Lowercase after colons in titles; restore MartEtal18; fix 13 entries

[jeremymanning]

Post-merge audit of #84 and #85, which were merged to master as 0d3b9ff while this
review was in progress. This PR is the result of that review: the colon rule applied
across the file, one entry restored that a merge dropped, and thirteen entries corrected
against their publishers' own records.

1. Colon rule — 186 titles

The word after a : in a Title is now lowercase unless it is a proper noun. Proper
nouns in this file are brace-protected ({C}alifornia, {M}arkov) or listed in
bibcheck/caps.txt; both are left alone, as is Gede24's : I ... (a pronoun).

184 of the 186 were the article "A", and bibcheck could never have flagged them:

• format_title short-circuits on w.lower() == "a" and appends the word unchanged
  (bibcheck/helpers.py:854), and
• A is line 1 of bibcheck/caps.txt.

So a capital A after a colon was preserved. It is only ever preserved, never
restored — verified directly:

format_title('Vision: A computational investigation') -> 'Vision: A computational investigation'
format_title('Vision: a computational investigation') -> 'Vision: a computational investigation'

That is why this change is stable under bibcheck and needs no change to bibcheck itself.

Only two of the 186 were not "A": BeckLim03 ("Strategic") and ContPrev24 ("Suicide").

Journal, booktitle, publisher and address fields are untouched — they go through
format_journal_name, not format_title, so Journal of Experimental Psychology: Learning, Memory, and Cognition stays correct.

2. Restored: MartEtal18

Dropped with no replacement by 4376a00 ("remove duplicate FitzEtal, re-add DOIs for
most missing") and not mentioned in #84's description. Restored, with the surname
corrected from Martinez-Rodrigues to Martinez-Rodriguez and the DOI added
(CrossRef 10.1016/j.eswa.2018.07.017).

3. Corrections

Each verified against CrossRef, DataCite, arXiv, OpenLibrary or the publisher directly.

Entry	Was	Now	Source
ChanEtal09b	title stopped at the main title	…eyewitness suggestibility: the reversed testing effect	CrossRef subtitle, 10.1111/j.1467-9280.2008.02245.x
Mars07	title stopped at the main title	…is not the same as recalling: implications for memory	CrossRef subtitle, 10.1111/j.1467-8721.2007.00467.x
NguyEtal19	…shared neural response	…shared neural responses	CrossRef 10.1016/j.neuroimage.2018.09.010
Schr00b	Pages = {96}	96--106	CrossRef 10.1037/0022-0663.92.1.96
XuEtal24a	no issue, no DOI	Number = {2}, DOI added	CrossRef 10.1038/s41928-023-01116-6
Vite67	A Viterbi	A J Viterbi	OpenAlex — Andrew J. Viterbi
VeraEtal25	two authors read Y Chen	2nd is Y-T Chen	arXiv 2509.20354, author 69 = Yi-Ting Chen
BarrEtal13	Doi = {https://doi.org/10.1016/…}	bare DOI	consistency with the other 64
MatuEtal17	Doi = {https://doi.org/10.1016/…}	bare DOI	consistency with the other 64
Fodo75	Harvard University Press, Cambridge MA	Crowell, New York NY	OpenLibrary — the 1975 edition is Crowell; HUP issued it in 1979/80
BeckLim03	whole title inside an extra brace pair	unbraced; {C}alifornia; + Number = {6}; + Doi	CrossRef 10.1162/089892903322370744
LairEtal87	no DOI	10.1016/0004-3702(87)90050-6	CrossRef record matches every field
Pyly73	no DOI	10.1037/h0034650	CrossRef record matches every field

BeckLim03 is worth calling out: the extra brace pair around the entire title makes
format_title pass the entry through untouched, which is how {c}alifornia survived.

SnijBosk11 → SnijBosk12, @incollection → @inbook

"More powerful tests for variance parameters" is a section inside chapter 6; chapter 6
is titled Testing and Model Specification. Snijders also states the official publication
year of the second edition is 2012 (it appeared November 2011) —
mlbook.htm,
table of contents.

Recast to match the repo's existing @inbook convention, where Chapter holds the
chapter's name (cf. Bord08, HallGree08). The key change is forced rather than
chosen: bibcheck derives the key from author + year. Pages dropped — the section's
range in the second edition could not be verified. R Bosker → R J Bosker.

4. doi → Doi

All 66 doi field names capitalized. helpers.write_bib emits k.capitalize()
(bibcheck/helpers.py:729), so every other field in the file is already capitalized;
these 66 were the only exceptions.

No citation key was re-pointed

Checked mechanically across all 6,301 keys shared with the pre-#84/#85 master: for every
key present in both, the title and author set still denote the same work. Every rename
from those two PRs was confirmed to carry its referent with it —
ChanEtal09→ChanEtal09a, ScheEtal08→ScheEtal08a, XuEtal24→XuEtal24a,
WangEtal10→WangEtal10a, Mann23a→Mann23, Mann23b→Mann24,
OwenMann23→OwenMann24, XuEtal23→XuEtal24b, FitzEtal23→FitzEtal26,
MontEtal12→MontEtal12a, RumeEtal86→RumeEtal86a, Krus98→Krus92,
ViswEtal17→VaswEtal17 — and no retired key was reused for a different work.

Checks

• bibcheck/test.py equivalent: 0 errors, 6,422 entries
• colon audit: 1 remaining hit, Gede24's : I (a pronoun — correct)
• structural: 0 non-uniform lines, 0 lowercase field names, every entry blank-line separated
• duplicate audit (suffix-aware keys, author+title, fuzzy titles): clean
• 157 new/changed entries re-verified against CrossRef after the edits

Also in this PR

Vint68 now cites the Russian original. The entry carried Springer's English-translation
record — Cybernetics 4(1):52–57, DOI 10.1007/BF01074755 — under the original's 1968 date,
so no source matched it as written and the DOI resolved to a record whose year contradicted it:

Source	Year	Journal	Vol/iss	Pages
CrossRef / OpenAlex (Springer's own deposit)	1972	Cybernetics	4(1)	52–57
Semantic Scholar	1968	Cybernetics	4	52–57
Wikipedia	1968	Kibernetika	4	81–88
arXiv 1001.2267	1968	Kibernetika	4(2)	81–88

Now Kibernetika 4:81–88, 1968 — what the speech-recognition literature cites, and it keeps
the key. The DOI is dropped (it belongs to the translation). Number is omitted deliberately:
Wikipedia gives no issue, arXiv 1001.2267 gives 4(2), so it cannot be asserted.

CrutEtal13 stays deleted, confirmed by @jeremymanning: it was the conference presentation
that became HealEtal14a, and HealEtal14a is the one that should be cited. Not touched here.

🤖 Generated with Claude Code

https://claude.ai/code/session_01MfEdd5PnqDbDppiYGnNNcg