25
ASPECTS OF CONNECTED ASPECTS OF CONNECTED SPEECH SPEECH

ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Embed Size (px)

Citation preview

Page 1: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

ASPECTS OF ASPECTS OF CONNECTED SPEECHCONNECTED SPEECH

Page 2: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

““Tale of a machine”Tale of a machine”

Scientists tried to develop a machine that Scientists tried to develop a machine that would produce speech from a vocabulary would produce speech from a vocabulary of pre-recorded wordsof pre-recorded words

Result: negative (limited, at best)Result: negative (limited, at best)Reason: ?Reason: ?

-differences between human -differences between human speech and “mechanical speech”speech and “mechanical speech”

Page 3: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

RHYTHMRHYTHMNoticeable event happening at regular Noticeable event happening at regular

intervals of time (heart-beat, flashing light, intervals of time (heart-beat, flashing light, piece of music)piece of music)

Rhythm in speech: regular occurrence of Rhythm in speech: regular occurrence of stressed syllables (still, regularity is relative)stressed syllables (still, regularity is relative)

English speech is rhythmicalEnglish speech is rhythmicalStress-timed rhythm: stressed syllables tend Stress-timed rhythm: stressed syllables tend

to occur at relatively regular intervals to occur at relatively regular intervals whether they are separated by unstressed whether they are separated by unstressed syllables or not (not characteristic for syllables or not (not characteristic for “mechanical speech”)“mechanical speech”)

Page 4: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Stress-timed rhythm theoryStress-timed rhythm theory

The time from each stressed syllable to The time from each stressed syllable to the next will tend to be the same, the next will tend to be the same, irrespective of the number of intervening irrespective of the number of intervening unstressed syllablesunstressed syllables

Page 5: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Syllable-timed rhythm theorySyllable-timed rhythm theory

All syllables, whether stressed or All syllables, whether stressed or unstressed, tend to occur at regular unstressed, tend to occur at regular intervals and the time between stressed intervals and the time between stressed syllables will be shorter or longer in syllables will be shorter or longer in proportion to the number of unstressed proportion to the number of unstressed syllables syllables

Japanese, Hungarian, Spanish are Japanese, Hungarian, Spanish are examples of such languagesexamples of such languages

Page 6: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Foot Foot

A unit of rhythm A unit of rhythm Begins with a stressed syllable and Begins with a stressed syllable and

includes all following unstressed syllables includes all following unstressed syllables up to, but not including the following up to, but not including the following stressed syllablestressed syllable

Page 7: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Tree diagramTree diagram

ss ww s w s w

twen tytwen ty pla cespla ces

ww s s

s w s ws w s wtwen ty pla cestwen ty pla ces

Page 8: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Longer phraseLonger phrase

w sw s

w s w s w s w s

s w s w s w ss w s w s w s twent ty pla ces fur ther backtwent ty pla ces fur ther back

Page 9: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

How strong is a syllable?How strong is a syllable?Strength of a syllable can be measured by Strength of a syllable can be measured by

counting up counting up the number of timesthe number of times the the symbol “symbol “ss” occurs above it” occurs above it

““metrical gridsmetrical grids” are correct for very slow ” are correct for very slow speech (“right-hand foot rule” is applied)speech (“right-hand foot rule” is applied)

EnglishEnglish tends towards a tends towards a regular alteration regular alteration of strong and weak syllablesof strong and weak syllables and tends to and tends to adjust stress levelsadjust stress levels to make this possible to make this possible(“(“stress-shiftstress-shift”): e.g. compact /k”): e.g. compact /kәәmm''ppæækt/ kt/ BUT compact disc /BUT compact disc /''kkּפּפmpæktmpækt ''ddıskısk//

Page 10: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

How rhythmically do we speak?How rhythmically do we speak?

