Ćwarmin has a rich system of cases, which to some extent form fusional forms with number and the definiteness system. Ćwarmin also has a rather rich system of postpositions which are not considered cases. What makes one set considered cases and the other set not?
The Ćwarmin cases can be found here. However, nearly all Ćwarmin postpositions also are suffixed (and the prepositions are prefixed) to words. There are, however, certain traits that distinguish them. Some suffixes are sometimes cases, sometimes not.
1. Possible carriers
Postpositions are less picky than cases. They go on the final word of a noun phrase, regardless of what that is. Cases can only go on nouns, adjectives, pronouns and numbers. However, some postpositions also have case suffixes on them, often showing their origin as nouns, adjectives or pronouns.
(Make up etymologies for "apart (from)", "together with", "between" and some others).
apart from: -źəd-ir (egəd 'roads' + plural general ablative)
"roads away (from each other)"
together with: -xədəŋ-ic
between: -xədəŋ-ijn (xədəŋ is related to the adaŋ-morpheme in numerals)
Similarly, only cases distinguish the paucal from the plural, or the specific from the definite. Postpositions tend to go on the accusative stem, but one also find examples of general ablative, the dative or the instrumental or negative, with the occasional comitative-to thrown in. Sometimes, a postposition will take a different case in the plural than in the singular. Postposition-case pairings are not entirely regular: there seems to be both a lexical component which is basically a parameter to a probabilistic parameter. In addition, many of the postpositions cause some morphological wear at the end of the suffix, sometimes making it unclear which case is involved. Factors which affect the postposition-case pairing are:
* the more individuated the noun, the more likely to be dative it is
* the more involved despite lack of volition, the more likely to be instrumental
* negations, absence, etc tend to favour the negative case
* the comitative-to sometimes is used to emphasize a noun as being central to the events or important
* the general ablative tends to be lexically determined by some nouns to appear in the plural
* the dative tends to be lexically determined by some verbs to appear with the postposition
Only in the case where a postposition goes on the accusative or general ablative stem is there any suspicion that they're actually cases.
2. Morphological position
Postpositions go after all the other suffixes of a word. Postpositions can, however, take their own suffixes. Cases go after derivational suffixes, and after case-number-definiteness*.
* Ćwarmin grammatical tradition is to consider the whole case-number-definiteness-complex a single marker, despite clear agglutination.
3. Behavior over gaps and ellipses
This, in particular, is the behavior where some cases in some constructions behave like cases, and in other constructions do not. Adpositions in Ćwarmin can mark a whole coordinated phrase, case can only mark one noun phrase each:
milt-əneś (ul) mar-ummona
[milti-əneś (ul) maruw-ummona]
"for liver (and) kidneys"
*milti ul mar-ummona
*wekre (ul) pokr-oku
wekr-iki (ul) pokr-oku
with garlic and onion
Two cases that have different behaviors in different contexts are the negative and the general ablative. The singular dative can also showcase both in a few particular constructions, e.g. with verbs for 'resemble x.dat', and other verbs of perception where something the stimulus is compared to is in the dative, it seems both case-like and case-unlike use is permissible for many speakers.
3.1 The negative
The negative, when communicating an abessive/caritive meaning, does not seem to be a proper case, but rather a postposition. Thus, you could get
wekre dar pokr-usta
garlic nor onion-without
without garlic and onion
but as subjects or objects,
*wekre dar pokr-usta ogm(o)-ur-ka źub-u
garlic nor onion stone-from-in grow-3sg
wekr-istə (dar) pokr-usta ogm-ur-ka źub-u
neither garlic nor onion grow from stone
4. Interactions with the reflexive accusative
Most postpositions can interact with the reflexive(ly possessed) accusative. Thus
wicxə-sin-rede (behind my/your/his/her/... (own) house)
*wicxə-sin-itite (of ... (own) house)
5. Congruence with numerals
The congruence with numerals is limited already, and is sometimes blocked by other case-marking constructions. However, the case suffixes do not participate in it at all, and only the case of the noun itself prior to case suffixing can affect case marking of numerals.
Fusionality with number and definiteness is generally not required to be a case, but if it is present, it is a case. All numbers that have such fusion have all the other hallmarks.
7. Coordination of postpositions
Postpositions can be coordinated, cases cannot:
*the roof-on or-onto
*on or onto the roof