MORPHOLOGY

Morphology
Class 4
Morphemes and their
properties II: fusion
FS 2014
Rik van Gijn
Goal of this class
This class is about the way morphology and phonology interact, and
you will learn more about non-concatenative morphology
√ Allomorphy (phonologically conditioned)
√ Morphophonology
√ Phonological word
√ Non-concatenative or non-linear morphology
Intermezzo: the International Phonetic Alphabet (consonants)
Voicing
Voicing is caused
by the vibration of the
vocal cords.
Manner
How the speech organs
are configured to
manipulate the air
stream, by constricting
the airflow in various
ways.
Place
The point in the vocal tract where two or more
articulators block or constrict the air flow.
Intermezzo: the International Phonetic Alphabet (consonants)
Speech organs (Arcadian)
http://training.seer.cancer.gov/headneck/anatomy/overview.html
Places of articulation (Instituto Lingüístico de
Verano
www-01.sil.org/mexico/ling/glosario/E005ciPlacesArt.htm)
Intermezzo: the International Phonetic Alphabet (vowels)
Rounding
Tongue body position II
Lips can be rounded to produce
rounded vowels.
The body of the tongue is pushed forward to
produce front vowels or backward to produce
back vowels.
Tongue body position I
Raising the body of the tongue
while letting the airflow out freely
produces high (close) vowels,
whereas pushing the body of the
tongue down produces low (open)
vowels.
Morphophonology
Languages have different types of phonological rules that apply to
certain domains. To see this, look at the following data from Yurakaré
lëtta kajun
tijajun
[læt:a kahun]
[tihahun]
‘one box’
‘my box’
tinkama
ajama
[tiŋkama]
[ahama]
‘He calls me.’
‘He is calling.’
Morphophonology
Languages have different types of phonological rules that apply to
certain domains. To see this, look at the following data from Yurakaré
lëtta kajun
tijajun
[læt:a kahun]
[tihahun]
‘one box’
‘my box’
tinkama
ajama
[tiŋkama]
[ahama]
‘He calls me.’
‘He is calling.’
Rule in Yurakaré:
If an underlying /k/ appears after a vowel, it is
changed to [h] within a word
/k/ → [h] / [V_]ω
Types of phonological rules
Minimality constraints: Yurakaré
püü
tapü
/pɨɨ/
/tapɨ/
‘road’
‘our road’
paa
tapa
/paa/
/tapa/
‘brother’
‘our brother’
too
tato
/too/
/tato/
‘bone’
‘our bone’
Types of phonological rules
Syllable structure
There are many differences
between languages in what
they allow in terms of syllable
structure.
Major constraints are:
No coda allowed (or only some sounds)
No complexity allowed
Types of phonological rules
Phonotactic constraints: epenthesis in Yuki (Tupí-Guaraní)
a-be-akiw
1SG-CAU-warm
‘I am warming up.’
[abeɾakiw]
o-ye-ire
3S-REF-wash
‘He washes himself.’
[ojejire]
Villafañe Lucrecia (2004) Gramática Yuki.
PhD thesis RU Nijmegen, p. 35
Types of phonological rules
Phonotactic constraints: epenthesis in Lenakel
rva
3sg
come
He comes = rɨva
Bickel & Nichols (2007) Inflectional morphology. In:
Shopen (ed. ) Language typology and syntactic
description III, p. 182
Types of phonological rules
Phonotactic constraints: vowel elision in Emérillon (Tupí-Guaraní)
o-paʔam-oŋ
ikiʔɨ
3.I-stand-PL.S
now
‘They are standing now.’
o-boʔi
baipuri-r-ɨe-ŋ
3.I-cut
tapir-REL-stomach-PL.S
‘They cut open the tapir’s stomach.’
Word-internal elision of sounds is also
called syncope.
Rose, Françoise (2011): Grammaire de l'émérillon teko, une
langue tupi-guarani de Guyane française. Louvain: Peeters.
175
Types of phonological rules
Consonant coalescence in Suruí (Tupí-Mondé)
/òn+pór/ → [òmor]
/èn+kánè → [èŋámè]
‘my brother’
‘(Somebody) wants you.’
