Synthesis and Isolation

[Terra]

I've some questions about synthesis and isolation.
1) Among languages of the world, does there seem to be any natural upper limit to the morphemes-per-word ratio?
2) Among languages that are descended from languages that displayed much synthesis, but that display only little or no synthesis, is there a common reason that made them change? (Like perhaps sound changes causing too many distant declensions/conjugations to collapse together, thus encouraging reanalyzation and innovation?)
3) What does one call the meanings indicated by something like "-o" in Spanish "amo"?, (Where it marks 3 different things: the 1st person, singular, present.)
4) Is there an upper bound to the ratio of these meanings-per-morpheme ratio too?

[zompist]

On 1), I dunno, but the hairest samples Whim and I found for the ALC were no more than 20. That's certainly not an upper limit; but e.g. I browsed through Valentine's immense Nishnaabemwin grammar and didn't find anything bigger.

My understanding is that polysynthesis, by its nature, doesn't really include indefinitely recursive procedures. Some recursion, yes: an incorporated noun may have several modifiers. But not an indefinite number, because incorporation is just not used for arbitrarily complex descriptions. Whim's polysynthesis thread has more on this.

On 4), Latin gets up to 7. (To the ones you found for -o, add mood, perfectivity, voice, and conjugation.) I wouldn't be surprised if that could be beaten; I'd be surprised if it could be doubled.

[Radius Solis]

On (4), go carefully. It can depend a lot on what you consider a distinct meaning or a distinct thing marked. Is past tense + perfective aspect a single thing, the "preterite"? Or is it two? Sometimes a grammar may give you reasons for wanting to call it one way or the other, but not necessarily. Or perhaps a suffix clearly marks person, number, and tense, but it only appears as part of the indicative paradigm; does it then mark mood too? Does it still mark mood if it's part of the otherwise-distinct paradigms for both indicative and, say, interrogative, but something else is found in the subjunctive? And so on and so forth.

[hwhatting]

On (2) - obscuring sound chnages have frequently been named as a reason for a shift from fusional to analytic, but I'm not sure whether there's real proof for that. It certainly doesn't need to lead to that outcome - there are many examples of fusional languages that have "reacted" by re-arraging their inflectional classes or morphonological patterns in order to make them more transparent without losing their fusional character. Quite often language contact seems to be the reason either by creolisation or due to the spread of areal features. Sometimes, they are hard to separate; consider English - a move towards analytic structures is typical for big parts of Western Europe, but there may also be more local reasons; I think it's no accident that the areas that most advanced from fusional to analytic during the middle English period were in the North-East, i.e. the Norse-English contact area.

[Terra]

On (2) - obscuring sound chnages have frequently been named as a reason for a shift from fusional to analytic, but I'm not sure whether there's real proof for that. It certainly doesn't need to lead to that outcome - there are many examples of fusional languages that have "reacted" by re-arraging their inflectional classes or morphonological patterns in order to make them more transparent without losing their fusional character. Quite often language contact seems to be the reason either by creolisation or due to the spread of areal features. Sometimes, they are hard to separate; consider English - a move towards analytic structures is typical for big parts of Western Europe, but there may also be more local reasons; I think it's no accident that the areas that most advanced from fusional to analytic during the middle English period were in the North-East, i.e. the Norse-English contact area.

I must ask then, what compelled Old Norse to become more analytic?

Heck, this whole thing raises the question of why languages change at all in the first place, and why they never settle on an certain set of features.

Also, is there any difference in meaning between the terms isolating/synthetic and analytic/fusional?

[hwhatting]

I must ask then, what compelled Old Norse to become more analytic?

The point is not that Old Norse went analytic and dragged English after it, but that there was a pidginisation / creolisation effect caused by two groups who spoke fusional languages trying to communicate in (Middle) English. As for why the North Germanic languages became analytic, I assume it's Western European areal influence.

Also, is there any difference in meaning between the terms isolating/synthetic and analytic/fusional?

AFAIK, "isolating" and "analytic" are mostly synonyms, "isolating" stressing the absence of syntactical role affixes and "analytic" the use of periphrastic constructions. "Synthetic" is often used as the general category for the subtypes "fusional / agglutinating / polysynthetic".

