Hand-held Conlanging Guide (WIP)

[Micamo]

M. Park's request gave me the idea to do this, so I decided to throw some ideas together.

Part 1: Vowel Systems

Section 1 - Inventory

To begin, pick one of the following vowel systems to start with:

Simple 3-vowel: /a i u/
Classic 5-vowel: /a e o i u/
Italian 7-vowel: /a e ɛ o ɔ i u/
Hungarian 7-vowel: /a e ø o i y u/

Next, add up to two "vowel splats." A vowel splat is one of two things:

1. The removal of a single vowel.

2. The addition of a single vowel.

The most likely vowels to be splatted, either added or removed, are central and/or high vowels. As an example, the Navajo vowel system /a e o i/ can be described as a classic 5-vowel with a /u/ splat.

If you chose the simple three-vowel system, only splat once.

This system of designing vowel systems seems rather limited, but actually, almost all the vowel systems of all the languages in the world can be described with this system. More complex vowel inventories that this system can't handle are by far the exception, not the rule. It's not recommended that newbies attempt to work with one of these more complicated systems for their first time: Vowel inventories that are much larger or much smaller than 6 vowels or so tend to be more difficult to work with.

Section 2 - Secondary Distinctions

There are three dimensions of secondary distinctions on vowels that are common enough that a beginner should worry about them: Nasality, Length, and Tone.

Nasality only ever occurs phonemically in binary values: Either the vowel is fully nasal or it is not. Keep in mind, however, that although nasality is reasonably common, it's not something every language should include. Only about a quarter of the languages of the world have them, and their distribution in your conlangs should be roughly similar.

Tone appears in a lot of variations, but you should only worry about two types of tone systems: Binary High/Low systems, and Ternary High/Medium/Low systems. More complex contour systems exist, but they're more difficult to work with. Most languages with tone have only 2-3 tones anyway, and only 40% of the world's languages have phonemic tone at all. So, like Nasality, don't throw it in everywhere. Languages with both tone AND nasality (though they exist) should be even less common in your work.

Length is fairly common (though I don't have exact numbers on its prevalence). Length is usually a binary short/long distinction, with a third, half-long distinction appearing only rarely. (Estonian, for example).



One final thing I wanted to discuss in this section is asymmetries within the secondary distinctions. Where only part of your vowel system distinguishes tone/length/nasality. You can do pretty much anything you want with this, but one thing I'd like to add is that mid/central vowels are far less likely to distinguish length than cardinal vowels.


(I'll add more when I feel like it. Likely never.)

[M. Park]

Since this was my suggestion, I feel obligated to ‘play test’ Should you go on and continue this, I will turn it in to a PDF and we can make millions of internet conlangdollars!

---

Language: Testoe
Vowels: /a e o i y/ (Spat: Swapped /u/ for /y/)
Secondary Distinctions: Length ( all vowels can be long or short.)
Nasalization ( /a/ and /i/ can be nasalized.)

---

[Omzinesý]

Hungarian 7-vowel: /a e ø o i y u/

Are You completely sure your Hungarian inventary is right?
Hungarian has /ɛ/ and /ɒ/ contrasting /e:/ and /ä:/.

[CMunk]

Simple 3-vowel: /a i u/
Classic 5-vowel: /a e o i u/
Italian 7-vowel: /a e ɛ o ɔ i u/
Hungarian 7-vowel: /a e ø o i y u/

You could add:

Turkish 8-vowel: /e ø a o i y ɯ u/
Scandinavian 10-vowel: /a ɛ œ ɔ e ø o i y u/
Estonian(+ɯ) 10-vowel: /æ a e ø ɤ o i y ɯ u/

Although these are probably a bit advanced for beginners.

Next, add up to two "vowel splats." A vowel splat is one of two things:

1. The removal of a single vowel.

2. The addition of a single vowel.

Other vowel splats:

3. The unrounding of a rounded vowel or the rounding of an unrounded vowel. /ɔ/ > /ʌ/; /i/ > /y/

4. The fronting of a back vowel or the backing of a front vowel. /u/ > /y/; /e/ > /ɤ/

[rickardspaghetti]

Here are the different vowel systems from the four conlangs I'm(hardly) working on.

1st:
/a V e i i: o u: }/
It's supposed to have started out as an Italian 7-vowel system with an additional /A/.