Van der Meer 1982: Fonologia da língua Suruí. São Paolo ,
PhD thesis Campinas, p.41
Allomorphs and affix competition
These interactions between phonology and morphology create
allomorphs.
Allomorphs
Lexically
conditioned
See flexivity
class
Phonologically
conditioned
Non-transparently
related
Transparently
related
Competing
affixes
Morphs
Transparently related phonologically conditioned allomorphs
English regular plurals
The Past Tense Rule
a.
b.
If the verb stem ends in [t] or [d] (the
alveolar stops), insert [ə] before the past
tense morpheme (e.g. defeated [dəfit +
d] → [dəfit + əd]).
Assimilate [d] to the voicing of an
immediately preceding consonant (e.g.,
licked [lɪk + d] →[lɪk + t])
Lieber (2009)
Transparently related phonologically conditioned allomorphs
English regular plurals
Past Tense English
-d
-ed
-t
morph
morph
morph
...
morph
Allomorphs of past
tense morpheme
Transparently related phonologically conditioned allomorphs
English plurals
Infinitive Irregular past Pattern
1
2
3
4
5
6
7
8
9
10
burn
keep
hit
feel
bleed
leave
sing
win
fight
come
Lieber (2009)
burnt
kept
hit
felt
bled
left
sang
won
fought
came
devoicing of suffix
vowel shortening
no change
vowel shortening with devoicing of suffix
vowel shortening and no suffix
devoicing of stem consonant
vowel ablaut (ɪ > æ)
vowel ablaut (ɪ > ʌ)
vowel ablaut (ai > ɔ)
vowel ablaut (ʌ > e)
Transparently related phonologically conditioned allomorphs
Dutch past tense suffixes
Booij, Geert (1999) The Phonology of Dutch.
Oxford: OUP, p. 61
Transparently related phonologically conditioned allomorphs
Turkish: Lewis (1967) in Lieber (2009, p. 162)
Abs. pl.
Gen. sg.
‘hand’ ‘measure’
el-ler ölçü-ler
el-in
ölçü-n-ün
‘evening’
akşam-lar
akşam-ɩn
‘fear’
korku-lar
korku-n-un
1. What are the different realizations (allomorphs) of the absolutive
plural marker and the genitive singular marker?
2. Is there a default form?
3. Can you think of a rule to describe the alternations?
<ü> = /y/
<ɩ> = /ɯ/
<ö> = /ø/
Transparently related phonologically conditioned allomorphs
Turkish: Lewis (1967) in Lieber (p. 162)
Abs. pl.
Gen. sg.
‘hand’ ‘measure’
el-ler ölçü-ler
el-in
ölçü-n-ün
‘evening’
akşam-lar
akşam-ɩn
‘fear’
korku-lar
korku-n-un
1. What are the different realizations (allomorphs) of the absolutive
plural marker and the genitive singular marker?
2. Is there a default form?
3. Can you think of a rule to describe the alternations?
<ü> = /y/
<ɩ> = /ɯ/
<ö> = /ø/
All non-high vowels have to agree in
backness with their base
Transparently related phonologically conditioned allomorphs
Belhare dissimilation
khat
-yu
kha t
+coronal
‘go’
‘non-past’
khaʔyu ‘s/he goes’
yu
+coronal
Bickel & Nichols (2007) Inflectional morphology. In:
Shopen (ed. ) Language typology and syntactic
description III, p. 182
Transparently related phonologically conditioned allomorphs
Belhare dissimilation
‘go’
‘non-past’
khat
-yu
kha t
+coronal
khaʔyu ‘s/he goes’
yu
+coronal
khaʔyu
Bickel & Nichols (2007) Inflectional morphology. In:
Shopen (ed. ) Language typology and syntactic
description III, p. 182
Non-transparently related phonologically conditioned allomorphs
Dutch diminutives (see data sheet)
Group a: -je
/lɪpjə/
Group b: -etje
/bɔmətjə/
Group c: -pje
/bodəmpjə/
Group d: -kje
/koniŋkjə/
Group e: -tje
/retjə/
Booij, Geert (1999) The Phonology of Dutch.
