How do languages stay agglutinative?

[Max1461]

How do languages stay agglutinative for long periods of time? It seems that after just a few conditioned changes, fusional aspects would almost immediately start to creep into the language, especially if it has loose restrictions on syllable structure. Yet the indigenous languages of North America are (correct me if I'm wrong) mostly agglutinative, and frequently have a huge range of allowed clusters. How is this possible? I can understand a bit more in the case of something like Japanese, where the syllable rules are strict enough that there are fewer possible conditioning environments that could to lead to weird stuff happening at morpheme boundaries, but it still seems hard to believe that the language could stay almost completely agglutinating for at least two thousand years. Is this due to a low number of conditioned changes (I doubt it), crazy amounts of leveling (seems more plausible), both, or something else entirely?

[Travis B.]

Analogy.

[Zaarin]

While you certainly have some bizarre clusters in languages like the Salishan family, other languages have much simpler syllable shapes, like CV(C) in Tlingit or Haida. Also, agglutinating languages, especially polysynthetic ones, are tolerant of a certain level of morphophonological fusion.

[8Deer]

Is this due to a low number of conditioned changes (I doubt it), crazy amounts of leveling (seems more plausible), both, or something else entirely?

Since Salish consonant clusters have been brought up, I will say that there very few conditioned sound changes in that family in general, and almost none that affect consonants. I doubt most agglutinative language familes are like this, but conditioned sound changes aren't always common.

[Frislander]

Is this due to a low number of conditioned changes (I doubt it), crazy amounts of leveling (seems more plausible), both, or something else entirely?

Since Salish consonant clusters have been brought up, I will say that there very few conditioned sound changes in that family in general, and almost none that affect consonants. I doubt most agglutinative language familes are like this, but conditioned sound changes aren't always common.

The same general lack of conditioning goes for Semitic, what with that tri-consonantalism and all.

[Travis B.]

It should be noted that an apparent lack of conditional sound changes is likely a consequence of analogy, due to analogy encouraging originally conditional sound changes to be either generalized or reversed, resulting in a language with only seemingly unconditional sound changes.

[8Deer]

It should be noted that an apparent lack of conditional sound changes is likely a consequence of analogy, due to analogy encouraging originally conditional sound changes to be either generalized or reversed, resulting in a language with only seemingly unconditional sound changes.

I think this is what has happened in Salish. Certain sound changes appear really irregular, possibly because the conditioning environment has been obscured through grammaticalization (especially with glottalization of resonants, stress, and ablaut).