#43 III
In the previous beanbag, we saw the contrast between Speech and Earth narration with regard to verb tense but noted there are some outliers.
The Space section 90 has an unusually high proportion of PAST tense verbs in its narration. Admittedly this is a short section (and has no speech at all) but of note is that the section consists primarily of summary rather than events unfolding: “The spin-up went as planned.” “Rocky selected...” “The g-forces ... were rough...” “Rocky delivered on the testing apparatus...”
The Earth section 4 has a low proportion of PAST tense verbs but this is the section with Dr. Petrova’s email which has been categorized as narration for these counts. The email contains a high number of PRESENT tense verbs: “My name is...” “I work at...” “I am...” “I have...” We should probably not treat this as narration for this (and other counts) nor as speech (given email is a distinct register from speech).
The Earth section 24 also has a low proportion of PAST tense verbs in its narration. This is because a lot of facts, rather than events, are presented: “CO₂ spectral emissions are...” “Astrophage are...” “Astrophage is...” “Its wavelength defines...” ”... is functionally nonexistent” “That’s why...” “...still absorbs light...”
The Space section 106 is unusual in having 100% PAST tense verbs in its direct speech (no other Space speech has more than 50%). But that’s just because other than saying “Hm” twice, Grace just rhetorically asks “Which one of you did this?”
The Earth section 23 is unusual in having 100% PRESENT tense verbs in its direct speech (no other Earth speech has more than 50%) but again, there is just very little speech here with only two verbs.
In short:
- the shifts in tense are not errors by the writer
- even in present-tense text, summary can be found in the past tense
- even in past-tense text, facts can be found in the present tense
- email should not be lumped with narrative (this may affect some other counts too, especially to do with attributing things to the narrator)
- variation is potentially less significant in shorter passages
The counts rely on the part-of-speech tagging performed by the NLP library, spaCy and some manual correction is still needed.
PAST here means the VBD
tag and PRESENT means the VBP
and VBZ
tags.