Oxford: OUP, p. 61
Non-transparently related phonologically conditioned allomorphs
Dutch diminutives (see data sheet)
Group a: -je
/lɪpjə/
Default -tje /tjə/
1
Group b: -etje
/bɔmətjə/
Group c: -pje
/bodəmpjə/
Group d: -kje
/koniŋkjə/
2
Group e: -tje
/retjə/
3
Non-transparently related phonologically conditioned allomorphs
English affix competition: -ize versus -ify
Based on Lieber (2009), p. 2
Speakers of English use the suffixes -ize (crystallize) and -ify (codify) to form
verbs from nouns. If you had to form a verb that means ‘do something the
way ex-Prime Minister Tony Blair does it’, which suffix would you use? How
about a verb meaning ‘do something the way ex-President Bill Clinton does
it’? What about former French president Chirac?
Non-transparently related phonologically conditioned allomorphs
English affix competition: -ize versus -ify
Based on Lieber (2009), p. 2
Speakers of English use the suffixes -ize (crystallize) and -ify (codify) to form
verbs from nouns. If you had to form a verb that means ‘do something the
way ex-Prime Minister Tony Blair does it’, which suffix would you use? How
about a verb meaning ‘do something the way ex-President Bill Clinton does
it’? What about former French president Chirac?
[...σ́]
[...σ]
→
→
-ify
-ize
Allomorphs and affix competition
Some researchers prefer to restrict the term allomorph to transparently
(in terms of form) related morphs.
Phonologically
conditioned
Non-transparently
related
Transparently
related
Competing
affixes
Allomorphs
An alternative term for
alternative realizations of
morphemes that are not
transparently related to each
other might then be called
competing affixes. Some
people talk about affix
suppletion.
We will use the terms
allomorph and morph in a
neutral way over all contexts.
Phonological word
Many languages seem to have a recurrent domain for certain rules that
often coincides with the grammatical word.
>> the phonological word
Whether or not the phonological word is a unit that can robustly be
shown to exist and be relevant in all languages of the world is still very
much a matter of debate and ultimately an empirical question.
Alignment
Three types of situations
g-word < p-word
g-word > p-word
g-word = p-word
The case of Yidiny: g-word > p-word
Alignment
The case of Yidiny: g-word > p-word
Yidiny stress assignment within the domain of the p-word:
1. a. Stress is assigned to to the first syllable with a long vowel,
b. If there are no long vowels - the first syllable of a word is stressed
2. From that basic stress point, every other syllable to the left and right of it is also
stressed.
waŋal-muday
bigu:n-muda:y-ɲdu
waŋal-muda:y-ɲdu
bigu:n-muday
‘‘boomerang-COMIT.ABS’’
‘‘shield-COMIT-ERG’’
‘‘boomerang-comit-erg’’
‘‘shield-comit.abs’’
How would you stress these words?
*Length is indicated by ‘:‘
**The internal morphological structure is given for each word.
Alignment
The case of Yidiny: g-word > p-word
Yidiny stress assignment within the domain of the p-word:
1. a. Stress is assigned to to the first syllable with a long vowel,
b. If there are no long vowels - the first syllable of a word is stressed
2. From that basic stress point, every other syllable to the left and right of it is also
stressed.
wáŋal-múday
bigú:n-mudá:y-ɲdu
waŋál-mudá:y-ɲdu
bigú:n-mudáy
‘‘boomerang-COMIT.ABS’’
‘‘shield-COMIT-ERG’’
‘‘boomerang-comit-erg’’
‘‘shield-comit.abs’’
Alignment
The case of Yidiny: g-word > p-word
Yidiny stress assignment within the domain of the p-word:
1. a. Stress is assigned to to the first syllable with a long vowel,
b. If there are no long vowels - the first syllable of a word is stressed
2. From that basic stress point, every other syllable to the left and right of it is also
stressed.
