I don't think you could claim that cran- is synchronically related to crane, since (1) there is no regular morphophonological process producing such "composition forms" (e.g. name + sake --> *namsake), (2) it is not *self-evident* -- even to a linguist, without investigating the history of the word and ruling out folk etymology -- that cran- has anything to do with crane diachronically (the semantic connection between cranes and cranberries is certainly obscure). By contrast, we are aware that -berry derives from berry even if there is a good deal of phonetic erosion (as in British English). Take mulberry or whorthleberry as still obscurer formations. I'd say that in such cases X- simply means 'X-berry' and -berry is semantically redundant. Bound morphemes may be recognisable as morphemes *only* because they are combined with a familiar element, e.g. overwhelm or unkempt. We know that unkempt means 'untidy, neglected' and that un- is a negative prefix, so we can associate -kempt with 'tidy', even if we don't use it on its own.
If both elements of a historical compound are obscured both formally and semantically, it is no longer analysable as a compound, even one containing bound morphemes, cf. gospel or gossamer. The latter derives from two familiar elements (goose + summer), but their form is distorted beyond recognition and semantically gossamer <-- goose-summer no longer makes sense.
----- Original Message -----
From: Mitko Sabev
To: phoNet@yahoogroups.com
Sent: Thursday, February 08, 2001 12:06 AM
Subject: [phoNet] Re: greater language units


{BE} and {CRAN} are morphemes for they do have meaning.

'be-' is a prefix and it has a variety of meanings; 'cran-' is
related to 'crane' and is one of the roots in a compound noun.

What might have made you assume they "aren't meaning-carrying" is the
fact that they are BOUND morphemes (as opposed to free morphemes),
i.e. they cannot be used on their own.