Linguistics

Phonemes and graphemes -
the sounds and spellings of English

September 2021

What do all these words have in common?

away

they

sleigh

cafe

matinee

cabaret

Each word rhymes by ending in the same sound, or phoneme, called AY. In this case, the phoneme is a vowel more commonly known as "long a".

aw-ay

th-ey

sl-eigh

caf-e

matin-ee

cabar-et

However, the groups of letters that produce the sound AY are all different. A single letter or group of letters that produces a phoneme is called a grapheme. In the examples, the graphemes are ay, ey, eigh, e, ee, and et.

Now, what do these words have in common?

cat

cello

ceiling

oceanic

accountancy

Each word contains the same grapheme, c. This single-letter grapheme, however, is responsible for different sounds, or phonemes, in each word.

c-at

c-ello

c-eiling

o-c-eanic

accountan-c-y

Each highlighted c corresponds to a different phoneme (K, CH, S, SH, T:S). In this case, the phonemes in blue are consonants. And the purple phoneme,T:S, is a consonant diphoneme, or merger of 2 phonemes.

In the previous examples, we saw a phoneme can map to multiple graphemes:

AY → ay, ey, eigh, e, ee, et

Vice versa, we saw a grapheme can map to multiple phonemes:

c → K, CH, S, SH, T:S

But where does this knowledge come from? How many phonemes are there? How many graphemes? And how many mappings are there between graphemes and phonemes?

To answer these questions, I looked at ~35K English words from the CMU Pronouncing Dictionary. Its data helps break words into their graphemes and phonemes, like the word box:

Graphemes

B

|

O

|

X

Phonemes

B

|

AA

|

K:S

Silent letters, like the c in czar, become part of the nearest sound-producing grapheme.

Graphemes

CZ

|

A

|

R

Phonemes

Z

|

AA

|

R

39 Phonemes

CMU modified ARPAbet's alphabet of sounds, resulting in 39 phonemes: 15 vowels & 24 consonants. It also has 21 diphonemes (2 phonemes merged).

All

NaN

AH

21,686

N

16,694

IH

15,992

T

15,950

S

13,095

L

12,966

R

12,664

K

10,597

IY

9,388

D

8,904

ER

7,750

M

7,385

EH

7,091

P

7,072

AE

5,944

B

5,130

AA

4,846

EY

4,434

F

4,134

OW

3,486

AY

3,460

NG

3,224

V

3,149

SH

3,080

AO

2,880

G

2,784

Z

2,428

W

2,036

JH

2,020

UW

1,829

HH

1,751

AH:L

1,308

CH

1,188

AW

888

TH

849

Y:UW

673

K:S

666

UH

528

Y

385

OY

323

Y:AH

290

DH

243

AH:M

231

ZH

193

HH:W

104

Y:UH

86

G:Z

77

AH:W

29

W:AH

21

K:SH

19

W:AA

11

T:S

7

W:EY

4

NG:K

3

Y:EH

3

G:ZH

3

Y:AE

2

Y:IH

2

AH:R

1

T:TH

1

The numbers show how many times a phoneme appears in the 35K CMU words. Some phonemes appear twice or more in the same word, like the N in banana.

275 Graphemes

From the mappings in the ~35K English words, there are 275 graphemes: 33 vowel graphemes, 101 consonant graphemes and 141 mixed.

