Conlang Random Thread

[bradrn]

In the SCA, is there some way to get, as output, a list of the pairs of pre-change words and post-change words?

Depends on the SCA. In zompist’s, select the input → output or output [input] options.

(I think it’s worth noting that my own SCA gives a bunch more options, but this may well cross the line into ‘blatant advertising’…)

Unrelated question: how do prefixes and suffixes marking things like cases and plurals and the like change over time? My impression from the few languages I know a little bit about is that after some time has passed, you might almost as well throw them out completely and introduce completely new ones, but I fully expect that people who know more about languages than I do will tell me that that's nonsense.

My impression from Indo-European languages is that they follow the same sound changes as everything else, thus fusing into the words they’re joined to; this can give rise to several paradigms, depending on the phonology of the ‘carrier’ word. However, I would welcome comments on this issue from people who know more about the subject.

[Raphael]

In the SCA, is there some way to get, as output, a list of the pairs of pre-change words and post-change words?

Depends on the SCA. In zompist’s, select the input → output or output [input] options.

Ouch. *slaps forehead* I really should have noticed that.

[bradrn]

In the SCA, is there some way to get, as output, a list of the pairs of pre-change words and post-change words?

Depends on the SCA. In zompist’s, select the input → output or output [input] options.

Ouch. *slaps forehead* I really should have noticed that.

Don’t blame yourself — that option isn’t exactly obvious.

If you haven’t noticed it yet, there is also a help button which may be useful to you — it links to this help page.

[bradrn]

Unrelated question: how do prefixes and suffixes marking things like cases and plurals and the like change over time? My impression from the few languages I know a little bit about is that after some time has passed, you might almost as well throw them out completely and introduce completely new ones, but I fully expect that people who know more about languages than I do will tell me that that's nonsense.

My impression from Indo-European languages is that they follow the same sound changes as everything else, thus fusing into the words they’re joined to; this can give rise to several paradigms, depending on the phonology of the ‘carrier’ word. However, I would welcome comments on this issue from people who know more about the subject.

Actually, I have something to add to my answer for this: it may be more useful to ask this question in the Sound Change Quickie Thread than in this thread, since it’s predominantly sound changes which change the forms of affixes.

[Salmoneus]

Affixes are affected by the same sound changes as everything else... in theory. So some sound changes can obliterate affixes quickly (like "loss of final unstressed syllables", which can quickly blast through suffixes), while others do little damage.

However, unlike most morphemes, affixes are part of very strong, recognisable patterns, which sound changes can break up (creating 'fusion'). To maintain those patterns, analogy can level some of the differences, which in some cases can mean restoring more conservative forms of the affix.

For instance, let's imagine a plural suffix -in, and a rule giving stress to the second syllable of a word. Let's take five words:

klamak + in = klamakin
vlasa + in = vlasain
rom + in = romin
kloruk + in = klorakin
kar + in = karin

Now, let's say that:
a) unstressed single vowels after the stress lenite to schwa
b) /n/ drops in the codas of unstressed syllables
c) /i/ (not in diphthongs) causes umlaut in preceding syllable
d) stressed open vowels lengthen
e) diphthongs > long vowels
f) /n/ drops after long vowels (i.e. no overlong syllables)
f) final schwa drops
g) stress shifts to the first syllable
h) palatalisation before /i/
i) unstressed short vowels are lost after a non-nasal sonorant

What do our singular/plural pairs look like now?
klamak vs klame:k
vlasa vs vlasa:
rom vs remin
klork vs klori:tch
kar vs kern

These plurals are a complete mess!

Now, a language may just put up with its plurals being a complete mess, at least in the short term. In this language, there are probably only half a dozen or so plural classes to have to remember. But very often, languages will try to simplify this mess, by moving some words into other classes. This is particularly likely if the irregular words are much fewer than the regular ones. In this case, that ghastly klork>kloriitch plural is probably a prime suspect for regularisation, particularly if it's not a common word.

One response a language might easily have to this system is to simplify it down to only three classes: bisyllables ending in a consonant get umlaut and vowel lengthening; bisyllables ending in a vowel just get vowel lengthening; monosyllables get umlaut and the suffix -in.

We could then imagine the loss of unstressed vowel length and quality; in that case, the first two classes stop indicating the plural at all. So maybe the language then analogically restores the only surviving ending, -@n, to all nouns (other than perhaps some common irregulars), in order to maintain the plural.

This would essentially mean that this suffix would seem to have 'skipped' a bunch of soundchanges that affected other morphemes in the language.


-----

Similalarly, imagine a past tense in ma-. Let's say that this is unstressed, and loses its vowel to become just m-. Before vowel-initial verbs, this is no problem: arak > marak. But before consonants, it results in a 'difficult' cluster, which is then simplified: kutu > mkutu > gutu (assimilation and then loss). Now, a languag can perfectly well cope with having these two different forms of the past tense. But it can also decide not to! So maybe it decides to analogically re-introduce the m- prefix of vowel initial words to ALL verbs. And this time it deals with the cluster through epenthesis. So now we have arak > marak and kutu > mugutu. And now again the prefixes are different, so maybe the language analogises again and gives us arak > muarak.

This doesn't have to happen. Indeed, it would probably be more likely to generalise the other way, and lose the tense markers entirely. But in this sort of way, language can sometimes 'fight back' to avoid losing marking that they find necessary.

[Raphael]

Thank you!

[masako]

On the left is an abstract representation of a face that includes all the strokes of Moj. I first drew an image like this almost 30 years ago. On the right is the updated version for Moya.

[bradrn]

[...]

A truly excellent summary Salmoneus — thank you!

