difference between Isolating (analytics) vs inflected (fusional) vs agglutinative languages

Question

It's not easy to grasp these concepts. I spent a lot of time perusing wikipedia articles but still can't really understand what makes a language: inflexed, isolating or agglutinative,

Background These are languages that I know, and I would love the answer to post some examples in these languages : - Polish - English - Spanish - French - Italian - Russian - Vietnamese - Chinese

As far as I understand the first 6 are inflexed languages(with Russian and Polish highly inflexed), and the last two are highly isolating.

Now, Wikipedia says that an inflexed language uses inflectional morphemes:

Inflectional morphemes modify a verb's tense, aspect, mood, person, or number, or a noun's, pronoun's or adjective's number, gender or case, without affecting the word's meaning or class (part of speech). Examples of applying inflectional morphemes to words are adding -s to the root dog to form dogs and adding -ed to wait to form waited.

Now I don't see a problem to appoint the Vietnamese các/nhũng and đã as the equivalent morphemes for respective English s and ed from the Wikipedia's excerpt. Namely:

a dog dies -> con chố chết

dogs died -> các con chó đã chết

"The only thing" that I see differs supposedly inflexed and supposedly isolating languages is that languages like English, Polish have many versions of a same morpheme, like for example there are at least more than 5 morphemes to express the past tense in Polish and English, while Vietnamese has only one, namely đã.

Polish to English to Vietnamese example:

Ja jem -> I eat -> tôi ăn

Ja jadłem -> I ate -> tôi đã ăn

Ja jadę -> I go -> tôi đi

Ja jechałem -> I went -> tôi đã đi

But then one could say that grammatical classifiers so abundant in Chinese and Vietnamese are morphemes that varies greatly depending on situation. Is it that linguistics was mainly developed in Europe and nowadays all world linguists try to look at other languages from the European standpoint?

1. Unfortunately I can't give any example of an agglutinative language as I don't know a word in any such language. But I would love the answer to address the distinction inflexed-agglutinative and isolating-agglutinative too
2. What are the problems with my reasoning with Polish English and Vietnamese?

Answer 1

Just to clarify matters a bit, the OQ seems to have a confusing presupposition, viz

Isolating (analytic) vs inflecting (fusional) vs agglutinating languages
(it's inflected, btw, not inflexed)

All three are not on the same level of abstraction. Instead of a three-way opposition, there are two dyadic oppositions, with one subordinate to the other:

• Analytic (isolating) are opposed to Synthetic (inflecting) grammars,
  or types of grammar. This is essentially the difference between

  • syntax (which predominates in the grammar of analytic languages) and
  • inflectional morphology (which predominates in the grammar of synthetic languages).

• Vietnamese and English are examples of Analytic languages.

• Turkish and Latin are examples of Synthetic languages.

As a subtype of Synthetic grammars, there are two distinct types of inflection.

1. Fusion (or amalgamation).
   Latin (and almost all other I-E languages) uses fusional inflection.

   Consider the paradigms of Latin verbs and Latin nouns; all are multidimensional, with one affix conveying case, gender, number, and root class for nouns, or person, number, tense, mood, voice, and root class for verbs. This results in a number of unique paradigms, each to be learned in context, but also in short inflections, with high information content.

2. Agglutination
   Turkish (and almost all other SOV languages) uses agglutinative inflection.

   Consider the verb forms in this Turkish puzzle, as described in this schematic solution. Notice that there are a number of affixes attached to a root, and that their paradigms are all one-dimensional. I.e, there is just one column of choices, representing only one characteristic -- number, person, tense --- instead of several; very low information content. The result is that the morphology is simple, and very regular, because the paradigms have only one dimension. However, there are many affixes per inflected word.

Executive summary:
1. There is a cline between Analytic (Isolating) and Synthetic (Inflecting) grammars.
2. There is a cline -- among synthetic grammars -- between Fusional and Agglutinative inflection.

Answer 2

@Eleshar's answer sums it up very well: “Good luck with separating some of the forms into morphemes”.

