-3

In the world of conlanging, its often said the best way to minimize the number of long words is to allow more short words. The main ways to do this is to have a large consonant inventory, liberal phonotactics, and/or phonemic tone.

However, I'm starting to question this. I made a word generator program for a conlang I was planning. After getting it to work properly, I found that I struggled to pronounce most of the words it was generating. I intentionally limited myself to phones I'm used to, so that's not he problem. The syllable structure is s[l,y,w]V(S,N) (S and N stands for fricatives and nasals).

Of course, I banned things like voiced and voiceless consonants appearing next to each other, or y next to front vowels, w next to rounded vowels, you get the idea. The words rarely look hard, but I struggle even with simple CVCV words. What?

Thinking about it, I think I see the problems. The language has front rounded vowels, which I'm used to, but there's no rhymne or reason to them. In German, the language that got me used to them, they normally only occur on stem vowels in response to an E appearing in the ending (mainly on verbs and plural nouns). I've found one of my problems is pronouncing a word which has both a front rounded vowel in it and a back vowel. Basically, its the complete lack of vowel harmony.

Of course, vowel harmony tends to reduce the number of words you can have. Look at Finnish. Words like 'muna' and 'münä' are possible, but 'munä' and 'müma' are not. I actually do find the latter two awkward funny enough. Basically, in such a system a=ä, u=ü, and o=ö. It essentially just gives you the same benefit of having phonemic tone. It just doubles the number of possible words. You can have 'front' and 'back' words, like you could have 'high' and 'low' tones on a word. Also, I find I tend to have the same problem with the distinction between long and short vowels. Its hard to pronounce a short vowel in an open syllable for instance.

There's a similar problem with consonants. Oddly, I have issues pronouncing the voiced fricatives in certain environments. Such phones are rare, and in languages that have them like English they're typically next to voiced consonants that actually make it hard to NOT pronounce them as voiced. As I said, I did ban voiced sounds occuring next to voiceless sounds, but my main issues are pronouncing them when they're alone at either the beginning or end of a word. So maybe a voicing distinction among fricatives doesn't help as much as it may look.

Also, thinking about it, English's phonotactics aren't as elaborate as they may look. Plosives aren't audibly released at the ends of words. This may seem to be an issue, but thinking about its hard to find minimal pairs. There's the word 'that', but no word 'tha'. There's 'mop', and no word 'mo'. The closest I can think of is 'cat' and 'cab', but in that cause there's redundancy, such as the vowel in 'cab' being long due to it being followed by a voiced consonant. If you wrote out English as its actually pronouncing, omitting final stops, its consonant clusters wouldn't look nearly as bad. In fact, this is why I chose to ban plosives in syllable codas in my conlang. There's the same problem with its long and short vowels. there's 'book', with a short vowel, but not 'booook' with a long vowel. Guess it does contrast with 'boo', but even with audible release that final 'k' is still technically doing something in terms of articulation. I'm more talking about how a language sounds here though, rather than how its produced.

Point is, large phoneme inventories tend to evolve out of things like harmony and similation, and sometimes dissimilation. This means that the extra phonemes normally occur only in specific environments that encourage them. If that's the case, why include them? Having vowel harmony does make things easier, yes, but it doesn't really do anything to increase the word count.

So, is it all a sham? Do things like large phoneme inventories and liberal phonotactics really help? They do for the written language yes, but not so much for spoken language where redundancy is more needed. And yes, I know there's languages that do pronounce plosives at the end of words (in particular Hungarian, which is a requirement because many of its markings are just a single plosive suffix). However, this is rare cross-linguistically, and even in languages that did have them in the past tend to lose them. They're not pronounced in English or French, and in the Sinitic languages most of them the final consonants were swapped out for phonemic tone. Point is, its rare and its hard to do. So perhaps these features aren't all they're cooked to be? Of course, that does beg the question as to why English words tend to be so short. If the final plosives aren't there, honestly English doesn't look much different from French or the other Romanic languages. How is English actually doing it then?