Very rhythmically – styles of public speechVery rhythmically – styles of public speechArhythmically – if hesitant or nervousArhythmically – if hesitant or nervousOne always speaks with some degree of One always speaks with some degree of

rhythmicality, but the difference is in rhythmicality, but the difference is in degreedegree

In a stress-timed language all the feet are In a stress-timed language all the feet are supposed to be of roughly the same supposed to be of roughly the same durationduration

Page 11: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

AssimilationAssimilationA feature of natural connected speech A feature of natural connected speech

onlyonlySounds belonging to one word can cause Sounds belonging to one word can cause

changes in sounds (phonemes) belonging changes in sounds (phonemes) belonging to neighbouring wordsto neighbouring words

More likely to be found in rapid, casual More likely to be found in rapid, casual speechspeech

The cases most often described are The cases most often described are assimilations affecting consonantsassimilations affecting consonants

Page 12: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Assimilation at word boundaryAssimilation at word boundary

- - - C- - - Cff CCii - - - - - -

wordword

boundaryboundary

Assimilation can be either Assimilation can be either regressiveregressive or or

progressiveprogressive

Page 13: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

3 types of assimilation:3 types of assimilation:1) assimilation of place of articulation1) assimilation of place of articulation

2) assimilation of manner of articulation2) assimilation of manner of articulation

3) assimilation of voicing3) assimilation of voicing

Page 14: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Assimilation of placeAssimilation of place The most common form involves the movement of

place of articulation of the alveolar stops /t/, /d/ and /n/ to a position closer to that of the following sound.

For instance, in the phrase ten cars, the /n/ will usually be articulated in a velar position, so that the tongue will be ready to produce the following velar sound /k/.

Similarly, in ten boys the /n/ will be produced in a bilabial position, /tem b/ to prepare for the articulation of the bilabial /b/.

This phenomenon is easy to find also in Italian: think of the different pronunciations of the ‘n’ in Gian Paolo, Gian Franco and Gian Carlo.

Page 15: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

BEFORE A VELAR (/k/, /g/)BEFORE A VELAR (/k/, /g/)

/n//n/ //ŋŋ//

e.g. ten cars e.g. ten cars

/d//d/ /g//g/

e.g.e.g. good girl //good girl //

/t//t/ /k//k/

e.g. that kide.g. that kid

Page 16: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

BEFORE A BEFORE A BILABIALBILABIAL (/ (/mm/, //, /bb//, /p/, /p/))

/n//n/ //mm//

e.g. ten men /tee.g. ten men /tem mm men/en/

/d//d/ //bb//

e.g. bae.g. bad bd boysoys

/t//t/ /p//p/

e.g. hoe.g. hot mt mushrooms /ushrooms / tha that pt person/ erson/ meameat pt pie/ lighie/ light bt bluelue

t + bilabial = pt + bilabial = pt + dental = dental plosive (e.g. that + dental = dental plosive (e.g. that tht thing / geing / get tht those / cuose / cut tht through)rough)t + velar = k (e.g. that case / bright colour / quit + velar = k (e.g. that case / bright colour / quite gte good)ood)d + similar contexts = b, d, gd + similar contexts = b, d, gn + similar contexts = m, n, n + similar contexts = m, n, ŋŋs and z + ∫ , j become ∫ and s and z + ∫ , j become ∫ and ζζ (e.g. this shoe, those years) (e.g. this shoe, those years)

Page 17: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Assimilation of mannerAssimilation of mannerThe change is most likely to be towards an The change is most likely to be towards an

easier consonant – one that makes less easier consonant – one that makes less obstruction to the airflowobstruction to the airflow

Final plosive becomes a fricative or nasal Final plosive becomes a fricative or nasal (tha(that st side, gooide, good nd night)ight)

A plosive / nasal + word-initial /A plosive / nasal + word-initial /ðð/ : i/ : in thn the, e,

geget tht them, reaem, read thd these (consonant ese (consonant ðð becomes identical in manner to the cons. becomes identical in manner to the cons. final, but with dental place of articulation) final, but with dental place of articulation)

Page 18: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Assimilation of voicingAssimilation of voicingThe vibration of the vocal folds is not something

that can be switched on and off very swiftly, as a result groups of consonants tend to be either all voiced or all voiceless. Consider the different endings of ‘dogs’ /dּכgz/ and ‘cats’ /kæts/, of the past forms of the regular verbs such as ‘kissed’ /kıst/ and ‘sneezed’ /sni:zd/.

