Tag Archives: conlanging

Review: Peterson, The Art of Language Invention

  • CLARIFICATION: David J. Peterson contacted me about providing an image featuring Ayeri’s script for his book in fall 2014, and it has subsequently appeared in the book (see the blog post of October 1, 2015, “You’ve Got Mail”). As a contributor, I received a signed specimen copy of the book after it had been launched in late September 2015 and decided myself to review it; all of the below is my own opinion. Even though I can’t be fully unbiased in my judgement anymore due to contributing to the book and receiving praise in it, I have still tried to be as fair as possible by highlighting both good and bad characteristics of the book in a constructive manner according to my own experience after over a decade in the community. I urge you to also read other reviews for a more balanced view.
'The Art of Language Invention' by David J. Peterson
‘The Art of Language Invention’ by David J. Peterson

From Horse Lords to Dark Elves to the language-affine TV watcher – David J. Peterson’s personal behind-the-scenes of language creation. A review.

Why is it that we make up languages, of all things? Why not write? Why not draw, paint or sculpt? Why not compose? For me at least, this recent tweet summarizes it nicely (although I also do photography):

I can’t draw well and I’m not confident in my writing, so creating #conlangs is my arts and crafts.
Robbie Antenesse (@robbieantenesse), October 4, 2015

Making up languages is maybe not a mainstream hobby (and that’s not necessarily a bad thing in my humble opinion), but it’s still a valuable creative outlet for a good few people, and also for me. And for David J. Peterson, who gained his laurels in the community with his creating the Dothraki language for the wildly popular HBO series Game of Thrones, based on the novel series A Song of Ice and Fire by George R. R. Martin.

In his freshly published book, The Art of Language Invention: From Horse-Lords to Dark Elves, the Words behind World-Building (292 pages, Penguin 2015, ISBN 978-0-14-312646-1, $17), David J. Peterson explores how the Craft came into being, from the 12th century German nun Hildegard von Bingen to today’s sprawling online community, additionally drawing on a stockpile of anecdotes from the early days of the internet and his own experiences with creating languages both for fun and for money, which form the heart of the book. The meat, then, is a basic introduction to linguistics for those interested bystanders and fans who want to dip their toes into deeper waters.

A thing I really liked about Peterson’s book is that Peterson is very honest from the get-go: you will not find quick success if you seriously intend to make up a language or several and want to do it well. It takes time. It takes work. And it’s totally worth it getting engrossed in descriptions of the myriad of fascinating twists and turns that languages make to get meaning across. A sense of the joy of building up, destroying, improving, and also of solving puzzles can be deeply felt especially in the “Case Studies” after every main chapter of the book, in which Peterson discusses his own approaches to the respective topics while creating his languages. These are also the parts that contain background anecdotes about his work with TV producers and are probably the most entertaining parts of the whole book, for seasoned language tinkerers and people interested in some insider information on the TV shows’ production alike.

But even in the ‘meat’ chapters and in spite of linguistics being a rather academic topic with lots and lots of technical vocabulary, you will find Peterson’s original humor breaking through even the most dire technical discussions by referencing Harry Potter, Soundgarden, Michael Jackson, werewolves, cranky printers, and discussing the absolute and utter awfulness of onions among other things. Oh, and cats. This is 2015, after all, and we all love cats on the internet. Needless to say, since this is an introductory book, discussions of more complex aspects of language are simplified, though usually in a reasonable way that allows you to expand on later. There is only so much information you can put between the two covers of a 290-page book, anyway, and Peterson says so himself. For the $17 the book sets you back, you will get a whole lot of information, though.

Another aspect worth noting from a long-term hobbyist’s point of view is that Peterson never fails to make it clear that English is not the standard average language, but that it has its own peculiarities (apart from its infamously baroque spelling system) which everyone who is serious about creating languages needs to take into consideration lest he or she unwittingly rip off English. In a similar vein, Peterson frequently encourages his readers to do their own research into languages both natural and artistic in order to broaden their understanding and to develop a feeling for what is natural and possible. After all, a master needs to know his or her tools, and for a creative use of linguistics, it’s no different.

While Peterson’s book is not the only guide to creating languages for beginners – the other big one is Mark Rosenfelder’s Language Construction Kit – he covers some crucial aspects that Rosenfelder only touches on briefly, if at all, with a good amount of depth for the purposes of this introductory book. These topics are, for example, syllable structure and language change regarding semantic drift and grammaticalization. Moreover, Peterson presents some basic considerations on the development of writing systems and the effects different writing implements crucially have on their evolution.

* * *

Language is devilishly complex. This cannot be denied, and especially the first chapter drives this point home with its insane density of information that would probably cover the contents of a few weeks of an Introduction to Linguistics class. It is not easy ploughing through this even as a seasoned conlanger, though things get more digestible once the reader has made his or her way through. Promised. Interestingly, the chapter on phonology is the longest of the book, taking up about twice the amount of pages of each the morphology-and-syntax chapter and the one on language change. Strictly speaking, the latter two are no less complex and interesting topics to explore, and I wonder if things could have been balanced better if the phononlogy chapter had been thinned out a little for better legibility. Sounds and sound structure are important to give a convincing impression, since that is what people will hear on screen or, by proxy, read on the page, but it’s by far not the only important part in language design. I think what the book would have generally benefitted from to keep a balance between information density and readability is more suggestions for where to read more on a given topic.

As far as clarity of writing goes, I think it’s fair to say Peterson writes as eloquently as he speaks in front of audiences. However, it is slightly peeving nonetheless that he regularly uses terminology like “subject” and “object” or “causative” but only introduces the respective concepts several to dozens of pages later. The book is equipped with a small glossary at the back (besides phrasebook sentences from his conlangs and a page index), but that unfortunately does not contain all technical terms he uses either, making the text a little hard to follow at times for those readers who have never had any formal training in language.

It is also very nice that the book contains a whole lot of examples, though experience has shown that even a minimal amount of glossing is very helpful in understanding what is going on, especially for created languages whose grammar is usually only known to their creator. Peterson dutifully urges his readers in the introductory chapter to remember that when they develop their own languages and want to discuss them with others. Unfortunately, though, he does not follow his own advice for the greater part of the book. Thus, it is sometimes hard to quickly glance the phenomenon he is referring to in his explanations of language examples just from the phonetic (or phonemic?) transcriptions of lines in foreign scripts he gives along with English translations, especially when the differences between two example sentences are subtle.

It would also have been very nice if more examples in the morphology chapter had come from reasonably accessible and well-documented natural languages instead of almost only Peterson’s own. I know that this is a book on constructed languages from Peterson’s point of view, but he mentions himself how it’s always good to look at what natural languages do. This point of criticism goes hand in hand with my request for more reading recommendations above, basically, since an absolute newcomer would probably first go and look up the referenced languages to learn more. At the very least it would have been helpful to name-drop these languages in the descriptions of phenomena; you often wouldn’t even have to resort to very exotic ones.

Another thing that slightly peeved me is how the morphology chapter makes it seem a little as though languages always morphologically express case and gender. It is not uncommon that they do, but it’s not always the case (Nice to meet you, English!). You don’t need tables upon tables of declensions and conjugations, and a fair share of the world’s languages doesn’t even make use of the concept of gender or noun classes. For a first step, it may be reasonable to not go wild immediately (from a Eurocentric point of view), but at least it should have been said more explicitly that not all the world’s languages work like classical Indo-European ones with maybe some feature erosion for modern descendants. I also missed a reminder that languages tend to leave holes in paradigms and that syncretism is most definitely a thing (for example, the difficult thing about German definite articles is that six word forms cover sixteen paradigm slots, not that articles decline as such).