[Vuvuzela]

AFAIK, "isolating" and "analytic" are mostly synonyms, "isolating" stressing the absence of syntactical role affixes and "analytic" the use of periphrastic constructions. "Synthetic" is often used as the general category for the subtypes "fusional / agglutinating / polysynthetic".

"Isolating" is sometimes stretched out to include derivational, as well as inflectionnal morphology. In which case, English, Indonesian and Afrikaans would be fairly anlalytic (relying heavily on word order to distinguish semantic roles) but not isolating (antidisestablishmentarianism, mempertanggungjawabkannyalah, and meervoudigeprobleemsiektetoestande being words with many morphemes.)

[clawgrip]

There is one thing I have wondered...isolating languages are supposed to have a low morpheme-per-word ratio, but this definition is unclear, because the definition of "word" is dependent on the orthographic conventions of specific languages. I see no reason why "football" should be considered one word of two morphemes while "bóng đá" should be considered two words of one morpheme each, since they are both clearly compounds referring to a single and specifically identifiable class of object. Does isolating then indicate a lack of bound derivational morphemes, such as anti-, -ous, -ment, etc., while analytic indicates a lack of bound inflectional morphemes, such as -ing, -ed, -s, etc.? What is the specific difference between them?

[Vuvuzela]

There are multiple working definitions of "word". For instance, "an apple" is one word, phonologically, because "an" is unstressed and vowel-reduced, and this is usually characteristic of unstressed syllables within phonological English word, but not in the words themselves. However, it's two words syntactically, because "underrated" is definitely it's own word, and "an underrated apple" is a phrase I can say freely in English, so it doesn't have a special attachment to the stem word. You can tell a compound word from a word which simply has an adjective stuck onto it in English by whether you can shove another word in between them and still have it make sense, for instance, I can say "an old, Greek house" and "A Greek, old house", and while the latter may sound just a bit odd, you know what I'm talking about. By contrast, I could never say "a foot oldball" and expect you to understand me. So, while there may be many definitions of "word", alan gua ge c an'tjus t p utspacesanywherean dgetreclassifiedtypologicallyforit.

Different languages approach orthographic words differently. While they tend to be based on some conception of a "word" they are often inconsistent. Spanish treats pronoun clitics as suffixes ("hacerlo") even though they are seperable from the verb ("lo hago"), however, it treats articles as their own words (un hijo). Vietnamese orthography is weird in this regard. There are things, albeit mostly loans, that are written with two orthographic words that have only one morpheme between them, like "cà phê", so I don't think where Vietnamese orthography draws the line between words should really be taken as the measure of what constitutes a word in this language.

[Cedh]

Does isolating then indicate a lack of bound derivational morphemes, such as anti-, -ous, -ment, etc., while analytic indicates a lack of bound inflectional morphemes, such as -ing, -ed, -s, etc.?

I can't say whether this phrasing would be entirely in line with how professional linguists specialising in morphology define the difference, but it's certainly the most concise summary of the distinction between "isolating" and "analytic" I have read so far, and probably the most useful one for conlanging purposes too, especially for beginners.

(Except that both terms describe not just "lack", but also "paucity".)

[zompist]

Actually I'd say that derivational morphemes are irrelevant to "isolating" as it's commonly used. The classic example is Chinese, after all, which has hundreds of bound morphemes.

Clawgrip is right to be suspicious of definitions of "word" that rely on orthography. For the purposes of estimating the degree of synthesis, a word can be loosely defined as a unit that can stand on its own and can't have additional arbitrary material inserted into it. E.g. using Quechua as an example, Puñu-chka-n-raq-mi 'He's still sleeping' is one word, as none of the morphemes can stand alone, and you can't add arbitrary things at any of the hyphens.

Compounds like "blackbird" or "White House" are best thought of as words, as they have a single lexical meaning and can't have additional material inserted in the middle.

(Quechua words obligingly also have a phonological feature— they have one stressed syllable— but this doesn't apply to all languages.)

In practice, none of this seems to cause any real problems, especially if you don't let yourself be swayed by where written languages put the spaces. Edward Sapir, at least, reported that native speakers of polysynthetic languages had no trouble agreeing on word boundaries even if they'd never written their language before.

