What is language? The side of sound: Phonetics and phonology and the beginning of morphology.

What is language?The side of sound:

Phonetics and phonologyand the beginning of

morphology

Linguistics

• The study of language may treat a language as a self-contained system; or it may treat it as an object that varies over space, time, and social class.

• We will consider only the first (and ignore diachronic linguistics and sociolinguistics).

Another distinction to bear in mind

• We can study the way in which language organizes thought and expresses statements about (perceived) reality; or,

• We can study the internal structure of language systems.

I'll focus on the second. We will aim to determine the distribution of items in particular languages; and to establish any universal principles that can be extracted from those to simplify the entire process.

Language

Perceived reality

Bear in mind...

…that English is an outlier among languages...

Language

English

(This is really because

almost all of the volume

of a hypersphere becomes

arbitrarily close to the skin, as the

dimensionality increases – so to

speak)

• We humans manage to analyze an extremely complex acoustic signal and translate it into an internal representation linked to meaning with little conscious awareness of the intermediate steps or the complexity of the operation.

Linguistics

Phonetics: sound, described as an acoustic and articulatory eventPhonology: the study of systems of discrete soundsMorphology: ... the internal structure of wordsSyntax: ...the principles governing combinations of words.Semantics:...the relationship between syntactic structures and meaning.

Some false statements about speech

• The speech stream can be divided into words on the basis of short pauses between words.

• Words can be analyzed as a sequence of phones of roughly equal length.

• Words can be analyzed as a sequence of syllables of roughly equal length.

• Words (syllables) have the same duration regardless of their context.

• Words of two syllables are longer than words of one syllable.

• dad 520 msec• daddy 420 msec.

Language is fast!

• Individual sounds can go by extremely fast (40 to 200 msec) and yet be easily grasped by the native speaker. There’s nothing else that I know of that we can do anywhere near that fast that appears to be under conscious control.

• Native speakers reconstruct sounds from extremely degraded sensory input:– Jeetjet? Nah, juw?

Fast? The word text: the k is 40 msec out of a total of 480 msec.

VowelS

t: closure + burstK

But what is language?

• A system of great complexity• Much of the complexity is learned

(we know that, because it is “language-specifïc”)

• It still eludes our attempts to accurately model it on computers (witness continuous speech recognition products).

1. Phonetics

1. Phonetics• We know more about how sound

is produced than how it is perceived, generally speaking.

• Source-filter model: Upon exhilation, the vocal cords vibrate freely if there is little blockage or obstruction through the mouth and nose. The frequency of that vibration is the fundamental frequency (50-200hz in males, double that in females).

Articulatory apparatussource: Kevin Russell

Vowels

• For vowels, the mouth/nose acts as an echo chamber, enhancing those harmonics that resonate there.

• These resonances are called formants. The first 2 formants are especially important in characterizing particular vowels.

“Hi” /haj/

we were away a year ago FORMANTS

/i/ green

/ae/ hat

/u/ boot

graphics thanks to Kevin Russell, Univ of Manitoba

Vowels, crudely…

• To identify a vowel is to identify its location in a 2-dimensional F1-F2 space.

Improvements:• … in 3-dimensional F1-F2-F3 space• …normalized by

3/1321 )( FFFF

See e.g. Harvey Sussman, The Neurogenesis of PhonologyPhonological processes and brain mechanisms 1988

Consonants

• Stops: p, t, k, b, d, g• Fricatives, affricates: ch, j, sh, th...• Nasals: m, n, ng (as in sing)• Stops and fricatives create their own

turbulence, and the oral shape determines what spectrum is enhanced.

Spectral character of sounds

• Stops show rapid change of formant frequency from their position to that of the neighboring vowel;

• Fricatives should wide band of noise• Vowels show 3 (major) bands of

formants whose energy is an enhancement of harmonics of the fundamental frequency (1st, 2nd, 3rd formant)

3 aspects of the signalThe linguistic signal can be divided into

three parts:• The fundamental frequency (intonation in

many languages, tone in others)• The cues to the oral gestures: energy

and formant structure: vowel and consonants

• Temporal (rhythmic) structure Big point:

Simultaneous and co-organized analysis of these aspects

Fundamental Frequency

• Intonation languages• Tone languages -- we’ll get to them

Cues to oral gestures: formants, formant changes, and spread-

out noise• Formants for vowels; • Long pauses inside of stops, followed

by rapid formant transitions to the following vowel

• Spread out regions of noise for fricatives

We recognize as many as 10 consecutive “objects” per second!

