya-kla (two syllables; kl-a fuses into one syllable kla)

Words are then resyllabified so that each syllable ends in a vowel:

Dwar-a-niʳ > Dwa ra niʳ

Consonant-initial suffixes are added to isolated forms. They can't be added to stems because they would result in complex clusters: e.g., yldnir with -dn- instead of ylnir which has no clusters (since syllabic -l- is like a vowel).

Next: Dwaranir vowels.

6.11.00:29: ADDENDUM: The notion of stems differing from isolated forms was inspired by Sanskrit: e.g.,

The isolated forms of Dwaranir are roughly analogous to the nominative singulars of Sanskrit; both lose stem-final consonants. The Dwaranir stem plus -a, -nir, and -a-nir forms are analogous to Sanskrit forms based on stems plus suffixes like -i (locative singular).

The English borrowings tend to be based on the nominative singular with the exception of the stem-based atman (instead of the expected nom. sg.-based atma). Conversely, the Hadanus literature uses the stem forms rather than the isolated forms.

09.6.9.23:59: THE LANGUAGES OF HADANUS: THE MELODIES OF DWARANIR

When Robinson described Dwaranir as "melodic", I immediately wanted it to be a Chinese-style tone language. The -r in dwar 'people' and other Dwaranir words could be interpreted as retroflexion rather than a final -r. Dwar [dwaʳ] with a retroflex vowel would not be far from Tangut

dwiaʳ 'skin'

dwiaʳ 'scabbard'

both with the so-called 'level' tone. (I am not claiming that Dwaranir is related to Tangut, but that the two languages could have had similar syllables by sheer coincidence. Tangut had no -waʳ after nonback consonants, and I wonder if original -waʳ shifted to -wiaʳ after nonback consonants.)

However, I ran into two problems.

First, Dwaranir names like Coen, Haen, and Meana could be polysyllabic roots which wouldn't be in a Chinese-style tone language. (Or they could be compounds: e.g., Meana could be Me-a-na 'na of me' since -a- is 'of', if I understand correctly. I believe Dwar-a-nir is literally 'people-of-the' = 'of the people'.)

Second, I assume that yuld is a Dwaranir word since the yuld are their "chief source of meat". The yuld are also bred by Superiors for food, but Superior has no final clusters and -ld or even -d are not possible final consonants in Chinese-style tone languages. (Their tones in part compensate for the loss of earlier final consonant contrasts. There is no need to have both tones and a rich system of final consonants.)

I did think of reinterpreting yuld as [yld] with a syllabic l and a rare yl-cluster (found in Japhug rGyalrong), but that still leaves -d.

So I considered imposing Swedish melodies* onto Dwaranir. Swedish has two melodies, rising and falling-rising. Applying what I know of Swedish melodic rules from Holmes and Hinchliffe (2nd ed., 2008):

Dwar 'people' is monosyllabic and would have a rising melody

Dwar-Cha 'people-home' is a disyllabic compound and would have a falling-rising melody

However, I thought Robinson might want something a bit original. I haven't fully worked out this scheme yet, but here are a few ideas:

All roots in Dwaranir either have a rising or falling melody.

These melodies are only fully realized in words of two or more syllables.

Otherwise the words only have a high or low tone (the first half of each melody).

Examples:

Dwár 'people' by itself could have a high tone (H) marked with an acute accent; this is the first half of the falling melody HL (L = low)

Dwár-à with -a 'of' attached has a falling melody: HL

Chà 'home' by itself could have a low tone (L) marked with a grave accent; this is the first half of the rising melody LH

Chà-á with -a 'of' attached has a rising melody: LH

Note that -a 'of' has no inherent tone; it just takes on the pitch required by the melody of the preceding root.

-nir 'the' also has no inherent tone. In words of three syllables or more, all syllables after the second have the same tone, so Dwár-a-nir becomes Dwár-à-nìr with a falling melody (HLL) and low tones on -a-nir 'of the'.

Further examples:

Dwár-nìr 'the people' (falling melody: HL)

Chà-nír 'the home' (rising melody: LH; I assume that -nir is 'the' after all nouns, but it's possible that Dwaranir has more than one 'the'-suffix like Swedish which has neuter singular -et and plural -na in addition to nonneuter singular -en)

Chà-á-nír 'of the home' (rising melody: LHH)

Dwár-Chà 'People-Home' (compound; each half retains its original half-melody: H + L)

Dwár-Chà-á 'of People-Home' (rising melody in second half: H + LH)

Dwár-Chà-á-nír 'of the People-Home' (rising melody in second half: H + LHH)

In any case, there's no need to mark the pitches in fiction about Hadanus, just as no one writes Swedish with tone marks.

6.10.3:30: ADDENDUM: The idea of variable tones for -a and -nir was inspired by Japanese: e.g., hana 'nose' and hana 'flower' have different melodies** in the standard dialect which are only apparent when the nominative case marker -ga is added:

鼻 hàná 'nose''

花 hàná 'flower'

(both have an LH melody in isolation)

鼻が hàná-gá 'nose-NOM' (LHH melody)

花が hàná-gà 'flower-NOM' (LHL melody)

(ga has no tone of its own; its tone is determined by the preceding word)

*Although these melodies are called 'Accent 1' and 'Accent 2' in the literature I've seen, I prefer to use the term 'melody' to avoid confusion with the meanings of accent in English.

