How much must morphemes be distinct from each other?

This is our main forum. Here, anything related to languages and linguistics can be discussed.

Moderator: Forum Administrators

xBlackHeartx
Posts: 91
Joined: 2017-02-23, 22:11
Country: US United States (United States)

How much must morphemes be distinct from each other?

Postby xBlackHeartx » 2018-12-26, 0:25

I have multiple times wondered how distinct words must be from each other to be practical. Obviously, having words that sound similar wouldn't matter much if the words are in separate semantic fields or in separate parts of speech. But what about things like case and conjugation endings? How different must they be from each other?

For example, lets say you had a series of endings that were all VC. How many could you have? Assuming they all served the same purpose, could you have the only distinction between them be the voicing of the last consonant? I kinda doubt it. Though I have seen plenty of systems where the only difference between certain endings was the vowel used.

Looking at actual case and conjugation endings across multiple languages, it would seem that natlangs prefer to keep the number of distinct consonants used to a minimum, often far less than the variety of consonants a word/syllable can end in within the language. Also, I have yet to find a language where every possible ending is unique to it just one meaning. Take Latin for instance, almost none of its case endings mark just one case, number, and gender combination. For an example I'm more familiar with, German marks case primarily through articles. It has four genders (which includes a universal plural) and four cases. In theory, this would mean that there should be a total of 16 definite articles, one for each possible combination of gender and case. In reality, there's only 6 varieties, and none are unique to one category. Only the masculine has distinct articles for the nominative and accusative, while the feminine uses the same article for the dative and genitive. The masculine and neuter use the same articles for the dative and genitive. The plural uses all the same articles as the feminine except in the dative. And then there's articles that are used for multiple unrelated combinations. For instance, the feminine dative and genitive is the same as the masculine nominative, and the masculine accusative is the same as the plural dative!

I don't get why I can't seem to find any languages where every possible inflectional ending holds just one meaning, but they seem to simply not exist. Do people just not like to have such a large number of distinct grammatical markers? I mean, German gets along fine with its case system, which interestingly enough allows free word order even though only pronouns and masculine nouns distinguish subject from object (and even then, not all pronouns do, the word for 'it' has identical nominative and accusative forms, and many don't distinguish the accusative from the dative).

I admit it, this is a question related to conlanging, but I figured this fit better in the 'general language' forum since is a question looking for examples from natural languages. In my mind, if no natlang does something, more often than not there's a very good reason for it.

h34
Posts: 928
Joined: 2014-12-16, 20:15
Gender: male

Re: How much must morphemes be distinct from each other?

Postby h34 » 2018-12-26, 11:10

I think some agglutinative languages are quite close to that ideal, e.g. Komi:
► Show Spoiler

xBlackHeartx
Posts: 91
Joined: 2017-02-23, 22:11
Country: US United States (United States)

Re: How much must morphemes be distinct from each other?

Postby xBlackHeartx » 2018-12-26, 15:50

You do give some good examples, though thinking about it, the distinction between German articles is also pretty minimal.

As I've said, German only has a total of six different forms of its definite article:

der
den
des
dem
das
die

As we can see, four start with 'de' and the only difference between them is the final consonant. Though I should note that the 'e' isn't pronounced the same in all of them. My German teacher (who was actually from west Germany) pronounced the 'e' in 'den' as an (ipa) 'e' while it was pronounced as 'ɛ' in the other two (though the 'e' in 'dem' as pronounced slightly higher, at least by my teacher).

A better example may be the indefinite articles, which don't have this obvious redundancy that the definite articles have:

ein
eine
eines
einem
einen
einer

Obviously, this is even more minimal than the definite articles. The only oddity I should note is that final 'er' is pronounced as an 'ʌ' (the word 'der' is pronounced dɛʀ). As we see, we have one article which has a null ending, while all the others (but one) have 'ɛ' either on its own or with a final consonant. Of note is that 'einen' and 'einem' are distinct, despite their final consonants being so close.

As for their usage, well, it largely mirrors the definite articles. Where you have 'der' you use 'einer' and where you have 'die' you use 'eine'. With a few exceptions of course (the masculine and neuter nominative article is 'ein' for both). Well, actually, you could find the charts over on wikipedia:

https://en.wikipedia.org/wiki/German_articles

So obviously, the indefinite articles are less differentiated than the definite articles. Though I should note that definite articles are used much more often, and also for the indefinite articles adjectives show more differences in case than definite articles (with definite articles, adjectives only take one of two endings, one for the nominative and masculine accusative, the other for everything else).