2nd:
/a i u @ a: i: u: a~ i~ u~ @~ a:~ i:~ u:~/
A simple 3-vowel system with an additional /@/. Length and nasalization is phonemic.


3rd:
/a e i o u a~ i~ o~ u~ A` @`/
A 5-vowel system with four nasals and two rhotics.


4th:
/a e i o u @ 1 a: e: i: o: u: @: 1:/
Don't know where to put this. Help anyone?

[CMunk]

4th:
/a e i o u @ 1 a: e: i: o: u: @: 1:/
Don't know where to put this. Help anyone?

Well, it could be a 5-vowel +@ +1. Or you could classify it as it's own 7-vowel system (we could call it the Romanian 7-vowel).

[rickardspaghetti]

4th:
/a e i o u @ 1 a: e: i: o: u: @: 1:/
Don't know where to put this. Help anyone?

Well, it could be a 5-vowel +@ +1. Or you could classify it as it's own 7-vowel system (we could call it the Romanian 7-vowel).

Didn't think of Romanian.

[CMunk]

4th:
/a e i o u @ 1 a: e: i: o: u: @: 1:/
Don't know where to put this. Help anyone?

Well, it could be a 5-vowel +@ +1. Or you could classify it as it's own 7-vowel system (we could call it the Romanian 7-vowel).

Didn't think of Romanian.

I cheated ;) I wikipedia'd the high central unrounded vowel, and looked through the languages that had it, and found that Romanian fit.

[Micamo]

Turkish 8-vowel: /e ø a o i y ɯ u/
Scandinavian 10-vowel: /a ɛ œ ɔ e ø o i y u/
Estonian(+ɯ) 10-vowel: /æ a e ø ɤ o i y ɯ u/

Although these are probably a bit advanced for beginners.

Yeah, I wanted to keep it simple for a reason. And as for Hungarian, I didn't intend that to mean "here's Hungarian's vowel system, exactly." I just meant to include the "five standard vowels plus two rounded front vowels" pattern. Same thing with the Italian 7-vowel.

Anyway, I have an idea on how to present Phonotactics, I just need to simplify it enough for the topic's purpose.

[Ossicone]

For the purpose of this beginner type guide I agree with Micamo that it's best to keep it simple. It seems to be designed for those that get overwhelmed by the many possibilities.

[Micamo]

Okay, now this still might be a little overcomplicated. But I'm gonna post it anyways and take suggestions on how to break it down.

Part 2: Phonotactics

Phonotactics are the rules about what phonemes are allowed to occur in what places in the word. No language allows for words to consist of any arbitrary list of phonemes the speaker likes, but the rules differ from language to language. You can tell that "strengths" couldn't possibly be a Japanese word because it violates Japanese phonotactic rules, but a word like that is just fine in English. You can also tell that "sxs" couldn't possibly be an English word, for the same reasons. But that kind of word is just fine in a language like Nuxalk.

Phonotactics are usually described in terms of syllable structure. English's syllable structure is CCCVCCCC, but this is quite inadequate for describing the complexities of English's phonotactics. Syllable structure like that alone cannot explain why, say, "strengths" is a valid word but "ktluvgbr" is not. Indeed no serious linguist would suggest that it could, and actual phonotactic descriptions are always more complex than that. And yet, most conlangers seem to think the syllable structure is all there is to phonotactic laws. As if CCCVCCCC really IS all there is to say about how English phonotactics works. They tend to just say "My conlang's CVC" and that's the end of it.

This has always struck me as a rather sad state of events. My suspicion has always been that conlangers who have trouble elucidating more complex phonotactic rules simply don't have an appropriate framework in which to think about them. Well, no more. In this section I'm going to share with you (a simplified version of) my personal system of phonotactic design. Hopefully beginners and intermediates alike can benefit from it.

Section 1 - Phonemic Sets

The first ingredient to a proper phonotactic system is the phonemic set. A phonemic set is a group of phonemes from your phoneme inventory. You can define these arbitrarily all you like, but for more natural results, make sure your sets consist of phonemes which all contain a single feature. Phonemic sets that contain both consonants and vowels are not recommended for beginners. You can make a set with a single member if you like, but this is never necessary (as explained later).

Let's take a look at a sample phoneme inventory and build a collection of phonemic sets from it:

Consonants: /p b t d k g f s v z m n r l y/
Vowels: /a e i o u/

It won't win any awards for creativity, but it'll do for our purposes.