(Also, a reminder: real languages are at least a little bit mixed. If we describe a language as "isolating/agglutinative/fusional", there's always an implied 'mostly'.)

[Salmoneus]

Just to point out: English also defines words phonologically, as well as syntactically. "Football", "blackbird" and "White House" are all phonologically distinct from "foot ball", "black bird" and "white house", thanks to the stress patterns. So, for instance, I'm not sure you can put anything between the two words in "Football League", but you can tell they're two words rather than one or three because of the stress patterns.

[clawgrip]

"Football league" has only one stressed syllable for me ('foot') and it basically functions as a single "word".

So if isolating doesn't have anything to do with bound morphemes, then what do these words even mean? How can we define them?

[zompist]

"Isolating" refers to a low degree of synthesis-- essentially the number of morphemes per word.

(We're usually thinking of inflectional morphology there, but compounding doesn't change much. Chinese is going to rival Nishnaabemwin even if you include compounds.)

"Analytic" is usually applied to constructions, not languages-- e.g. we might say that English has both an analytic and a synthetic comparative.

[clawgrip]

"Isolating" refers to a low degree of synthesis-- essentially the number of morphemes per word.

(We're usually thinking of inflectional morphology there, but compounding doesn't change much. Chinese is going to rival Nishnaabemwin even if you include compounds.)

This is why it's confusing. People always claim Chinese to be an isolating language, yet it has many bound morphemes and makes heavy use of compounding, e.g. 牛海绵状脑病, which is arguably one word with six morphemes. I guess it just means that Chinese is more isolating than many other languages. A 100% isolating language would require a separate morpheme for every single concept.

[zompist]

Whether a language does compounding or not is not a useful criterion, because as far as I know all languages do (not that they all do it the same way). That's why we look mostly at inflections.

To put it another way, if you wanted to define "isolating" as "no compounding, no bound morphemes", you won't find any examples at all and it'd be a useless term.

It's true that there's a lot of confusion when it comes to East Asian languages. Some of this is due to Western unfamiliarity, but some is due to the native speakers, thanks to the Chinese writing system, which is so syllable-oriented that the native intuition is that language is built out of syllables (zì). (This attitude carries over even to languages that have abandoned hanzi: han'kul is still written in syllables, and Vietnamese prefers to separate syllables with spaces.)

Chinese linguistics didn't even itemize bound vs. free morphemes till Yuen Ren Chao's 1947 dictionary.

But linguistics considers the spoken language primary— if nothing else, the vast majority of languages are unwritten (by their users). So we really do have to ignore the structure of the writing system.

[clawgrip]

I recognize that Chinese characters cause a distortion in perceptions. I recognize also that there is no language that does not employ compounding of some sort. What I meant was that since compounding is a form of synthesis, Chinese is less isolating than people think.
Also, when you say that derivational morphemes are irrelevant to the common definition of "isolating", what then is relevant? Just compounding? How do you wind up with more than one morpheme per word in a way that is relevant to "isolating"?

[Magb]

Just to point out: English also defines words phonologically, as well as syntactically. "Football", "blackbird" and "White House" are all phonologically distinct from "foot ball", "black bird" and "white house", thanks to the stress patterns. So, for instance, I'm not sure you can put anything between the two words in "Football League", but you can tell they're two words rather than one or three because of the stress patterns.

I disagree with your analysis of "football league". It's stressed like "blackbird", not like "black bird", i.e. it's [ˈfʊtˌbɒlˌliːg], not [ˈfʊtˌbɒlˈliːg]. It's a single word.

Also, you can put the infix "fucking" between any two syllables in any English word. Foot-fucking-ball-fucking-league.

[Yng]

What? No you can't. -fucking-'s position is determined by stress and sometimes by morphemic constraints: *ab-fucking-solutely doesn't work, but abso-fucking-lutely does.

[Magb]

What? No you can't. -fucking-'s position is determined by stress and sometimes by morphemic constraints: *ab-fucking-solutely doesn't work, but abso-fucking-lutely does.

I may have exaggerated.

[Yng]

What? No you can't. -fucking-'s position is determined by stress and sometimes by morphemic constraints: *ab-fucking-solutely doesn't work, but abso-fucking-lutely does.