Only regressive assim. of voicing is found:cons.final voiced + cons.initial voiceless devoices

cons.final: e.g. get them /get tәm/Progressive assim. of voicing with suffixes s/z

Page 19: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Elision is very simply the omission of certain sounds in certain contexts. The most important occurrences of this

phenomenon regard: 

1     Alveolar consonants /t/ and /d/ when ‘sandwiched’ between two consonants (CONS – t/d – CONS), e.g.

The neThe next dxt day….ay….

The laThe last cst carar……

HoHold thld the dog!e dog!

SeSend Fnd Frank a card.rank a card.

Page 20: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Cases of elisionCases of elision

1)1) Loss of weak vowel after p/t/k: potato, tomato, Loss of weak vowel after p/t/k: potato, tomato, perhaps, etc.perhaps, etc.

2)2) Weak vowel + n/l/r = syllabic n / l / r : tonight, Weak vowel + n/l/r = syllabic n / l / r : tonight, police, correct, etc.police, correct, etc.

3)3) Avoidance of complex cons. clusters:Avoidance of complex cons. clusters:

““George the SiGeorge the Sixth´s thrxth´s throne”one”

4)4) 2/3 plosives + a fricative : the middle plosive 2/3 plosives + a fricative : the middle plosive disappears: acts, looked back, scripts, etc.disappears: acts, looked back, scripts, etc.

5)5) Loss of final /v/ in Loss of final /v/ in of of before consonants: lots before consonants: lots of of them, waste them, waste ofof money, etc. money, etc.

Page 21: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Elision = disappearance of sounds Elision = disappearance of sounds

This can also take place within affricates /t∫/ and /dζ/ when preceded by a consonant, e.g.

lunchtime  strange days

A phoneme may have zero realisation or be deleted.

Page 22: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Are Are contractionscontractions a case of elision? a case of elision?

Had, would : “´d”Had, would : “´d” Is, has : “´s”Is, has : “´s”Will : “´ll”Will : “´ll”Have : “´ve”Have : “´ve”Not : “n´t”Not : “n´t”Are : “´re”Are : “´re”

Page 23: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Linking (Liaison)Linking (Liaison)Words in English can be linked in special Words in English can be linked in special

ways: “linking r” and “intrusive r”ways: “linking r” and “intrusive r”Linking r: “here are” vs. “here”Linking r: “here are” vs. “here”

“ “four eggs” vs. “four”four eggs” vs. “four”

“ “law and order” vs. “law” law and order” vs. “law” Intrusive r: “Formula A”Intrusive r: “Formula A”

“ “media event”media event”

These are special cases of These are special cases of juncturejuncture..

Page 24: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words

Juncture Juncture Refers to the relationship between one sound Refers to the relationship between one sound

and the sound/s that immediately precede it or and the sound/s that immediately precede it or follow itfollow it

E.g. “my turn” vs. “might earn”E.g. “my turn” vs. “might earn”Example of Example of close junctureclose juncture: m + ai: m + aiExternal open junctureExternal open juncture: m, n are in such posit.: m, n are in such posit. Internal open junctureInternal open juncture: how are the two above : how are the two above

mentioned examples differentiated?mentioned examples differentiated?Answer: due to aspiration or non-aspir. of /t/ Answer: due to aspiration or non-aspir. of /t/

thanks to its position at word boundarythanks to its position at word boundary

Page 25: ASPECTS OF CONNECTED SPEECH. “Tale of a machine” Scientists tried to develop a machine that would produce speech from a vocabulary of pre-recorded words