Introduction
Literature Review
Theoretical Preliminaries
Theory and Review
Endnotes

Experiment 1
Experiment 2
Experiment 3
Experiment 4
Experiment 5
Experiment 6
Experiment 7
Experiment 8
Experiment 9
Experiment 10
Experiment 11
Experiment 12
Experiment 13
Experiment 14

---

4. Phonosemantic Experiments

4.1 Experiment 1 -- Classification First by Phoneme Sequence and then by Semantic Domain

See Appendix I for full data and results.

4.1.1 Methodology

· All the monosyllables familiar to me were extracted from Houghton Mifflin's American Heritage Dictionary.

· These words were divided into 24 classes based on the consonantal phonemes that they were composed of. The consonants in my dialect of American English are /b/, /d/, /g/, /p/, /t/, /k/, /v/, /D/ (as in bathe), /z/, /Z/ (as in beige), /f/, /T/ (as in bath), /s/, /S/ (as in wash), /h/, /J/ (as in jump), /C/ (as in cheese), /m/, /n/, /G/ (as in hang), /r/, /l/, /w/, /j/. (Nonstandard (non-IPA) symbols are employed due to technical limitations.)

· An attempt was made to find a Phonosemantic Classification for each of these subclasses.

· The words within each of these resulting phonesthemes were then subdivided again according to position in the syllable. The following positions were identified:

1. initial position.

2. second position

3. third position

F1. pre-pre-final position

F2. pre-final position

F3. final position

· 114 of the 3485 monosyllables (or 3%) did not fall easily into a Phonosemantic Classification. These were placed in a different Natural Classification. All of these exceptional words fell into one of the following natural semantic classes. I will refer to these as the Concrete Noun classes:

Concrete Noun Classes (Nouns with Rigid Referential Domains)
people, titles, body parts, clothing, cloth, periods of time, games, animals, plants, plant parts, food, minerals, containers, vehicles, buildings, rooms, furniture, tools, weapons, musical instruments, colors, symbols, units of measurement.

Notice that very few people disagree on what constitute the referents for a word in one of these classes. That is, people largely agree on which trees are oaks, which tools are hammers, which rooms are kitchens and so forth. This is not as true of other semantic domains. (In addition to being less ambiguous and more impervious to Clustering than other semantic domains, the Concrete Noun classes seem to be more nearly universal cross-linguistically than other semantic domains.)

---

4.1.2 Example

I endeavored to make the lists in Appendix I exhaustive.

The format of the output is as follows:

Classification Type (Phonesthemes or Concrete Nouns)

Relevant Phoneme

 

Superclass # Superclass Descriptor # words in SC, % -- words in SC/all words containing RPh

Phonestheme # Phonestheme Descriptor Position Indicator

Word List words in phonestheme, % -- words in phonestheme/all words in this superclass

_________________________

A sample entry:

A1. Consonantal Phonesthemes

/r/

A1 Walk, Run and Ride 133 13.3%

______________________________________

1. Walk, Run (No Vehicle) 1

race, raid, range, reach, rip, roam,

romp, rove, run, rush

10 7%

______________________________________

1. Walk, Run (No Vehicle) 2

break, crawl, creep, cross, cruise, drag, drift,

drop(by), frisk, prance, press, prowl, thread,

trace, track, trail, tramp, tread, trek, tromp,

troop, trot, trudge

23 6%

______________________________________

1. Walk, Run (No Vehicle) 3

scram, scream, spread, spring, sprint, stray,

streak, stream, stride, strike, stroll, strut

12 15%

______________________________________

1. Walk, Run (No Vehicle) F2

barge, charge, course, curve, dart, ford, forge,

fork, forth, hurl, march, part, storm, swarm,

swerve, warp

16 6%

______________________________________

1. Walk, Run (No Vehicle) F3

fare, near, scour, tear

4 3%

______________________________________

Explanation:

· The A1. Consonantal Phonesthemes indicates that this is the section of phonesthemically classified words

· /r/ is the relevant phoneme in this case

· A1 is the superclass number

· Walk, Run and Ride is the superclass designator

· 133 is the total number of unique words in this superclass

· 13.3% is the percentage of words in this superclass as compared to the total number of monosyllabic words containing /r/. That is, there are 1003 monosyllabic words in my vocabulary which contain /r/ and 133= .133*1003.

· 1. is the phonestheme number

· Walk, Run (No Vehicle) is the phonestheme designator

· 1, 2, 3, F2, F3 refer to the relevant phoneme's position within the syllable

· In the first phonestheme: 'race, raid, range, reach, rip, roam, romp, rove, run, rush' is, I believe, the list of all monosyllables with /r/ in initial position and which have at least one sense which refers to non-vehicular motion with a human agent.

· 11 is the number of words in the first class

· 7% is the percentage of words in the phonestheme as compared to all the monosyllables starting with /r/. There are 140 such monosyllables.

---

4.1.3 Discussion of Findings

4.1.3.1 Overview

The most important result of this experiment is, of course, that the phonology of a word affects its meaning. Furthermore, it has a much more specific effect on meaning than is generally supposed.

Much of my effort over the last years has been directed at trying to find a Phonosemantic Classificational system for each consonant for which the classes were as clear and indisputable and as obviously interrelated as possible. I do this in an effort to make the fundamental meaning underlying each phoneme very accessible, and of course in an effort to make the phonosemantic data as incontestable as I can. I have devoted myself primarily to English in part, of course, because English is my native tongue. But I also use English because there is a very common attitude -- even among those who accept linguistic iconism -- that it's not productive and therefore occurs only in obscure vocabularies of obscure languages that have undergone relatively little change over time. My findings show that iconism runs throughout the most basic vocabulary of at least one language in very broad usage... a language that has been as overwhelmed as any by foreign influences and radical and sudden diachronic changes.

This experiment provides evidence for criteria 1-6 of the Phonosemantic Classification, which as the reader will recall, constitute the criteria required to prove the Phonosemantic Hypothesis -- that all phonemes have an identifiable meaning:

Criterion 1. Very nearly every word with the given phonological characterization fits in some semantic class.

Criterion 2. Each semantic class contains a large percentage of the words which match that phonological characterization.

Criterion 3. There are relatively few semantic classes in the classification.

Criterion 4. The semantic classes in the classification are distinct

Criterion 5. Each word fits into an average of a fairly large number of classes.

Criterion 6. The semantic classes are narrowly defined. By a 'narrowly defined' semantic class, I mean one which encompasses a small percentage of words in the language as a whole.

In addition it provides strong evidence for the basic claim regarding the relationship between concrete reference and iconic meaning

The Arbitrariness of Reference
The salience of iconic meaning in a word is related inversely to the concreteness of its reference.

It provides only indirect evidence for general character of Phonosemantic Association and of Iconism:

Phonosemantic Association
When semantic domain S is associated disproportionately frequently with phoneme X, then people will be inclined to associate semantic domain S with phoneme X productively.

Iconism
The connotation of a word is affected directly by its phonological form.

In addition, this experiment:

1. provides a general idea of the semantic domains to which each phoneme is restricted and the percentages of words that fit into these semantic domains.

2. allows us to observe indirectly the effect that phoneme position has on the semantics of the word.

Appendix I sums up data compiled and analyzed over many years. Naturally, in the course of a project of such large scope, one makes many more specific observations than can be written down. Only the fundamental results are summarized here. In my more detailed discussion of this first test, I will limit myself to the following topics:

1. I will outline the major phonesthemes provided in Section A1 to give the reader an indication of the semantic domains associated with each of the English consonants. (4.1.3.2)

2. I will discuss the mechanism whereby concrete reference obscures the manifestation of phoneme meaning. (4.1.3.3)

3. I will discuss the nature of the 'senses' of the word and give a brief overview of the structure of word semantics suggested to me by the phonosemantic data. (4.1.3.4)

4. I will discuss how the position that a given consonant occupies within the syllable affects the semantics of the word. (4.1.3.5)

 

---

4.1.3.2 Semantic Domains of the Consonants

This Phonosemantic Classification shows that words containing each of the consonantal phonemes fall within the semantic domains listed below in the quantities and percentages indicated. I am not hereby suggesting that this is the 'right' phonesthemic classification. This data only provides one profile of the semantic domains to which each of the English consonants are constrained. If word-meanings were insensitive to phonological form, we would anticipate that these profiles would all be the same. But they are not. And though this remains to be proved, in the course of formulating these phonesthemes, it becomes obvious that it is impossible to make them the same.

Notice that the semantic characterizations of the consonants seem to resemble the articulation of the consonants themselves. For example, to pronounce /b/, one creates a barrier by closing together the lips; one builds up pressure behind the lips causing the cheeks to bulge slightly; and one releases the barrier to produce a sort of explosion. I think it's not coincidental that 14% of words containing /b/ can be described as barriers and interferences, 6% involve binding, 11% concern bulging and 6% exploding. /b/ is one of the most 'high pressure' or subjectively 'big' consonants. Neither /d/, /g/ nor /p/, /t/, /k/ among the stops seems to involve as much air under pressure. And /b/ also involves disproportionately many words of bigness, money and large quantities. This similarity between a phoneme's articulation and its semantic characterization as formed by Clustering is indirect evidence for True Iconism, i.e. that there is a direct, unmediated effect of form on semantic content in a word.

