In search of isolating conlangs

[clawgrip]

Yiyenmore, which I made for the Kool Map Game (remember that?), is mostly isolating. It has some inflectional plurals, but these are avoided in favour of reduplication, when possible.

[Yng]

How is that isolating then?

Isolating and compounding aren't exclusive.

Exactly: Fym is itself what I term "polyanalytic": Highly synthetic - though the phrase is the basic unit of speech, not the "word", but it entirely lacks bound morphemes:

Zī ňí p̃ejá xét rà tàs ŝycȳt cēp nsé hwā ([t͡sʰi˧ ɲɪ˥ k͡pʰeˑja˥ ħet˥ ɹä˩ tʰas˩ ʂəˑʈ͡ʂʰəʈ ʈ͡ʂʰep˧ nze˥ hwɑ˧])
face.eyes.A EXCL color flax.linen arm.hand.fingers.P good garb.clothing speech.word.phrase.SECUNDATIVE MIDDLE.SPEECH.LEVEL cloth.fabric.textile
“S/he weaves my colorful linen into good clothing.”

mate what are you doing.

Your choice of awful unreadable colour coding for unclear reasons is making my head feel unusually explosive but uh... what? Compounding and isolating are indeed not mutually exclusive, but you're not doing compounding as far as I can see, you're just bizarrely overglossing individual words (why have you given three synonyms for cloth in your gloss of the word for 'cloth'?!) and... well, how does this sentence even work? Let's gloss it readably:

3p EXCL color linen hand good clothing speech.SEC MID cloth

Not sure what your .a and .p are marking, but unless there's actually a morpheme there indicating agent and patient (?) roles weird to gloss them. Also strange to gloss 'face' as 'face' given that it's presumably 3p anaphora here? What is the last 'cloth' doing? Is it a verb? If so what is hand doing? Is it an instrumental? More importantly, speech.SEC and MID look like morphemes to me and certainly not independent ones, so why are you analysing them as independent morphemes? Is it because you're writing them separately and you really want your language to be isolating? I hope not! In any case, what is your case for calling it polyanalytic?

[chris_notts]

mate what are you doing.

Your choice of awful unreadable colour coding for unclear reasons is making my head feel unusually explosive but uh... what? Compounding and isolating are indeed not mutually exclusive, but you're not doing compounding as far as I can see, you're just bizarrely overglossing individual words (why have you given three synonyms for cloth in your gloss of the word for 'cloth'?!) and... well, how does this sentence even work? Let's gloss it readably:

3p EXCL color linen hand good clothing speech.SEC MID cloth

Not sure what your .a and .p are marking, but unless there's actually a morpheme there indicating agent and patient (?) roles weird to gloss them. Also strange to gloss 'face' as 'face' given that it's presumably 3p anaphora here? What is the last 'cloth' doing? Is it a verb?

I'm going to guess that it might be some kind of classifier of the verb's object, like in e.g. Navajo. Of course I don't know, but it would explain why we have one redundant looking "noun" floating at the end. If that's so, then for me the biggest question is around 'hand', which I think might be an instrumental, i.e. somehow 'weave' is 'speak by hand'. But if it's an instrumental, what does '.p' mean?

Is it because you're writing them separately and you really want your language to be isolating? I hope not! In any case, what is your case for calling it polyanalytic?

Perhaps it depends what you mean by 'word'? For example, orthographies sometimes write clitics as separate words for grammatical reasons even though phonologically they aren't really. Prepositions / role markers in a lot of languages fail to behave like independent phonological words but are still written separately. And then there are all kinds of odd real-world decisions to write things like unstressed agreement markers as separate words in languages from Bantu to Polynesian.

In my own conlang Mɛdíṭṣai I do write a number of verbal clitics as separate words. The main reason I chose to do this is because (a) I felt like it, and (b) I thought it was a bit clearer given the context in which the clitics are used. Mɛdíṭṣai has SVCs whose component verbs retain independent stress / tone, and these serialised verbs can then be preceded or followed as a group by certain verbal clitics. As a multi-word phrase is in the middle of a clitic 'sandwich', I though that writing these clitics as if they were part of the first or last verb would be less clear than writing them like they were separate words. An example:

Written: ní θò ɕīp wɛ

Structure: ní=[θò ɕīp]=wɛ
1.sg.dir=[want go]=stative

Pronounced: /niˈθòˈɕīpwɛ/
- stress on verbs only (causing e.g. difference in vowel length and quality)
- loss of tone from ní - would migrate if following verb had mid-tone, but as it is the tone is lost

The same ní is also used as a possessive prefix, and in that case I do write it as if it were part of the noun it attaches to, again since (a) I felt like it, and (b) I felt that there were fewer confusing scope issues in that context. For example:

pʰiaī = sister
nipʰiaí = my sister

I guess what I'm saying is that a lot of real word orthographies don't have a strict alignment between words and written units with spaces around them, and also some morphemes have some properties of words but not others. And I don't think that a conlang necessarily has to be written in what a linguist would consider a perfectly designed system.

