Identifying function words

[MarcoP]

whether a word, clitic, shorthand or otherwise, the portion of the script that represents "and" is presumably to be found in between labels of homogenous objects

It would be interesting to make some experiments along those lines, but how do we define "homogeneity"?

Let's say we want to investigate if common prefixes correspond to "and".
For each prefix p, we could considers words W1,W2 that occur before the prefix and attached to the prefix: W1.pW2
Maybe we should only consider W2 if it is a valid word (that occurs without the p- prefix).

We could then measure, say, the average Levenshtein distance between all W1 / W2
Doing this for several prefixes p1,p2,p3,etc would allow us to see if some generate a significantly lower average distance.

A problem I see is the massive presence of quasi-reduplication.
okchey.okedy.qokchedy.chedy.
pchedar.opchedy.qokedy.opchedy.
ysheod.sheo.sheody.qokeody.qoky.chees.

These look homogeneous:
okedy q-okchedy
opchedy q-okedy
sheody q-okeody
But couldn't quasi-reduplication be different from conjunction?