This is an advanced topic in Common syntax. Although Common has free phrase order, and often fronts elements to make them more prominent, it does have a clear preference - SVO in main clauses, unusual for an ergative language. However, Common's syntax is clearly accusative (see attached article on Ergativity and Syntactic Pivot in Common), and one consequence is that the case of the subject can be either ergative or absolutive, with the choice being governed by the declared valence of the verb. The subject of transitive verbs is the ergative argument. However, there is a small class of verbs, called 'reverse sense' (rokéulysno pufas) verbs, that buck this trend.

It can sometimes be unclear whether the concept of subject really and truly has a place in Common, because it can be difficult to apply common syntactic tests for a subject to Common. Common does not have non-finite verbs which are involved in a lot of syntactic tests for subjects, it is generally impossible to omit relativisers, and subjects manifestly do not share a single grammatical case.

Word order gives some evidence. Common reorders phrasal elements of a sentence from the preferred SVO order at will for emphasis, although in dependent clauses, the verb overwhelmingly goes at the end of the clause (dependent clauses are SOV). Subject ellipsis as we see with syntactic pivots provides one of the best clues, indeed, this is one of the main pieces of evidence supporting the fact that Common is syntactically accusative.

Common has a small class of 'reverse sense' transitive (skurun) verbs that are perceived by speakers as having a 'backwards' argument structure. Take the skurun verb 'pex', know. Its argument structure is the reverse of what you would expect in English - the knower is in the absolutive case, and the thing known is in the ergative case, almost as though information has agency and forces people to know things. Here is an example of how that works:

We te pex xi zu se ikky hitaj.
I(ABS) hit(NP.IM) know RELV.ERG you(ABS) stand(NP.IM) not sleep.
'I know that you're not sleeping.'

The clause 'xi zu se ikky hitaj', 'that you're not sleeping', is governed by the relativiser 'xi', which is in the ergative case. In Common's normal, preferred word order, this sentence would be 'Xi zu se hitaj te pex we', and indeed you could say this, but Common speakers will overwhelmingly prefer the other order. This will even be seen if the thing known isn't a dependent clause:

We te pex ja posa.
I(ABS) hit(NP.IM) know the(ERG) answer.
'I know the answer.'

There is a solid argument that in these reverse sense verbs, the absolutive argument is actually the subject, rather than the ergative argument. This can be seen in coordinated propositions with the putative syntactic pivot bolded:

1. We(ABS) te pex ja posa.
2. Je(ERG) hyp zisse iju.
We te pex ja posa hanja ___ hyp zisse iju.
I(ABS) hit(NP.IM) know the(ERG) answer and ___ give(NP.IM.IR) give you(DAT)
'I know the answer and would tell you.'

Although it can be hard to discern amidst the noise of Common's generally pro-drop syntax, Common speakers will think of the omitted argument in the second proposition as referring back to the absolutive argument of the first proposition for these types of structures. Reverse sense verbs bring a touch of ergativity to Common's accusative syntax, where the absolutive argument is acting as the pivot. Taking another example, let's use the reverse sense verb 'suka', to like, which has an absolutive liker and an ergative thing liked.

1. A Can te suka ja Mari.
'John likes Mary'
2. Ja Can te matta a spe'n.
'John is helping her'

If the hypothesis that the subject of the reverse sense verb is the absolutive argument is true, then the following coordinated construction will work:

A Can te suka ja Mari hanja ___ te matta a spe'n.

This form does work, Common speakers will tend to associate the missing term with John, not Mary.

Where this can be interesting is if the reverse sense verb is in the second coordinated clause. In this case, the antipassive can actually see some rare use in helping to facilitate an S/O pivot.

1. Ja Mari te matta a Can.
'Mary is helping John'
2. A Can te suka ja spe'n.
'John likes her'

The coordinated sentence does not work with the second subject omitted:

*Ja Mari te matta a Can hanja ___ te suka ja spe'n.

This form gives the undesired impression to Common speakers that Mary likes John instead of the other way around. But if the second sentence is reframed with the antipassive:

2. A Mari se suka na Can.
The(ABS) Mary stand(NP.IM) please ∅ the(NOM) John.

This mean literally 'John likes Mary', but is read more literally something like 'Mary is pleasing of John'. Now the coordinated form can be expressed like:

Ja Mari te matta a Can hanja ___ se suka na spe'n.
Mary helps John and ___ pleases of him.

This is hard to express properly in English, but in Common, it works - a Common speaker understands that the omitted argument refers to Mary and that John likes Mary.