grammar - Attributive form in Japanese narration

I have recently started reading novels as a means to improve my Japanese after reaching a comfortable enough level to do so.

I am familiar with the use of the non-past form (dictionary form) in Japanese narration to convey a sense of the present (the action occurs at the time of utterance) in which case it feels closer to the English use of the past tense than the present progressive, in the sense that the action is complete.

My question is about the interpretation of when these verbs appear before nouns (attributive verbs). More specifically when the verb is an action rather than a state or emotion, for example :

俺の肩を掴む手をそっとはずす.

差し込む日差しは春のように柔らかで...

倒れている俺と剣を構える兵士の間に...

泣きじゃくる少女の背中を優しくさする.

Now from the context as I remember it, all of these should have been ongoing actions (or states resulting from an action) rather than actions that occurred at the time of utterance (I say that because I remember clearly the action starting a couple of lines earlier in most of these cases), so it should have been something like 掴んでいる手... 剣を構えている兵士... 泣きじゃくっている少女... etc

Another way of interpreting them (especially if they come up in a conversation rather than narration) is as habits, which is clearly not the case here. Can you please tell me how I should subtly understand this pattern? I am reasoning based on a conclusion I reached that apart from verbs expressing states or internal emotions the dictionary form never expresses something that's ongoing right now.

Thanks in advance!

Answer

It sounds like you already have the correct understanding, and you're just looking for confirmation/better motivation to believe what you suspect to be true.

For attributive verbs, the present progressive (～している) is considered informal, and is therefore forced to be recast as the plain form (～する). So you're essentially correct -- in a colloquial/informal setting, your example sentences would likely be phrased as 掴んでいる手、差し込んでいる日差し、構えている兵士、and 泣きじゃくっている少女。

If you're looking for a theory-intense approach, you can view it as a context-driven merger of attributive verb forms, where both underlying ～する and ～している surface as ～する in formal contexts. Accordingly, the habitual interpretation is only attainable when the underlying form is also plain, and the instantiating interpretation when the underlying form is progressive.