On Sat, 20 Oct 2012 17:05:45 +0200, G. van der Vegt <[log in to unmask]> wrote:

>Has anyone ever designed a conlang where most (or all) morphemes fell
>into two main classes, 'objects' and 'modifiers' and where you could
>slot any modifier into any object, and you couldn't have any modifiers
>or objects that stood alone without the other (though a 'null'
>modifier or object could exist, of course.)
>This might be a little abstract, so I'll try to give a few examples of
>hypothetical morphemes for such a language in the section that
>This hypothetical conlang has a CV structure, all object morphemes are
>C_C_C_ and all modifier morphemes are _V_V_V (this is probably more
>restrictive than if I decided to create a more full-fledged conlang,
>but should suffice for the purposes of demonstration)

Using an example based on three consonants, there was your mistake!  Now everyone's gone off about Semitic, and missed what I take to me your actual point, namely that both the "object" and "modifier" halves of a word have independent fairly lexical meanings, which is a lot more compositional (dare I say Legolike?) than Semitic is.  All ye listmembers who have cited Semitic, tell me, what semantic contrasts distinguish say the noun-forming vowel frames 1a23 and 1i23 and 1u23 ?

But I do have an approximate ACADEW, namely Bleackley's iljena (  There the consonant-frames are nouns and the vowel-frames are intransitive verbs expressing semantic roles, more or less.  Granted, the verb thing isn't quite what you're talking about, but I don't actually see how you meant to provision for (/ avoid) verbs in your sketch.