A thing that conlangers go into much too rarely is writing systems like Chinese, and while Peterson touches on that topic, a little more in-depth explanation about phonetic parts and semantic parts of Chinese characters in the fashion of Mark Rosenfelder’s page on ‘Yingzi’ would have been helpful, especially in comparison to the attention given to other kinds of writing systems. Korean presents the very interesting case of a featural alphabet made to mimic Chinese characters, but has unfortunately been omitted completely.

* * *

So, what is the takeaway of my ramblings? For one, you get a whole lot of bang for the buck. Peterson’s book is chock-full of information and a good way to get you going if you’re interested in creating your own language. For everyone else, it’s a nice round trip through the engine room of language, though for this purpose it’s maybe a little too technical at times. As a seasoned conlanger, I was more interested in the anecdotes behind Peterson’s work than the actual explanations and definitions of linguistic concepts, though, and would have liked more information in this regard, for example about George R. R. Martin’s reactions to Peterson’s work and the slight changes Peterson made to details in Martin’s work. The book could use some more clarity in places and more suggestions on literature, but should overall be a good toolbox that nicely complements similar works, such as Mark Rosenfelder’s Language Construction Kit, which is also recommended reading for fledgling conlangers.

  • Peterson, David J. The Art of Langauge Invention. From Horse-Lords to Dark Elves, the Words Behind World-Building. New York: Penguin Books, 2015. Print.

You’ve Got Mail

A while ago, David J. Peterson – prolific creator of languages for Hollywood – asked me about maybe providing him with a sample of my fictional language’s script for his new book, The Art of Language Invention. Since I’ve known David from fictional-language venues on the internet for a while and admire his relentless spreading the word that languages are fascinating things to explore (and also to build), I was happy and flattered about his request and sent something. And look what arrived in the mail today! In his book, David found some very kind words on Ayeri (among others):

To date, the best languages ever created were not created for television series or movies, but were created just for the joy of it – languages like […] Carsten Becker’s Ayeri […] — Peterson 15

Goes down like oil, as we say in German … (Heads up conlangers: any idioms in your languages that express feeling flattered?) And on Ayeri’s script, Tahano Hikamu, he writes:

I was aiming to create something that could sit alongside some of the other outstanding conscripts that had been created by conlangers, like Carsten Becker’s Tahano Hikamu […] — Peterson 249

Athdavrazar, David, thanks a ton!

  • Peterson, David J. The Art of Langauge Invention. From Horse-Lords to Dark Elves, the Words Behind World-Building. New York: Penguin Books, 2015. Print.

Conlanging and the Real Life

There hasn’t been a lot going on here recently. This is because even though I have accumulated a bunch of papers I want to read for getting inspiration for conlanging, especially regarding the fledgling conlang I sketched out earlier …

A few papers
Some linguistics papers I want to read for conlanging. Currently, there are more important things to do, though.

… I’m currently caught up in the middle of preparing work on my M.A. thesis on adjective morphosyntax in Middle High German as exhibited by a bunch of late-13th-century texts both pragmatic and literary.

I am officially supposed to start working on it on August 18; my thesis has to be submitted in February 2016. It’s probably not like I won’t be doing anything but working on my thesis, but if there’s not a lot of discussion of fictional languages going on here for the next 6 months, this is the most likely reason why. Plus, conlanging is not the only thing I do in my spare time.

Notes on a Vaporware Conlang VIII: Sound Changes, Part 5

Review of Stage IV

The last set of changes saw the Great Palatalization in the course of which also was lost – which basically palatalized everything around it as it merged with *i. This, however, also caused the consonant inventory to look rather tidy again, besides the fact that there is no velar nasal when there are nasals for both the bilabial and the alveolar points of articulation:

MOA bilabial alveolar palatal velar glottal
nasals *m *n
*nʲ
plosives *p *b *t *d *k *g
*tʲ *dʲ
taps/flaps *r
fricatives *s *z *ʃ *ʒ *x *h
approximants *w *l *j

The vowel system didn’t change as much as the consonant one, except that there is no , *ɨː anymore, as I said above, reducing the 6-vowel system[1. “T6Cb” according to the classification attempted in the Vowel Systems thread on the ZBB, which says that /i ɨ e a o u/ is typical of many Slavic languages, so at least my result isn’t as weird as I thought it would be.] to a 5-vowel one. *uʊ also opened up to *uə, as a complement to *iə.

Height Front Central Back
high *i, *iː
*iə
*u, *uː
*uɪ, *uə
mid *e, *eː *o, *oː
low *a, *aː
*aɪ, *aʊ

Stage IV to Stage V

1. Reduce final vowels:

  • *V → *ə / V $ [–stress] _ #

This change reduces all unstressed final vowels in polysyllabic words to schwa if the stressed syllable is open; diphthongs aren’t affected. Example: *taka*takətak.

2. Apocope of final schwa:

  • *ə → Ø / !V _ #

Subsequently, except in diphthongs (i.e. *iə), lose final schwa. Example: *huta*hutəhut.

3. The Great Monophthongization

  • *iə → iː

This schwa is now lost as well, but the monophthongization continues:

  • *aɪ → eː[1. Originally, I wanted to make this one /aː/, but that doesn’t happen a lot apparently (Southern US English is an example of it, though). /aɪ/ turning into something /e/-like seems way more common.]
  • *aʊ → oː
  • *uɪ → iː
  • *uə → uː

All diphthongs resolve into long monophthongs, for now even at the end of words. Examples:

  • *dasiə*dasiː*daːsi 〈dāsi〉,
  • *taɪgateːg 〈tēg〉,
  • *rasaʊ*raso:raːso 〈rāso〉,
  • *nuɪganiːg 〈nīg〉,
  • *huənʲhuːnʲ 〈hūņ〉.

4. Further apocope of final syllables:

  • V → Ø / CSON $ [–stress] [–voiced]SON _ #

After two consonants, the final vowel of an unstressed open syllable disappears if the sonority of the first one is higher than that of the second one and the second one is voiceless.[1.

Originally, I was not deleting final consonants and wanted to contract final CVC syllables as well as doing the change outlined in this section according to this huge table:

p t k s ʃ h m n l
1 1 1 1 1 2 2 4 8 8 8 16
p 1 ! ! ! !
t 1 ! !
1 !
k 1 ! !
s 2 (✕)
ʃ 2
h 4
m 8 (✕) (✕) (✓)
n 8 (✕) (✕) (✓)
8 (✕) (✕) (✕) (✕) (✓)
l 16
j 32
r 64

This table is supposed to indicate that the possible combinations roughly follow a sonority hierarchy – a sound in a row may only precede the matching one in a column if its index is larger than the other sound’s (“✓”). However, plosives and sibilants are an exception to this (“!”), and even though a combination is theoretically allowed, it might either not be desirable (“(✓)”) or not permissible according to euphony (“(✕)”). Of all these combinations, only *mp, *ns, *nt, *ntʃ, *ntʲ, *sp, *st, *stʲ, *ʃk, *ʃp, *ʃt, *ʃtʲ occur in my sample now, though. Note to self: Make a sister language or dialect that doesn’t delete final consonants and then carries out the originally intended change wholesale, metathesizing CVC and CC combinations to make contraction possible. The reason why I decided to delete final consonants in stage 4 here is that without metathesis, not too many final syllables actually contract. And I like the current result, too.] Some examples:

  • *hanʲtʃihantʃ 〈hanč〉,
  • *rinʲtʲarintʲ 〈rinț〉,
  • *sustasust 〈sust〉,
  • *waspawasp 〈wasp〉.

5. Shorten long monophthongs in last syllable:

  • [+stress] V … [–stress] Vː → [+stress] Vː … [–stress] V

This shortens the long monophthongs from diphthongs in the last syllalbe and transfers the length to the stressed syllable. Don’t create overlong vowels, though. For example:

  • *haːruɪ*haːriːha:ri 〈hāri〉,
  • *nantiə*nantiːnaːnti 〈nānti〉.