Also, Germans normally don't play around with their word order as much in everyday speech; you largely only see them make use of their free word order in poems and songs (German isn't topic-prominent, unlike most free word order languages).

xBlackHeartx
Posts: 91
Joined: 2017-02-23, 22:11
Country: US United States (United States)

Re: How much must morphemes be distinct from each other?

Postby xBlackHeartx » 2018-12-26, 20:38

Going on to Spanish, its verb endings also tend to be rather minimal. Each personal ending tends to be differentiated purely by which vowel is used. Stress is of course phonemic in Spanish, but I don't think there's any tense where stress alone indicates person. Though you do have minimal pairs across different tenses. For instance, the -o ending either marks the first person present or the third person past, depending on whether the stress is placed on the final syllable or not.

For an example of a verbal paradigm, here's the endings for the simple present tense of an -a verb:

-o
-as
-a
-amos
-ais
-an

As we see, there are three persons (-as, -a, and -an) which are distinguished purely by a single consonant, though they aren't that close to each other, and there may be some kind of allophonic redundancy going on here like there is in the German definite articles (I only know really basic Spanish, and I've never had a teacher who was a native speaker unfortunately).

Linguaphile
Posts: 2096
Joined: 2016-09-17, 5:06

Re: How much must morphemes be distinct from each other?

Postby Linguaphile » 2018-12-26, 22:00

In some situations the difference between case forms is only the length of the phoneme.

Finnish: metsä (nominative: "forest"); metsää (partitive: "forest")
Lule Saami: miehttse (nominative: "forest"); miehtse (genitive: "forest")
Northern Saami: meahcci (nominative: "forest"); meahci (genitive: "forest")
Inari Saami: mecci (nominative: "forest"); meeci (genitive: "forest")
Estonian: metsa (genitive: "forest"); me'tsa* (partitive: "forest") (*normally not indicated in writing but heard in spoken language)
Estonian: maja (nominative: "house"); majja (illative: "house")

Most of these languages do have different endings for each distinct case declension, like h34 mentioned for Komi (different vowels, different consonants, and/or different phoneme lengths). Usually there are a few exceptions, such as genitive and partitive having the same ending for some words, etc. But for languages that have between six and twenty cases, that's still a lot of different endings. For example, Veps (again, the meaning here is "forest") - you can see that there are 18 different singular endings, even though mecan is used for more than one case:

Nominative: mec
Genitive & accusative: mecan
Partitive: mecad
Illative: mecha
Inessive: mecas
Elative: mecaspäi
Allative: mecale
Adessive: mecal
Ablative: mecalpäi
Translative: mecaks
Terminative: mechasai
Essive: mecan
Abessive: mecata
Comitative: mecanke
Prolative: mecadme
Propinquitive: mecanno
Approximative: mecannoks
Aditive: mechapäi
Egressive: mecannopäi

Then there are the plural forms (only mecad is repeated here):
Nominative: mecad
Genitive: mecoiden
Accusative: mecad
Partitive: mecoid
Illative: mecoihe
Inessive: mecoiš
Elative: mecoišpäi
Allative: mecoile
Adessive: mecoil
Ablative: mecoilpäi
Translative: mecoikš
Terminative: mecoihesai
Essive: mecoin
Abessive: mecoita
Comitative: mecoidenke
Prolative: mecoidme
Propinquitive: mecoidenno
Approximative: mecoidennoks
Aditive: mecoihepäi
Last edited by Linguaphile on 2018-12-27, 5:13, edited 1 time in total.

kevin
Language Forum Moderator
Posts: 2019
Joined: 2012-03-29, 11:07
Gender: male
Country: DE Germany (Deutschland)
Contact:

Re: How much must morphemes be distinct from each other?

Postby kevin » 2018-12-26, 22:09

xBlackHeartx wrote:Also, Germans normally don't play around with their word order as much in everyday speech; you largely only see them make use of their free word order in poems and songs (German isn't topic-prominent, unlike most free word order languages).

Depending on what you mean by "play around", this sounds questionable to me. In practice, SVO sentences aren't all that common. You get almost always something else than the subject that takes the first position. Often adverbials, but OVS word order isn't unusual either.

User avatar
Dormouse559
Language Forum Moderator
Posts: 6542
Joined: 2010-05-30, 0:06
Real Name: Matthew
Gender: male
Country: US United States (United States)

Re: How much must morphemes be distinct from each other?