The percentages do not add up to 100%, of course, because for each consonant, each word fits on average in several phonesthemes. These profiles are compiled for all English consonants in all the positions in the word:

A1. Consonantal Phonesthemes

/b/

A1 Bulging, Brushy 64 11.4%

A2 Big, Much, Many 109 19.4%

B1 Barriers, Interference 76 13.5%

C1 Emptiness 35 6.2%

D1 Binding, Contact, Connection 33 5.9%

E1 Foundations, Carrying and Balance 50 8.9%

F1 Explosion, Blowing and Breaking 35 6.2%

F2 Departure 19 3.4%

F3 Hitting, Battling, Games 50 8.9%

F4 Bizarre and Chaotic 8 1.4%

G1 Noises and Music 36 6.4%

G2 Effusive Language and Writing 74 13.2%

G3 Bother and Bargain 19 3.4%

H1 Birth and Beginnings 45 8.0%

I1 Badness 31 5.5%

I2 Pain 43 7.7%

I3 Error 22 3.9%

J1 Water 68 12.1%

J2 Alcohol 16 2.9%

J3 Boats 27 4.8%

K1 Fire, Light 18 3.2%

K2 Saturated Color 20 3.6%

L1 Boards and Bricks 18 3.2%

/d/

A1 End, Death, Sleep, Drug 50 10.4%

A2 Diminishment, Smallness 83 17.3%

A3 Breadth, Dragging On 41 8.5%

A4 Scarcity, Danger 60 12.5%

A5 Confusion, Discord and Barriers 69 14.4%

A6 Dark, Dirty and Dreary 97 20.2%

B1 Divisions, Groups, Amounts 130 27.1%

C1 Execution of Pending Process 91 19.0%

C2 Motion 31 6.5%

D1 Down 154 32.1%

E1 Good, Dear 33 6.9%

F1 Water 62 12.9%

G1 Light and Color 11 2.3%

/g/

A1 The Gullet 27 7.2%

B1 Sound and Talk 33 8.8%

B2 Voice 14 3.7%

C1 Containers and Valleys 49 13.0%

C2 Quantity 50 13.3%

D1 Getting, Holding and Greed 77 20.4%

D2 Blockage 38 10.1%

E1 Giving 6 1.6%

E2 Going 55 14.6%

E3 Growing 26 6.9%

F1 Goodness 15 4.0%

G1 Light (Generally Indirect) 18 4.8%

H1 Understanding 23 6.1%

I1 Grids and Grains 12 3.2%

J1 Death and Gloom 30 8.0%

K1 Too Much Where It Doesn't Belong 112 29.7%

K2 Not Enough Where It's Needed 106 28.1%

L1 Hidden Source or Goal 117 31.0%

/p/

A1 Prongs, Peaks, Points 105 15.8%

B1 Puffy and Plane 84 12.6%

C1 Containers and Enclosed Areas 74 11.1%

D1 Parts and Pictures 91 13.7%

E1 Groups, Units, Levels, Size 94 14.1%

F1 Picking, Pulling 107 16.1%

G1 Patrolling 20 3.0%

H1 Pamper, Pester, Prepare 88 13.2%

I1 Pouring, Putting, Pushing and Punching 183 27.5%

J1 Stepping and Paths 48 7.2%

K1 Endings, The Past 22 3.3%

L1 Talk 72 10.8%

M1 The Problem 62 9.3%

N1 Heat 2 0.3%

/t/

A1 Travel 125 15.0%

A2 Cast, Blow, Flow 60 7.2%

A3 Sprout, Bloat 19 2.3%

A4 Coat, Cover 11 1.3%

B1 Time and Counting 65 7.8%

C1 Fast, Bright, Lively 75 9.0%

D1 Up, Down, Around 77 9.3%

E1 Trying, Tending, Tiring 82 9.9%

E2 Trickery, Error, Nervousness 96 11.5%

F1 Teach, Tame 54 6.5%

F2 Language, Sound 60 7.2%

G1 Traits, Timbres, Tastes 42 5.0%

H1 Touch and Take 214 25.7%

H2 Tie, Tight, Still 130 15.6%

I1 Tips and Tops 146 17.5%

J1 Groups, Area 71 8.5%

K1 Tininess, Ending 148 17.8%

L1 Unpleasantness 55 6.6%

/k/

A1 Containers, Closure and Crevasses 167 20.2%

A2 Corners and Crinkles 105 12.7%

A3 Closeness, Catching, Collecting and Contact 243 29.3%

B1 Carrying, Crawling 64 7.7%

C1 Cutting, Ending, Weakness 141 17.0%

C2 Cruelty, Ache and Irritation 104 12.6%

D1 Kings and Commoners 33 4.0%

D2 Care, Control 66 8.0%

D3 Clumsy, Queer 74 8.9%

E1 Speaking and Throaty Sounds 31 3.7%

E2 Knowledge 67 8.1%

/v/

A1 Containers, Narrow Opening 29 20.0%

B1 Carving 5 3.4%

C1 Vying, Evil 20 13.8%

D1 Veering 7 4.8%

E1 Solving and Serving 8 5.5%

F1 Energy 27 18.6%

G1 Have and Empty 32 22.1%

/H/

A1 Function Words, Definite, Distant 18 56.3%

B1 Coming Close Up Against 7 21.9%

C1 Smooth and Flexible 4 12.5%

D1 Loathe and Soothe 7 21.9%

E1 Causatives 4 12.5%

/z/

A1 Grammatical Function 20 17.1%

A2 Ways and Means 5 4.3%

B1 Energy 19 16.2%

B2 Vibration 27 23.1%

C1 Pause, Fuse 38 32.5%

D1 Altered Consciousness, Smarts 22 18.8%

E1 Ease and Irritation 19 16.2%

/Z/

A1 Highfalutin 2 100%

/f/

A1 Full and Fuzzy 63 15.6%

A2 Fizz, Fountain 20 5.0%

A3 Narrow Opening, Limitation 107 26.6%

A4 Foundations, Fuel 19 4.7%

B1 Weakness, Failure 53 13.2%

B2 Falling, Floating 28 6.9%

B3 Flight 20 5.0%

B4 Freedom, Fate 15 3.7%

C1 Face, Deceive 46 11.4%

D1 Fight and Fuss 52 12.9%

E1 Flap, Flick 29 7.2%

F1 Fire 14 3.5%

G1 Fun, Fine, Fast, Fable 46 11.4%

H1 Family, Sex 22 5.5%

/T/

A1 Theme 3 3.2%

B1 Through 14 14.9%

C1 Thick, Thin, With 28 29.8%

C2 Viscous, Frothy 5 5.3%

D1 Thrust, Thud 9 9.6%

E1 Thrill 7 7.4%

F1 Thrive 23 24.5%

G1 Heat, Thirst 5 5.3%

H1 Earth 6 6.4%

H1 Theme 6 6.4%

I1 Three 3 3.2%

/s/

A1 Smooth Movement 100 9.2%

A2 Walk 40 3.7%

A3 Sink 40 3.7%

A4 Smooth and Fast 37 3.4%

B1 Long 90 8.3%

B2 Circular 33 3.0%

B3 Small 50 4.6%

B4 Spread 39 3.6%

B5 Secrete 43 4.0%

C1 Source, Start 86 7.9%

C2 Stop, Stash 254 23.4%

C3 Seize, Seduce, Mix 81 7.5%

D1 Seek, See 56 5.2%

D2 Swallow 27 2.5%

E1 Struggle, Strike 103 9.5%

E2 Sever 65 6.0%

E3 Scrub 21 1.9%

F1 Strong, Spirited 102 9.4%

G1 Serve, Support 103 9.5%

H1 Several, Series, Size 151 13.9%

H2 Single, Symbol 40 3.7%

H3 Uncountably Many 41 3.8%

H4 Sex 20 1.8%

I1 Surface 85 7.8%

J1 Heat, Light and Fire 50 4.6%

K1 Nose 17 1.6%

L1 Money -- Spend, Save, Steal 75 6.9%

M1 Speak, Seduce 92 8.5%

N1 Soul, Spirit 112 10.3%

O1 Dirt, Spoilage, Sorrow, Sickness, Evil 239 22.0%

/S/

A1 Shake and Shatter 45 23.2%

B1 Shout 15 7.7%

C1 Sheet 27 13.9%

D1 Gush, Brash, Lush 20 10.3%

E1 Shelter 38 19.6%

E2 Shake Off 24 12.4%

F1 Shallow 36 18.6%

G1 Should 24 12.4%

/h/

A1 Have, Hold, Home 62 23.1%

A2 Halt 50 18.7%

A3 Hunger 19 7.1%

A4 Haste 21 7.8%

A5 Hosts, Heavy 15 5.6%

B1 Help and Hear 20 7.5%

C1 Center, Half 11 4.1%

C2 Holy, Health 10 3.7%

C3 High 27 10.1%

D1 Harm 47 17.5%

D2 Difficulty 21 7.8%

D3 Happy 24 9.0%

E1 Happen 6 2.2%

F1 Who and He 9 3.4%

G1 Containers 2 0.7%

/J/

A1 Join 25 14.7%

A2 Jab 14 8.2%

A3 Jutting, Jumping 36 21.2%

A4 Journeying 17 10.0%

B1 Joy, Jazzy 23 13.5%

C1 Judgement, Subtlety 6 3.5%

C2 Smallness 10 5.9%

D1 Giant and Gems 13 7.6%

D2 Junk 36 21.2%

E1 Job 8 4.7%

/C/

A1 Challenge, Forward Motion 119 62.3%

A2 Chew, Scratch 55 28.8%

A3 Scrunch 18 9.4%

B1 Chanting, Charm, Chum 15 7.9%

C1 Much, Money, Quantity 31 16.2%

/m/

A1 Measure 168 29.8%

A2 Match 28 5.0%

B1 Mask, Frame 57 10.1%

C1 Make and Maintain 20 3.6%

D1 Move and Mix 112 19.9%

E1 Must and May 12 2.1%

F1 Mash 55 9.8%

G1 Flames, Earth, Moisture 56 9.9%

H1 Boom 19 3.4%

H2 Mouth 25 4.4%

I1 Mistake 19 3.4%

I2 Mad and Monstrous 106 18.8%

J1 Mirth and Magic 41 7.3%

K1 Mind 22 3.9%

L1 Man 24 4.3%

/n/

A1 Number 45 8.4%

B1 None 99 18.5%

C1 Narrow, Near, Nudge 216 40.3%

D1 Distant 36 6.7%

E1 Bumps and Small Amounts 72 13.4%

E2 Nose 14 2.6%

F1 Line and Plane 66 12.3%

G1 Now, Nave, Knowledge 162 30.2%

H1 Fun, Fine 48 9.0%

I1 Nasty 89 16.6%

K1 Burn, Shine 17 3.2%

L1 Water 17 3.2%

/G/

A1 Noises 21 21.4%

B1 Strong, Bonk 17 17.3%

C1 Sting 4 4.1%

C2 Fling, Bring 11 11.2%

D1 Long, Sink, Hang 22 22.4%

E1 Wrong 21 21.4%

F1 Blank, Mysterious 13 13.3%

G1 Ring, Rink 21 21.4%

H1 Thing 3 3.1%

/l/

A1 Little 89 10.7%

A2 Long 76 9.1%

A3 Levels 13 1.6%

B1 Loop, Curl, Ball 89 10.7%

C1 Flat 93 11.2%

D1 Large, Prolonged 122 14.7%

D2 Prolonged Sound 24 2.9%

D3 Prolonged, Smooth Motion 51 6.1%

E1 Live, Hold, Lock 173 20.8%

F1 Lead, Lunge 43 5.2%