6. Reinforce final *h:

  • *h → x / _ #

Example: *gruhi*gruhgrux 〈gruh〉.

7. Fix: Homorganic assimilation:

  • *k, *g → tʲ, dʲ / m, n(ʲ) _
  • *n(ʲ)p, *n(ʲ)b → mp, mb
  • *mt, *md → nt, nd

8. Fix: Straighen out some of the palatalized clusters:

  • *n(ʲ)(t(ʲ))ʃ → ntʃ
  • *n(ʲ)(d(ʲ))ʒ → ndʒ
  • *nʲt(ʲ) → ntʲ
  • *nʲd(ʲ) → ndʲ

Phonemic Inventory for Stage V (present day)

We’re now at the final and current stage. A romanization is given in angular brackets where the letters deviate from the IPA symbols. Note that those commas and cedillas should all be commas, in fact, but fonts handle them inconsistently.

Consonants

There hasn’t been any change in the consonants at all this time around, actually.

MOA bilabial alveolar palatal velar glottal
nasals m n
nʲ 〈ņ〉
plosives p b t d k g
tʲ 〈ț〉 dʲ 〈d̦〉
taps/flaps r
fricatives s z ʃ 〈š〉 ʒ 〈ž〉 x 〈h〉 h
affricates tʃ 〈č〉 dʒ 〈dž〉
approximants w l j

Vowels

Diphthongs were all conflated with the monophthongs, so there’s just this five-vowel system with vowel length now:

Height Front Central Back
high i, iː 〈ī〉 u, uː 〈ū〉
mid e, eː 〈ē〉 o, oː 〈ō〉
low a, aː 〈ā〉

The list of sample words

Several times before, I mentioned a list of automatically generated words I tested the sound changes with. To give an overview of the respective outcomes of the various stages, I made a Google Doc of it: List of 2,000 Sample Words.

Name of the language

I’m still undecided and to create a self-name, I guess I will have to wait until I worked out some basic things on the language’s or even the proto-language’s adjectives. For now, I guess it might be best to go with a exonym, maybe from Ayeri. However, that name also still needs to be found. Though now that I’m writing this, how about Ikami? I find that a little cheesy, though, for some reason. Or how about Turayi?

  • Added “Turayi” as a possible name. “Hillandic”. I think I like that better.

Notes on a Vaporware Conlang VII: Sound Changes, Part 4

Review of Stage III

The thing that changed most last time was shifting stress to the first syllalbe
generally, while it was lexical before and could be either on the first or the
second syllable of words. Other than that, we’re stuck with a somewhat unbalanced
phoneme inventory. For example:

  • there is no *pʰ when there are *tʰ and *kʰ;
  • there is no *tʲ when there are *kʲ and *gʲ;
  • there are no *pʲ and *bʲ either;
  • there might also better be a *zʲ accompanying *sʲ;
  • there is no *iʊ when there are diphthongs for all other
    vowels on the apices of the original vowel triangle.
MOA bilabial alveolar palatal velar glottal
nasals *m *n
plosives *p *b *t *d *k *g
*tʰ *kʰ
*dʲ *kʲ *gʲ
taps/flaps *r
fricatives *s *z *x *h
*sʲ
approximants *w *l *j
Height Front Central Back
high *i, *iː
*iə
*ɨ, *ɨː *u, *uː
*uɪ, *uʊ
mid *e, *eː *o, *oː
low *a, *aː
*aɪ, *aʊ

These irregularities, on the other hand, are perfect for applying further
changes towards the phonology I sketched out at the very beginning of this series.

Stage III to Stage IV

I’m now leaving stress marks and syllabification out since words are now stressed on the first syllable in all cases.

1. Open *uʊ to *uə:

  • *uʊ → *uə

2. Lose final consonants in polysyllabic words:

  • *C → Ø / $ _ #

I’m doing a French here, in some way. It’s basically a continuation of the change in stage III, section 2.[1. I was thinking of either chopping off all final consonants or none, but getting rid of all final consonants seemed too radical. This is basically the in-between solution.] Example: *askʲas*astʲa.

3. Change the aspiration distinction to a voicing one:

  • *k → *g / [+sonorant] $ _
  • *t → *d / [+sonorant] $ _
  • *kʰ → *k
  • *tʰ → *t

If the previous syllable ends with a sonorant, change *k and *t to their voiced counterparts; change all *kʰ and *tʰ to their unaspirated counterparts, thus partially merging old *k, *t with old *kʰ, *tʰ. Examples:

  • *saŋkuɪn*saŋguɪ*sanʲdʲuɪ,
  • *gakʲa*gagʲa*gadʲa;
  • *santas*sanda,
  • *ditul*didu;
  • *drakʰis*draki;
  • *hutʰas*huta.

4. The Great Palatalization

4a. Palatalization from merger of with *i:

  • *[+alveolar] ɨ → *[+alveolar]ʲ i / # _
  • [+alveolar] → *i [+alveolar]ʲ (elsewhere);
  • *[+velar] ɨ → *[+velar]ʲ i / # _
  • [+velar] → *i [+velar]ʲ (elsewhere)

As changes to *i, alveolar and velar consonants palatalize at the beginning of words before , and after word-internally. Examples:

  • *sɨm*sʲim*ʃim,
  • *jɨːdan*jiːdʲa;
  • *kɨŋkriŋ*kʲiŋgri*tʲinʲdʒi*tʃinʲdʒi,
  • *rɨkdal*rikʲda*ritʲa.

4b. All remaining merges with *i:

  • *ɨ > *i

No side effects here. Example: *kusɨs*kusɨ*kusi.[1. None of the conditions in section 4a apply here because at the stage where they could apply, there is no *s following the anymore, and the *sɨ is word-internal, not at the beginning!]

4c. Fronting of palatalized velar plosives:

  • *kʲ → *tʲ
  • *gʲ → *dʲ

A merger of palatalized velars with their alveolar counterparts happens. Examples:

  • *saskʲa*sastʲa;
  • *uːgʲik*uːdʲ*uːtʲ.

4d. fronts to *nʲ:

  • *ŋ(ʲ) → *nʲ

Probably as a result of the change in section 4c, fronts to *nʲ, whether palatalized or not. Example: *haŋkʰɨn*haŋki*hanʲtʲi*hanʲtʃi.

4e. Homorganic assimilation after new *nʲ:

  • *[+plosive –voiced] → *tʲ / nʲ _
  • *[+plosive +voiced] → *dʲ / nʲ _

Stops other than alveolar ones assimilate their point of articulation after *nʲ. Example: *haŋkʰɨn*haŋki*hanʲtʲi*hanʲtʃi.

4f. Alveolars palatalize after *nʲ:

  • *[+alveolar] → *[+alveolar]ʲ / nʲ _

Further palatalization assimilation happens after *nʲ with alveolars. Example: *huŋsa*hunʲsa*hunʲsʲa*hunʲʃa.

4g. Merger of *sʲ, *xʲ to and *zʲ, *rʲ to :

  • *sʲ, *xʲ → *ʃ
  • *zʲ, *rʲ → *ʒ

Examples:

  • *husʲaŋ*huʃa;[1. I couldn’t find any examples for *xʲ probably because I’ve already eradicated *x too thoroughly.]
  • *kʰɨz*kizʲ*kiʒ*kiʃ,
  • *snɨraŋ*snirʲa*sniʒa*sinʒa.

4h. Palatalization of *tr and *dr:

  • *r → *ʲ / t _ [–front]
  • *r → *ʲ / d _ [–front]

In the aftermath of the changes in section 4g, *tr and *dr palatalize to *tʲ and *dʲ, except before front vowels, i.e. *i and *e. Examples:

  • *tʰɨkrak*titʲra*titʲa,
  • *kaŋgra*kanʲdʲra*kanʲdʲa.