wáŋal-múday
bigú:n-mudá:y-ɲdu
waŋál-mudá:y-ɲdu
bigú:n-mudáy
‘‘boomerang-COMIT.ABS’’
‘‘shield-COMIT-ERG’’
‘‘boomerang-comit-erg’’
‘‘shield-comit.abs’’
wáŋal-múday
bigú:n-mudá:y-ɲdu
wáŋal-mudá:y-ɲdu
bigú:n-múday
‘‘boomerang-COMIT.ABS’’
‘‘shield-COMIT-ERG’’
‘‘boomerang-comit-erg’’
‘‘shield-comit.abs’’
Alignment
The case of Yidiny: g-word > p-word
Yidiny stress assignment within the domain of the p-word:
1. a. Stress is assigned to to the first syllable with a long vowel,
b. If there are no long vowels - the first syllable of a word is stressed
2. From that basic stress point, every other syllable to the left and right of it is also
stressed.
wáŋal-múday
bigú:n-mudá:y-ɲdu
wáŋal-mudá:y-ɲdu
bigú:n-múday
‘‘boomerang-COMIT.ABS’’
‘‘shield-COMIT-ERG’’
‘‘boomerang-comit-erg’’
‘‘shield-comit.abs’’
(wáŋal)ω (múday)ω
(bigú:n)ω (mudá:yɲdu)ω
(wáŋal)ω (mudá:yɲdu)ω
(bigú:n)ω (múday)ω
Disyllabic affixes form phonological words of their own, i.e. they form a domain in themselves
(plus monosyllabic suffixes attached to them) for stress assignment. It is sometimes also said
that such affixes are non-cohering (rather than cohering) because they don’t cohere
phonologically with their morphological host.
Alignment
The case of (British) English
a. -age, -al, -ant, -ance, -ary, -ate, -ic, -ion;
b. -able, -er, -en, ful, -hood, -ish, -ism, -less, -like, -ment.
Can you think why we would group the English suffixes as we did above?
Alignment
The case of (British) English
a. stress-shifting suffixes: -age, -al, -ant, -ance, -ary, -ate, -ic, -ion;
b. stress-neutral suffixes: -able, -er, -en, ful, -hood, -ish, -ism, -less, -like, -ment.
We could also say that the affixes in (b) are non-cohering, because they are not ‘seen’
by the stress rule. However, they are non-cohering in a different way than was the
case with Yidiny.
Why?
Scholars often talk about strata (pl of stratum) in the affix lexicon.
>> See week 8 when we talk about templates
Typological parameter II: fusion
From the perspective of the morph:
If you can identify a morph, what can you say about the degree of
fusion it has with its host?
Three basic values:
Isolating - Concatenative - Nonlinear
Isolating morphology
Isolating
Lai Chin (Tibeto-Burman)
Tsew Máŋ
niʔ
Tsew Máŋ
ERG
ω
ω
ω
‘Tsew Mang hit me.’
ʔa-ka-t̪hoʔŋ
3SG.A-1SG.P-hit
ω
(Although spelling suggests differently, this is a situation equivalent to Yidiny)
Bickel & Nichols (2007), p. 173
Concatenating morphology
Concatenating: segmentable dependent morphs
Yurakaré
dúla
dulá-ni
dula-ní-shta
‘He makes it’
‘He is going to make it.’
‘He will be going to make it.’
Concatenating morphology
Concatenating: segmentable dependent morphs
Also think of the rules discussed above for Yurakaré (lenition) Yuki and
Lenakel (epenthesis), Emérillon (elision), Belhare (dissimilation),
Turkish (vowel harmony) and Dutch and English (phonologically
conditioned allomorphy).
They all show evidence of phonological integration of affixes with their
host.
Concatenating morphology
Concatenating: segmentable dependent morphs
Fusion of concatenative morphs seems to be a matter of degree
(remember the English strata)
Movima (Isolate)
onarana=us
[ʔɔnaɾaꞌnaʔus]
know=MASC.ABST
‘He knows X.’
iye:ni=as
[iꞌjɛ:niʔas]
move=NEUT.ABST
‘It moves.’