All

NaN

I

19,637

A

18,757

N

16,101

T

14,235

E

14,051

O

12,537

R

11,973

L

11,289

S

10,817

C

8,315

D

7,580

M

7,001

P

6,508

U

6,407

Y

5,232

ER

4,923

B

4,919

G

3,446

F

3,175

NG

2,638

V

2,374

H

1,727

LL

1,616

TI

1,529

K

1,427

W

1,390

OU

1,368

LE

1,327

SS

1,304

EA

1,164

TE

1,155

ED

1,153

TH

1,068

CH

1,033

SH

1,014

OR

896

CK

771

VE

764

CE

762

X

726

OO

716

EE

641

AI

622

SE

548

TT

542

NE

525

UR

520

OW

505

PH

503

AR

462

RE

452

J

430

Z

428

GE

411

PP

408

Q

383

FF

358

AU

354

RR

348

DE

331

AY

315

NN

303

IE

302

MM

297

WH

290

ZE

271

IR

244

OA

243

OI

224

ME

223

GG

203

IGH

196

KE

190

DD

189

EY

180

AW

174

CI

166

BB

162

NT

150

SI

146

TCH

146

CA

143

EW

140

PE

135

SC

133

CC

124

EI

118

URE

113

OY

112

SSI

98

EU

93

WR

87

URR

86

UE

83

KN

74

GN

72

ST

71

DGE

64

GU

61

UI

57

EAR

57

ZZ

56

AE

55

OE

52

OUR

51

OUGH

51

DG

47

FE

46

BE

46

GH

45

LK

45

EIGH

44

MB

42

PS

40

TTE

39

GI

38

QUE

36

ORR

36

XC

34

RH

34

ERR

33

GUE

32

CHE

30

THE

29

QU

28

LM

26

EYE

24

AH

23

ET

22

AUGH

21

SCH

21

HO

21

DJ

20

ARR

20

SHI

20

EUR

19

CQ

17

ND

17

IA

17

BT

17

CT

16

MN

16

EAU

16

AA

14

ON

14

PT

14

YE

14

YR

14

HI

13

LLE

13

SCI

13

FT

12

UO

11

SSE

11

EO

11

PA

10

KH

10

XI

10

NNE

10

HL

10

FFE

10

SW

9

HER

9

CCH

7

LF

7

HAU

7

OH

7

SCE

6

AUR

6

IS

6

ORRH

6

PPE

6

HOU

6

HU

6

AIGH

6

NM

5

AWE

5

IRR

5

UY

5

OT

5

LD

5

MME

5

EWE

5

PN

5

SLE

4

LVE

4

DI

4

AO

4

TW

4

CHT

4

IEU

3

ERE

3

STH

3

ULL

3

EAUX

3

RRE

3

RPS

3

GM

3

HA

3

NGUE

3

HEI

3

SL

3

IU

2

AER

2

ARRH

2

XE

2

DA

2

OIS

2

UGH

2

UOY

2

UA

2

OL

2

MBE

2

OUP

2

PB

2

OAR

2

AGH

2

CZ

2

UT

2

IL

2

DH

2

VV

2

EH

2

IT

2

OWE

2

GL

2

IO

2

NGE

2

TG

2

TS

2

OS

1

AWR

1

AYE

1

AT

1

AS

1

BH

1

IN

1

AIS

1

GNE

1

IRRH

1

OLO

1

ZI

1

OUE

1

NGH

1

ALL

1

ELL

1

XS

1

SZ

1

LV

1

UH

1

JH

1

UER

1

JU

1

KK

1

ES

1

URRE

1

YRRH

1

EOU

1

IER

1

AIT

1

PF

1

NK

1

OOH

1

RRH

1

UEUE

1

RT

1

EZ

1

OUS

1

OUX

1

PPH

1

PHO

1

BP

1

AL

1

LLA

1

EAH

1

The numbers show how many times a grapheme appears in the corpus. Some graphemes also appear twice or more in the same word, again like the N in banana.

Phonemes ↔ Graphemes

The graphic on the following page lets you dive more into all the unexpected mappings between sounds and letters in English.

You can discover oddities like the phoneme P mapping to the bizzare grapheme GH in hiccough. The OED labeled this folk etymology relic "a mere error," yet its use continues.

Following this intro to linguistics, I'll make a couple more articles diving into alternate spellings and the poetic features of words. Lastly, if you're interested in the data, check out the accompanying Github repo.

Phonemes ↔ Graphemes

Phoneme

Count

Graphemes

AH	21,686
N	16,694
IH	15,992
T	15,950
S	13,095
L	12,966
R	12,664
K	10,597
IY	9,388
D	8,904
ER	7,750
M	7,385
EH	7,091
P	7,072
AE	5,944
B	5,130
AA	4,846
EY	4,434
F	4,134
OW	3,486
AY	3,460
NG	3,224
V	3,149
SH	3,080
AO	2,880
G	2,784
Z	2,428
W	2,036
JH	2,020
UW	1,829
HH	1,751
AH:L	1,308
CH	1,188
AW	888
TH	849
Y:UW	673
K:S	666
UH	528
Y	385
OY	323
Y:AH	290
DH	243
AH:M	231
ZH	193
HH:W	104
Y:UH	86
G:Z	77
AH:W	29
W:AH	21
K:SH	19
W:AA	11
T:S	7
W:EY	4
NG:K	3
Y:EH	3
G:ZH	3
Y:AE	2
Y:IH	2
AH:R	1
T:TH	1

PHONEME

GRAPHEME

# Occurrences in 35K Word Corpus

15K

10K

5K

1K

Vowel

Consonant

Diphoneme

NaN%

FREQUENCY

COUNT