C=pbtdkgfsvzmnrly (all consonants)
S=fsvz (all fricatives)
P=pbtdkg (all plosives)
N=mn (all nasals)
L=rly (all liquids/glides)
V=aeiou (all vowels)
F=ei (all front vowels)
B=ou (all back vowels)
H=iu (all high vowels)
M=eo (all mid vowels)


Of course, your own phonemic sets don't need to conform to this set exactly. It should be adapted to fit both your own phoneme inventory and the set of phonotactic rules you want.

Section 2 - Constraints

My system of phonotactic designs works through a set of constraints. A constraint is simply something not allowed. A valid word in the language is anything that doesn't violate any* of the phonotactic constraints. It sounds like more work than simply listing what's allowed, but surprisingly it's a lot less work in most cases. Plus, it gives an easy and convenient way to handle loanwords and is a lot closer to how natlangs actually work.

Before I can tell you how to design your own constraints, however, first we must go over how to read a constraint. Constraint rules are composed of a few elements:

# - Indicates a word boundary. The start of a word if placed at the left, the end if placed at the right.
A - An upper-case letter, indicates any member of a phonemic set.
a - A lower-case letter, indicating a single, specific phoneme. (This is why you never need to define a phoneme set with only one member.)
* - Placed to the right of an element with a number N attached, indicates that element is repeated at least N times. C*3 means "at least 3 consonants."
() - Placed around a (set of) elements, indicates that the element need or need not be present.
_ - An underscore, indicating any number of any elements, including nothing.

A few examples, using the phonemic sets defined in our example above:

_HC*1H_ - Two high vowels separated by at least one consonant. This disallows "mini", "puki", but still allows for "pikati" or "lumanu".

_PN_ - A nasal immediately preceded by a consonant. Note that this bans "pnak" just as it bans "pakna".

_VC*3_ - Any sequence of 3 consonants following a vowel.

_F# - A word-final front vowel.

_F_B_ - A back vowel occuring anywhere after a front vowel has already occurred.

The possibilities are close to endless here. Be creative!

Section 3 - Violation Marks

The final ingredient of our phonotactic system is violation marks. A violation mark is an abstraction that lets us judge how "far" a particular word candidate is from the ideal of a word that obeys all of the language's phonotactic constraints. This is the handy-dandy system of handling loanwords I talked about earlier: In most natlangs, loanwords are often exceptions to phonetic rules. The violation mark system allows us to imitate this increased flexibility for loanwords.

For every constraint, apply a value of how many violation marks (always at least 1) this constraint assigns if the word-candidate matches the pattern described by the constraint. Next, decide a value of how many violations a loanword is allowed to have. Native words should almost always have 0 violations: The situations where it's acceptable for native words to have violations is beyond the scope of this guide.

When a loanword candidate has too many violations, modify the word somehow to make it fall within the threshold. Don't worry about this process being too regular for the moment: The pronunciation of a foreign word is often something subject to much individual interpretation. It would be best, however, if you held on to some possible regular ways of neutralizing the violations for later when we move onto morphology.

Section 4 - The Finale

At last, we have all the ingredients we need to bake up our phonotactic system! The final step is to build our list of constraints. First let me say you're by all means encouraged to come up with your own list of constraints for your own language. This list is intended as a kickstarter for those who are having trouble coming up with constraints on their own.

--------

Any C within the following may be replaced by a consonant set.

Any V within the following may be replaced by a vowel set.

Any N within the following may be replaced by whatever number you desire.

Cluster Length Constraint - _C*N_ (N > 1)

Vowel Hiatus Constraint - _V*N_ (N > 1)

Intervocalic Cluster Constraint - _VCCV_

Final Consonant Constraint - _C#

Initial Consonant Constraint - #C_

Final Vowel Constraint - _V#

Initial Vowel Constraint - #V_

Adjacent Vowel Constraint - _VC*1V_ (_aC*1e_ forbids "pake" but not "panike")

Adjacent Consonant Constraint - _CV*1C_

--------

Now, with all of our tools in hand, let's put them all together to build an interesting phonotactic system with our example phoneme inventory:

_C*3_ (3 violations)
_V*2_ (3 violations)
#C*2_ (2 violations)
_FC*1B_ (2 violations)
_BC*1F_ (2 violations)
_PP_ (1 violation)
_FF_ (1 violation)
_NN_ (1 violation)
_LL_ (1 violation)
_PF_ (1 violation)
_PN_ (1 violation)
_PL_ (1 violation)
_FN_ (1 violation)
_FL_ (1 violation)
_NL_ (1 violation)


With a threshold of 5 violations.