G1 Leave, Lose 66 7.9%

H1 Lone 35 4.2%

I1 Lend, Dole Out 27 3.2%

J1 Loot, Call, Blend, Collide 91 10.9%

K1 Lousy, Negative 154 18.5%

L1 Lash, Kill 95 11.4%

M1 Lie, Fall, Limp 111 13.3%

N1 Lift 41 4.9%

O1 Liking 106 12.7%

P1 Liquid 65 7.8%

P2 Light, Color 44 5.3%

P3 Heat and Cold 31 3.7%

P4 Land 31 3.7%

P5 Air 13 1.6%

Q1 Learning, Law 60 8.2%

/r/

A1 Run and Ride 135 13.9%

B1 Word, Ruckus 114 11.7%

C1 Emotion 228 23.4%

D1 Fire, Dark 46 4.7%

E1 Rot, Wrong 125 12.8%

E2 Rid, Ruin 220 22.6%

E3 Parts 57 5.9%

F1 Strength, Quantity, Intensity 185 19.0%

G1 Rise, Drop, Rank, Peer 68 7.0%

H1 Linear, Round, Wrinkle 162 16.6%

I1 Support, Hard, Work 116 11.9%

J1 Rule 50 5.1%

K1 Room, Where 110 11.3%

K2 Closeness, Connections, Taking 139 14.3%

L1 Prepare, Raw, Beginnings 117 12.0%

/w/

A1 Function Words, Not Known or Present 27 8.7%

B1 War 31 10.0%

C1 Wrong and Wild 29 9.4%

D1 Want 53 17.2%

E1 Work 16 5.2%

F1 Know 19 6.1%

G1 Away, Fro 56 18.1%

H1 Wee 21 6.8%

I1 Wind and Water 41 13.3%

J1 Wail, Whish, Wheeze 37 12.0%

K1 Waves 59 19.1%

L1 Walking, Whizzing 32 10.4%

M1 Whole and One 81 26.2%

/j/

A1 Extent 65 39.2%

A2 Young, Die 17 10.2%

B1 Try 21 12.7%

C1 Use, Yield, Pay 10 6.0%

D1 Protected, Secretive 10 6.0%

E1 Yay, Nay 46 27.7%

E2 Spirituality 3 1.8%

F1 Pronouns 13 7.8%

---

4.1.3.3 'Exceptional' Words and Concrete Noun Classes:

It was mentioned that 3% of English monosyllables did not fit in some phonosemantically defined class. It was also mentioned that all the dictionary senses of those 'aberrant' words fit in the (natural) Concrete Noun classes itemized above. These exceptional 3% of the monosyllables for this particular Phonosemantic Classification are:

People -- bach, bub, chef, gal, Jew, pa, senate, thane, vet, yid

Body Parts -- beak, jowl, thigh

Clothing -- drawers, gown, jeans, pants, togs

Games -- craps, golf, whist

Animals -- chimp, coon, cub, daw, deer, doe, drake, ewe, flea, foal, gnu, goat, hake, hare, hart, hen, loon, mare, moose, newt, pooch, prawn, pup, scrod, squid, stag, stork, swan, tern, thrush, tom, trout, wren

Plants -- beet, chard, chive, clove, cress, dill, kale, larch, maize, pear, phlox, plum, rice, rye, sedge, soy, tea, thyme, wheat, yew

Food -- beet, bran, chard, chive, clove, coke, dill, ghee, kale, kirsch, knish, lox, pear, plum, quiche, quince, rice, roe, rum, rye, schnapps, scone, scrod, slaw, soy, squid, steak, tea, thyme, torte, trout, veal, wheat, wine, wurst, yam

Materials -- jean, lye, myrrh, quartz, teak, zinc

Time -- June

Color -- mauve, roan, taupe, teal

Symbols -- dah, five, four, pi, schwa

Units -- ton

To some extent, a different Phonosemantic Classification would result in a different list of exceptions, but whenever I have formed a Phonosemantic Classification all of the words which don't conform to the classification end up being Concrete Nouns.6 In addition to these 3% that don't fit in my phonosemantic classes, there are hundreds of words that fit in both the phonosemantic classes and the Concrete Noun classes. Very broadly, these words listed here that fit in only the Concrete Noun classes (not the phonesthemic classification) tend to have a single narrow and well-defined function in the language. The word is rarely used metaphorically or poetically.

A much higher percentage of polysyllabic than monosyllabic monomorphemes fail to fit in the Phonosemantic Classification. The reason for this is that the large majority of common monosyllables in English have been in the language for some time and have acquired a broad range of usages. Polysyllabic monomorphemes tend to a much higher degree to be more recent borrowings and to have very concrete reference.

By way of example, I include here a summary of my Concrete Noun classification for people words. A complete summary of all Concrete Noun classes, and a complete listing of the words themselves can be found in Section B1 of Appendix I. Notice that words for people are fairly evenly divided among the phonemes. Those phonemes which occur less frequently in the language in general also occur proportionally less frequently in People words. This is typical of the Concrete Noun classes.

Some of the classes are not marked sequentially, because the classificational system was set up to include also polysyllabic words. When the polysyllables were deleted from the list, some classes fell away altogether:

People

878 words

53% of Concrete Nouns

 

/b/

63 words 7% of people words

1 Beautiful, Handsome, Sexy People 22 35%

2 Mean, Criminal People 11 18%

3 Big, Loud People 22 35%

4 Ugly, Stupid People 17 27%

5 Professions 8 13%

6 Children 4 6%

7 Smart, Enthusiastic People 3 5%

8 Other People 2 3%

9 Groups of People 12 19%

 

/d/

26 words 3% of people words

1 Dear People 4 15%

2 Ladies, Gentlemen 7 27%

3 Titles 3 12%

4 Dummies 7 27%

5 Negative Women 3 12%

6 Mythical Beings 3 12%

7 Other People 2 8%

 

/g/

29 words 3% of people words

1 Socially Inept People 9 31%

2 Mythical Beings 3 10%

4 Grumpy People 2 7%

5 Going People 4 14%

6 Gracious People 6 21%

7 Directing People 4 1%

8 Sexual Orientation 1 3%

9 Groups of People 5 17%

 

/p/

37 words 4% of people words

2 Small People 6 16%

3 Mythic Beings 2 5%

4 Two People 2 5%

5 Substitutes 2 5%

6 Endearing Terms for Women 7 19%

7 Powerful People 3 8%

8 Priests 2 5%

9 Papas 4 11%

10 Prudes 3 8%

13 Unpleasant People 8 22%

15 Groups of People 2 5%

16 Other People 2 5%

 

/t/

26 words 3% of people words

1 Teams 12 46%

3 Tyrants 1 4%

4 Young, Small People 5 19%

5 Groups of People 4 15%

6 Two People 2 8%

7 Travellers 6 23%

8 Sexually Appealing Women 4 15%

9 Unpleasant People 5 19%

 

/k/

52 words 3% of people words

1 Kin 4 8%

2 Clique, Club, People with Special Knowledge 15 29%

3 Other Groups 1 2%

4 People of High Position 9 17%

5 Commoners 13 25%

6 Queer People 7 14%

7 Clowns 2 4%

9 Grouchy People 2 4%

10 Derogatory Terms for Nations 2 4%

12 Other People 3 4%

 

/v/

2 words 0% of people words

3 Other People 2 100%

 

/z/

1 word 0% of people words

1 People 1 100%

 

/f/

28 words 3% of people words

1 Friends, Family 3 11%

2 Groups of People 3 11%

4 Women 3 11%

5 Gay, Effeminate Male 4 14%

6 Mythological Beings 2 7%

7 Contemptible People 8 29%

8 Criminals 2 7%

9 Enemies 3 11%

11 Flirts 2 7%

 

/T/

6 words 1% of people words

1 People 6 100%

 

/s/

66 words 8% of people words

1 Soul 2 3%

2 Mythological and Holy People 6 9%

3 Spirits, Spooks 3 5%

4 Sir, People of High Position 9 14%

5 Groups of People 7 11%

6 Servants 6 9%

8 Snobby People 2 3%

9 Contemptible People 8 12%

10 Sneaky People 4 6%

11 Slow People 4 6%

12 Stiff People 2 3%

13 Sloppy People 4 6%

14 Small People 4 6%

15 Drunk People 4 6%

16 Relatives 4 6%

17 Professions 5 8%

18 Other People 1 2%

 

/S/

13 words 2% of people words

1 Pronouns 1 8%

2 Contemptible People 8 62%

3 Protectors 3 23%

4 Other People 1 8%

 

/h/

31 words 4% of people words

1 General Person 3 10%

2 Negative People (Mostly Secretive and Evil) 12 39%

2 Judges 1 3%

3 Married Partner 2 7%

4 Unmarried Partner 2 7%

5 Comic People 2 7%

8 Hired People 2 7%

9 Other People 1 3%

10 Who 3 10%

11 He, She 5 16%

 

