NYSIIS Algorithm Extraction

Terminology. A phoneme is a contrastive unit of sound; a phonetic algorithm maps spelling to an approximation of pronunciation so that sound-alike strings collide. Place of articulation (where the vocal tract is constricted), manner of articulation (how airflow is shaped), and voicing (whether the vocal folds vibrate) are the dimensions that distinguish consonants. See ../GLOSSARY.md for fuller definitions.

Algorithm Overview

NYSIIS (New York State Identification and Intelligence System) was published by Robert L. Taft for the New York State Identification and Intelligence System in 1970. It was designed for matching American names, particularly for law enforcement and social services applications. Each extracted rule carries a weight in [0, 1] expressing the residual edit cost of treating the two spellings as equivalent.

Key Features

1. Designed specifically for American name patterns
2. Retains more vowel information than Soundex
3. Special handling of common American name prefixes
4. Focus on phonetic accuracy over code brevity

Algorithm Steps

1. Translate first characters of name:
   • MAC → MCC
   • KN → N
   • K → C
   • PH, PF → FF
   • SCH → SSS
2. Translate last characters:
   • EE, IE → Y
   • DT, RT, RD, NT, ND → D
3. Translate remaining characters with context rules
4. Remove trailing S and A
5. Remove trailing AY, replace with Y
6. Collapse duplicate characters
7. Optionally truncate to fixed length

Extracted Rules

Initial Prefix Transformations

[id: 2800, name: "mac to mcc nysiis", weight: 0.10, group: nysiis_prefix]
mac -> mcc / #_;  // MacDonald → McCDonald

[id: 2801, name: "mc to mcc nysiis", weight: 0.10, group: nysiis_prefix]
mc -> mcc / #_;  // McDonald → McCDonald

[id: 2802, name: "kn to n nysiis", weight: 0.10, group: nysiis_prefix]
kn -> n / #_;  // Knight → Night

[id: 2803, name: "k to c nysiis", weight: 0.10, group: nysiis_prefix]
k -> c / #_;  // Katz → Catz

[id: 2804, name: "pf to ff nysiis", weight: 0.10, group: nysiis_prefix]
pf -> ff / #_;  // Pfeiffer → FFeiffer

[id: 2805, name: "ph to ff nysiis", weight: 0.10, group: nysiis_prefix]
ph -> ff / #_;  // Phelps → FFelps

[id: 2806, name: "sch to sss nysiis", weight: 0.10, group: nysiis_prefix]
sch -> sss / #_;  // Schmidt → SSShmidt

Final Suffix Transformations

[id: 2810, name: "ee final to y nysiis", weight: 0.10, group: nysiis_suffix]
ee -> y / _#;  // McGee → McGy

[id: 2811, name: "ie final to y nysiis", weight: 0.10, group: nysiis_suffix]
ie -> y / _#;  // Christie → Christy

[id: 2812, name: "dt final to d nysiis", weight: 0.10, group: nysiis_suffix]
dt -> d / _#;  // Schmidt → Schmid

[id: 2813, name: "rt final to d nysiis", weight: 0.10, group: nysiis_suffix]
rt -> d / _#;  // Hart → Hard

[id: 2814, name: "rd final to d nysiis", weight: 0.10, group: nysiis_suffix]
rd -> d / _#;  // Howard → Howad

[id: 2815, name: "nt final to d nysiis", weight: 0.10, group: nysiis_suffix]
nt -> d / _#;  // Grant → Grand

[id: 2816, name: "nd final to d nysiis", weight: 0.10, group: nysiis_suffix]
nd -> d / _#;  // Roland → Rolad

EV to AF Transformation

[id: 2820, name: "ev to af nysiis", weight: 0.15, group: nysiis_vowel]
ev -> af;  // Steven → Stafan

K/Q Transformations

[id: 2825, name: "k to c nysiis internal", weight: 0.10, group: nysiis_k]
k -> c;  // Uniform K → C

[id: 2826, name: "q to g nysiis", weight: 0.15, group: nysiis_q]
q -> g;  // Quentin → Guentin

SCH Internal

[id: 2830, name: "sch to s nysiis", weight: 0.15, group: nysiis_sch]
sch -> s;  // Internal sch simplification