Now let's look at some of the effects of this constraint set:

- It allows clusters of mixed voicing pairs. "Pazk" is no problem, as is "Bofd."
- It doesn't allow for geminates.
- It does allow unrestrictedly heterorganic clusters, like "komd" and "gaft."
- It doesn't allow onset clusters of any type, only coda clusters.
- It has front-back vowel harmony, with /a/ being neutral.


Fixing the first and third problems, mixed-voicing pairs and heterorganic clusters, would require defining other sets and making many more rules. Notice a further limitation of the system is a lot of copy-and-paste of similar rules to cover everything of a simpler idea, like "in a cluster, the left consonant must always be more sonorous than the right." Both problems are fixable, but require making the system far more complicated than I think is appropriate for this guide. Hopefully, this simplified system should be good enough to allow you to get started with defining an interesting phonotactic system. Happy conlanging!

[M. Park]

---

Language: Testoe
Vowels: /a e o i y/ (Spat: Swapped /u/ for /y/)
Secondary Distinctions:
Length ( all vowels can be long or short.)
Nasalization ( /a/ and /i/ can be nasalized.)
Consonants: /p b t d k g f s v z m n r l y/
Phonotactics: CVCCC*
Constraints(Violation Threshold: 5):
#P_ (1)
_PS_(4)
_V#(1)
_FH_(3)
_g#(1)

---

*Though you mention it in the constraints part, Im assuming in the phonotactics CV line that it allows for everything from V to CCVCCC and everything inbetween, with out putting it like: (c)(c)V(c)(c)(c)

I had no trouble understanding this, and my vocabulary is somewhat small as far as linguistics terms come to. I think they key parts to this section will be noting not to get too hung up on it as it can always be reworked, and that this is where you should spend your time to really get the ‘sound’ of your language defined. I wrote about 50 words in my current language before I realized I was allowing something I didn’t, so then I went back and rewrote the rule.

There is no consonant chart, but I think, as you mention something’s are over the threshold of this guide, I think we can say an understanding up the sum of an ipa chart could be under the threshold of this guide.

[Micamo]

Oh, I fully intend to do a consonant inventory guide. Eventually. I just got an idea on how to present phonotactics first.

[Omzinesý]

You could menation vowel triangles - most languages (all you mentioned)
and vowel squares - English, Finnish, most North-American languages

[Chagen]

Thanks for the phonotactics thing. I was always somewhat intimidated by how truly complex the best descriptions were.

However:

Native words should almost always have 0 violations: The situations where it's acceptable for native words to have violations is beyond the scope of this guide.

Even if this isn't part of the guide, I'd be interested in knowing when a native word could have a violation.

[Micamo]

Even if this isn't part of the guide, I'd be interested in knowing when a native word could have a violation.

Occasionally, archaic terms can survive into the modern language while changing little, without ever having apparently fallen out of use and been readopted. This can result in a closed-class handful of (technically) native terms that still apparently violate the rules of the language. The key examples I have to give of this are Tlingits discourse particles and kinship terms: These contain clusters, especially onset clusters, that are otherwise illegal in Tlingit. Exactly what caused these words to stay around in their more archaic forms, however, is something I don't understand and thus, can't write about.

[M. Park]

Could the english word "Dreamt" be a somewhat an example of this happaning? Though its CCVVCC, it is the only english word that follows __Ct#

[Micamo]

Could the english word "Dreamt" be a somewhat an example of this happaning? Though its CCVVCC, it is the only english word that follows __Ct#

It could well be, as evidenced by "dreamed" being a perfectly valid (and standardized) alternative.

[Ceresz]

Could the english word "Dreamt" be a somewhat an example of this happaning? Though its CCVVCC, it is the only english word that follows __Ct#

It could well be, as evidenced by "dreamed" being a perfectly valid (and standardized) alternative.

I think the mentioned pn QI that dreamt is the only word in the English language that ends in mt or something like that. Wiktionary has an alternative transcript of dreamt with mpt, which I think I've heard. It also says this:

"Dreamt" is less common in both US and UK English in current usage, though somewhat more prevalent in the UK than in the US.

[eldin raigmore]

... the english word "Dreamt" ... is the only english word that follows __Ct#

No, it isn't. There are many.