/J/

12 words 1% of people words

3 Wonderful People 3 25%

4 Jerks 4 33%

5 Outsiders 1 8%

6 Guys 2 17%

 

/C/

12 words 1% of people words

1 Chiefs and Champs 4 33%

2 Groups of People 1 8%

4 Informal, Friendly Words for People 3 25%

5 Children 2 17%

6 Derogative Words for People 2 17%

 

/m/

25 words 3% of people words

1 Mothers 4 16%

3 Men 3 12%

4 Mates 4 16%

7 Gods 2 8%

8 Mavericks 2 8%

12 Small People, Servants 3 12%

13 Mutes 3 12%

16 Mobs 2 8%

18 Me 3 12%

 

/n/

9 words 1% of people words

1 Small People 2 22%

2 Insignificant People 2 22%

3 People Who are Near 2 22%

5 Noble People 3 33%

6 People who Renounce Something 1 11%

 

/l/

14 words 1% of people words

1 Lord, Lady 7 50%

4 Louts 3 21%

5 Other People 2 14%

 

/r/

15 words 2% of people words

1 Rabble 10 67%

2 Royalty 2 13%

3 Other People 3 20%

 

/w/

19 words 2% of people words

1 Common or General Words for People 3 16%

2 Abandoned People 2 11%

3 Children 3 16%

4 Unpleasant People 6 32%

5 Women 3 16%

6 Watchful People 4 21%

7 Competent People 3 16%

8 Other People 2 11%

 

/j/

6 words 1% of people words

2 Naive or Inexperienced People 2 33%

3 You 3 50%

4 Other People 1 17%

 

*************

I suggest that it is the specificity of reference that is interfering with the more obvious manifestation of iconic semantics. Let me provide an example to clarify why this is so. The consonant /b/ appears in initial position in words referring to loud sounds, explosions, irreverent behavior, bulging and large quantities much more frequently than one would expect statistically if phonology and semantics were completely unrelated. The phoneme /b/ also appears in initial position in a lot of words referring to large animals: bear, boar, bull, buck, behemoth, buffalo, etc. And even when the animal is small, it still tends to be among the largest or most irritating in its phylum: bee, beetle, bug. In some cases, like 'bug' and 'beast' and even 'bear', 'boar' and 'bull', the loud, irreverent connotative meaning is prevalent enough that The American Heritage Dictionary lists it as a separate sense. But the more obscure and specific the animal, the less likely this is to be the case. We don't say, "*He's such a bandicoot," probably in part because this animal isn't part of the average English speaker's everyday experience. In addition, the less specific the animal is (bug, beast, brute, animal, creature, critter), the more likely it is to be used metaphorically. It is these common or general terms that fall most easily in the Phonosemantic Classifications.

Strictly speaking, a 'buck' and a 'boa' and a 'bison' do not fit in the phonesthemes. At least according to most dictionaries, the word 'bison' has no alternate sense involving loudness, strength and obstreperousness in general. It only has what we think of as its 'basic' sense -- that of a bovine. Observe how this quality of concrete reference interferes with Phonosemantic Classification.

'Reference' answers the question, "What is word X?" That is, in general, the Natural Classification is organized along the semantic axis of metonymy/hyponymy. The phonological component answers the question, "What is X like?" In the case of non-concrete Natural Classes, these two questions often overlap. For example, "What is 'bungle'?" A bungle is a clumsy aggressive action -- that is its referent. That's also what a bungle is like. If we ask, "What is a 'bull'?" The answer is that it is a mammal, a male, a bovine. That's how it fits in its Natural Classification. It does not answer the question, "What is this animal like?" So what is this animal like? Speaking objectively, it is large, powerful, hairy, horned, etc. But folk mythology also makes it aggressive and clumsy. We speak of 'a bull in a china shop'. Is a bull factually clumsier or more aggressive than a hippo? That's not at all certain. Therefore, whereas 'bungle' is a clumsy action by its very definition, 'bull' is by definition a large mammal, and is only perceived as clumsy. If a bull is only perceived as clumsy, we are already on slipperier ground classifying it phonosemantically as 'clumsy' than we are classifying 'bungle' as 'clumsy'. In this case, American Heritage provides a 'clumsy' definition which allows us to classify 'bull' phonosemantically. In a similar manner, strictly speaking, a bison is a mammal with a certain DNA sequence, and that's all it is. 'Bisons' are close neighbors of 'bulls' and 'buffaloes', which also both begin with /b/. Does the /b/ in 'bison' predispose us to think of it as clumsy analogously with 'bull' and 'buffalo'? Perhaps. But there's no direct 'official' evidence for that within the language. To determine whether that is true, we cannot use dictionaries; we have to resort to psycholinguistic experiments.

Why is it that 'bull' and buffalo' are perceived as more clumsy than 'bison'? One possibility is that they both contain an /l/. The phoneme /l/ appears in a disproportionately large number of 'clumsy' words. The phoneme /s/ on the other hand occurs disproportionately frequently in words of competence: smart, snappy, sassy, swift, smug, style,... Most sloppy words that contain an /s/ also contain an /l/. If Phonosemantic Association happens on the level of the phoneme, then the evidence in fact suggests that the aggressiveness in 'bull' and 'buffalo' comes from the /b/, and the 'clumsiness' comes from the /l/. 7

So some of the words like 'bison' and 'pi' fall outside the Phonosemantic Classification, and one is initially inclined to think that the Phonosemantic Hypothesis cannot therefore be wholly maintained. The /p/ in 'pi' cannot directly be placed in a /p/ phonestheme for 'precision', though there is such a phonestheme. 'Pi' has no metaphorical usage meaning 'precise' or any other connotative usage for that matter -- it means 3.14159..., and that's all it means. Still I hope to demonstrate that there is good evidence to suggest that the precision so prevalent in /p/ otherwise is also likely to influence the English speaker's dynamic usage of the word 'pi', and that therefore even those words that don't fit in a Phonosemantic Classification are influenced by their sound. If this proves to be the case, that is, if the /p/ is having a semantic effect even in very Concrete Nouns (pi, spinach, piccolo, piranha), then it is most effective to view Iconic meaning as the fundamental level of word semantics, and to view referential semantics as superimposed on it.

---

4.1.3.4 Theoretical Status of Senses and Phonesthemes

Unlike a Natural Classification, a Phonosemantic Classification is not right or wrong unless it also violates the Natural Classes. Some Phonosemantic Classifications bring out certain semantic aspects of a given phoneme, and others bring out other aspects of the phoneme. The phonesthemes are therefore epiphenomena. I find that they are not psychologically real in the way that Natural Classes are. That is to say, I do not think they are part of 'langue'. I experience this subjectively as I work devising phonosemantic classes for a given set of data. I find that to some extent, I'm free how I would like to organize and present the phonosemantic data, and to some extent, I'm not free. Certain classifications are, so to speak, ungrammatical. Those which are ungrammatical are those which fail to do deference to the 'Natural Classes' built into English. But although I do not think of Phonosemantic Classifications as psychologically real, I do find them to be an extremely important device to make the phonosemantic data accessible and readable to the average researcher.

I think the senses of a word -- like the phonesthemes -- are useful, but not psychologically real. The reader will notice that the various dictionary senses or referents of the words in the Phonosemantic Classification in Appendix I are not explicitly marked. For example, 'cross' has many senses other than the motional one intended in the /r/ phonestheme used as an example in the previous section (4.1.2): 'to cross the street'. Yet nowhere in this experiment have I made a list of the senses for each of these words and then cross-referenced them with the occurrences of these words in the phonesthemes. It may be presumed that the reason for this is that the sense intended is obvious from the superclass and phonestheme designators, and this is in part true, but the reason for omitting references to the intended sense also runs deeper than this.

I distinguish, as mentioned, three levels of semantics: the iconic, the classificational and the referential. I will discuss them in much more detail in the final chapter of this dissertation. Suffice it to say for now that in my view, the iconic level of word semantics is fundamentally what makes word semantics sensitive to the phonological structure of the word. The classificational level is the level of Natural Classes, we have so frequently discussed. Each successive layer is superimposed on the previous. So the iconic level in semantics is primary. Superimposed on it is the classificational and superimposed on that again is the referential.

The senses of a word arise as a side effect of filtering the iconic meaning of a single word through the level 2 classificational system and then providing them with a referent at level 3. The phonesthemes arise analogously as a side effect of filtering the iconic meaning of all the words containing a given phoneme through the classificational system. Word senses therefore are epiphenomena, and like the phonesthemes themselves are not psychologically real -- not a part of 'langue', or the 'grammar'. And a brief comparison of the definitions provided in 3 different dictionaries of a few basic English words like 'get' or 'take' will convince anyone that lexicographers in no way agree on what constitute the correct 'senses' of a word.

Put another way, the sense of 'cross' that I have in mind in the above sample entry is actually defined as that sense which fits in the level 2 class of non-vehicular motion verbs with a human agent. Further giving it a sense definition would not only be redundant, but misleading, because it implies that word 'senses' are not analyzable in terms of more basic concepts. I believe that they are.

---

4.1.3.5 The Positional Effect

The process of coming to understand what specific effect each aspect of the word's phonology is having on its semantics is something like teasing apart a puzzle. One first uncovers what semantic domains are disproportionately represented in which phonemes. Then one goes into specific Natural Classes such as the 'light' words discussed above and observes how these more general tendencies manifest within a specific semantic domain. In addition to the pure effect of the phoneme on the meaning of the word, there is a positional effect that at least for me is much harder to discern at first. In order to see it, one has to first abstract away from the Natural Classes, and then abstract away from the semantics of each individual phoneme. It may seem trivial, but I find it to be no small matter to uncover which aspects of meaning are attributable to what.

As discussed above, non-vehicular motion is well represented in /r/ no matter what position it appears in. But in words of 'light' containing /g/ and /l/, both consonants must precede the vowel. Why? And what are the differences between the various /r/s for motion in the various positions? I will give some examples here to try to make clear what effect I believe the position of the relevant phoneme to have.