Rhythm and timing

• Japanese: based on moras (‘haku’ in Japanese )

• A mora is: • a CV: ka zo ku ‘family’ wa ta shi ‘I’ • the V in CVV: ko-o-ko-o ‘high school’• ŋ at end of syllable: o-ba-a-sa-n

‘grandmother’

• C at end of syllable: cho - t - to ‘a little’The length of individual moras varies greatly in

duration. BUT -- the length of an entire word varies linearly with the number of moras!

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1 2 3 4 5 6

Japanese

English

Japanese

English

Japanese Moras

English Syllables

2 Phonology

Main points

• Phonology of a language imposes highly and tightly structured organization.

• Languages differ greatly from one another, but there are many deep generalizations relating them. (That is, the range of possible phonologies is large; but the range is also much smaller than it might be logically.)

• The main principle to bear in mind is simultaneous signals and simultaneous constraints.

Sounds and sound inventories

1. The phonemic principle in languages2. Categorization into vowels and

consonants3. More refined analysis along sonority

hierarchy4. Strong universal (anthropophonic)

tendencies in selection of vowel and consonant inventories

5. Strong symmetry tendencies: which means that sounds are composed of parts...

1. The phonemic principle

• Humans perceive sound chunks (“phonemes”) in discrete categories; hence ability to discriminate between exemplars is extremely good at the boundary between phonemes, and poor for within-category cases.

Example

• Difference between /b/ and /p/ is voicing, realized phonetically as Voice-Onset Time

50 msec

“voiceless”“voiced”

Voice Onset Time:length of time between

opening the mouth and theonset of vocal fold

vibrations

-10

0

10

20

30

40

50

60

70

/bi/ /di/ /gi/

Vo

ice

On

set

Tim

e (m

sec)

-10

0

10

20

30

40

50

60

70

/bi/ /di/ /gi/

Vo

ice

On

set

Tim

e (m

sec

)

http://convention.asha.org/2005/handouts/293_McCrea_Christopher_073466_112805070408.ppt

Three different F0s

Each language has its own inventory of phonemes

• English distinguishes /b/ from /v/• Spanish does not

3. Sonority hierarchy

• Vowels a > i, u• Liquids: l, r• Nasals: n, m , ŋ (angma)• Fricatives: s, f, v, z, θ, …

h• Affricates: č [US], ʤ

[IPA]• Stops: b,d,g…p,t,k

SONORANTS

OBSTRUENTS

Sonority plays a very important role in determining what

sequences of sounds are permissible in a language

• It’s not the case that a word is just a sequence of sounds permitted in a language.

• The set of permissible sequences is much smaller than the set of imaginable sequences...

Syllables

• Words are sequences of permissible syllables, and in general,

• Syllables are waves of sonority:

peak: the vowel

increasing sonority

decreasing sonority

Syllables• The most basic syllable structure: CV• Most languages put very heavy

restrictions on what consonants can appear after the vowel, in the coda:

syllable

onset rhyme

s p r nucleus coda

I n t

English syllable

b l a c k is OK, but

l b a c k:

l b a ckNot a permissible

sonority sequence

b u m p is OK, but b u p m is not.

b u p m

Competition for sonority...

All phonemes must be organized into syllables;

an segment will ‘capture’ a less sonorous segment on its immediate left.

Limitations on the syllable

Many languages permit no more than three items in a syllable:

Consonant + Vowel + 1 thing• C V• C V V• C V C

Strong symmetry tendencies...

labial coronal velar

p t k

b d f

m n ŋ

Fundamental domains of phonology

• Theory of gestures (actions)• Theory of rhythm• Theory of information (contrasts,

redundancies)• Theory of audition: not much here

Prime effect: synchronization

• Speech is not the linear concatenation of atomic units (phonemes);

• It is the organization over time of units on a large number (~15) of independent tiers

• Just like the production of an orchestra: each instrument’s production is autonomous vis-à-vis the other instruments...

Orchestral score

• One instrument may be silent for a while;

• Another may play 2 notes over the same period that a third plays only 1 note;

• But all the instruments are locked onto a over-all guidance metronome-- the conductor’s baton

“pin” [pIn]

• A labial gesture aligned with glottal widening;

• A rise of the tongue body combined with a narrowing of the glottis, leading eventually to spontaneous vibration of the vocal folds

• a raising of the tongue to the top of the mouth together with a drop of the velum to permit air to flow through the nose.

Let’s focus on tone

• First, a simple system like that of English!

• English assigns a tonal melody such as HL or LH to certain specific syllables.

• This melody is then stretched out or squeezed into the time available, given the syllables of the utterance...

Question versus statement

• Did you go to the store?

I went to the store yesterday.