14
  • 5
    While syllables ending in æ aren't generally allowed in English, there are plenty of minimal pairs like "rack", "rat", "rap", "rag", "rad" showing that final consonants do matter.
    – Draconis
    Mar 25 at 17:54
  • 2
    Vowel harmony doesn’t double the number of possible words… it halves it. Yes, you can have doubled parallels like muno and münö, but that’s because you’ve got those rounded vowels to begin with. The vowel harmony doesn’t enable you to ‘double’ muno (the only option English would have) by adding münö – rather, it prevents you from also having munö and müno, both of which are perfectly reasonable. Also remember that even languages with vowel harmony can break it (e.g., Finnish miljonääri ‘millionaire’, Turkish rüya ‘dream’). Mar 25 at 19:10
  • 2
    »Point is, large phoneme inventories tend to evolve out of things like harmony and similation, and sometimes dissimilation. This means that the extra phonemes normally occur only in specific environments that encourage them.« — Where are you getting this from? It’s entirely untrue. (And as for your ‘book/booook’ example, consider cook /kʊk/ vs kook /kuk/.) Mar 25 at 19:13
  • 1
    Yes, specific sounds in specific languages can be the result of as- or dissimilatory sound changes (though the ich and ach sounds are generally still considered allophones of the same phoneme), and this can cause phoneme inventories to grow or shrink. But that doesn’t mean that’s how complex phonological systems arise by default. In most cases, we don’t know exactly how a given language ended up with its phonetic inventory. We know that nearly all Austronesian languages have very small inventories and Caucasian languages very large ones, but that’s true as far back as we know. Mar 25 at 22:12
  • 2
    »the reason s before a stop is pronounced as sh is because it evolved from a common 'si' prefix. The s became sh, then deleted the vowel.» — Are you talking about German here? If so, no. /s/ becoming [ʃ] before various other consonant sounds is not due to any erstwhile /i/ that was subsequently deleted. Consider words like stehen (from the PIE root *steh₂-) or Schmerz (from the PIE root *(s)merd-): in both cases, the /s/ and the following consonant have been right up against each other for thousands of years – much, much longer than the [ʃ] pronunciation has been present. Mar 26 at 1:18

1 Answer 1

5

There's a similar problem with consonants. Oddly, I have issues pronouncing the voiced fricatives in certain environments. Such phones are rare, and in languages that have them like English they're typically next to voiced consonants that actually make it hard to NOT pronounce them as voiced. As I said, I did ban voiced sounds occuring next to voiceless sounds, but my main issues are pronouncing them when they're alone at either the beginning or end of a word. So maybe a voicing distinction among fricatives doesn't help as much as it may look.

Be careful to distinguish what's useful for a certain purpose (in this case, increasing the number of possible syllables) from what's easy for you to pronounce. The latter is shaped almost exclusively by the language(s) you spoke growing up; it doesn't necessarily convey any fundamental truths about language. It's important not to end up like that guy a while back who claimed Proto-Indo-European couldn't possibly have had uvular stops because he couldn't pronounce a uvular stop himself! I fully believe he couldn't pronounce the uvular stops he was talking about, but that doesn't stop uvular stops from existing in Arabic, Inuktitut, etc.

Plosives aren't audibly released at the ends of words. This may seem to be an issue, but thinking about its hard to find minimal pairs. There's the word 'that', but no word 'tha'. There's 'mop', and no word 'mo'. The closest I can think of is 'cat' and 'cab', but in that cause there's redundancy, such as the vowel in 'cab' being long due to it being followed by a voiced consonant.

Rap, rat, rack, rad, rag. English phonotactics forbid certain vowels at the end of words, mostly due to historical constraints (at one point those vowels were only allowed in closed syllables), but final consonants still very much matter.

If you wrote out English as its actually pronouncing, omitting final stops, its consonant clusters wouldn't look nearly as bad. In fact, this is why I chose to ban plosives in syllable codas in my conlang.

See above.

There's the same problem with its long and short vowels. there's 'book', with a short vowel, but not 'booook' with a long vowel.

Vowel length is not, as a general rule, phonemic in English.

So, is it all a sham? Do things like large phoneme inventories and liberal phonotactics really help?

Take a look at the table on page 44 here. "Phonological system" indicates how many consonants, vowels, and tone/stress features there are, "Syllable structure" is a metric of how many phonemes can be in a single syllable, and "Inventory size" counts how many distinct syllables are found in the 20,000 most frequent words of the language.

Overall, "it is observed that the size of phonological system (number of vowels and consonants) is positively correlated with the size of syllable inventory among the 18 languages". So it seems there is in fact a correlation between large phoneme inventories and liberal phonotactics and the number of distinct syllables found in the most common words of the language.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.