Where are the analytic and nonconcatenative conlangs?

[Vardelm]

Tone and stress are absolutely considered morphology.

Well, I learned something today!

[Das Public Viewing]

I don't know. Are tone & stress considered "morphology"? I assumed they weren't, and therefor wouldn't be "non-concatenative morphology".

Tone and stress are absolutely considered morphology. They're distinct phonological segments which can bear meaning - there are morphemes and morphological processes in many languages that consist of nothing but tone or stress.

For a semi-productive example of morphological stress in English, consider the difference between the verb recórd and noun récord.

Oh, there's tons of examples like that in English.
2-syllable alternating noun/verb stress: alynnidalar already gave the example of récord/recórd, but there's also the classic áddress/addréss, plus pérmit/permít, áffect/afféct, rédo/redó, ínvite/invíte, úpset/upsét...heck, it's got a wikipedia page!
Roots never surpass the antepenultimate: This is pairs like fántasy/fantástical, anthropólogy/anthropológical, eléctrical/electrólysis, álchemy/alchémical, xénophobe/xenophóbia, vérify/verifíable/verifiabílity, dúrable/durátion/durabílity, phárynx/pharýngeäl(or even pharyngéäl). I also could've sworn I saw something similar with a germanic word and the -ery suffix, but I can't recall it for the life of me. The only exceptions to this one that I can think of are irréparable and pálatalize, the former of which is usually said as "irrépprable" and the latter of which I've seen as palátalize on occasion. I've never once heard a native speaker mess up on this one. I do believe I've been schooled. Ah, how I missed being thouroughly wrong...
Emphatic stress: the one everyone already knows about

This actually makes me wonder, though: how many conlangs use stress in a similar way? I think I remember a romlang or two that had a regular stress contrast somewhere in the verb system, but otherwise, has anyone made a "stress lang" (so to speak)?

[zompist]

Roots never surpass the antepenultimate: This is pairs like fántasy/fantástical, anthropólogy/anthropológical, eléctrical/electrólysis, álchemy/alchémical, xénophobe/xenophóbia, vérify/verifíable/verifiabílity, dúrable/durátion/durabílity, phárynx/pharýngeäl(or even pharyngéäl). I also could've sworn I saw something similar with a germanic word and the -ery suffix, but I can't recall it for the life of me. The only exceptions to this one that I can think of are irréparable and pálatalize, the former of which is usually said as "irrépprable" and the latter of which I've seen as palátalize on occasion. I've never once heard a native speaker mess up on this one.

abominable, accompaniment, accuracy, adequacy, agglutinating, agriculture, apothecary, architecture, brother-in-law, candlemaker, caricature, cauliflower, cemetery, countertenor, dedicated, dictatorship, discriminating, doppelgänger, egalitarianism, gentlemanly, intimacy, inimitable, isolating, interesting, invigorating, laboratory (in America), lesbianism, literature, mercilessness, metalworker, miserable, moneylender, nominative, oligarchy, oratory, orthodoxy, partisanship, patriarchy, seriousness, solitary, testimony, trustworthiness, veritable, voluptuousness, witticism, womanizer

[Salmoneus]

Tone and stress are absolutely considered morphology.

Well, I learned something today!

Well, tone and stress are just phonology, same as any other phonology. It's only when they're used to mark grammatical distinctions that they become, like any other phonology, morphological.

[Salmoneus]

Roots never surpass the antepenultimate: This is pairs like fántasy/fantástical, anthropólogy/anthropológical, eléctrical/electrólysis, álchemy/alchémical, xénophobe/xenophóbia, vérify/verifíable/verifiabílity, dúrable/durátion/durabílity, phárynx/pharýngeäl(or even pharyngéäl). I also could've sworn I saw something similar with a germanic word and the -ery suffix, but I can't recall it for the life of me. The only exceptions to this one that I can think of are irréparable and pálatalize, the former of which is usually said as "irrépprable" and the latter of which I've seen as palátalize on occasion. I've never once heard a native speaker mess up on this one.

Aside from the fact that, as Zompist points out (though some of his counterexamples seem to be mistaken*), this isn't true, it's also worth pointing out that the stress plays no real morphological role here - the suffix is morphologically contrastive, and the stress shift is conditioned by that, rather than the stress itself being significant (which is part of the reason these words so easily vacillate between stress patterns - cóntroversy vs contróversy, for instance).

Although there are some cases where secondary stress placement is phonemic and derivational, I think - precípitate vs precípitáte, for instance. It's marginal, though.




*I've never in my life, for example, heard 'countertenor' without a strong stress on the penult.

