It seems that for content recognition of delimiter seperated values the content doesn't matter to the 
import routine. OOo seems to rather only consider the file-suffix.

I have tried the same content packed into different files with different suffixes.

---

Part before suffix = Indicates content and used delimiter.
Suffix = Just copied the same file and added different suffixes.

contacts-semicolon
contacts-semicolon.csv -> SUCCESS
contacts-semicolon.dsv
contacts-semicolon.tsv
contacts-semicolon.txt
contacts-tab
contacts-tab.csv -> SUCCESS
contacts-tab.dsv
contacts-tab.tsv
contacts-tab.txt

---

Only when the file ends in ".csv" the routine works correctly, it asks the user for how to interpret the 
content (which charset, which delimiter, column datatype setting, etc).
The other endings "tsv,dsv,txt" or none seem to get ignored, before content analysis even starts.

I was using NeoOffice 2.2.5 Patch 6, and I am not sure the OOo equivalent, hence filed it under 2.2.1