Briefly, a consonant which appears before the vowel has a function of 'setting the stage' for the action that plays itself out in the word. A consonant which occurs after the vowel constitutes a sort of conclusion or 'punch line'. The vowel is somewhat analogous to a verb in a sentence. It defines the nature of the flow or action. The initial consonants are like subjects, and the final consonants like objects. In addition, the positional effect depends on whether the consonant appears in absolute initial position or after another consonant. If it appears in absolute initial position, it is given free reign in a manner of speaking over the backdrop of the word. If a semi-vowel follows another consonant but precedes the stressed vowel, its effect is mitigated or modified by the initial consonant. The phonemes /l/ and /r/ are initially the most useful consonants to look at in English to get a sense for the effect of position on the semantics of the word, because they can occur in the most positions within English syllables.

Consider the words containing /r/ and referring to some kind of noise. There are 114 of them making up 13% of English monosyllables containing /r/:

1 Ruckus, Sound

1

rage, rant, rap, rasp, rave, ring, roar, rout, row

2

bray, breath, crack, crash, creak, croak, croon, crunch, cry, drawl, drone, drum, frog, groan, growl, grunt, shriek, shrill, thrum, trill, troll

3

screak, scream, screech, strain, strike, stroke, strum

F2

bark, birl, burp, chirp, chord, dirge, fart, hoarse, horn, snarl, snort, storm

F3

birr, blare, chirr, churr, purr, roar, snore, whirr

 

2 Word, Speak

2

brag, bring, broach, greet, grill, gripe, grouse, phrase, praise, prate, pray, prayer, preach, prove, threat, thresh, train, trope

3

screen, spread, spring, stress, stretch

F2

blurt, harp, spurt, word, yarn

F3

air, pour, prayer, share, square

 

3 Read, Write

2

braille, browse, draw, graph, phrase, press, print, proof, prose, trace

3

scrawl, scribe, script, scroll

F2

card, chart, clerk, forge, mark, term, verse, word

F3

score

 

4 Hear

F2

hark, learn, mark

F3

ear, hear

Notice that those words which have an /r/ in initial position tend to have a loud, devil-may-care quality about them. This quality runs throughout the phoneme /r/. Notice that /r/ does not occur in initial position in words of sound which require more focus or precision, specifically words involving coherent language. The phoneme /r/ can provide raw energy to something, but it implies no inherent control over this energy -- that control must be provided from without. Notice that nearly all the /r/ words of speaking have a great energy about them. Notice that /r/ doesn't occur in the most receptive of sound words -- those of hearing -- except after the vowel. When the energy that /r/ provides, in other words, happens at the receiving end of the speech event, /r/ appears after the vowel in English. Notice that if /r/ occurs in absolute final position in words for noises, the noise is prolonged. If /r/ occurs in pre-final position, the noise is cut short. Once again, this little exposition does not of itself prove anything. But hundreds of other examples of this nature can be found by looking at the data in Appendix I.

---

4.1.3.6 Summary of Results of Experiment 1 and Outline of Resultant Theories about Language

· Monosyllabic words in English which contain a given consonant fall within much narrower semantic domains than one would expect if the relationship between phonology and semantics were arbitrary. This semantic domain resembles the articulation of the consonant in question, and this is one piece of circumstantial evidence that the relationship between phonology and semantics is essentially 'Iconic' in the Peircean sense. The phonology of a word has a much more pervasive and specific effect on its semantics than is generally supposed.

· There is a type of psychologically real classificational system which I term a 'Natural Classification'. The phonesthemes are not in general psychologically real in the way that the Natural Classes are. Phonesthemes are subsets of the Natural Classes. They are epiphenomena resulting from the combination of Natural Classifications with the semantics of sound. Such a combined classificational system I call 'phonosemantic'. Because of the epiphenomenal nature of the phonestheme, there is no one right Phonosemantic Classification. Various Phonosemantic Classifications make various aspects of phoneme semantics more accessible to analysis. The possibility of creating Phonosemantic Classifications as defined in this dissertation is one of the primary means I use for testing the Phonosemantic Hypothesis stated in the introduction. This particular experiment tests for criteria 1-6, but not criteria 7-9 of a Phonosemantic Classification.

· I find an additional process called 'Clustering' or Phonosemantic Association. Clustering is the tendency for phonemes and phoneme sequences to become even more narrowly limited than their iconic semantics demands -- the tendency to try to assign a coherent referent to every phoneme or phoneme sequence. (The whole becomes more narrow than the sum of the parts.) For example, words beginning with /gl/ are inherently limited to a certain semantic domain by the very nature of the semantics of /g/ and the semantics of /l/. We find that a fairly large range of this potential is factually represented in the vocabulary of English, but disproportionately so. A much larger percentage of these /gl/ words fall into the sub-domain 'reflected light' than one would expect if the semantics of /gl/ simply were a combination of the semantics if /g/ plus the semantics of /l/.

· At the third level of semantics, a specific referent is assigned to each word. The more concrete and unambiguous the referent for the word, the less salient is its phonosemantics. The reason for this concerns what the word 'is'. If the referent for a word by its very nature is connotative or interpretive, then the word's phonosemantics can cooperate with its referent. If, however, the word refers to some concrete object in the world, and no room is left for connotation or interpretation, then the phonosemantics of the word seems to impose a connotation or interpretation to the word rather than affecting what the word actually refers to. For example, the verb 'bungle' is an aggressive, clumsy act by its very definition. Therefore the verb falls easily in phonesthemes for bungling and aggression. But the primary sense of 'buffalo' is merely a mammal with a certain DNA sequence, so it falls less readily into phonesthemes for bungling and aggression. The bungling and aggression of a buffalo is merely a cultural interpretation. Because 'buffalo' exists as a verb of bumbling and aggression, it can still be classified into these phonesthemes. But since no such formal usage for 'bison' exists, it falls outside the Phonosemantic Classification. This does not, however, mean that sound has no effect on its meaning, as can be demonstrated by other kinds of tests for the psychological reality of sound-meaning among native informants.

· Like the phonesthemes, the 'senses' of a word are not stored as part of langue. They are epiphenomena resulting from the combination of all three levels of word semantics -- the iconic, the classificational and the referential.

· The position that a consonant occupies in a syllable also affects its meaning. Consonants that appear before the vowel form the backdrop for the action of the word, and consonants that appear after the vowel express the result of the action implicit in the word.

 

---

---

4.2 Experiment 2 -- Classification First by Phoneme Sequence, Subclassification by Semantic Domain and then Regrouping of Different Phonemes by Semantic Domain

See Appendix II for full data and results.

4.2.1 Methodology

· Find all monomorphemic or root words containing a given consonant in a given position in the word. In this case all monomorphemes in my active English vocabulary in which /r/ occurs in second position were used.

· Divide the words according to another phonological characteristic. In this case, the monomorphemic English words containing /r/ in second position were sub-classified by initial consonant.

· For each of these resultant classes, create a Phonosemantic Classification. If necessary, ignore Concrete Nouns.

· Now match up phonesthemes cross-phonemically that fall in the same Natural Classes. For example, the 'breaking' phonestheme for /br/ is aligned with the 'fracturing' phonestheme for /fr/ and the 'cracking' phonestheme in /kr/.

· Identify how these matching phonesthemes differ semantically.

---

4.2.2 Example

Rupture and Fractioning

/r/ appears in many words of destruction. The ruptures that are non-iterative show up when a stop consonant is in initial position. If the stop is voiced, we find an additional and related class of words which are fractioned into many pieces. If a fricative is in initial position, the result is broken into uncountably many fine particles. We can more or less characterize the effects of the phonological features in this classes follows:

[+stop, +voiced] -- many distinct but countable parts

[+stop, -voiced] -- snip off an end or pierce at a point

[+fricative] -- mashed into single consistency, pieces are uncountably many

[+labial] -- ends, points, tips

[+dental] -- lines

[+velar] -- rupture in a surface

The verbs are listed first followed by related words which are the results of the actions of these verbs.

a. Rupture

[+stop]

/b/

Break -- something hard broken off or severed into two or more pieces

bran, branch, breach, break, brief, brittle, brook, browse, bruise

 

/d/

Dig -- regular breaking downward through dirt

dredge, drill

Dirt -- that which remains from digging

dreck, dredge, dregs

 

/g/

No verb exists

Groove -- an open indentation in a surface, the deepest point is typically not visible

grave, groin, groove, grotto

 

/p/

Prick -- a long hard object with a point which pierces a surface at one point

prick, prickle, probe, prod, prong

 

/t/

Trim -- something linear and often growing the tips of which are cut back just slightly

trim

Trifle -- a small thing which has been made out to be bigger than it is

trifle, trinket, trite

 

/k/

Cut -- to cut a surface

crack, crop

Crack -- deformities in a surface

crack, cranny, crater, crease, crevasse

Crunch (Crinkling sounds) -- the sound of deforming a surface

crack, crackle, crash, crinkle, crunch

 

b. Fractioned/Many Pieces

[+stop, +voiced]

/b/

Branchy -- radiating lines from a base

bracken, braid, brake, bramble, branch, briar, bristle, broom, brow, brush

Breed -- offspring of a single source

brace, breed, brood, brother

Type -- a group which all fit a specific characterization and have a common source

brace, bracket, branch, brand

 

/d/

Drip -- liquid flowing linearly cut into drops

dribble, drip, drizzle, drop

Drop -- particles of liquid resulting from dripping

dribble, drip, drivel, drizzle, drop

 

/g/

Grind -- to push through a grid

grind, grate

Grid -- a network of lines crossing at 90 degrees to form squares

grate, grid, grill, grille, graph

Grainy -- small bits resulting from grinding

grain, gravel, grit

 

c. Broken into a Mass of Uncountably Many Tiny Particles

[+fricative]

/f/

Fray -- to split the tips of something soft into a mass of fuzz or foam

frizz, frizzle, fray, froth, fry

Frill -- intricate decorations at the edge

frill, frieze, fringe, frock

Froth -- foam, uncountable, small bubbles or bits, usually in liquid

(freckle), frost, froth

 

/T/

Thresh -- to flail something flexible and linear fairly violently

thrash, thresh

Thread -- a long piece of materials thinner than a string

thread

 

/S/

Shred -- to cut something solid into many small strips or particles

shred

 

The non-concrete monomorphemic words which have an /r/ in second position and which are not mentioned in Appendix II are:

 

brawn, bribe

graze, greet

prey, prowl, price, prairie

trace, trait, trend

crux

frail, fraught, frisk

---

4.2.3 Discussion of Findings

4.2.3.1 Evidence this Experiment Provides for the Major Theses in this Dissertation

Evidence for the Phonosemantic Hypothesis

This experiment does not so readily give one a general overview of the semantics of the whole phoneme as did the first experiment (4.1 -- Appendix I), but it does provide a better view of what specific role each phoneme has within a given semantic domain. This is the best test I've come up with for identifying the semantics of phonetic features.