I'm also not sure though that I would claim that Mɛdíṭṣai is isolating (if his situation is similar to mine). It is certainly mostly agglutinating / non-fusional, but it has a number of clitics, especially verbal clitics, which are basically obligatory in many contexts. And I think that's probably the most interesting thing to me - whether a word is phonologically independent or not and whether it's written with a space or not is less interesting than how often and in what contexts it is obligatory. An 'isolating' language with a well-defined paradigm of obligatory pre-verbal subject 'pronouns' is really not much different to a language with subject agreement, regardless of phonological facts such as ability to be independently stressed.

[CatDoom]

More importantly, speech.SEC and MID look like morphemes to me and certainly not independent ones, so why are you analysing them as independent morphemes? Is it because you're writing them separately and you really want your language to be isolating? I hope not! In any case, what is your case for calling it polyanalytic?

Well, it looks like the morphemes in question have there own tones, and are presumably not distinguished prosodically from content words, making them phonetically independent. It looks something like a Chinese language, which make use of a lot of grammatical particles rather than affixes or clitics. It's like the difference between the English "to Hell!" and the Lithuanian "Velniop!", which uses the allative case suffix -op to express the same meaning as the English preposition "to."

[chris_notts]

More importantly, speech.SEC and MID look like morphemes to me and certainly not independent ones, so why are you analysing them as independent morphemes? Is it because you're writing them separately and you really want your language to be isolating? I hope not! In any case, what is your case for calling it polyanalytic?

Well, it looks like the morphemes in question have there own tones, and are presumably not distinguished prosodically from content words, making them phonetically independent. It looks something like a Chinese language, which make use of a lot of grammatical particles rather than affixes or clitics. It's like the difference between the English "to Hell!" and the Lithuanian "Velniop!", which uses the allative case suffix -op to express the same meaning as the English preposition "to."

The problem is that "particle" often just means a clitic that isn't recognised as such. For example, it's my understanding that many so called "particles" in Mandarin bear the "neutral tone", which is actually more like the lack of a distinct tone. The actual surface tone on these particles is basically determined by the tone of surrounding syllables in content words. From Wikipedia:

Also called fifth tone or zeroth tone (in Chinese 轻声 [輕聲] qīng shēng, literal meaning: "light tone"), neutral tone is sometimes thought of as a lack of tone. It is associated with weak syllables, and thus usually comes at the end of a word or phrase, and is pronounced in a light and short manner. The pitch of the neutral tone depends almost entirely on the tone of the preceding syllable. The situation is further complicated by dialectal variation; in some regions, notably Taiwan, the neutral tone is relatively uncommon. The following table shows the pitch at which the neutral tone is pronounced after each of the four main tones.[31] (These values refer to Beijing dialect; other dialects may be slightly different.)

So even in Mandarin, there is a phonological difference between content morphemes and some grammatical morphemes. Are these particles really phonologically independent words, or are they clitics? And if they aren't independent phonologically, does this make a difference to whether Mandarin counts as an isolating language or not?

Of course, there may be languages where there is no difference in phonological behaviour between grammatical morphemes and content morphemes. But a lot of the time "particle" means "clitic", and the choice to insert a space or not is an arbitrary orthographic choice rather than reflecting actual independence between morphemes.

To take another example, you compare Lithuanian case marking and English prepositions. Let's add in Basque, shall we? Basque has a number of role markers which are written without a space and change shape (a bit) depending on the word they attach to:

nire etxe-an
my house-LOC
"in my house"

So, similar to Lithuanian, right? So let's see what happens when we add an adjective:

nire etxe txikian
my house small-LOC
"in my small house"

The case markers are actually noun phrase final clitics rather than suffixes, despite the orthographic choice. Despite that, they form an obligatory small closed class and mark core relations (absolutive, ergative), and some of them show more phonological integration than English prepositions do. So in some ways they are like Lithuanian cases, and in some ways like English prepositions.

