Agglutinative vs. Analytic. What's the difference?

Question

First of all, I understand that these typological distinctions are not absolute and almost all languages show signs of almost all morphological strategies but most display a certain tendency towards one or the other.

I also understand that agglutinative languages are said to use bound morphemes called affixes for many grammatical functions where isolating/analytic languages prefer independent function words like adpositions and particles.

What I'm not sure about is how an affix is differentiated from a particle. As a result, I'm not sure about how to distinguish (mostly) agglutinative languages from (mostly) analytic languages.

I'm a native Turkish speaker. In Turkish, affixes undergo certain (mostly predictable) phonological assimilations (like vowel harmony) according to the base morpheme they're attached to. Less often the base morpheme also undergoes similar (again, mostly predictable) changes. This is how I differentiate, say, a case suffix from a postposition. The former undergoes the assimilations while the latter doesn't. For example:

    yaprak + den  = yapraktan   ~ from (the) leaf
    yaprak + gibi = yaprak gibi ~ like (a) leaf

Here, I can tell that the first one is a case suffix because it undergoes vowel harmony and voice assimilation while the second is a mere postposition because it doesn't undergo similar changes.

Are these criteria applicable to other agglutinative languages? If not, what other methods are used to distinguish an affix from an independent function word?

Answer 1

Your analysis is correct - the typical criterion is whether or not the affix/particle seems to act as part of the word it attaches to. When nice phonological demonstrations like vowel harmony aren't available, prosody is the next recourse. If prosody isn't helpful, you're pretty much left with an open question - you have to reason from the rest of the language's typology, and if that provides no further help, you're basically stuck.

Many languages with a long history of being written don't follow the above criteria exactly, and often divide words from affixes based on either the etymology of the morphemes in question or on historical spelling patterns. Some languages, in either their native systems or in romanisations, largely ignore the criteria, though - Japanese particles seem very much like they ought to be considered affixes (certainly they're not separable prosodically, and they've undergone some word-medial sound changes to their initial consonants), but the prevailing non-scientific romanisations (and several scientific romanisations) treat them as separate words.

Answer 2

In addition to the criteria in Sjiveru's answer, syntactic criteria are also important for distinguishing affixes from adpositions (or determiners, or any other syntactic class whose function might be equivalent to that of affixes; but I'll focus on adpositions here).

Affixes attach at the word level, while adpositions attach at the phrase level. If what they're attaching to is a single word, you might not be able to tell the difference: in a phrase like to Rome, to looks like it could be analyzed a a prefix. But this is no longer true when a longer phrase is involved: in to the beautiful city of Rome, to Italy's capital, to that place we went last summer, it's clear that to can't be a prefix since it's attaching to a whole phrase, not just the nearest word.

Another test that's sometimes useful is coordination: you can say to and from Rome, but affixes can't typically be coordinated in that way.