[Xwtek]

Do you know any way to use analogy to make the declension/conjugation less regular. My language is inspired by navajo where the ablaut is not regular at all. There is some semitic inspiration, because occassionally two vowels are mutated at once. (Usually only one vowel is mutated, because of monosyllabicity). However, the main difference between this language and both Navajo and Semitic language is that the vowels still have a lexical distinction. Also, there is only 4 forms distiguished by ablaut.

[bradrn]

Do you know any way to use analogy to make the declension/conjugation less regular.

I’m not sure this is possible. The whole point of analogy is that it makes a system more regular, not less! If you do want to make something less regular, try applying some sound changes to it.

[Vijay]

Do you know any way to use analogy to make the declension/conjugation less regular.

I’m not sure this is possible. The whole point of analogy is that it makes a system more regular, not less!

Not really. If you extend an analogy to forms that already follow a regular pattern, then it does make it less regular. Lyle Campbell in Historical Linguistics mentions as his first example dove (in North American English) instead of dived; it's a weak verb (following the regular pattern of adding -ed to make the past tense), but in NAE at least, the strong verb pattern has been extended to it, resulting in less regularity.

[bradrn]

Do you know any way to use analogy to make the declension/conjugation less regular.

I’m not sure this is possible. The whole point of analogy is that it makes a system more regular, not less!

Not really. If you extend an analogy to forms that already follow a regular pattern, then it does make it less regular. Lyle Campbell in Historical Linguistics mentions as his first example dove (in North American English) instead of dived; it's a weak verb (following the regular pattern of adding -ed to make the past tense), but in NAE at least, the strong verb pattern has been extended to it, resulting in less regularity.

Good point Vijay!

[Vijay]

Thanks!

[Salmoneus]

Do you know any way to use analogy to make the declension/conjugation less regular.

I’m not sure this is possible. The whole point of analogy is that it makes a system more regular, not less!

Not really. If you extend an analogy to forms that already follow a regular pattern, then it does make it less regular. Lyle Campbell in Historical Linguistics mentions as his first example dove (in North American English) instead of dived; it's a weak verb (following the regular pattern of adding -ed to make the past tense), but in NAE at least, the strong verb pattern has been extended to it, resulting in less regularity.

Depends what you mean by 'regular'. Turning 'dive' into a strong verb IS regularising it. There are very few verbs in -ive - the only really common ones are "drive" and "strive", both of which are strong; "shrive" is also strong while "thrive" I think is "meant" to be weak but in practice is already strong. "Skive" I assumed used to be weak, but is very often strong, while "rive" certainly has the strong participle, although to be honest I don't think I've ever heard its past tense. That only really leaves "dive" (and I guess "jive") as "irregularly" weak verbs.

So what's really happening isn't just random irregularisation - it's regularising a new class of verbs in which eventually all verbs ending in -ive will conjugate the same way.

What makes English verbs irregular isn't the fact that there's more than one conjugation class, some of which are less common than others, but the fact that the conjugation class is not predictable from the shape of the verb. So moving a verb fom a common, but unanticipated class into an uncommon, but predictable class is regularisation, not irregularisation.

More fundamentally, because analogy makes one form the same as another, it's very difficult for it to introduce genuine irregularity - what's happening here isn't the creation of an irregular class, it's just the movement of words from one existing class to another existing class.

What can happen however, as here, is that analogy can be only partial, progressing form-by-form, which can create new mixed classes, at least temporarily - so we have drive/drove/driven, shrive/shrove/shriven, strive/strove/striven, but I've only heard dive/dived/dived or dive/dove/dived, not dive/dove/*diven. In this case, it's probably only a matter of time before we get 'diven', but if enough verbs, or common enough verbs, go through this process at the same time, a new class can get 'stuck' partway through the process.

But what this process doesn't give us is irregular forms - we have the same number of forms as before. At most it gives us partially-irregular class membership. And I think what Akanga wanted was diversity of forms...

[jal]

Though of course one could argue what "more" or "less" regularity is. If you have two patterns, and switch over a certain word or conjugation to another pattern, does it make it less regular? If you have more "weak" verbs and less "strong" verbs, do you have a more "regular" grammar? If you take "irregular" to mean "deviation from the norm", then you have to think about what the "norm" is - the most commonly used pattern, or any sufficiently used pattern?

EDIT: Seems Salmoneus ninja'd me. What happened to the old board's warning of post in the meantime?

JAL

[bradrn]

What happened to the old board's warning of post in the meantime?

It still works for me on this board. (I never posted much on the old board, so can’t say if it works exactly the same, but if someone posts while I’m still writing a post, it shows me the new post while I’m still writing.)

[Pabappa]

But it's supposed to explicitly say "Someone has posted in this thread since then". I dont know. Maybe it only does that if you both are quoting someone?

[Vijay]

I’m not sure this is possible. The whole point of analogy is that it makes a system more regular, not less!

Not really. If you extend an analogy to forms that already follow a regular pattern, then it does make it less regular. Lyle Campbell in Historical Linguistics mentions as his first example dove (in North American English) instead of dived; it's a weak verb (following the regular pattern of adding -ed to make the past tense), but in NAE at least, the strong verb pattern has been extended to it, resulting in less regularity.

Depends what you mean by 'regular'.

We were talking about making the system more regular, not making a specific pattern more regular.

[Xwtek]

The reason why I ask for using regularity is because Navajo conjugation is also formed this way. Regularization that makes the verb less regular. Or at least according to Proto-Athabaskan Verb Stem Variation by Leer.

[jal]

Regularization that makes the verb less regular.

Examples? Because like was explained above, most here don't think that's possible with the regular definition of regular.


JAL