4i. Creation of *tʃ and *dʒ:

  • *tʲ → tʃ / _ i
  • *dʲ → dʒ / _ i
  • *r → ʃ / t _ [+front]
  • *r → ʒ / d _ [+front]

As a continuation of section 4h, *tʲi and *dʲi further palatalize to *tʃi and *dʒi, resepctively.

Similarly, *tʲr and *dʲr further palatalize to *tʃ and *dʒ, respectively, before front vowels. Examples:

  • *kɨŋ*tʲinʲ*tʃinʲ,
  • *huŋ.gik*hunʲ.dʲi*hunʲdʒi;
  • *kɨŋkriŋ*kʲiŋgri*tʲinʲ.dʲri*tʃinʲ.dʲri*tʃinʲdʒi.[1. I couldn’t find an example for the *tr*tʃ change, even though I don’t see offhand why it should’t be possible in principle.]

After we’ve just palatalized the hell out of alveolar sounds, let’s now turn to other things …

5. Final *h to *x:

  • *h → *x / _ #

This is more of an allophonic thing. I’m including it anyway. Example: *stuh*stux.

6a. Simplification of *sk across syllable borders:

  • *k → Ø / s $ _

Example: *suska*susa.

6b. Simplification of *sk elsewhere:

  • *sk → *ʃ / $ _, _ $ … #

*sk changes to in syllable onsets and syllable-finally inside words. Examples:

  • *skaːsal*ʃaːsa,
  • *ruskda*ruʃ.ta.

7. Simplification of *ʃx:

  • *ʃx → ʃ

Where, due to sound changes, we’ve ended up with *ʃx now, level to , like for example: *kʰasʲxi*kaʃxi*ka.ʃi.

8. Metathesis and simplification of *sn:

  • *snV → *sVn

Example: *snagʲ*santʲ.

9. Homorganic assimilation in case of resulting *nk, *ng:

  • [+plosive –alveolar][+plosive +alveolar] / n _

Example: *snakʰun*sanku*santu (since we don’t have *ŋk anymore).

10. Drop resulting final *r, *l in clusters that are due to section 8:

  • *V C r # → Vː C
  • *V C l # → Vː Cʲ

Examples:

  • *snar*sanr*saːn,
  • *snul*sunl*suːnʲ.

11. Assimilation of voiced sounds after unvoiced:

  • [+voiced][–voiced] / [–voiced] _

This applies to all pairs of consonants for which a voicing distinction exists. Examples:

  • rakdan → rakda → rakta,
  • hɨsʲ.gal → hiʃ.ga → hiʃ.ka.

12. Final devoicing:

  • [+voiced][–voiced] / _ #

Again, this applies only if there is a voiceless counterpart of a voiced sound. Example: *tʰaːgʲ*taːtʲ.

Phonemic Inventory for Stage IV

Consonants

After the Great Palatalization and some leveling in its aftermath, as well as leveling out the three-way distinction in voicing and aspiration, the whole system looks a lot tidier and balanced:

MOA bilabial alveolar palatal velar glottal
nasals *m *n
*nʲ
plosives *p *b *t *d *k *g
*tʲ *dʲ
taps/flaps *r
fricatives *s *z *ʃ *ʒ *x *h
approximants *w *l *j

Vowels

There is no , *ɨː anymore, since it merged with *i, *iː, resulting in a nice five-vowel system with a length distinction. However, there is a *uə now, instead of *uʊ.

Height Front Central Back
high *i, *iː *u, *uː
mid *e, *eː *o, *oː
low *a, *aː
Height Front Central Back
high *iə *uɪ, *uə
mid
low *aɪ, *aʊ

Notes on a Vaporware Conlang VI: Sound Changes, Part 3

Review of Stage II

Last time, we changed quite a few things from Stage I. The phonemes *p, *b, *d, *g, *z and *j were created while got lost, and also *x changed to other things or was lost in many places. The phonemic difference between aspirated and unaspirated *t, *k and *tʰ, kʰ has been weakened in unstressed with the emergence of the voiced consonants *b, *d and *g.

MOA bilabial alveolar palatal velar glottal
nasals *m *n
plosives *p *b *t *d *k *g
*tʰ *kʰ
taps/flaps *r
fricatives *s *z *x *h
approximants *w *l *j

We also created the vowel phonemes *e, *o and while weakening , as well as creating long versions of all of them (except for ) and the diphthongs *iɪ, *ɨɪ, *uɪ and *aɪ:

Height Front Central Back
high *i, *iː
*iɪ
*ɨ, *ɨː
*ɨɪ
*u, *uː
*uɪ
mid *e, *eː
*əɪ
*o, *oː
low *a, *aː
*aɪ

Stage II to Stage III

1. Palatalization of *Cj sequences:

  • *j → *ʲ / C _

2. Simplification of word-final coda clusters:

  • *C₂ → Ø / VC₁ _ #

Word-final coda clusters get simplified by losing the last consonant, e.g. like

  • *a.kúsk*a.kús*kús,
  • *ga.táns*ga.tán and
  • *pu.kʰánt*pu.kʰán.

3. Apocope in post-tonal unstressed open syllables:

  • *V → Ø / [–stress] C _ #

Vowels in post-tonal open syllables get lost but only under the condition of not creating new final clusters. Examples:

  • *í.daíd,
  • *ká.gakág and
  • *náː.dʲanáːdʲ.

This change may have gone through a stage akin to Latvian weak final vowels or by a weakening of the vowel to ?ə, which was subsequently lost, as e.g. in many Upper German dialects (Kariņš 15–34; König 159, Paul 109–111). This change goes hand in hand with the next one:

4. Vowel reduction in unstressed open monosyllables:

  • *V → *ə / # [–stress] (C) C _ #

Example: *ga*gə.

5. Loss of short-vowel only and schwa-containing simple initial unstressed syllables:

  • *V[–long] → Ø / [–stress] # _ $
  • *Cə → Ø / [–stress] # _ $
  • *əC → Ø / [–stress] # _ $

Examples:

  • *u.kʰɨ́ɪt*kʰíət;
  • *gə.krú*krú; and
  • *əŋ.sú*sú.

6. Loss of *r with compensatory lengthening:

  • *VCr → *VːC

Long vowels and diphthongs don’t get extra length, again. Example: *káː.dra*ká:.da.

7. Weakening of initial *x:

  • *x → Øː / _ a
  • *x → *j / _ [–back]
  • *x → *w / _ [+back]

*x weakens to a labial and palatal approximant respectively before back and non-back vowels; it drops out completely before *a (possibly going through *x?ɰ → Ø). In all cases, it leaves compensatory lengthening behind. Examples:

  • *xá.gas*áː.gas;
  • *xís.kri*jíːs.kri; and
  • *xú.gʲu*xúgʲ*wúːgʲ*úːgʲ.

8. *ns simplifies:

  • *ns → *s

Example: *sáns.kun*sás.kun.

9. Generate diphthongs with :

  • *(w)u → *ʊ / V _
  • *w → Ø / _ u

Again, don’t create long diphthongs. Examples:

  • *da.wús*dáʊs;
  • *tɨ́ws.tal*tə́ʊs.tal.

10. Avoid vowel hiatus:

  • *ɪ → *j / V _ V
  • *ʊ → *w / V _ V

A diphthong in with another vowel following is turned into a *VjV sequence to avoid hiatus. The same goes for *VʊV, where turns into *w.[1. This is not attested as happening in my list of 2000 generated words, so I don’t know if it actually happens.] Example: rúɪ.wuk?rúɪʊk*rú.juk.