If Basque were an obscure language spoken in the middle of the Amazon and the linguist who wrote the grammar decided to put spaces before the case clitics, maybe labelled them particles or adpositions instead of case endings, then we'd think Basque was more like English and less like Lithuanian. As it is, people don't speak of Basque case "particles", they speak of case "endings".

My point is that while the end points might be well defined, for a lot of languages in the middle the difference is more a matter of analytic choice than it is of underlying differences between languages.

[Trebor]

chris_notts--good points about Basque. Shouldn't the cases really be analyzed as clitics, since they don't always and invariably attach to nouns?

And cool to bump into you directly (I remember conversing with you after meeting on the Conlang mailing list, of which I was an active member a decade ago).

[CatDoom]

My point is that while the end points might be well defined, for a lot of languages in the middle the difference is more a matter of analytic choice than it is of underlying differences between languages.

That's a good point; even the English preposition I used in my example can't always be said to be phonologically independent. At least in casual speech it often behaves (in the dialect I'm most familiar with) as if cliticized to the preceding word, such that "go to the market" is realized as [goɾə ðə mɑɹkɪ̈t].

This does, however, raise the question of whether any language can really be described as isolating, and if so what criteria we want to use to distinguish such languages. I tend to think think that if we exclude a language like Mandarin from our definition of "isolating," it renders the term nearly useless.

[chris_notts]

chris_notts--good points about Basque. Shouldn't the cases really be analyzed as clitics, since they don't always and invariably attach to nouns?

Well, yes they are clitics, I think. But they are more interesting clitics than English prepositions as they form a small closed class, and some of them mark core roles rather than just obliques.

And cool to bump into you directly (I remember conversing with you after meeting on the Conlang mailing list, of which I was an active member a decade ago).

Yes, you actually told me about the ZBB. I only visit occasionally now, since life gets in the way a lot.

[chris_notts]

That's a good point; even the English preposition I used in my example can't always be said to be phonologically independent. At least in casual speech it often behaves (in the dialect I'm most familiar with) as if cliticized to the preceding word, such that "go to the market" is realized as [goɾə ðə mɑɹkɪ̈t].

This does, however, raise the question of whether any language can really be described as isolating, and if so what criteria we want to use to distinguish such languages. I tend to think think that if we exclude a language like Mandarin from our definition of "isolating," it renders the term nearly useless.

I think this is kind-of my point though - the class of "isolating" vs "agglutinating" languages are hard to define well. I think (although I have no proof) that in almost all "isolating" languages there is a class of high frequency function words/particles/morphemes that are phonologically unlike content words in some way, e.g. lack of stress, lack of tone, tendency towards reduction or integration with neighbouring words, and different people will make different choices about how to represent those differences orthographically. So we get "particles" from some descriptive linguists and "clitics"/"affixes" from others when often there isn't much difference.

This is a problem for our proposed typology because not only is it hard to define the boundary, but for the majority of lesser known languages we talk about, we don't even have enough material to understand the phonological behaviour of the morphemes in question. We just have to take the analysis of the person who wrote the grammar at face value. Even if there was a possibility of a well defined boundary, or even a well defined cline (which is the importance of different features to the property "isolating"?), we don't have enough information to classify most actual languages firmly.

And then because this typology exists, people will create conlangs that are isolating or agglutinating. But most of those people with "isolating" conlangs won't describe well the behaviour of their "particles" - are they always stressed? Sometimes stressed? Never stressed? What about tone? Duration? Possibility of vowel contraction / reduction? Do they behave any differently to content words? The assumption would be, if there is no statement to the contrary, that all their particles behave exactly like any other morpheme, but I really think that this is quite rare in real natural languages, the rareness is just hidden by incomplete documentation.

For me, the more interesting question than "how much of the grammatical machinery consists of dependent morphemes?" are:

- how complex is the grammatical machinery, regardless of whether it is particles/clitics/affixes, e.g a language where the most common clause contains a bare verb root is very different to one where every clause contains obligatory agreement + TAM + evidentiality, even if those are marked by "particles". Isolating in the sense of "there are few obligatory grammatical categories marked" is probably better defined than "there are few dependent morphemes".

A good example here would be a number of Polynesian languages with fairly obligatory subject and TAM "particles" - I don't really see much difference with, say, some Bantu languages apart from an orthographic convention to insert some extra spaces. But those languages with their subject and TAM particles probably mark more obligatory distinctions in their clauses than most Chinese languages do.

Let's call this synthesis2 to distinguish it from the more common question about dependent morphemes (synthesis1).