I may have exaggerated.

Best not to, it only makes you look stupid.

[Magb]

What the hell, man? I made a serious comment and added what I thought would be taken as a lighthearted, not very serious aside at the end of it. I know a lot of people in this forum are extremely fond of correcting people (myself included), but you're going pretty far.

[Gulliver]

What the hell, man? I made a serious comment and added what I thought would be taken as a lighthearted, not very serious aside at the end of it. I know a lot of people in this forum are extremely fond of correcting people (myself included), but you're going pretty far.

Yes, that was mean, now let us please move on.

Just to point out: English also defines words phonologically, as well as syntactically. "Football", "blackbird" and "White House" are all phonologically distinct from "foot ball", "black bird" and "white house", thanks to the stress patterns. So, for instance, I'm not sure you can put anything between the two words in "Football League", but you can tell they're two words rather than one or three because of the stress patterns.

I disagree with your analysis of "football league". It's stressed like "blackbird", not like "black bird", i.e. it's [ˈfʊtˌbɒlˌliːg], not [ˈfʊtˌbɒlˈliːg]. It's a single word.

Both /ˈfʊtˌbɒlˌliːg/ and /ˈfʊtˌbɒlˈliːg/ sound okay to me, so maybe one's a true compound noun and one's a modifier noun and a head noun.

As others have said, though, the word word means different things and is not a fantastically useful analytical concept. Most linguistics courses at some point involve a discussion on what a word is, and there are no clean answers.

2) Among languages that are descended from languages that displayed much synthesis, but that display only little or no synthesis, is there a common reason that made them change? (Like perhaps sound changes causing too many distant declensions/conjugations to collapse together, thus encouraging reanalyzation and innovation?)

Language change is influenced by so many factors, I doubt that there's a one-size-fits-all answer for this. In English, the move from inflecty Old English to not-that-inflecty Middle English was probably a mixture of contact with Brythonic language (a high density of OE L2 speakers learning the language imperfectly and forming new linguistic norms), contact with Old Norse (fairly similar language, probably with a high rate of mutual intelligibility but different inflections in some cases, with the two languages rubbing together until the sticky out bits fell off), sound changes that weakened endings that were not made up for by syntax, and contact between different dialects of Old English (as late as Early Modern English, present tense verbs varied considerably across regions, with some not taking any inflection, some taking -s on all counts, and some following rules that are suspiciously Celtic-looking). There are some theories that a tight social network or cohesive speech community can hold back language change, with a trend for more isolated and unified speech communities being somehow inherently more conservative.

3) What does one call the meanings indicated by something like "-o" in Spanish "amo"?, (Where it marks 3 different things: the 1st person, singular, present.)

I'd just call it a bound morpheme that indicates three things. Morphemes with multiple meanings aren't uncommon enough to warrant their own specific vocabulary, I think, and the modern study of linguistics comes largely from speakers European languages where it's commonplace and hence unremarkable.

[Magb]

I disagree with your analysis of "football league". It's stressed like "blackbird", not like "black bird", i.e. it's [ˈfʊtˌbɒlˌliːg], not [ˈfʊtˌbɒlˈliːg]. It's a single word.

Both /ˈfʊtˌbɒlˌliːg/ and /ˈfʊtˌbɒlˈliːg/ sound okay to me, so maybe one's a true compound noun and one's a modifier noun and a head noun.

As others have said, though, the word word means different things and is not a fantastically useful analytical concept. Most linguistics courses at some point involve a discussion on what a word is, and there are no clean answers.

I'm sure there's some variation on this point in the English-speaking world, but when I sound out the version with the stressed "league" I can only hear it as "football league", as in "No, not the football tournament, the football league."

I completely agree with your statement about the word "word", by the way. I originally had a paragraph about it in my post but I took it out for some reason.

[merijn]

3) What does one call the meanings indicated by something like "-o" in Spanish "amo"?, (Where it marks 3 different things: the 1st person, singular, present.)

I'd just call it a bound morpheme that indicates three things. Morphemes with multiple meanings aren't uncommon enough to warrant their own specific vocabulary, I think, and the modern study of linguistics comes largely from speakers European languages where it's commonplace and hence unremarkable.

It's called a portmanteau morpheme.