Nonlinear morphology
Nonlinear: morphs that are not segmentable in linear strings
We have seen a number of examples already last week
-
Vowel mutation
Consonant mutation
Subtraction
Transfixation
Nonlinear morphology
Replacive mophology
Yurakaré
bata
duta
bëjta
wilita
‘go’
‘burn (itr)
‘see’
‘return’
bache
duche
bëjche
wiliche
Alternative: weak suppletion
‘send’
‘burn (tr)’
‘show’
‘bring back’
ba = ?
du = ?
bëj = ?
wili = ?
Nonlinear morphology
Suprasegmental rules: tonal morphemes
Hausa (Chadic)
sháa
cí
‘to drink’
‘to eat’
shâa
cîi
‘drinking’ (N)
‘eating’ (N)
Tone languages make use of pitch level to distinguish words from each other. In some of
these languages, tone is used systematically to perform morphological (derivation,
inflection) operations.
Two parameters are of particular importance for
tone languages:
1. The number of distinctive pitch levels
2. Whether there are only level tones or also
contour tones
Nonlinear morphology
http://wals.info/feature/13A#2/28.6/152.8
Nonlinear morphology
Suprasegmental rules: stress
English
convíct
contrást
incréase
permít
recórd
adréss
vs.
vs.
vs.
vs.
vs.
vs.
cónvict
cóntrast
íncrease
pérmit
récord
áddress
Reduplication
Reduplication is a morphological operation whereby part of a base or
the entire base is copied and attached to that base.
Javanese (Austronesian)
full reduplication:
baita ‘‘ship’’
sesupe ‘‘ring’’
omaha ‘‘house’’
baita-baita ‘‘various ships’’
sesupe-sesupe ‘‘various rings’’
omaha-omaha ‘‘various houses’’
partial reduplication:
geni ‘‘fire’’
gegeni ‘‘to warm oneself by the fire’’
jawah ‘‘rain’’
jejawah ‘‘to play in the rain’’
tamu ‘‘guest’’
tetamu ‘‘to visit’’
Uhlenbeck 1978, cited in Booij (2007)
Reduplication
Reduplication
Partial reduplication: what is copied?
Ponapean (Austronesian)
duhp ‘dive’
mihk ‘suck’
wehk ‘confess’
du-duhp ‘be diving’
mi-mihk ‘be sucking’
we-wehk ‘be confessing’
Rehg 1981, cited in Haspelmath & Simms (2010)
Reduplication
Partial reduplication: what is copied?
Mangap-Mbula (Austronesian)
kuk ‘bark’
kel ‘dig’
kan ‘eat’
kuk-uk ‘be barking’
kel-el ‘be digging’
kan-an ‘be eating’
Bugenhagen 1995, cited in Haspelmath & Simms (2010)
Reduplication
Partial reduplication: what is copied?
Mangap-Mbula (Austronesian)
kuk ‘bark’
kel ‘dig’
kan ‘eat’
kuk-uk ‘be barking’
kel-el ‘be digging’
kan-an ‘be eating’
Bugenhagen 1995, cited in Haspelmath & Simms (2010)
Reduplication
Is reduplication linear or non-linear?
It seems to involve both nonlinear (material from the
base) and linear (concatenative adjunction) aspects
Continuum
The fusion parameter seems to have a continuous character
nonlinear
base
modifications
isolating
concatenative
reduplication
inner stratum
affixes
outer stratum
affixes
Non-cohering
affixes
Duplifixes
Duplifixes
Somali (Afro-Asiatic)
buug
fool
koob
jid
‘book’
‘face’
‘cup’
‘street’
buug-ag
fool-al
koob-ab
jid-ad
Berchem 1991, cited in Haspelmath & Simms (2010)
‘books’
‘faces’
‘cups’
‘streets’
Duplifixes
Duplifixes
Tzutzujil (Mayan)
saq
rax
q’eq
tz’iil
‘white’
‘green’
‘black’
‘dirty’
saq-soj
rax-roj
q’eq-q’oj
tz’il-tz’oj
Berchem 1991, cited in Haspelmath & Simms (2010)
‘whitish’
‘greenish’
‘blackish’
‘dirtyish’