All of the languages I've made so far are placed on the same continent of my conworld, and it's a different continent from where humans evolved, so they would have had to migrate into it. For basically worldbuilding reasons (e.g. having certain mythologies preserve a deep cultural memory of this migration and how one initial population gave rise to the myriad of current ethnic groups), I want to limit the number of migration events/number of initial hunter-gatherer populations everyone else eventually descends from. This maybe, but not necessarily, suggests that all the current language families descend from just a handful of ancestral macrofamilies.
Now, a lot of my languages (from different families) have enough morphological similarity to suggest... some kind of connection, but if a phylogenetic one, it's not at all clear which ones are supposed to be grouped together, and it's slowly driving me insane. There don't seem to be two that are obviously more related to each other than to the others.
To take 4 example languages from 4 different families, which I'll just call A, B, C and D:
- *-k derives adjectives in B, genitives in C, and some sort of oblique marker in D that could conceivably come from a genitive. It has no discernable analog in A though.
- *-(V)S (probably *-ʃ) reflexes as an productive accusative + non-productive nominative in A, but a productive masculine singular nominative in B, productive masculine nominalizer in C, and some sort of core argument marker in D
- *-Vr in C is an ablative that became an ergative-genitive, an agentive marker in B, and a participle marker in A that might derive from one of those other functions. D has no obvious analog.
- Instead, *-(a)r shows up as a plural marker in D, and sort of in A too
- C and D both have *-om as a core patientive argument marker, plus the related *-(V)n-om where I haven't yet quite decided what the *-n- does
- A, B, and D all have a 2nd person pronoun derived from *-ɣ-
- A and B both have 1st person verb subject marker derived from *-(V)S
- A and B both have a 3rd person verb subject marker derived from *-o/-a
- B has a bunch of obligatory verb infixes, some of which like *-ɣ- and *-β- match up with verbalizer suffixes in A, others of which like *-l- match up with verbalizers in D
- B and C both use *-l- in a benefactive or related oblique case; A has *-l- show up genitives/adjectives and it's uncertain whether this is related
- A, B and D all seem to derive an oblique case from *-ʕə or something like it
- A has a dative -t and D has a locative -ta, perhaps both derived from an allative
- A and D both have a definite marker derivable from *-jə-
- B, C and D all mark gender from something derived from *-jə-, but D marks the opposite gender with it that B/C do
etc. There are plenty of parallels, but seemingly no consistent grouping to be made.
Grouping them all under the umbrella of one macrofamily seems to either require 1) way more convergent evolution than seems likely, or 2) case markers - not like, the rules of role marking, but the actual morphemes themselves - getting constantly loaned between different families, which I thought was highly unnaturalistic.
On the other hand, not grouping under the same macrofamily makes it seem really suspicious why all these families seem to share the same morphemes - something I thought even Sprachbunds can't accomplish, only converging on the same structure and not literally the same phonetic value for analogous morphemes.
So... what does that leave? How do I explain a bunch of families being too similar to not be related, but also not consistently similar enough to be related?