Postby Dormouse559 » 2018-12-27, 7:04

For a broader look, WALS has chapters on syncretism (identical forms) in case marking and verbal person/number marking. I don't have much time to dig into them right now, but it looks like both languages with no syncretism and languages with at least some syncretism are common in WALS' sample.
N'hésite pas à corriger mes erreurs.

xBlackHeartx
Posts: 91
Joined: 2017-02-23, 22:11
Country: US United States (United States)

Re: How much must morphemes be distinct from each other?

Postby xBlackHeartx » 2018-12-27, 16:04

I clicked on a link on that website to find a list of languages that lack case syncretism. The only ones I found that I was even vaguely familiar with was Hungarian and Yoruba.

Looking through Hungarian's case system, it would appear that many are only one morpheme off. Though a lot of this is brought about by the fact that its spacial case endings follow an obvious pattern. This pattern is particularly evidence in the endings indicating 'movement away from' (whose structure is alwasy C+ól/ől) and the 'interior' case endings which all use B for the initial consonant. Though again, there may be some kind of allophonic redundancy going on here. Also, its obvious that some of the endings are related to each other, meaing that they probably once looked as regular as the 'movement away from' cases.

Note that I'm looking at the case chart on a wikipedia article:

https://en.wikipedia.org/wiki/Hungarian ... n_suffixes

Linguaphile
Posts: 2096
Joined: 2016-09-17, 5:06

Re: How much must morphemes be distinct from each other?

Postby Linguaphile » 2018-12-27, 17:33

xBlackHeartx wrote:I clicked on a link on that website to find a list of languages that lack case syncretism. The only ones I found that I was even vaguely familiar with was Hungarian and Yoruba.

The map and list include only 198 languages, so there are many languages that aren't included in this particular study. The point it makes is that both types of languages do exist in a fairly widespread way: languages with syncretism (of various types) and languages without syncretism.
I'm not sure what you mean by being "vaguely familiar with" languages listed there; that you have studied them a bit? that you have heard of them? I don't know which languages those are in either case. But in terms of widely-spoken languages from various language families, among the languages that lack case syncretism on the map besides Hungarian and Yoruba there is, for example, also Turkish, Khalkha (= Mongolian), Kannada, Hausa, and Quechua.

Quechua (example meaning "house"):
nominative: wasi
genitive: wasipa
dative: wasipaq
accusative: wasita
instrumental: wasiwan
abessive: wasinaq
locative: wasipi
inclusive: wasipiwan
allative: wasiman
ablative: wasimanta
terminative: wasiyaq
transitive: wasinta
comitative: wasintin
immediate: wasiraq
intrative: wasipura
exclusive: wasillam
comparative: wasihina
causative: wasirayku
benefactive: wasipaq

xBlackHeartx wrote:Looking through Hungarian's case system, it would appear that many are only one morpheme off. Though a lot of this is brought about by the fact that its spacial case endings follow an obvious pattern. This pattern is particularly evidence in the endings indicating 'movement away from' (whose structure is alwasy C+ól/ől) and the 'interior' case endings which all use B for the initial consonant. Though again, there may be some kind of allophonic redundancy going on here. Also, its obvious that some of the endings are related to each other, meaing that they probably once looked as regular as the 'movement away from' cases.

Yes, for many Uralic languages at least, there are often patterns to the locative cases. (And something similar for Quechua in, for example, the allative wasiman and ablative wasimanta).
For example, Estonian:
Inessive: -s (metsas) 'in the forest'
Illative: -sse (metsasse) 'into the forest'
Elative: -st (metsast) 'out of the forest'
Adessive: -l (metsal) 'at the forest'
Allative: -le (metsale) 'to the forest'
Ablative: -lt (metsalt) 'from the forest'
So you can see, the basic pattern is a single letter for a static location (-s for 'in', -l for 'at') with the addition -e for movement toward (-se for 'into', -le for 'to') or the addition of -t for movement away from (-st for 'out of', -lt for "from").
Similar patterns exist in many other Uralic languages. I'm not sure if you're interested, but you can find lists of case decensions for about two dozen Uralic languages here where we've collected them at Unilang (also conjugation of the verb 'to be' in most of the same languages here, although generally that's not the best verb if you want examples of regular conjugations).

It's occurred to me that looking at lists of cases is probably not the best way look at this, though. Let's take Estonian as an example again:
Nominative: mets
Genitive: metsa
Partitive: metsa (lengthened, so the pronunciation is not the same as the genitive form)
Illative: metsasse (although an alternative form is the same as the partitive, the lengthened metsa)
Inessive: metsas
Elative: metsast
Allative: metsale
Adessive: metsal
Ablative: metsalt
Translative: metsaks
Terminative: metsani
Essive: metsana
Abessive: metsata
Comitative: metsaga
No syncretism except for the alternative illative form (the so-called 'short illative') which matches partitive.