Still, there's one important difference that makes it impossible to draw a straight parallel between classifiers (of isolating languages) and morphemes of inflexed languages.

This is because in fusional languages, the modifier morphemes conjugate as well!

Here's the story.
When Lewis Carroll was traveling to Russia, he saw an interesting Russian word: защищающихся. It means, "people who defend themselves". He wrote this word in his diary according to the English phonology: zаshchееshchауоushchееkhsуа.

Most likely, Carroll was marveled about the consonant /shch/ which sounds unusual to an English speaker; we, instead, will dig into this word's morphology:

защищающихся
  └─┘        щит            1. /shchit/  n. shield         
├───┘        защит[а]       2. /zashchita/ n. defence, lit. "behind shield"
├────┘       защища[ть]     3. /zashchishchat'/ v. defend
├──────┘     защищающи[й]   4. /zashchishchajushchij/ adj. one who defends
├───────┘ └┘ защищающи[й]ся 5. /zashchishchajushchijsya/
                                    adj. one who defends self
└──────────┘ защищающихся   6. /zashchishchajushchikhsya/
                                    adj. one who defends self +PL +GEN

(in square brackets I put morphemes that do not exist in the final word)

While steps (2), (4), and (5) can be directly understood in terms "morpheme → morphologic change", several steps can't be translated that easy. Look what happens here:

1. At step (3), the root consonant shifts: /t/ → /shch/; an impossible thing for isolating languages;
2. At step (6), the inflection particle inflects itself:
   "-ий-" /ij/ (SNG.NOM) → "-их-" /ikh/ (PL.GEN)

---

So, even if a morpheme-to-morpheme parallel can be drawn between an arbitrary pair of isolating and fusional languages, there's still a considerable amount of cases where this parallel does not work.

Answer 3

The main difference according to my understanding is that in inflective languages, one usually declines, or change morphemes (which are closely integrated into the word) to inflect meaning of the word. In agglutinative languages, one append prefixes/suffixes to add meaning to a word - the word's stem is largely untouched. In analytic languages, those morpheme changes are largely absent - instead, one combines independent morphemes to convey meaning.

However, few or none languages are purely inflective/agglutinative/analytic (isolating). Proto-Indo-European, and thus many Indo-European lanaugages, are heavily inflective. Romance languages are mainly inflective, but modern Romance languages have developed many analytic features as well (especially French). Mandarin Chinese is basically analytic, but it does have a few agglutinative features. Modern English is also mainly analytic, but it has some notable inflective features (e.g., what's left of verb inflections). Modern Japanese is both significantly analytic and significantly agglutinative.

I don't think any answers have yet given examples of agglutinative languages, so I'll give some examples of the agglutinative features of Japanese. Let's take the Japanese verb 食べる taberu "to eat" as an example:

食べる taberu "eat"
食べられる taberareru "able to eat"
食べられたい taberaretai "want to be able to eat"
食べられたくない taberaretakunai "do not want to be able to eat"
食べられたくなかった taberaretakunakatta "did not want to be able to eat"
食べられたくなかったら taberaretakunakattara "If I did not want to be able to eat"

You could see that each suffix appended to the verb taberu does exactly one thing: -rareru expresses the potential to perform an action, -tai expresses the wish to do some action, etc. They are largely independent from the verb stem, but do not form independent words on their own. Comparatively, English is basically analytic: to convey the same meaning, one use independent morphemes like "able to" and "want" to modify "eat". While in a inflective language, it will be hard or nearly impossible to extract those morphemes out - not only because they are so closely integrated into the word stem itself, but also because it is simply not true that each prefix/suffix has one function.