11. Open *iɪ, *ɨɪ, lower *iʊ, *ɨʊ:

  • *iɪ, ɨɪ → *iə
  • *iʊ, ɨʊ → *əʊ

Examples:

  • *nas.kɨ́ɪs*nas.kíəs,
  • *síɪ.sa*síəs;
  • *i.kʰɨ́wk → … → *kʰɨ́ʊ*kʰə́ʊ,
  • *kɨ́w.ku*kɨ́ʊ.ku*kə́ʊ.ku.

12. Initial stress

Probably under foreign influence:[1. Or more accurately, because reasons.] change all stress to initial.[2. My Python script also has a thing here to clean up weird syllabifications: VCCVVC.CV; CCCCC.C everywhere else; VCVV.CV. This doesn’t always work perfectly, but well enough to require very little manual fixing.]

13. Move length from long unstressed vowel to preceding short stressed vowel:

  • *V́ … *Vː → *V́ː … *V

This is in parallel with the stress shift, though not limited to words that experienced it. Again, no long diphthongs or doubly long vowels. Examples:

  • *su.túːs (→ ?sú.tuːs) → *súː.tus;
  • *táŋ.goːm*táːŋ.gom.

14. Apocope of :

  • *ə → Ø / V C _ #

Where it still exists, drop final , but only so that there are no final consonant clusters. Example: *té.srətéː.sə*téːs.

15. Lowering of remaining :

  • *ə → *a (except in diphthongs *iə, *əʊ)

This gets finally rid of all , basically.[1. Not sure if this is believable, but let’s go with it for now. I guess, you could also interpret this as (〈a〉 on its head if it doesn’t appear as such), rather.] Example: *sá.gəs*sá.gas.

Phonemic Inventory for Stage III

In this stage, we did … things! Most radically, a stress shift to first syllables. We also kind of killed off schwa and created diphthongs with . The consonant inventory hasn’t changed much, but there are palatalized variants of *d, *k, *g and *s (at the very least) now. This feels kind of imbalanced, so we will see some more regularization in the next step.

Consonants

MOA bilabial alveolar palatal velar glottal
nasals *m *n
plosives *p *b *t *d *k *g
*tʰ *kʰ
*dʲ *kʲ *gʲ
taps/flaps *r
fricatives *s *z *x *h
*sʲ
approximants *w *l *j

Vowels

Height Front Central Back
high *i, *iː *ɨ, *ɨː *u, *uː
mid *e, *eː *o, *oː
low *a, *aː
Height Front Central Back
high *iə *uɪ, *uʊ
mid
low *aɪ, *aʊ
  • Kariņš, A. Krišjānis. “Vowel Deletion in Latvian.” Language Variation and Change 7.1 (1995): 15–34. Print.
  • König, Werner. DTV-Atlas Deutsche Sprache. 16th ed. Munich: Deutscher Taschenbuch Verlag, 1994. Print. 159.
  • Paul, Hermann. Mittelhochdeutsche Grammatik. Eds. Thomas Klein, Hans-Joachim Solms et al. 25th ed. Tübingen: Niemeyer, 2007. Print. 109–111.
  • Corrected the sound change in section 11 to also include *ɨʊ as per my commenting – I made a mistake with the regex again. There are still no attested cases of *iʊ*əʊ in my testing word list, but there are definitely cases of *ɨʊ*əʊ. I also corrected the list of resulting phonemes accordingly.

Notes on a Vaporware Conlang V: Sound Changes, Part 2

Review of Stage I

Last time, I assigned a bunch of phonemic values to my underspecified proto-phonemes , *q, *x and , and they are currently in complementary distribution. What we’re starting out with for the changes from Stage I to Stage II is the following phoneme inventory:

MOA bilabial alveolar velar uvular glottal
nasals *m *n
plosives *t *k *q
*tʰ *kʰ
taps/flaps *r
fricatives *s *x *h
approximants *w *l

The vowels stayed all the same:

Height Front Central Back
high *i *u
mid
low *a

Stage I to Stage II

For the record, I’m leaving out the syllable delimiter $ when it is not strictly part of the condition of the sound change. It doesn’t matter for most of the changes.

1. Generating bilabial plosives in partial analogy to *m:

  • [+plosive +oral] > *p / _ w

The oral plosives *t, *k, *q produce *p when before *w; remains . This is maybe stretching things somewhat, but let’s just go with *tʷ merging with the result of a merger of *kʷ and *qʷ, all into *pʷ, and then becoming *p. Example words that get affected by this: *kwəs*pəs; *twá.kwa*pá.pa; *qwá.tra*pá.tra.

2. Homorganic nasals (assimilation of nasals’ POA to the following plosive):

  • [+nasal] > *m / _ [+bilabial +plosive]
  • [+nasal] > *n / _ [+alveolar +plosive]
  • [+nasal] > *ŋ / _ [+velar +plosive]

Currently, the relevant plosives are just the voiceless ones, so we get *mp, *nt, *ŋk from this change for combinations that aren’t congruent, e.g. *húwm.txa*hó:n.dja.

3. Retract *r to *x after *q:

  • *r → *x / q _

This means, *qr merges with *qx (see section 6 below). This would be a possible move through a [+back] allophone of *r (maybe via ?[ʐ] for *r), let’s call it .[1. Regrettably, 〈ρ〉 looks too much like 〈p〉 to stay with the Greek letters I’ve used before.] Then, devoices after the voicless stop, leaving us with what we might reasonably refer to as *qx₂ if we assume that due to the merger with original *qx₁, the association of *x₂*ʀ̥ with *r disappears over time.

4. Resolve *Vw sequences into monophthongs or diphthongs, respectively:

  • *iw → *eː / _ … a
  • *iw → ɨː elsewhere
  • *uw → *oː / _ … a
  • *uw → *uː elsewhere
  • *aw → *oː

Examples:

  • *tʰíw.qak*tʰé.kak;
  • *tra.qíws*dra.kɨ́ːs;
  • *húwm.txa*hóːn.dja;
  • *ruws*ruːs and
  • *aʔ.ráwk*a.róːk.

5. Simplify coda clusters in unstressed syllables:

  • C → Ø / [–stress] _ C $

In unstressed syllables, clusters of two consonants in final position simplify to leave only the latter, e.g. *ask.tʰíln*as.tíɪn.[1. Originally, I had this result in vowel reduction: *i, *u and *a; I decided against this, though, probably because too many syllables would reduce to zero too quickly. I didn’t take notes besides commenting this part out.]

6. Partial reduction of word-initial *q, *qx:

  • *q(x) → *x / # _ [+high]

Examples: *qis*xis, *qxú.kʰu*xú.ku.

7. Merger of remaining *q with *k:

  • *q → *k

Examples: *qə.rán*gə.rán.

8. Vowel lowering before glottal stop:

  • *iʔ → *e
  • *uʔ → *o
  • *əʔ → *a
  • *aʔ → *a / [–stress] _
  • *aʔ → *aː

All of the original non-low vowels drop to the next lower tier[1. Conlanger Ragnar K. noted in another place: “Pharyngeals I would say are the most likely candidates to do that [i.e. lowering of previous vowels], and one could argue that /q/ could be thought of as a ‘pharyngeal’ /k/ in some languages, so that works too, and /ʔ/ also occurs.” He cites Klallam as an example of where this happens, cf. Montler 300–310.] while *a gets extra treatment: in unstressed syllables it just remains *a while it gets lengthened everywhere else. The glottal stop gets elided in all cases. Examples include:

  • *síʔ.stu*sé.stu;
  • *ti.súʔ*di.só;
  • *tʰu.qə́ʔ*du.ká;
  • *wák.tʰaʔ*wá.ta and
  • *háʔ.ska*háː.ska.

