Word boundaries

I'm just been thinking... how "real" or how artificial are word
boundaries? Especially for languages that don't have word boundaries in
their (original) writing systems. Why must we treat every language
(nat/conlang) as if they have units called "words"?

For example, Mandarin doesn't really have "words" in the English sense of
"word", because most "words" are single-syllables, and sometimes double or
triple syllables become identifiable units that usually don't occur
isolated. And then you have clitics like (classical) Greek prepositions,
which often can merged into a unit with verbs. Even the description
"merged into the verb" sounds a bit too artificial to me: isn't that
"merging" just a result of Greek speakers habitually saying the
preposition before the verb, so that they become treated as a unit?

And who's to say that word endings are part of the word? I don't see why
they aren't considered as separate clitics that just happen to be always
attached to another word. I'm thinking of original Greek writing, which
didn't have embedded spaces. Why are word boundaries drawn where they are,
since they aren't even represented originally?

Then you have agglutinative languages... why is a particular sequence of
affixes considered as a "word"? Why not consider the affixes themselves
as "words", and a sequence of affixes as a "phrase"?

I know there are probably semantic reasons why word boundaries are the way
they are, but I'm not sure that things are quite so clear-cut as we seem
to normally think of them. The conlang Odonien comes to mind, where noun
case-endings are treated like clitics/particles that can be dropped when
unnecessary. So are these clitics/particles part of the word or not? They
don't mean anything on their own, and they only occur next to a noun if
they do occur at all. Why not consider them as part of the noun -- so the
noun without the particle is simply a "general case" of the noun which can
function as any case (but the context narrows it down to a specific case),
whereas the forms with the attached particle can be considered "specific
forms" of the noun which the speaker chooses to use in order to clarify
what function he meant the noun to be.


T