I went to the store yesterday.

But wh-questions are different:

• When did you go to the store?

When did you go to the store?

Hence:

• The High-Low melody is a “thing” in itself -- an intonational melody -- but to understand the sentence, you must know how it lines up with the words.

• Tone and words: separate, autonomous, interbraided. Good word: symplectic structure.

Flap (D) in American EnglishWe find the flap of water (wa[D]er) under these

conditions strictly inside a word:

Followingvowelstressed

Followingvowelunstressed

Precedingvowelstressed

rare orimpossibleBeethoven

obligatoryatom

Precedingvowelunstressed

impossibleattire,atomic

optional:sanity

But across words:

• Word initial t never flaps, regardless of stresses before or after*; eat my tomato, see Topeka...

• Word-final t followed by a vowel-initial word normally does flap, regardless of stresses before or after. at all, sit on it...

*But in the words to, tonight, today, tomorrow, the to acts as if it were linked to the preceding word.

Generalization• English permits phonemes to

belong simultaneously to two syllables ( = be ambisyllabic) under certain conditions.

• Ambisyllabic t's convert to flaps.Generally speaking:Within a word: • C becomes part of syllable with a

following onset ("maximize syllable onset"):

...within a word:

C V

This also applies across words --in English, and in many languages,

but not (e.g.) in German

V C [#

Within a word, ambisyllabification before

an unstressed vowel

V VC

-stress+stress

e.g., atom

But not across word boundaries

we don't say my tomato my [D]omato

Linguistics

Phonetics: sound, described as an acoustic and articulatory eventPhonology: the study of systems of discrete soundsMorphology: ... the internal structure of wordsSyntax: ...the principles governing combinations of words.Semantics:...the relationship between syntactic structures and meaning.

Classic distinctions in morphology:

Analytic (isolating) languages:– no morphology of derivational or

inflectional sort.

Synthetic (inflecting) languages:– Agglutinative: 1 function per morpheme,

clear divisions between morphemes– Fusional: more than 1 function per

morpheme, and/or segmentation into morphemes is uncertain

Agglutinative:Finnish Nominal Declension

talo 'the-house' kaup-pa 'the-shop'

talo-ni 'my house' kaup-pa-ni 'my shop'

talo-ssa 'in the-house' kaup-a-ssa 'in the-shop'

talo-ssa-ni 'in my house’ kaup-a-ssa-ni 'in my shop'

talo-i-ssa 'in the-houses’ kaup-o-i-ssa 'in the-shops'

talo-i-ssa-ni 'in my houses’ kaup-o-i-ssa-ni 'in my shops' Courtesy of Bucknell Univ. web page

Fusional: LatinLatin Declension of hortus 'garden'

Singular Plural

Nominative (Subject) hort-us hort-i

Genitive (of) hort-i hort-rum

Dative (for/to) hort-o hort-is

Accusative (Direct Obj) hort-um hort-us

Vocative (Call) hort-e hort-i

Ablative (from/with) hort-o hort-is

Incorporation: syntax inside morphology

let’s night-long ball play

“let’s play ball all night long”

example thanks to Bucknell web page

Morphology

The internal structure of words:• phonological characteristics• interaction with syntax

Most words in most languages are composed of

several morphemes

• A verb from Tonga (Bantu):tu - la - ba - bon – a

we Present them see - indicative.

English

• Has a morphology most similar to that of the other Germanic languages; and Germanic languages are similar to other Indo-European languages.

• A small number of prefixes, quite a few suffixes.

Inflectional and derivational morphology

I sing, you sing, he sings, we sang; all different inflectional forms of the same lexeme sing. Different forms are required in different contexts (syntactic contexts, we say).

Derivational morphology:I read; I am a reader. You out-wit me. You're

a mind-reader.derivational morphology: creates new

lexemes.

• That's one functional dimension along which morphemes differ; they also differ with regard to the effects they have on the stem to which they attach.

• Some affixes leave the base unchanged, while in other cases, the base + affix is modified so as to better satisfy the phonotactics (=well-formedness conditions, high-frequency patterns) of monomorphemic lexemes.

Two layers of morphology in English

Layer 1: causes change in stress pattern, vowel-shortening, vowel-deletion, c->s; Layer 2: little change in the base.Catholicism: the religion; versusCatholic-ism: speech forms particular to

Catholics?

Linguistics

Phonetics: sound, described as an acoustic and articulatory eventPhonology: the study of systems of discrete soundsMorphology: ... the internal structure of wordsSyntax: ...the principles governing combinations of words.Semantics:...the relationship between syntactic structures and meaning.