**Similarly, I use the term 'melody' for Japanese instead of the usual term akusento. I also use 'tone' instead of 'pitch' so that I can have one terminology for all the languages in this post.

09.6.8.23:59: THE LANGUAGES OF HADANUS: SYLLABIC COMPLEXITY AND WORD LENGTH

Comments from David Boxenhorn led me to reconsider my continuum which I now see as a multidimensional table that can't possibly be depicted in HTML. Let me just focus on two dimensions, syllabic complexity and word length. Before using any actual linguistic examples, compare how zero through nine are represented in binary and decimal:

Notice how binary has fewer distinct symbols (just 0 and 1) but more digits. Conversely, decimal has more distinct symbols (0-9) but fewer digits.

Languages like Hawaiian are like binary. They have relatively few distinct syllables forming long, polysyllabic words. Conversely, languages like English, Polish, or Mandarin are like decimal. They have many distinct syllables and some may consititute complex monosyllabic roots and words: e.g.,

English strength

Polish Gdańsk

Khmer phteəh 'house'

Mandarin hǔ 'tiger' (with a phonemic tone)

Each of these syllable-rich languages is complex in different ways:

(For comparison, Hawaiian has very few consonants, no consonant clusters, more vowels than Polish, and no tones.)

No syllable-rich language is rich or poor in all four ways. A language with few consonants and vowels and no clusters or tones could not have many syllables and therefore could not be syllable-rich. A language full of all of those things would be more complex than necessary. It would be like having a base-64 number system. Imagine memorizing 64 different symbols so you could write 0-63 with one digit each; the equivalent of '10' in base-64 would be 64 in decimal.

Alien names and words in English-language SF and fantasy seem to be either English-like or what I call 'asdf': letters (and apostrophes!) strung together to look odd and inhuman. (asdf is a highly unlikely syllable in human languages.) The languages of Hadanus seem to be English-like in terms of their sound structure, though as we have seen, Superior is considerably simpler: e.g., the only permissible consonant clusters contain s- in syllable-initial position. But little of the phonological diversity of Dead Earth seems to have survived on Hadanus: e.g., there are no languages with Polish consonant clusters, Khmer complex vowels, Zulu clicks, etc. I initially wanted Dwaranir to be a Chinese-style tonal language, but that won't work for reasons I'll reveal next time.

09.6.7.23:59: THE LANGUAGES OF HADANUS: THE COMPRESSION CONTINUUM

Before examining the two other major languages of Hadanus (Dwaranir and Urdreh), I want to say a few things about language typology.

Languages can be characterized using many different criteria: e.g., what I call the compression continuum. A language's position on this continuum reflects how much meaning is expressed at what length:

This is an extremely crude chart, so many real languages don't fit neatly on it.

There is a general inverse correlation between word length and complexity of syllable structure. Languages with low compression have long words with simple syllables, whereas languages at the medium and high points of the continuum have short words with complex syllables or simple syllables with phonemic tones. Here's how a language can go through all three stages:

Low compression: Hadanusu (four simple CV syllables: ha-da-nu-su)

Low to medium compression: H'danus > Dhanus (two syllables CV.CVC; the cluster Hd- is hard to pronounce; Dh- found in Indian languages is relatively easier to pronounce and counts as a single consonant)

Medium compression: Dhnus (complex CCVC syllable; dh-)

Medium to high compression: Dnus > Nus (CVC syllable)

High compression: Nù (simple CV syllable again but with a phonemic falling tone represented by the grave accent; the tone compensates for the loss of the final consonant -s)

A language can also cycle back to low compression. Nù could be attached to affixes that make the word longer, and the longer the word gets, the less important tones become.

Superior is probably low compression unless names like Ladane are composed of three roots.

I've thought the musical Dwaranir language could be high compression and tonal, but next time I'll reveal what I now think it might be.

6.8.0:54: ADDENDUM: Real, probable, and potential examples of compression:

1. Proto-Slavic Gudanisku (4 syllables; low compression) contracted into Polish Gdańsk (1 syllable; medium compression). Perhaps someday this will end up being Dàń or even just Dã (both with tones). To disambiguate Dã, other monosyllables meaning 'city', etc. could be added to it, making it longer and reducing the need for tones to zero.

2. ʔakalaʔ (3 syllables; low compression) might have contracted to Old Chinese 虎 *xlaʔ (1 syllable; medium compression; x here is a sound roughly halfway between k and h) which then contracted further to Mandarin hǔ (1 syllable; high compression with a tone). But in spoken Mandarin, the word for 'tiger' has gained a prefix 老 lǎo 'old': 老虎 lǎohǔ 'tiger' (lit. 'old tiger'*). Eventually the short form may be forgotten, and lǎohǔ may lose its tones. And even further in the future, toneless laohu may be reduced to laoh or lù (with a tone).

*The phrases for 'young tiger' and 'old tiger' both contain the long word for 'tiger' with this prefix:

年輕的老虎

lit. 'year-light(-suffix) old-tiger'

'young tiger' (not 'young old tiger')

年老的老虎

lit. 'year-old(-suffix) old-tiger'

'old tiger' (not 'old old tiger')