But here's the problem: accusative is not listed because accusative is often not considered a separate case in Estonian. And it's often not considered a separate case because it does not have a unique case ending. (Depending on which linguists you believe, Estonian accusative is either the partitive case, or is the same as genitive in the singular and nominative in the plural.) But the point here is that often a particular case is only considered a true case in the first place if it does have a unique form. As I mentioned earlier, most Uralic languages have between six and twenty cases; when they have a lower number of cases compared to related languages, it's generally because those fewer cases have taken on multiple roles. Where forms overlap, they're simply not considered separate 'cases'.
So, it's got me wondering: is the lack of the existence of a case also considered a form of syncretism? :para:

User avatar
voron
Language Forum Moderator
Posts: 4834
Joined: 2007-07-15, 3:29
Real Name: Igor
Gender: male
Country: TR Turkey (Türkiye)

Re: How much must morphemes be distinct from each other?

Postby voron » 2018-12-27, 18:14

Linguaphile wrote:there is, for example, also Turkish

Turkish is marked in the WALS data as having no syncretism, which is technically correct: all of its cases are distinct.
nominative: ev (house)
accusative: evi
genitive: evin
locative: evde
dative: eve
ablative: evden

However, some of these endings are homonymous with some possessive affixes:
evi - ev+i, can be accusative or 3SG possessive
karın - kar+ın, can be genitive of snow, or 2SG possessive your snow
(another two possible interpretations are karı+n your wife, or karın, nominative of stomach).

Even more homonymy occurs when case affixes and possessive affixes are combined:
evini - can be ev+in+i, accusative of your house, or ev+i+n+i, accusative of his/her house
arabanın - can be araba+nın, genitive of car, or araba+n+ın, genitive of your car

Linguaphile
Posts: 2096
Joined: 2016-09-17, 5:06

Re: How much must morphemes be distinct from each other?

Postby Linguaphile » 2018-12-28, 1:59

Something similar happens with a few Estonian suffixes. My 'favorite' (the one with the most potential for confusion) is the fact that the suffix -tu (-less) in the plural ends up being identical to the past participle -tud. You therefore end up with words and phrases that can have two meanings, which are the exact opposites of each other.
For example:
Värvitud seinad can be "colorless/unpainted walls" or "colorful/painted walls", depending on whether värvitud is the past particle of värvima (to color, to paint) or is the noun värv (color) with the suffix that means the same as "-less" in English.
Likewise litsentsitud terviseasutused either means "licensed health institutions" (from the verb litsentsima) or "unlicensed health institutions" (from the noun litsents) and so on.
:ohwell:

User avatar
md0
Posts: 7533
Joined: 2010-08-08, 19:56
Country: CY Cyprus (Κύπρος / Kıbrıs)

Re: How much must morphemes be distinct from each other?

Postby md0 » 2018-12-28, 5:41

I would imagine recoverability is a factor. If you can recover the proper meaning from the rest of the clause, the language doesn't mind merging a few morphemes here and there. I can see how that would make sense from an acquisition perspective.

What comes to mind is several noun pairs in Greek where the feminine singular and the masculine plural of nouns from the same root sound the same.
Eg /i politi'ci/ is either η πολιτική (Nom. Sg. F. 'the politics/the policy') or οι πολιτικοί (Nom. Pl. M. 'the politicians').
That's not at all a problem though, because in a phrase, both will trigger different number agreement on other words, and on other nominals they will also trigger different gender agreement.

Still, in political speeches (!), they love the rhetorical device where they take a word like /i politici/ and unnecessarily disambiguate it by saying "politics, written with heta, not omicron-iota" and then continue with a verb in 3rd person singular which is a sufficient cue for any native speaker anyway. Despite Greek having more-or-less 6 ways to spell /i/, we do not usually spell out words in conversation to disambiguate -- what politicians are going for in that context is to differentiate themselves, humble servants of the people, from either the precise art of governing or the bureaucratic machine that eats the servants of the people alive (depending on their views on government).
"If you like your clause structure, you can keep your clause structure"
Stable: Cypriot Greek (el-cy)Standard Modern Greek (el)English (en) Current: Standard German (de)Elementary Finnish (fi)
For fun: Legacy: France French (fr)Japanese (ja)Standard Turkish (tr)


Return to “General Language Forum”

Who is online

Users browsing this forum: No registered users and 1 guest