This experiment provides evidence for criteria 1-6 and criterion 9 of the Phonosemantic Classification, which is required to prove the Phonosemantic Hypothesis:

Criterion 1. Very nearly every word with the given phonological characterization fits in some semantic class.

Criterion 2. Each semantic class contains a large percentage of the words which match that phonological characterization.

Criterion 3. There are relatively few semantic classes in the classification.

Criterion 4. The semantic classes in the classification are distinct

Criterion 5. Each word fits into an average of a fairly large number of classes.

Criterion 6. The semantic classes are narrowly defined. By a 'narrowly defined' semantic class, I mean one which encompasses a small percentage of words in the language as a whole.

Criterion 9. Any class in a Phonosemantic Classification can be defined narrowly enough that words not matching the relevant phonological characterization are excluded from it.

Evidence for Clustering

The fact that this test can be conducted at all is, of course, indirect evidence for Clustering. But there is additional evidence for Clustering on another level. Consider this 'rupture' class above for voiced stops in initial position. The phoneme /r/ tends to rip, break, part and tear no matter where it's positioned in the word, and /d/ is downward and linear throughout the English lexicon (as well as frequently wet), but /dr/ in this rupture class is not just a vertical line that is fractioned into several pieces -- which is what the /d/ combined with the /r/ alone optionally predispose the word toward; the large majority of the /dr/ words specifically concern a vertical line of water that is fractioned into many pieces -- in other words, dripping water. The tendency is therefore to attribute an identifiable referent to /dr/ which is narrower than the semantic range formed by /d/ and /r/ alone. This Clustering manifests not as a single invariable referent, but only as skewed distributions -- a tendency to prefer dripping water over other potential referents.

Evidence for the Interference of Reference

The above example can be used to explain why one finds information about the semantics of phonetic features more readily by subdividing words into small groups defined by two phonemes rather than one first, and then recombining them, as I do in this experiment. The alternative would have been to create a Phonosemantic Classification for all the words containing /r/ in second position and then subdividing all the 'fracture' words according to initial consonant afterwards. Had I done this, I would have found no words for multiple fractioning beginning with /d/, because it wouldn't have seemed to me that 'dripping' involves a fracture, since it refers to a motion of water, and the dotted line is only how the water moves. However, having seen that there is a very obvious 'branching' class in /br/ and 'grid' class in /gr/, and having seen from the previous 'rupture' class that in those cases as well, labials seem to imply a point and velars a surface, I am already asking myself, "If this pattern holds, then I would expect to find the form of a dotted line somewhere in words beginning with /dr/. Do I find such a thing?" Well, as it turns out, I do... dripping water. Had I, however, done the experiment the other way around, I wouldn't have thought of 'dripping water' as a dotted line, but rather as downwardness and water. When I ask myself what other occurrences of 'dotted lines' I find commonly in the world around me, I'm hard pressed to think of any besides dripping water. So from this perspective, it's not surprising that English has chosen /dr/ for dripping water. Performing the experiment in the order I suggest, in other words, helps focus attention away from the referential aspects of meaning and toward those aspects which are determined by sound, and this is why I think it is so effective in bringing out the meanings of the phonetic features.

Again, the characterizations of the phonetic features were all derived from the non-concrete word classes. The more concrete the semantic class, the more the referential aspects of the meaning -- like the 'water', as opposed to the linearity, in 'dripping' -- impose themselves on the researcher. So this aspect of the experiment also provides evidence that reference interferes with the salience of sound-meaning.

---

4.2.3.2 Common Semantic Domains for /r/ in Second Position

The natural domains which /r/ in second position was found to occur in frequently were:

· Rupture and Fractioning

· Garbage

· Negative People

· Iteration

· Deception

· Containers

· Verge, Brim

· Directed Movement Verbs

· Pressure

· Receiving

· Support

· Future

· Groups

· Grab/Crave

· Three

These are similar to the classes that were found for /r/ in Experiment 1 (Appendix I). Words in the following Natural Classes were also classified for this experiment:

· Heat

· Water

· Sound

· Emotion

· The Mind

· Materials

· Pretty

---

4.2.3.3 Characterizations of the Phonetic Features

By performing this experiment for this set of data, one arrives at the following characterizations of the phonetic features:

[+voiced]

many distinct but countable parts

dirty, angry

heavy duty

creative source, but little concern for results

[-voiced]

specific intention or result

an ongoing, preexisting or pending process

[+stop]

emphasis on a thing or product as opposed to a process

specific path, starting point, boundaries

receiving, support

end, point, boundary, container

[+fricative]

mashed into single consistency

soft

uncountably many

emphasis on the activity or process itself

release, no concern for the path

hysteria

[+labial]

a narrow opening

selected for a purpose

ends, points, tips, edges, initiation

senseless, empty waste of time or energy

completed, clear

[-labial]

sadness

fear

group selected for a purpose

[+dental]

linearity

natural motion, sleep/trance

implicit goal or direction

mid-stream, process

[-dental]

large size

[+velar]

surface

a mature process

gathering, grabbing, craving, excess

something hidden, unclear, unexamined

---

4.2.3.4 Characterizations of the Phonetic Features Sorted by Semantic Class

Iterative/Nonbreaking

[+stop, -voiced]

Containers

[+stop, -voiced]

Edge of Something

[+labial]

Directed Movement Verbs

[+voiced] -- no concern for the result

[-voiced] -- specific intention or result

[+labial, +stop] -- pressure onto something, often from within a container with a narrow opening like a well or the lungs

[+dental] -- natural linear motion, against resistance in /d/ and generally with little resistance in /t/ and /T/

[+voiced] -- motion over (/k/) or rooted in (/g/) a surface or terrain

[+stop] -- specific path

[+fricative] -- no concern for the path

Pressure

[+voiced] -- focus on the process, heavy

[-voiced] -- focus on the point of contact

[+labial, +stop] -- support or preparation from behind

[+dental, +stop] -- natural linear motion

[+voiced] -- pressure against a surface out in front

[+stop] -- pressure causes a permanent effect

[+fricative] -- solidify into a mass, effect in place only as long as the conditions maintain

Support

[+stop]

Future

[+labial] -- initiation

[+dental] -- propelling a process in mid-stream

[+voiced] -- a mature process

[+stop] -- emphasizes a starting point, boundaries

[+fricative] -- emphasizes the process itself

[+voiced] -- creation of something new

[-voiced] -- implies an ongoing, preexisting or pending process

Groups

[+labial, +stop] -- group selected for a purpose

[+dental] -- gathered by following a common goal

[+voiced] -- general gathering

Size

[+stop, -dental]

Grab/Crave

[+voiced]

Receiving

[+stop]

Garbage

[+labial] -- senseless, empty waste of time or energy

[+dental] -- that which is thrown or drained away

[+voiced] -- greasy or crumbly texture, excess from something eaten or used

[+stop] -- emphasis on the waste itself

[+fricative] -- emphasis on the activity of discharging and its subsequent release

[+voiced] -- dirtier, more heavy duty garbage

Derogative Terms for People

Was not able to see semantic patterns across the phonetic features

Deception

[-voiced]

The Mind

[+labial] -- completed, clear

[+dental] -- process, implicit goal or direction

[+voiced] -- unclear, unexamined

 

---

---

4.3 Experiment 3 -- Natural Classes for Arbitrary Sets of Words

See Appendix III for full data and results.

4.3.1 Methodology

· Choose a random set of words. In this case, every 10th English monosyllable in alphabetical order was used. This resulted in a random set of 342 words.

· Find a Natural Classification for this set of words.

· Separate off the Concrete Noun classes resulting from that classification.

· Look for phonological commonalities in the remaining classes. Break larger classes down into smaller ones if necessary.

· Compare the classification of random words to a classification found for a similar number of words sharing a common phonological trait. In this case, the 295 monosyllables with initial /b/ were used.