PH Handling

[id: 2835, name: "ph to f nysiis", weight: 0.10, group: nysiis_ph]
ph -> f;  // Phonetic standard

H Handling

[id: 2840, name: "h after vowel nysiis", weight: 0.15, group: nysiis_h]
h ->  / [aeiou]_;  // Drop H after vowel

[id: 2841, name: "h before consonant nysiis", weight: 0.15, group: nysiis_h]
h ->  / _[^aeiou];  // Drop H before consonant

W Handling

[id: 2845, name: "w after vowel nysiis", weight: 0.15, group: nysiis_w]
w ->  / [aeiou]_;  // Vowel absorbs W

AW to A

[id: 2850, name: "aw to a nysiis", weight: 0.10, group: nysiis_aw]
aw -> a;  // Hawkins → Hakins

M/N Before Consonant

[id: 2855, name: "m to n before consonant nysiis", weight: 0.15, group: nysiis_mn]
m -> n / _[^aeiou];  // Thompson → Thonpson

Vowel Transformations

[id: 2860, name: "a to vowel nysiis", weight: 0.20, group: nysiis_vowel]
a -> a;  // Keep A

[id: 2861, name: "e to vowel nysiis", weight: 0.20, group: nysiis_vowel]
e -> a;  // E → A

[id: 2862, name: "i to vowel nysiis", weight: 0.20, group: nysiis_vowel]
i -> a;  // I → A

[id: 2863, name: "o to vowel nysiis", weight: 0.20, group: nysiis_vowel]
o -> a;  // O → A

[id: 2864, name: "u to vowel nysiis", weight: 0.20, group: nysiis_vowel]
u -> a;  // U → A

Double Letter Simplification

[id: 2870, name: "double letter nysiis", weight: 0.10, group: nysiis_double]
([a-z])\1 -> \1;  // Any doubled letter → single

SH/SCH to S

[id: 2875, name: "sh to s nysiis", weight: 0.10, group: nysiis_sh]
sh -> s;  // Simplified

[id: 2876, name: "sch to s internal nysiis", weight: 0.10, group: nysiis_sh]
sch -> s;  // Internal position

Z to S

[id: 2880, name: "z to s nysiis", weight: 0.10, group: nysiis_z]
z -> s;  // Uniform sibilant

GHT to GT

[id: 2885, name: "ght to gt nysiis", weight: 0.10, group: nysiis_ght]
ght -> gt;  // Knight → Knigt

DG to G

[id: 2890, name: "dg to g nysiis", weight: 0.10, group: nysiis_dg]
dg -> g;  // Badge → Bage

Common American Name Patterns

// -son endings
[id: 2895, name: "son ending nysiis", weight: 0.05, group: nysiis_patronymic]
son -> san / _#;

// -sen endings (Scandinavian)
[id: 2896, name: "sen ending nysiis", weight: 0.10, group: nysiis_patronymic]
sen -> san / _#;

// -stein endings (Germanic)
[id: 2897, name: "stein ending nysiis", weight: 0.10, group: nysiis_patronymic]
stein -> stan / _#;

// -ski/-sky endings (Slavic)
[id: 2898, name: "ski ending nysiis", weight: 0.10, group: nysiis_patronymic]
ski -> scy / _#;

[id: 2899, name: "sky ending nysiis", weight: 0.10, group: nysiis_patronymic]
sky -> scy / _#;

Comparison with Soundex

Feature	Soundex	NYSIIS
Output Length	4 (fixed)	Variable (typically 6)
First Character	Retained	Encoded
Vowel Handling	Removed	Normalized to 'A'
Prefix Rules	None	MAC, KN, PH, etc.
Suffix Rules	None	EE→Y, DT→D, etc.
Target Population	General English	American names

Use Cases

NYSIIS is particularly effective for:

• Law enforcement databases
• Social services record matching
• Immigration record matching
• Hospital patient matching
• American genealogical research

References

1. Taft, Robert L. (1970). "Name Search Techniques". New York State Identification and Intelligence System, Special Report No. 1, Albany, NY. (No DOI; canonical source of the NYSIIS algorithm.)

---

← Documentation Index