Note that the list above does not include ?*eʔ and ?*oʔ, since they do not occur. So far, we get our *e and *o from *iw and *aw, respectively (see section 4), which means that since *q only turns into syllable-finally after vowels, and *w is considered a consonant for this purpose, *iwq and *awq stay *iwq and *awq until *q eventually merges with *k (see section 7).

9. Metathesis of *xs:

  • *xs → *sx / _ [–alveolar] and !#

Except before alveolar consonants and word-finally (*t, *d, *n, *r, *s, *l), *xs switches around into *sx, e.g. in *ílx.san*íls.xan*íɪ.sjan (*rat.síxs*ra.zíːs).

10. Merger of initial *ts and *ks:

  • *ts, *ks → *s / $ _

Examples of this simplification include: *tsa.káls*sa.káɪs and *tʰíx.ksə*tʰíː.sə.

11. Reduction of *ts:

  • *ts → *z / [+voiced] _ [+voiced]
  • *ts → *s elsewhere

Between voiced sounds in general (both voiced consonants and vowels), *ts turns into *z; all remaining instances of *ts now also become *s (see section 10). Examples:

  • *rát.sra*ráz.da;
  • *kált.su*káɪ.zu and
  • *húlt.stu*húɪs.tu.

11. Metathesis of *ks:

  • *ks → *sk

Note that *ks doesn’t appear at the beginning of syllables anymore at this stage, as per section 10. Example: *rúqs.ra*rúks.ra*rúsk.da.

12. Reduction of *sx:

  • *x → *j / s _

*sx is now further eroded to *sj (see section 9), exemplified by the previously quoted *ílx.san*íls.xan*íɪ.sjan.

13. Elision of *x:

  • *x → *Øː / V _ C

As *x is dropped between a vowel and a consonant, it only leaves compensatory lengthening behind. Note that this is not supposed to lead to long diphthongs like ?*aʊː! Also, we now have created ourselves a bunch of long vowels. Example: *ruʔ.tʰíxk*ro.tʰíxk*ro.tʰíːk.

Furthermore, *x gets lost at the end of syllables – though without triggering lengthening – where the previous change hasn’t yet deleted it:

  • *x → Ø / _ $

Example: *rux.tʰálx*ruː.tʰálx*ruː.tʰál.

14. Reduction of word-initial *kx:

  • *kx → *x / # _

Example: *kxí.kʰal*xí.kal.

15. Simplification of two plosives in succession:

  • *[+plosive -aspirated] → Ø / _ [+plosive +aspirated]
  • *[+plosive ±aspirated]₁ → Ø / [+plosive ±aspirated]₁ _
  • *[+plosive -aspirated] → Ø / _ [+plosive -aspirated]

An unaspirated plosive (*p, *t, *k) gets elided before an aspirated plosive (*tʰ, *kʰ). Generally, if two instances of the same plosive are in succession, they simplify to one. If two unaspirated plosives follow each other, only the second one is kept while the first drops out. Examples:

  • *ít.kʰa*í.kʰa*í.ka;
  • *kúsk.qaxt*kúsk.kax*kúsk.ka*kús.ka;
  • *sák.tri*sá.tri*sá.dri.

16. Reduction of *x after plosives:

  • *x → *j / [+plosive] _

The same thing as in section 12, except now with plosives as well. Example: *krí.txə*krí.tjə*krí.djə.

17. *l vocalization:

  • *l → *ɪ / V _ (C) !#

After vowels, vocalize *l to in syllable codas (except at the end of words if no consonant is present at the end), creating rising diphthongs, for example: *qál.sku*kál.sku*káɪ.sku.

18. raising, lowering:

  • *ə → *ɨ / [+stress] _
  • *ɨ(ː) → *ə / [-stress] _

In stressed syllables, raise to ; do the opposite in unstressed syllables and reduce long *əː to short . Examples:

  • *sa.kə́s*sa.kɨ́s and
  • *tʰílq.tiws*tʰílq.tɨːs → … → *tʰíɪ.dəs

19. Make *ŋx become *ŋk (and simplify *ŋkŋ):

  • *x → *k / *ŋ _

Example: *saŋ.xáŋk*saŋ.káŋk. Where we now get *ŋkŋ (or originally already had it), simplify further to *ŋk:

  • *ŋkŋ → *ŋk

Example: *táŋk.ŋawm*táŋk.ŋoːm*táŋ.koːm*táŋ.goːm.

20. Generate *b from *w:

  • *w → *b / m _

Example: *kʰám.wəʔ*kʰám.wa*kʰám.ba.

21. Dissimilation of *r … r:

  • *r → *d / r … V _
  • *r → *j / r … C _

If *r precedes in the previous syllable, turn *r into *d after a vowel and *j after a consonant. Examples: *srá.rat*srá.dat; *trá.sra*trá.sja.

22. Abolish phonemic aspiration in unstressed syllables:

  • *t → *d / ([+sonorant]) [–stress] _ !#
  • *k → *g / ([+sonorant]) [–stress] _ !#
  • *tʰ → *t / [–stress] _
  • *kʰ → *k / [–stress] _

Since the introduction of the voiced unaspirated plosives *b and *d, we’re stuck with a three-way distinction between [+voiced –aspirated] (*b, *d), [–voiced +aspirated] (*tʰ, *kʰ) and [–voiced –aspirated] (*p, *t, *k). We’re now gradually simplifying this into a [±voiced] distinction, starting with unstressed syllables.[1. Seriously, this. Very much this. I had a problem, tried to solve it with regular expressions and wound up with two problems instead. I think I may have spent an hour testing and fixing the regular expressions I used for the first two rules in this part as I was writing this up for the website. While I was doing so, I noticed that I had gotten something wrong and my original regex was overly greedy, i.e. not well formed for what I intended it to do. I eventually arrived at these ugly, ugly expressions to match *t (analogous for *k) in the right environments:

^(?<![hkpqstx])t(?!ʰ))

This matches ‘t’ not preceded by ‘h’, ‘k’, ‘p’, ‘q’, ‘s’, ‘t’, ‘x’ at the start of a line if it is also not followed by ‘ʰ’. For the word-medial case, I have:

(?<![hkpqstx])(\.)(?<!')t(?!(?:ʰ|$))

This matches a full stop (i.e. syllable divider) not preceded by ‘h’, ‘k’, ‘p’, ‘q’, ‘s’, ‘t’, ‘x’, then ‘t’ not preceded by an apostrophe (i.e. stress marker) and not followed by ‘ʰ’.]

Examples for this particular change are:

  • *tʰáxt.xuln → … → *tʰáː.tjul*tʰáː.djul;
  • *tsə́l.kaʔ → … → *sɨ́ɪ.ga;
  • *tʰa.kwá*tʰa.pá*ta.pá and
  • *kʰa.kʰít*ka.kʰít*ga.kʰít.

23a. Fix: Undo newly created plosive-plosive sequences:

  • *[+voiced]₁ > Ø / [–voiced]₁ _
  • *[–voiced]₁ > Ø / [+voiced]₁ _

Example: *rást.ras*rást.das*rás.tas.

23b. Fix: Another round of homorganic assimilation:

  • *[+nasal] > *m / _ [+bilabial +plosive]
  • *[+nasal] > *n / _ [+alveolar +plosive]
  • *[+nasal] > *ŋ / _ [+velar +plosive]

See section 2.

23c. Fix: Delete instances of gemination:

  • *C₁ > Ø / C₁ _

Example: *rís.tsa*rís.sa*rí.sa.

Phonemic Inventory for Stage II

A whole lot has changed (and maybe too much for cramming this all into one stage), and especially *x was extremely unstable.