[zompist]

I've never in my life, for example, heard 'countertenor' without a strong stress on the penult.

Really? Even a UK dictionary has /ˈkaʊn tə ˌten ə/. Another one. You really have /kaʊn tə ˈten ə/?

(There's a secondary stress on 'tenor', sure, but primary stress should be on 'counter'.)

[chris_notts]

I've never in my life, for example, heard 'countertenor' without a strong stress on the penult.

Really? Even a UK dictionary has /ˈkaʊn tə ˌten ə/. Another one. You really have /kaʊn tə ˈten ə/?

(There's a secondary stress on 'tenor', sure, but primary stress should be on 'counter'.)

I think I have the primary stress on the penultimate syllable as well... This gives both stress patterns as possible for British English with recordings, but the one with the initial stress sounds really weird to me:

https://www.oxfordlearnersdictionaries. ... untertenor

[missals]

This actually makes me wonder, though: how many conlangs use stress in a similar way? I think I remember a romlang or two that had a regular stress contrast somewhere in the verb system, but otherwise, has anyone made a "stress lang" (so to speak)?

I've speculated about making something like this before, where stress is the primary or exclusive morphological operator - partially inspired by Solresol, in fact, which has some stress-based inflection and derivation (although it's essentially broken, along with the rest of the language).

E.g. nominal roots might be disyllabic, inflecting for number: /ˈliko/ 'dog' vs /liˈko/ 'dogs' and verbal roots might be trisyllabic, inflecting for different categories: /ˈkabato/ 'kill.nonpast' /kaˈbato/ 'kill.past' /kabaˈto/ '(a) killing/(a) kill'

I don't think such a system is impossible for a natlang, but it would likely have evolved from stress changes induced by earlier affixes, which would probably rear their heads in various contexts; it would also be prone to evolving into some kind of ablaut or Semitic-ish consonantal root system - or at the very least, to be accompanied by English-type vowel reduction as a secondary realization of stress. Not necessarily, but likely.

And with regard to tone, it's definitely plausible and possible for a language's entire morphological system to rely completely or almost completely on tone.

[Salmoneus]

I've never in my life, for example, heard 'countertenor' without a strong stress on the penult.

Really? Even a UK dictionary has /ˈkaʊn tə ˌten ə/. Another one. You really have /kaʊn tə ˈten ə/?

(There's a secondary stress on 'tenor', sure, but primary stress should be on 'counter'.)

I have primary stress on 'counter', but secondary full stress also on 'tenor'; I've also heard primary stress on 'tenor' with secondary full stress on 'counter'; I've never heard it without full stress of some sort on 'tenor'.

By which I mean, it's a full lexical stress, not the automatic weak secondary stress seen in most words, so it's not really a counterexample to the antepenult rule. Otherwise you could break the antepenult rule any time you just compounded one word to another - "water ocelot", for instance, has stronger stress on 'water' than on 'ocelot', but it's not really what people mean when they talk about the antepenult rule.

[Alon]

E.g. nominal roots might be disyllabic, inflecting for number: /ˈliko/ 'dog' vs /liˈko/ 'dogs' and verbal roots might be trisyllabic, inflecting for different categories: /ˈkabato/ 'kill.nonpast' /kaˈbato/ 'kill.past' /kabaˈto/ '(a) killing/(a) kill'

I don't think such a system is impossible for a natlang, but it would likely have evolved from stress changes induced by earlier affixes, which would probably rear their heads in various contexts; it would also be prone to evolving into some kind of ablaut or Semitic-ish consonantal root system - or at the very least, to be accompanied by English-type vowel reduction as a secondary realization of stress. Not necessarily, but likely.

Spanish conjugation relies on this extensively: amo = I love, amó = s/he loved.

And then there's umlaut, which is very common in conlangs because it's common in European languages. But more extensive systems of vowel alternations seem uncommon.

[Kuchigakatai]

My question is, where are the analytic conlangs? Where are the isolating ones, using word order, particles, and free morphemes? (Hebrew is of course not analytic, but because it has very little concatenative morphology, derivation that uses common European affixes that Semitic languages don't have weights for uses free morphemes: lucky is "son_{con.} luck," reelection is "election from-new," insensitive is "lacking sensitivity," etc.)

Are there good examples of analytic and nonconcatenative conlangs? If I'm right and there are very few of them, why do you think that is?

You can do a lot of interesting things with analytic syntax, though, like having coverbs and relational nouns rather than adpositions, or playing with word order, or not really using pronouns.