However, Japanese is heavily analytic in many other areas of its grammar, compared to languages generally considered primarily agglutinative (e.g., Turkish). I don't know any Turkish, but I saw this on Wikipedia:

ev "(the) house" evler "(the) houses"
evin "your (sing.) house"
eviniz "your (pl./formal) house"
evim "my house" evimde "at my house"

While Turkish nouns seem to be generally agglutinative, Japanese nouns are generally not. A rough translation of the above mentioned expressions into Japanese would be:

家 ie "(the) house/(the) houses" (there is no explicit plural in Japanese)
あなたの家 anata no ie "your (sing.) house" (literally "house of you")
あなた達の家 anata-tachi no ie "your (pl.) house" (literally "house of you (pl.)")
私の家 watashi no ie "my house" (literally "house of me")
私の家に watashi no ie ni "at my house"

This pattern is very analytic. Instead of using prefixes/suffixes, Japanese uses independent words and particles in this case. The only thing that could be categorized as an agglutinative suffix is tachi (which is equivalent to Chinese 們 men). Of course, one could analyze particles such as no and ni as prefixes/suffixes, but this makes about as much sense as considering English of and at, or Mandarin 的 de and 在 zai as such. So, no, I would consider Japanese nouns as analytic.

Answer 4

To be honest, this classification comes from 19th century and it is based mostly on verb, occasionally nominal inflections as the primary criterion of comparing languages (and subsequently, also establishing which one is better, more evolved, more reflecting the overall more advanced national spirit of its speakers etc.).

This corollary aside, the principle is that isolating languages have the morphemes typically as separate words, sometimes mere clitics, sometimes full-fledged words but in any case pretty much separate. As a result of this, indeed there is typically just one way of expressing the grammatical meaning, because being a separate word, the morph does not undergo contextual changes.

On the other hand, in fusional languages, the morpheme is typically merged with the content morphemes it is related to to a rather high degree and it cannot be conceivably separated from them and the morphemes can carry more than one grammatical function at once.

Consider sanskrit word for father. It declines as follows (sg./du./pl.):

nom.: pitá / pitaráu / pitarah

gen.: pituh / pitróh /pitrnám

dat.: pitré / pitrbhyám / pitrbhyah

acc.: pitaram / pitaráu / pitrn

voc.: pitah / pitaráu / pitarah

loc.: pitari / pitróh / pitršu

instr.: pitrá / pitrbhyám / pitrbhih

abl.: pituh / pitrbhyám / pitrbhyah

Now this is not an irregular declension, it is a whole paradigm. Good luck with separating some of the forms into morphemes, let alone finding what constitues the case in the ending and what constitutes the number.

Of course there are varying degrees of this and as indicated in the corollary at the top, in Indo-European languages the fusionality hits mostly case endings for declensions and person endings for verb conjugation. Otherwise the derivation and even most inflection is based on the same principle of one morpheme per one function and in this respect may be considered agglutinative.

Agglutination is between these two extremes and you can argue that many of the endings are actually somewhat more clitics than parts of the word. E.g. when you decline an adjective phrase, the case endings are appended frequently just to the very last element of the sequence (ADV ADJ NOUN-nr-case) but if you rearrange the order, they are not bound to a specific word class (NOM ADV ADJ-case endings). The fusional languages often require that the grammatical functions are expressed on all the elements of the phrase (ADV ADJ-nr-case NOUN-nr-case).

Answer 5

I find that question interesting. I've noticed that in some (?) analytic languages a phrase which consists of -let's say- a preposition, an article and a noun can be just one phonological word. The string 'for the lake' is listed as one. If we followed different conventions these morphological words could have been written as one.

In Modern Greek (which is fusional) the string 'ο άνθρωπος' ('definite article' in nominative and noun in nominative) is one phonological word /o'ɐn̪θropos/. The accusative with the article would be /ton'ɐn̪θropo/.

I guess someone who reads that may consider that there's a prefix and a suffix for each case. But if we add an adjective the article would form a phonological word with the adjective and there are phonological changes too in that case. /tonkɐ'lon 'ɐnθropo/ → /to(ŋ)gɐ'lon 'ɐn̪θropo/. (n+k → (ŋ)g)

I'm not a linguist. Sorry for any wrong term used, although I think I was careful.