Consonants

MOA bilabial alveolar palatal velar glottal
nasals *m *n
plosives *p *b *t *d *k *g
*tʰ *kʰ
taps/flaps *r
fricatives *s *z *x *h
approximants *w *l *j

Vowels

Height Front Central Back
high *i *u
mid *e *o
low *a

These all can appear as rising diphthongs, however, in my list of 2000 generated and processed words, ?*eɪ and ?*oɪ did not appear, since there is no ?*ʔλ coda in the Proto Language. For diphthongs we thus get this smaller chart:

Height Front Central Back
high *iɪ *ɨɪ *uɪ
mid *əɪ
low *aɪ

Similarly, there are long versions of all vowels, except for ?*eː, however this seems only due to my word generator not generating ‘Ciwx.Ca sequences by chance (*Cíwx.Ca*Ce:.Ca), so ?*eː should be possible. The chart for long vowels looks like this, accordingly:

Height Front Central Back
high *iː *ɨː *uː
mid *eː *oː
low *aː

Note the absence of ?*əː.

  • Montler, Timothy: “Vowel Retraction before Glottal Stop in Klallam.” Studies in Salish Linguistics in Honor of M. Dale Kinkade. Ed. Donna B. Gerdts and Lisa Matthewson. Missoula: U of Montana P, 2004. 300–310. Print. Occasional Papers in Linguistics 17. Web. 28 Apr. 2015. ‹http://montler.net/papers/Montler-KlallamVowelRetraction.pdf›.
  • Seems like I can’t publish anything without having to edit it later … 😩 I updated section 4 and 18 to hopefully make more sense now.
  • More corrections in section 11: my regexes weren’t covering all cases here either. I also decided to just have *ts turn *s everywhere that remains instead of just before voiceless sounds in analogy to the change in section 10.
  • There’s no *əː coming out of this because of the update in section 18. Corrected the vowel chart at the end accordingly.
  • Reader Hallow XIII suggested to me that a change like *kʰák.tʰa*kʰá.ga (“A plosive gets elided after another plosive.”) seems strange and he would rather expect *kʰá.tʰa as the outcome. I modified the rule in section 15 accordingly by adding some alternations based on the original rule. Aspiration is still lost in unstressed syllables, though, so it becomes *kʰá.ta now.

Notes on a Vaporware Conlang IV: Sound Changes, Part 1

So, if you remember, some time earlier this spring I came up with some designs for another conlang after being prodded a bit by some people I know online for a while:

Now I’ve spend the better part of the week obsessing over coming up with a bunch of sound changes that would turn my musings about a proto language sound inventory as sketched out in part III to the recent version sketched out in part II. It seems like this is not so much vaporware anymore after all … I’ve been thinking about maybe giving this project a name some time soon.

Believe it or not, this is the first time I’m actually seriously applying sound changes to something,[1. I know, I know, there is Playing with Sound Changes, but I consider that play. This time is for real.] so if there are any overly weird things that can’t even be hand-waved away, tell me. I think most of the changes are a little on the cautious side, though.

Some Notation Conventions

I will assume you’re familiar with the usual way sound changes are notated:

X → Y / Z

This means that X turns into Y under the conditions of the environment Z.

I will also use the following additional abbreviations:

  • C: consonant
  • V: vowel
  • _ (underscore): position of X in the environment
  • # (hash sign): start or end of a word, i.e. a word boundary
  • $ (dollar sign): start or end of a syllable, including word boundaries
  • ! (exclamation mark): negation
  • … (ellipsis): somewhere before or after

Sometimes, feature notation is being used as a shortcut for groups of vowels or consonants: [±feature].

Proto Lang Allophony (Stage I)

The (re)constructed phoneme inventory for the Proto Language looked like this in my previous post (I’m copying these things over for convenience, adding changes that occurred while working on the sound changes themselves):

MOA labial alveolar velar guttural
nasals *μ [m~n~ŋ]
plosives *t *k *q [k~q~ʔ]
*tʰ *kʰ
fricatives *s *x [x~h]
liquids *w *λ [r~l]

And the vowels:

Height Front Central Back
high *i *u
mid
low *a

The sound changes below apply a degree of allophony so that the underspecified proto-phonemes , , *q, *x actually receieve an assumed phonemic value:

1. Position-based alternation between [r] and [l] for :

  • *λ → *r / $ (C) _
  • *λ → *l / _ (C) $

Basically, what happens here is that will be realized as [r] in syllable onsets and as [l] in syllable codas, i.e. there is a complementary distribution at this stage.

2. Position-based alternation between [m], [n] and [ŋ] for :

  • *μ → *m / _ [+bilabial]
  • *μ → *n / _ [+alveolar]
  • *μ → *ŋ / _ [+velar]
  • *μ → *ŋ / _ [+guttural]

Here, is assimilated to the point of articulation of the following consonant. “Guttural” refers to *q and *x here. This assimilation can also happen across syllable boundaries:

  • *μ → *m / _ $ [+bilabial] (unless C _)
  • *μ → *n / _ $ [+alveolar] (unless C _)
  • *μ → *ŋ / _ $ [+velar] (unless C _)
  • *μ → *ŋ / _ $ [+guttural] (unless C _)

Unless a consonant is preceding, further assimilates in POA to the next consonant across a syllable boundary. The assimilation continues (resistance is futile *badum tsh*):

  • *μ → *m / [+bilabial] $ _ (unless _ C)
  • *μ → *n / [+alveolar] $ _ (unless _ C)
  • *μ → *ŋ / [+velar] $ _ (unless _ C)
  • *μ → *ŋ / [+guttural] $ _ (unless _ C)

also assimilates in POA to preceding consonants at the beginning of syllables, unless followed by a consonant. And what’s more:

  • *μ → *m / _ V [+bilabial]
  • *μ → *n / _ V [+alveolar]
  • *μ → *ŋ / _ V [+velar]
  • *μ → *ŋ / _ V [+guttural]

In this set of changes, assimilates in POA to a following consonant even if a vowel is in between them. The next set now takes care of almost all remaining instances of :

  • *μ → *m / [+front] _
  • *μ → *n / [+center] _
  • *μ → *ŋ / [+back] _
  • *μ → *m / _ [+front]
  • *μ → *n / _ [+center]
  • *μ → *ŋ / _ [+back]

The above set is maybe somewhat of a weird change: changes according to the height of vowels here: *m ~ *i/*u, *n ~ , ~ *a. If this is too unnatural, I will change this – or maybe there is a way to build in another condition to achieve the same outcome. Anyways, just for good measure let’s also do:

  • *ŋ → *n / # _

Instances of word-initial are replaced with *n.[1. No, this is not just because English doesn’t allow initial /ŋ/.]

3. Position-based alternation between [q], [ʔ] and even [k][1. A change to my original designs!] for *q:

  • *q → *ʔ / V _ $
  • *q → *k / _ ŋ
  • *q → *k / ŋ ($) _

Syllable-finally after vowels, *q is realized as a glottal stop [ʔ]; before and after [ŋ], it has a velar release [k], and remains [q] in all other places.

4. Position-based alternation between [x] and [h] for *x:

  • *x → *h / V ($) _ ($) V
  • *x → *h / # _ V

*x will have a glottal release [h] anywhere between vowels and initially before a vowel.

Intermission: Some Example Words

These words are randomly generated with Wharrgarbl, as mentioned in the previous article in this series. Many of these changes are very unspectacular so far.

*it *it
*kwas.ku *kwas.ku
*kwu.wakt *kwu.wakt
*sawk *sawk
*suq *suʔ
*sλas.tu *sras.tu
*sλu.sə *sru.sə
*sμi.kxəq *sni.kxəq *sni.kxəʔ
*tʰas.kʰuμ *tʰas.kʰuŋ
*tλi.taλs *tri.taλs *tri.tals

Applying Stress Rules

The Proto Language has syllable-weight based stress rules and I wrote a little script to apply this. What weird missyllabifications of CCC clusters occasionally appear are due to Wharrgarbl lacking checks for sonority-hierarchy related issues. If I handcrafted all proto words, this wouldn’t be an issue in this case. I decided to use a word generator, though, to get a taste of how my words may look like and turn out with all the changes applied, en masse.