Honestly, when I conlang I always seem to be more interested in working out the morphology (especially the inflectional morphology) than the syntax. So making a very isolating language (like Vietnamese or Cantonese) would mean skipping what I find most fun. My conlangs aren't even that morphologically heavy (nouns typically distinguish only 2 numbers and 2-4 cases, and verbs only 1-3 TAMs and 2-4 valencies, derivational affixes don't add up to more than 15 per word class), so they're very analytic while not isolating, but nevertheless it's the declension/conjugation categories and morphophonological processes that I concentrate on.

Besides, you can have all that sweet syntax stuff with non-isolating languages too, whether we're talking about strict word order, or particles, free morphemes corresponding to European derivational affixes, coverbs (maybe with limited tense-agreement, to boot!), relational nouns, implied pronouns, etc.

[Das Public Viewing]

excrement

obvious things

First of all, I'd like to formally apologize for completely forgetting that secondary stress exists.

Second, a question about the scope of nonconcatenativity: do any of the following count?
Truncation - Is this nonconcatenative? If so, why is any one form of the word considered the "base form"? Isn't it just that everything else has an affix, while the truncated form is the base (a-la the Slavic a-stem gen.pl)? What's the criteria of what the base morpheme is?
Infixation - It's an affix, sure, but it's inside the word, and often dependent on the phonological constraints of the word. If there's a difference between apophony and infixes, the which category do IE nasal infixes fall under? In Arabic iftaa'ala, is there an infix and what is it? Is regular metathesis of an external affix equivalent to an infix? For vowel length, is the extra mora an infix? What about internal gemination?
Reduplication - I kinda thought this was cut-and-dry concatenation, since you're adding on additional elements (even if it's the same element). However, wikipedia seems to disagree. Is there an argument in favour of it being considered nonconcatenative?

You can do a lot of interesting things with analytic syntax, though, like having coverbs and relational nouns rather than adpositions, or playing with word order, or not really using pronouns.

...you can have all that sweet syntax stuff with non-isolating languages too, whether we're talking about strict word-order, particles, free morphemes corresponding to European derivational affixes, coverbs (maybe with limited tense-agreement, to boot!), relational nouns, or implied pronouns, etc.

But aren't there still analytic features that are hard to justify putting elsewhere? Do any nonanalytic languages have an equivalent to the Chinese classifiër system, for instance? (where specific quantities of something need to be accompanied by a particle)

[Pabappa]

Japanese , but they may have borrowed it from Chinese

[missals]

Second, a question about the scope of nonconcatenativity: do any of the following count?
Truncation - Is this nonconcatenative? If so, why is any one form of the word considered the "base form"? Isn't it just that everything else has an affix, while the truncated form is the base (a-la the Slavic a-stem gen.pl)? What's the criteria of what the base morpheme is?

Well, firstly, one can distinguish between "truncation" as in "clipping", a derivational process that produces new words by shortening already existing ones - e.g. fridge from refrigerator - and the process of "disfixation", which is usually applied to inflectional processes. I think clipping can straightforwardly be termed "nonconcatenative" - refrigerator > fridge isn't knocking off a meaning-bearing phonological unit, it's just knocking off some phonological material.

Disfixation is harder to pin down since it's quite uncommon and can often be analyzed in different ways in the languages it (arguably) appears in. But to address your comment about the truncated form being the base, no, that isn't necessarily the case - the shortest form of a word is by no means necessarily the most basic form. Commonly, linguists identify the form of a word from which the other forms of a word can be predicted as the base form. When something like disfixation is at play, this is generally not the truncated form. For example, if we have the following pairs of words, the one on the left the masculine, and the one on the right the feminine, which form is the base? Which is the one which, if you know it, you can predict the other?

epu, epuz
fʀɛ, fʀɛʃ
fʀwa, fʀwad
blɑ̃, blɑ̃ʃ

The form on the right, the feminine. If you only know the masculine is epu, you have no way of knowing that the feminine is epuz. But if you know that the feminine is epuz, you can apply the general rule observable in the other pairs - knock off the last consonant - and correctly predict the masculine form, epu. Moreover, it is simpler to state that "The masculine is generated via deletion of the final consonant from the feminine" than to posit that -z, -ʃ, -d, etc., are all realizations of a feminine suffix whose variant phonological forms are unpredictably distributed across different vocabulary terms. This is why it is possible to describe French (the language above, of course) as having a type of disfixation.

Infixation - It's an affix, sure, but it's inside the word, and often dependent on the phonological constraints of the word. If there's a difference between apophony and infixes, the which category do IE nasal infixes fall under? In Arabic iftaa'ala, is there an infix and what is it? Is regular metathesis of an external affix equivalent to an infix? For vowel length, is the extra mora an infix? What about internal gemination?