---

4.3.2 Example

I compare the random set of 342 with the 295 monosyllables beginning with /b/:

Bump

Random -- [+labial]: bulge, dune, heap, lobe, nub, paunch, rough

B-Words: bag, bale, ball, bay, bead, belch, bell, bilge, blimp, blip, bloat, blob, blouse, blow, boil, boll, boob, breast, bud, bug(eye), bulb, bulge, bum, bump, bunch, bun/s, burl, burst, bust, butt, butte

 

Incline/Fall

Random: cline, cock, prone, sheer, step, stoop, swoon

B-Words:

 

Float/Bounce

Random:

B-Words: ball, bank, bath, bathe, beach, bilge, birl, blimp, bloat, boat, bob, boil, bounce, bound, breach, breeze, buck, bulge, bump

 

Long/Thin

Random: flue, knife, oar, peg, pole, rake, saw, screw, shot, strand, thorn, trunk

B-Words (Sticks, Building Materials): balk, bar, bat, bead, beam, birch, birl, blade, bloom, board, bone, boom, bough, brace, branch

 

Foamy, Frilly

Random -- [+fricative]: frill, froth, shag

B-Words (Brushy): bang, barb, beard, bosk, braid, brake, branch, broom, brow, browse, brush, bur, burr, bush

 

Cry/Talk

Random -- /b/, /p/: bawl, beg, bill, bode, mot, pitch, preach, squib, weep, yawp

B-Words (Loud, Effusive): bah, barb, bark, bash, baste, bat, beard, beck, beef, beg, bend, bet, bid, bilge, bill, bis, bitch, blab, blame, bless, blot, blow, bluff, blunt, blurb, blurt, boast, bode, bolt, book, bore, bosh, boost, boss, bounce, bout, brag, brand, bull, bunk, butt, buzz

 

Exclamations

Random:

B-Words: bad, bah, bam, bang, bash, blah, blast, blaze/s, boo, boom, bosh, boy, bud, bull, bye

 

Noise

Random -- [+liquid]: bawl, blare, clang, cluck, grunt, hark, horn, peal, roar, taps, ti, tune, tweet

B-Words (Loud, Sudden): baa, bam, bang, bark, bawl, bay, beep, belch, bell, birr, blare, blast, bleat, bleep, blow, bomb, bong, boo, boom, brawl, bray, burp, burr, buzz

---

4.3.3 Discussion of Findings

This experiment provides evidence for all the criteria 1-9 of the Phonosemantic Classification, which is required to prove the Phonosemantic Hypothesis:

Criterion 1. Very nearly every word with the given phonological characterization fits in some semantic class.

Criterion 2. Each semantic class contains a large percentage of the words which match that phonological characterization.

Criterion 3. There are relatively few semantic classes in the classification.

Criterion 4. The semantic classes in the classification are distinct

Criterion 5. Each word fits into an average of a fairly large number of classes.

Criterion 6. The semantic classes are narrowly defined. By a 'narrowly defined' semantic class, I mean one which encompasses a small percentage of words in the language as a whole.

Criterion 7. A much smaller percentage of the words which do not match the relevant phonological characterization fit into any class.

Whenever words chosen at random are classified, they fall into the Natural Classification. Among these Natural Classes one will find the Concrete Nouns, but also others, many of which are represented in this data: water, fire and light; sound and language; big, medium, small; beginning, middle, end; strong, weak; good, bad; crime and deception; quantities and emptiness; long, round, flat; bumps and indentations; strong, weak; smooth, fuzzy, bumpy; dirt and washing; verbs of motion (vehicular, non-vehicular), verbs of contact, verbs of destruction, verbs of creation; verbs of wiggling and turning; happy, sad, angry, irritated; give, get; boundaries, containers, groups, impediments; etc.

However, one finds a number of classes which appear in the /b/ classification and therefore are very prevalent among words beginning with /b/, but which accept no words from the random list: These are boundaries(5% of monosyllables beginning with /b/ have at least one sense which refers to a boundary vs. 0% of random monosyllables), impediments(8% /b/ vs. 0% random), interference(11%! /b/ vs. 0% random), stopping and waiting(5%), binding and fastening(5%), floating and bouncing(7%), breaking(6%), exploding(5%), blowing(4%), departing(7%), badness(2%), crime(2%), emptiness and blindness(11%), carrying(6%), future(7%), immersion(4%), growth(6%). In all of these classes, there are, of course, some words which don't contain /b/, but the disproportions are great enough that in a random sampling, I came up with no matches in many cases.

The converse also holds. There were several classes which are quite common in the language generally, but which are relatively rare in words beginning with /b/: eating, taking and receiving, throwing and giving, spending, slowness, surfaces, ability. There were no words starting with /b/ that didn't fit in the Concrete Nouns or the /b/ phonesthemes. There were, however, 12 non-concrete random words which fell into a Natural Class which they shared with no other words in the random selection. These words were: air, mend, quark, sky, snide, stint, toy, troth, west, yep, yon, yum

Criterion 8. Those words that do not match the relevant phonological characterization but which nevertheless do fit in the phonesthemic classification fit on average in a smaller percentage of classes, than those words which do match the phonological characterization.

The average word containing a /b/ in initial position falls into about 3 classes (which is typical of phonemes which begin about 300 monosyllabic words), and words in the random classification fit into an average of only about 1.5 classes.

Criterion 9. Any class in a Phonosemantic Classification can be defined narrowly enough that words not matching the relevant phonological characterization are excluded from it.

Clustering will cause words with common phonological traits to be unevenly distributed among the natural semantic domains. In Appendix III some classes were marked to indicate which phonemes appeared there most frequently. For example, a disproportionately large number of the verbs of physical contact start with /b/. One can get an approximate picture of how the semantics of words containing /b/ is biased by looking at the variations in distribution between the words in the classification. In many cases, very few or no words beginning with /b/ appear in certain classes which are quite common in the language otherwise. Instead, words beginning with /b/ fall in other classes which are similar. Examples are slowness ( from the randomly selected words) vs. interference and blockage (in words starting with /b/), ability and possibility (random) vs. the future (/b/), beginnings, middles and endings (random) vs. only beginnings (/b/), negativity and loss (random) vs. emptiness and blindness (/b/), weak (random) vs. blocked (/b/), touch (random) vs. beat (/b/), cut or chafe (random) vs. break (/b/), frills and froth (random) vs. brushy (/br/), take and receive (random) vs. bind and fasten (/b/), areas (random) vs. boundaries (/b/), etc.

Words which have common phonological traits and which fit into narrower semantic domains than those covered by the entire vocabulary will fit into subsets of the Natural Classes. For example, the language as a whole will have many words for people, and /b/ also has many words for people. But the words for people which begin with /b/ are confined to a subset of people. Words for people beginning with /b/ are outrageous, sexy, bad and beautiful. There are also many children and many groups of people in /b/. People in /p/ tend to go into the priesthood; they are often prudish or work in professions which give them authority and control over others. In the above classification, I have tried to indicate in one or two words what semantic trait distinguishes the words in a given class which begin with /b/ from all the other words in that class.

Once again, all the exceptions to the Phonosemantic Classification are Concrete Nouns, which constitutes evidence that the salience of iconic meaning in a word is related inversely to the concreteness of its reference.

There is evidence for Iconism as well. In addition to these disproportions in the Natural Classes, each phoneme will deliver to the word within a given Natural Class a specific element of meaning. For example, of the /b/ words of physical contact, essentially all are violent (bat, beat, bash, bonk,...). And unlike their counterparts starting with /p/ (prick, pike, pin, poke,...), they rarely pierce the surface. The disproportions we observe are due to Clustering or Phonosemantic Association. But the specific meaning which each phoneme provides within a given semantic domain is a reflection of what I have called Iconism.

 

---

---

4.4 Experiment 4 -- Classify Words Containing a Phoneme Sequence X into a Classification Designed for Words Containing Phoneme Sequence Y

See Appendix IV for full data and results.

4.4.1 Methodology

· Choose a natural set of words which have some common phonological feature. In this case, all the English monosyllables beginning with /b/ were chosen.

· Find a Phonosemantic Classification for this set of words.

· Choose a different natural set of words which have some common phonological feature. In this case, all the English monosyllables beginning with /l/ were chosen.

· Try to fit these words into the Phonosemantic Classification found for the first set of words.

---

4.4.2 Example

A1 Bulging, Brushy

1 Bulging

/b/ Words -- bag, bale, ball, belch, bell, bilge, blimp, bloat, blob, blouse, blow, boil, boob, bulge, bum, bun/s, burl, burst, bust

/l/ Words -- lung

 

2 Bump

/b/ Words -- ball/s, bay, bead, blip, bloat, blob, boil, boll, boob, breast, bud, bug(eye), bulb, bulge, bum, bump, bun, bunch, bun/s, burl, bust, butt, butte

/l/ Words -- lobe, lump

 

3 Round

/b/ Words -- bale, ball, bead, bell, blimp, blip, bloat, blob, blotch, bowl, bulb, bulge

/l/ Words -- loop

 

4 Bend

/b/ Words -- bay, belt, bend, bight, bow

/l/ Words -- lens

 

5 Brushy

/b/ Words -- bang, barb, beard, bosk, braid, brake, branch, broom, brow, browse, brush, bur, burr, bush

/l/ Words -- lace, lash

---

4.4.3 Discussion of Findings

This experiment provides evidence for criteria 1-9 of the Phonosemantic Classification, which is required to prove the Phonosemantic Hypothesis:

Criterion 1. Very nearly every word with the given phonological characterization fits in some semantic class.

Criterion 2. Eeach semantic class contains a large percentage of the words which match that phonological characterization.

Criterion 3. There are relatively few semantic classes in the classification.

Criterion 4. The semantic classes in the classification are distinct

Criterion 5. Each word fits into an average of a fairly large number of classes.

Criterion 6. The semantic classes are narrowly defined. By a 'narrowly defined' semantic class, I mean one which encompasses a small percentage of words in the language as a whole.

Criterion 7. A much smaller percentage of the words which do not match the relevant phonological characterization fit into any class.

Criterion 8. Those words that do not match the relevant phonological characterization but which nevertheless do fit in the classification fit on average in a smaller percentage of classes, than those words which do match the phonological characterization.

Criterion 9. Any class in a Phonosemantic Classification can be defined narrowly enough that words not matching the relevant phonological characterization are excluded from it.

· Criteria 7 and 8 There are 295 monosyllables beginning with /b/ and 148 beginning with /l/ in my monosyllabic vocabulary. There were 4 words beginning with /b/ which did not fit in the Phonosemantic Classification designed for /b/, and all of these fit in the Concrete Noun classification. There are 4 words beginning with /l/ which do not fit in the Phonosemantic Classification designed specifically for /l/, and all of these were Concrete Nouns. Once again, all the exceptions to the Phonosemantic Classification are Concrete Nouns as evidence that the salience of iconic meaning in a word is related inversely to the concreteness of its reference. There were, however, 77 words beginning with /l/ which did not fit in any superclasses designed for /b/. Of these, 23 fit in the Concrete Noun classification. In addition, 24 words beginning with /l/ did not fit in any superclass designed for /b/. There are therefore 101 /l/-words -- or 68% -- that fit in no /b/ phonestheme.