Syllable stress basically counts how many consonants there are in a word and increases a counter by 1 whenever it finds one. The syllable that is more consonant-heavy receives stress. In case of a tie, the first syllable will be stressed. Monosyllabic words currently don’t receive stress; depending on them being content words or particles, this will of course change. I will have to sketch out some grammar for the Proto Language in that case, though. If I want the Proto Language to be mildly inflecting, long bisyllabic words are awkward anyway, but generated bisyllabic words still give a taste of how the final outcome could look like.

For the words above, applying the stress rules results in: *it (1), *kwás.ku (3:2), *kwu.wákt (2:3), *sawk (3), *suʔ (2), *srás.tu (3:2), *srú.sə (2:1), *sni.kxə́ʔ (2:3), *tʰás.kʰuŋ (2:2), *tri.táls (2:3).

Prospect

In the next installment, we’ll start eroding things and also generate a bunch of vowels, giving us *it, *pás.ku, *pu.wák, *saʊk, *so, *srás.tu, *srú.sə, *sni.kjá, *tʰás.kuŋ, and *tri.táɪs.

  • I decided that it makes at least a little more sense to have the nasals pattern with vowel height rather than frontness, i.e. not *m ~ *i, *n ~ /*a, ~ *u, but *m ~ *i/*u, *n ~ , ~ *a. In the word list of 2000 generated words that I’ve been using to test the sound changes laid out here, only 8 words were affected by this decision, though, i.e. a low 0.4%, so for example we now have *μu.λuktmu.rúkt instead of *μu.λukt*nu.rúkt. I also had to correct a rule above that I misunderstood from my own notes.

Notes on a Vaporware Conlang III: A Tentative Proto-Language Phoneme Inventory

A thing that’s always bothered me a little regarding Ayeri is that, so far, I haven’t come up with a proto-language which to draw irregularities and other neat and natural idiosyncracies from. For this conlang idea, I want to right those wrongs from the start, and by experimenting with a “fresh” language I can also learn how I could maybe retrofit Ayeri a little while avoiding noob mistakes … Since any sound can change to any other at least in principle, I thought I should just

  1. assume some phoneme inventory for the proto-language, maybe with some underspecification à la PIE’s laryngeals;[1. If you like some not totally serious but still good conlanging, look at Pthag’s Shapshiruckish with its series of voiceless fricatives *Ⅰ, *Ⅱ and *Ⅲ, with their sound value “reconstructed” as *​h~ʔ, *f, *θ, respectively.]
  2. make up some phonotactics of course;
  3. make up some words in either system accordingly;
  4. morph the proto words to their recent shape and thus gain a bunch of sound changes to draw on and extend.

Furthermore, of course, I can also go backwards from the current shape and make assumptions on what may have happened to generate the current surface form. For example, given words like boď and šaňt, there may very plausibly have been a high front vowel following the palatalized sounds which was elided by apocope or syncope, so e.g. *bodiboď and *šanitšaňt.

For a proto-language inventory, how about this, for a start:[1. That is, this may be subject to changes, depending on whether I can make it work or not …]

Consonants

MOA labial dental velar guttural
nasals *μ [m~n~ŋ]
plosives *t *k *q [q~ʔ]
*tʰ *kʰ
fricatives *s *x [x~h]
liquids *w *λ [r~l]

Vowels

Height Front Mid Back
high *i *u
mid
low *a

Some possible words

Given the above phonemes and the phonotactic rules I sketched out here,[1. This is a rule file for a little word generator I wrote some time ago, called Wharrgarbl. You can check that out, too, if you like.] here is a list of some possible words:

*i.ksu, *kak.tʰast, *kaks.μis, *kaλq.sa, *ksa, *ksaμ.tʰas, *ksi.λuq, *ksiq.sak, *ksə.taμ, *kuμs, *kə, *kʰak.xu, *kλa.sta, *kλu.sλi, *quλ.ksa, *qλa.tʰuλq, *qλax.wat, *saq.kiμs, *saλ.kʰəxt, *stu, *stə, *sxi.sa, *sλa.taλt, *sλa.λakt, *sμa.kəws, *ta.tʰaks, *tas.takt, *taλ.kʰaμt, *ti.kux, *tu.kʰaλt, *tuλx, *twu.μit, *tʰa.tʰuλk, *tʰaxk.tλu, *tʰu.kʰa, *tʰəλμ.kʰaμs, *tλu.tʰaqs, *tλu.λaxt, *waμq.λuk, *wikt.kʰu, *xax, *ə, *λa.μaμs, *λast.skə, *λat.μu, *λaλk, *λu.txu, *λus.tuq, *μa, *μaq.kλa.

Obviously, in order to reach the phoneme inventory posited in my previous blog article in this little series, there has to be some splitting of phonemes, and the underspecification of some of them will be useful there, I suppose, in that allophony (e.g. *λ → *r / #(C)_ but *λ → *l / _#, so *λəλk.λaλ*rəlkral) provides different contexts for sound change to operate on as complementary distributions get undermined. Also, there will have to be some playing around with vowels to generate /e/ and /o/ and to get rid of /ə/, and also to generate a length distinction.

Notes on a Vaporware Conlang II: A Phoneme Inventory and Some Thoughts on Grammatical Features

OK, the bug has mildly bitten me. Here are some further notes:

Phonemic inventory

Consonants

m n nʲ 〈m n ň〉
p t tʲ k 〈p t ť k 〉
b d dʲ ɡ 〈b d ď g〉
s ʃ x 〈s š h〉
z ʒ 〈z ž〉
r 〈r〉
w l j 〈w l ľ〉

Vowels

i iː u uː 〈i ī u ū〉
e eː o oː 〈e ē o ō〉
a aː 〈a ā〉

Feature ideas

  • Split is based on [±volition]: [+volition] S/A, P; [-volition] S/P, A
  • Different definite articles and/or pronouns for S/A and S/P?
  • Maybe the same for all S, but differing object A and P depending on Ergative or Accusative?
  • Wackernagel position affects definite articles (N Art, but Adj Art N) and maybe modal verbs (S-Mod-O-V – yes, I’m looking at you, German!)
  • Wackernagel position also for possessive pronouns: stressed/unstressed forms here as well
  • no distinction between A series of pronouns and possessive pronouns (I = my)
  • maybe only direct vs. oblique case marking (if that works with split-A); otherwise small system with case markers from adpositions like Romance (SOV correllates with postpositions, however)
  • definite article from “this” and “that”
  • proclitic definite article full form; enclitic definite article unstressed form
  • left-branching preferred, i.e. modifier-head order: Adj Noun and Possessor Noun
  • The definite article may serve as a subordinator for chains of Poss-N (goes counter to the above rule of wackernageling the definite article, though! Either irregularity, or use a different morpheme …)
    • John car of ‘John’s car’
    • car I ‘my car’
    • John car of the motor of ‘of the motor of John’s car’
    • car I the motor of ‘of my car’s motor’
  • inanimate nouns can’t serve as A
  • non-S can’t be relativized: need for passives and applicatives
  • most common verbs (e.g. be/sit, go, get, put) have suppletive or irregular forms, e.g. for past or plural
  • have = be with/on?

Maybe it could look a bit like this:

[gloss]Srāt -m boď na marh ūt šaňt.
NOUN.A =DEF SOMETHING DEF.ACC NOUN.P done VERB[/gloss]