I can't say if all forms of infixation are concatenative, but many are straightforwardly concatenative - in many languages that have infixes, the "infixes" are really just prefixes or suffixes that undergo metathesis in order to not violate phonotactic constraints - e.g. a prefix ik- added to adoma becomes ikadoma, but added to tosila it produces tikosila.

Reduplication - I kinda thought this was cut-and-dry concatenation, since you're adding on additional elements (even if it's the same element). However, wikipedia seems to disagree. Is there an argument in favour of it being considered nonconcatenative?

Concatenative morphology refers to morphemes being strung together like beads on a string. If a language with a word luka 'dog' reduplicates nouns to mark the plural, does the second /luka/ in lukaluka 'dogs' mean 'plural' in the way English plural -s does? Does the first? Which phonological sequence /luka/ is the one that means 'dog' and which means 'plural'? More sensible is to say that there was no morpheme that was added to the word - rather, the word's phonological transformation itself, involving a complete duplication of the whole phonological sequence, imbues it with a plural meaning, with the whole word standing for 'dog.plural'

But aren't there still analytic features that are hard to justify putting elsewhere? Do any nonanalytic languages have an equivalent to the Chinese classifiër system, for instance? (where specific quantities of something need to be accompanied by a particle)

There might be some features found mainly or nearly exclusively in heavily analytic languages, but I do know that there are plenty of non-analytic languages that have nominal/count classifiers - Japanese and Mayan languages both use them, for example.

[Kuchigakatai]

Infixation - It's an affix, sure, but it's inside the word, and often dependent on the phonological constraints of the word. If there's a difference between apophony and infixes, the which category do IE nasal infixes fall under? In Arabic iftaa'ala, is there an infix and what is it? Is regular metathesis of an external affix equivalent to an infix? For vowel length, is the extra mora an infix? What about internal gemination?

I can't say if all forms of infixation are concatenative, but many are straightforwardly concatenative - in many languages that have infixes, the "infixes" are really just prefixes or suffixes that undergo metathesis in order to not violate phonotactic constraints - e.g. a prefix ik- added to adoma becomes ikadoma, but added to tosila it produces tikosila.

As for the Arabic verbal form VIII, iCtaCaCa (note that all the /æ/s are short, no "aa"), it is definitely an interesting thing. I would say that, since the /t/ stays attached to the first consonant of the root throughout, there is a concatenative derivational process where an infix /t/ is added to the root, C-C-C -> Ct-C-C (EDIT: with a change in meaning of the derivational kind that I imagine you're aware of), and then the appropriate transfixes are added non-concatenatively as inflectional morphology: Ct-C-C -> iCtaCaCa, uCtuCiCa, yaCtaCiCu, yuCtaCaCu, iCtiCaaC, muCtaCiC... It's up for debate though.

However, missals's tikosila example seems purely concatenative, in as much as the phonological segments -ik- are added to the established string t-adoma.

[mae]

-

[Kuchigakatai]

It's "possible" to analyze French as involving subtractive morphology, but that analysis is pretty much universally rejected now, especially since the final consonants aren't absent from the masculine forms, they're just repressed in certain phonological contexts (those where liaison doesn't apply).

I suspect the analysis would still apply very well to some or many speakers though: French singular nouns never undergo liaison (so you'd never hear the underlying [z] at the end of e.g. masculine [epu]), and French adjectives only do so when they're found before a noun, which would mostly happen when reading poetry aloud when it comes to adjectives such as [fʁwa] ("[fʁwad]"), [fʁɛ] ("[fʁɛz]") or [blɒ̃] ("[blɒ̃k]")--an activity that many French native speakers do not do. I find it surprising you report relevant linguists "universally" reject the analysis with truncation.

[mae]

-

[Znex]

The liaison evidence is only one issue. Another is that despite the common contention that it's "implausible" or excessively "complex" for people to memorize a multitude of essentially arbitrary added segments, the data from acquisition, etc. shows that people do memorize all those things anyway. The subtractive analysis seemed at least reasonable to me when I first looked at the data, so like you I was also surprised to find that no one really uses that analysis anymore.

I suppose the more basic solution is to presume that both forms still have an underlying final segment? So froid still has |frwad| [fʁwa], and froide has |fʁwadE| [fʁwad]. Final segment deletion is hence a synchronic change rather than diachronic; the feminine form still has an archiphoneme marking the gender.

[mèþru]

I think Znex's analysis is correct, although instead of putting a "gender archiphoneme" I'd say it is the morphophonological realisation of final consonants in specific syntactic and morphological contexts (liaison and gender, as liaison has syntactic restrictions)