Let me provide an example to clarify what I mean by /l/-words that fit in the /b/ superclasses, but not the /b/ phonesthemes. There is a /b/ superclass for explosions and breakage. There is also a /b/ phonestheme for verbs of breakage. The phoneme /l/ starts two words -- lance and lathe -- which are not verbs of cutting like the /b/ words, but nouns referring to tools or weapons which cut. These words therefore fit in the natural 'superclass' of breakage and cutting, but not in the particular phonestheme to which words beginning with /b/ are confined. Thus 'lance' and 'lathe' constitute 2 of these 24. Similarly, the verb 'leak' is a verb related to breakage, but unlike all the /b/ verbs, it is not itself a type of breaking or cutting. Words beginning with /b/ fit in the /b/-based classification an average of 3.3. times. Words beginning with /l/ fit in the /l/-based classification an average of 3.2. times. Words beginning with /l/ fit in the /b/-based classification an average of 3/10 times, or one tenth as frequently.

The example given above is typical of what happens. Notice that although most of the bulging, circular and bumpy words beginning with /b/ do contain an /l/, very few words which actually begin with /l/ refer to anything bulging, circular or bumpy. Most of the /l/-words that are in this bulging class contain either an /b/ or a /p/. The one counter-example is 'lung', which happens to be a Concrete Noun.

· Criterion 9 Notice that the /l/-words which do fit in this classification fit differently. This is easiest to see in comparing the word 'loop' with the circular words beginning with /b/. The 'loop' is a ring with a hole in the middle. The circular /b/ words that are hollow are 3-dimensional (and coincidentally contain an /l/). This 'loop' shape is typical of words containing /l/ in conjunction with other sounds beside /b/: leap, lip, lob and lop all involve that same 'loop'-shaped motion, though they are not themselves circles. (Leap, lip, lob and lop do not fit in this particular phonestheme with 'loop', because this phonestheme was reserved for nouns, not that there are any /b/ words involving circular motion.) When /b/ occurs before the /l/ in a 'bulging' word, the word tends to refer to some membrane or cover which is pushed outward from within by air or water pressure: ball, bloat, boil, bulge. Exceptions are blip, blob, bulb and boll. Three of four of these end in a labial stop and are semantically similar to 'lobe' in that they do not imply air or water pressure. The last two of these (bulb and boll) are Concrete Nouns.

· Criterion 9 One can see this loop-shape also when /l/ is not in initial position: claw, coil, curl, fleece, flounce, fold, kilt, plait, pleat, reel, roll, scroll, sleeve, sling, spool, swirl, twirl, whirl, whorl. It is less obvious in comparing 'lobe' and 'lump' with the 'bumpy' /b/-words. The words 'lump' and 'bump' both fall in the same Natural Class. What then is, after all, the difference between words like 'lump' and 'bump' and how do we learn that difference?

Because there is so little which distinguishes 'lump' and 'bump' on the classificational and referential levels, then if I am at all correct in my hypotheses regarding word semantics, much of the semantic difference between the two can be attributed to True Iconic meaning differences, that is, to the unmediated effect that the phoneme /l/ vs. /b/ has on the semantics of these words. Every English speaker subconsciously recognizes the difference between a 'bump' and a 'lump'. A 'bump' is harder, more immobile and attached to the surface. A lump is moister, softer, more mobile and tends to be below the skin or in the cookie dough. These tendencies toward moistness, mobility and softness are quite generally typical of /l/ vs. /b/. It therefore sounds strange to talk of a 'bump' in the cookie dough or a 'lump' in the road.

· Concrete Nouns: It is typical that only 4 Concrete Nouns beginning with /l/ do not fit in the Phonosemantic Classification designed for /l/, but 23 Concrete Nouns beginning with /l/ do not fit in the classification designed for /b/. A couple of examples may make clear why this is so. The animal 'leech' fits in an /l/ phonestheme of sucking and slurping, but not in any /b/ phonesthemes. 'Lamb' fits in an /l/ phonestheme for gentle things, but not in any /b/ class. 'Leaf' fits in an /l/ phonestheme for flat things, but not in any /b/ phonestheme... and so forth.

· Concrete Nouns: There is a considerable number of /l/ words which neither fit in the Concrete Noun classes nor in any of the /b/ superclasses. These do, however, fit in classes typical of /l/:

Little: least, less, light, lint

Lead, Late, Follow: last, late, lax, lead, left, lest, lorn

Land: land, lawn, lea, loam

Fall, Lay: land, lay, lean, leap, lie, log, low, lug

Launch: leap, lunge, lurch

Lazy/Limp: limp, lithe, loaf, loll, lop, lounge

Lift: leap, lift, lob, loft

To the Side: lean, left, limp

Flat: lawn, lay, leaf, ledge, lie

Get, Eat, Take, See: lap, learn, leech, leer, lick, lunch

Run/ Walk/ Jump: leap, lick, lilt, look, lope, lug

Long: lane, limb, line

Happy: life, lift, light

Attraction: like, love

Life: life, live, live

lewd, lie (fib), like (similar), loom, lunge

By looking at this classification, one can begin to see what types of things /l/ conveys that /b/ does not. The phoneme /l/ conveys elements of linearity, light, laziness and loving where /b/ conveys bumpiness, burdens, business and brutality.

 

---

---

4.5 Experiment 5 -- Monolingual Classification First by Semantic Domain, then by Phoneme -- Words of Motion on Foot

See Appendix V for full data and results. -- Results are also listed in full below.

4.5.1 Methodology

· Locate all the words in a language which fit some narrowly defined semantic characterization. Try to insure insofar as possible that all these words fit in the same natural subclass, so that their referents, part of speech, argument structure, semantic class, etc. differ as little as possible. This tends to be easier to do with words other than the Concrete Nouns.

· In this case, I have used the monosyllabic words in my English vocabulary which in at least one of their senses refer to motion necessarily on foot. In this case I have excluded many words which may be verbs of walking, running, etc. and included only those for which the movement must be with the feet. For example, the verb 'stalk' is omitted, because it is grammatical to say that one stalks someone in a car as well as on foot. All the verbs of departure beginning with /b/ are omitted, because although in most cases the departure can be on foot, the mode of leaving is not specified inherently in the word, and any means of transport is possible. However, verbs of stamping, hopping and dancing which are not verbs implying motion from one point to another, but which necessarily involve the feet are included.

· Classify these by common phonological traits and attempt to determine whether individual phonemes are contributing specific aspects of meaning.

---

4.5.2 Example

This is a small scale test, so I include the relevant data here in its entirety. 1 -- initial position, 2 -- second position, 3 - third position, F3 -- final position, F2 -- pre-final position, F1 -- 3rd from last position:

/H/, /z/, /Z/, /f/, /T/, /S/ -- no verbs of motion on foot contain these phonemes

/b/

1

Run, Jump - bound

/d/

1

Dance - dance

F3

Walk - plod, pound, stride, tread, wade, wend

Step - pound, tread

Run, Jump - bound

/g/

F3

Walk - slog

Run - jog

/p/

1

Walk - pace, plod, pound, prance

Step - pound

2

Run - sprint

Jump - spring

F3

Walk - tramp, trip, tromp, troop

Crawl - creep

Limp - limp

Step - stamp, step, stomp, tamp, tramp, tromp

Run - lope, romp, skip

Jump - hop, jump, leap, skip

/t/

1

Walk - tramp, tread, trek, trip, tromp, troop

Step - tamp

Run - trot

2

Walk - steal, stomp, stray, stride, stroll, strut

Step - stamp, step

F2

Walk - waltz

Dance - waltz

F3

Walk - strut

Run - sprint, trot

Skate, Ski - skate

/k/

1

Walk

Crawl - crawl, creep

2

Walk

Climb - scale

Run - skip

Skate, Ski - skate, ski

Jump - skip

F3

Walk - hike, trek

/v/

F3

Walk - rove

/s/

1

Walk - slog, steal, stray, stride, stroll, strut

Climb - scale

Step - stamp, step, stomp

Run - skip, sprint

Skate, Ski - skate, ski

Jump - skip, spring

F3

Walk - pace, prance, waltz

Step - prance, trounce

Dance - dance, waltz

/h/

1

Walk - hike

Jump - hop

/J/

1

Jump - jump

F3

Walk - trudge

Step - trudge

/C/

F3

Walk - march

/m/

1

Walk - march

F2

Walk - tramp, tromp

Limp - limp

Step - stamp, stomp, tamp, tramp, tromp

Run - romp

Jump - jump

F3

Walk - roam

/n/

F2

Walk - pound, prance, wend

Step - pound, trounce

Run - bound

Dance - dance

Jump - bound

/G/

F3

Run - spring

Jump - spring

/l/

1

Walk - limp

Run - lope

Jump - leap

2

Walk - plod

F1

Walk - waltz

Dance - waltz

F3

Walk - steal, stroll

Crawl - crawl

Climb - scale

/r/

1

Walk - roam, rove

Run - romp, run

2

Walk - prance, tramp, tread, trek, trip, tromp, troop, trounce, trudge

Crawl - crawl, creep

Step - tramp, tread

Run - trot

3

Walk - stray, stride, stroll, strut

Run - spring, sprint

Jump - spring

F2

Walk - march

/w/

1

Walk - wade, walk, waltz, wend

Dance - waltz

F2

Walk - rove, stroll, troop, trounce

Step - trounce

/j/

F2

Walk - hike

Crawl - creep, leap, steal, stride

Climb - scale

Jump - leap

Skate, Ski - skate

F3

Walk - stray

Skate, Ski - ski

---

4.5.3 Discussion of Findings

None of the words in this experiment have concrete reference, so this test provides no evidence for or against the inverse relationship of concreteness to the salience of iconic meaning. The experiment also provides only indirect evidence for general character of Phonosemantic Association. However it provides direct evidence for Iconism proper. It does not do much toward affirming the criteria for the Phonosemantic Classification, which