Fix 874 entries: N before velar K/G should be NG

[ptarjan]

Summary

In English, /n/ always assimilates to the velar nasal [ŋ] before /k/ or /g/. The CMU dictionary already uses NG correctly for most entries (e.g., THINK → TH IH1 NG K), but 874 entries still had plain N before K or G.

This is a systematic phonological rule with no exceptions in English — the N phoneme cannot surface before velar consonants without assimilating.

Examples

Word	Before	After
AANCOR	AA1 N K AO2 R	AA1 NG K AO2 R
ANGLOPHILE	AE1 N G L AH0 F AY2 L	AE1 NG G L AH0 F AY2 L
ANCHETA	AA0 N K EH1 T AH0	AA0 NG K EH1 T AH0
WINGATE	W IH1 N G EY0 T	W IH1 NG G EY0 T

Note that many entries were already correct (e.g., ANCHOR → AE1 NG K ER0, THINK → TH IH1 NG K). This PR fixes the remaining inconsistencies.

Methodology

Scanned all entries for sequences where the N phoneme is immediately followed by K or G, and replaced N with NG in those positions. The change is purely mechanical — no judgment calls required, since velar nasal assimilation before velar stops is exceptionless in English.

Reference

Ladefoged & Johnson (2014), A Course in Phonetics: /n/ assimilates to [ŋ] before velar consonants /k, g/ in all English dialects.

🤖 Generated with Claude Code

[jsalsman]

According to the official CMUDICT home page, the GitHub source for the CMUDICT file should be at https://github.com/Alexir/CMUdict -- are these (very good and much appreciated) changes being submitted there?

Please see #53