+ - 0:00:00
Notes for current slide
Notes for next slide

Morphology

Ponrawee Prasertsom
LOT Training Camp 2025

1 / 30

Me

Ponrawee Prasertsom [pʰonráwiː pràsɤ̀ːtsǒm]

  • PhD student, Centre for Language Evolution, University of Edinburgh
2 / 30

Morpheme

  • Morphology is the study of words and their structures
  • Words in languages are formed by combining morphemes
  • Morphemes are the smallest meaningful units.
  • These are all morphemes:
    • ไก่ (Thai)
    • pig (English)
    • re- (English)
    • =ni/に (Japanese)
    • jan/जन् (Sanskrit)
3 / 30

Morpheme-to-word ratio

  • Isolating = Low ratio ~ 1:1 (Thai)

    ฉัน

    1SG

    อยาก

    want

    จะ

    IRR

    ไป

    go

    เที่ยว

    travel

    ฉัน อยาก จะ ไป เที่ยว

    1SG want IRR go travel

    ‘I want to travel.’

  • Synthetic = Mid-range ratio (Japanese)

    Tarou-tachi=wa

    Taro-ASSOC=TOP

    gohan=o

    rice=ACC

    tabe-mashita

    eat-POLITE:PFV

    Tarou-tachi=wa gohan=o tabe-mashita

    Taro-ASSOC=TOP rice=ACC eat-POLITE:PFV

    ‘Taro and his group ate rice.'

  • Polysynthetic = High ratio (Central Yup’ik) ← IOL loves this type.

    tuntu-ssur-qatar-ni-ksaite-ngqiggte-uq

    reindeer-hunt-FUT-say-NEG-again-3SG.IND

    tuntu-ssur-qatar-ni-ksaite-ngqiggte-uq

    reindeer-hunt-FUT-say-NEG-again-3SG.IND

    ‘He had not yet said again that he was going to hunt reindeer.’

4 / 30

Making words from morphemes

  • Affixation
    • Prefix: ⟨me-⟩muni 1pl.excl.subj-drink ‘(We) drink.’ (Lewo)
    • Suffix: lil⟨-its⟩ cry-caus ‘make (sb) cry.’ (Chichewa)
    • Circumfix: ke-⟩besar⟨-an nmlz-huge-nmlz ‘hugeness’ (Malay)
    • Infix: k⟨-amn-⟩aət born:nmlz ‘birth’ (Khmer)
  • Compounding: demir road + yol-u iron-poss → demiryolu ‘railway (lit. road of iron)’ (Turkish)
  • Base modification: káʔba ‘filth’ → káʔbá filth:adjlz ‘filthy’ (Chalcatongo Mictec)
  • Reduplication
    • Full reduplication: แดง → แดงๆ ‘kind of red’ (Thai)
    • Partial reduplication: kuk → kuk-uk bark-prog ‘be barking’ (Mangap-Mbula)
    • ‘‘Duplifix’’: jid → jid-ad street-pl ‘streets’ (Somali)
5 / 30

Languages have many ways to build words out of morphemes.

Discontinuous morphology

  • Morphemes are usually combined in a (superficially) linear fashion.
    • nat- (root) → nat-ion → nat-ion-al → nat-ion-al-ity → nat-ion-al-iti-es (English)
  • ... But Semitic languages such as Arabic, Amharic, Hebrew and Maltese have consonantal roots (usually 3) and transfixes.

Examples from Arabic:

  • kataba ‘he wrote.’
  • katabat ‘she wrote.’
  • yaktubu ‘he writes.’
  • taktubu ‘she writes.’
  • kitaab ‘book’
  • kaatib ‘writer (masculine)’
  • kaatibat ‘writer (feminine)’
6 / 30

IOL 2003, Problem 2: Arabic Arithmetic

tumn + tumnēn = talatt itmān
sabaʕt itlāt + suds = ʕašart irbāʕ
tusʕēn + tusʕ = sudsēn
xamast ixmās + subʕ = tamant isbāʕ
subʕēn + xumsēn = 2435

All words and sums are fractions. Words and sums' numerators and denominators ≤ 10 and there is no x1.

7 / 30

IOL 2003, Problem 2: Arabic Arithmetic

tumn + tumnēn = talatt itmān
sabaʕt itlāt + suds = ʕašart irbāʕ
tusʕēn + tusʕ = sudsēn
xamast ixmās + subʕ = tamant isbāʕ
subʕēn + xumsēn = 2435

All words and sums are fractions. Words and sums' numerators and denominators ≤ 10 and there is no x1.

8 / 30

IOL 2003, Problem 2: Arabic Arithmetic

tumn + tumnēn = talatt itmān
sabaʕt itlāt + suds = ʕašart irbāʕ
tusʕēn + tusʕ = sudsēn
xamast ixmās + subʕ = tamant isbāʕ
subʕēn + xumsēn = 2435

All words and sums are fractions. Words and sums' numerators and denominators ≤ 10 and there is no x1.

Roots

  1. t-m-n
  2. t-l-t
  3. s-b-ʕ
  4. s-d-s
  5. ʕ-š-r
  6. r-b-ʕ
  7. t-s-ʕ
  8. x-m-s

Transfixes

  1. Ø-u-Ø-Ø
  2. Ø-u-Ø-ēn
  3. Ø-a-a-t
  4. i-Ø-ā-Ø
9 / 30

IOL 2003, Problem 2: Arabic Arithmetic

tumn + tumnēn = talatt itmān
sabaʕt itlāt + suds = ʕašart irbāʕ
tusʕēn + tusʕ = sudsēn
xamast ixmās + subʕ = tamant isbāʕ
subʕēn + xumsēn = 2435

All words and sums are fractions. Words and sums' numerators and denominators ≤ 10 and there is no x1.

All denoms ≤ 10
∴ subʕēn, xumsēn denoms = 5 and 7.
∴ {subʕēn, xumsēn} = {25, 27}
∴ Ø-u-Ø-ēn = 2root
{s-b-ʕ, x-m-s} = {5, 7}

∴ Transfixes = xroot

10 / 30

IOL 2003, Problem 2: Arabic Arithmetic

tumn + tumnēn = talatt itmān
sabaʕt itlāt + suds = ʕašart irbāʕ
tusʕēn + tusʕ = sudsēn
xamast ixmās + subʕ = tamant isbāʕ
subʕēn + xumsēn = 2435

All words and sums are fractions. Words and sums' numerators and denominators ≤ 10 and there is no x1.

Ø-u-Ø-Ø+2t-s-ʕ=2s-d-s

  • Consider constraints
  • Assume integer roots
  • Assume unique words for integers

We get

  • Ø-u-Ø-Ø = 1root
  • t-s-ʕ = 6 or 9
  • s-d-s = 4 or 6
11 / 30

IOL 2003, Problem 2: Arabic Arithmetic

tumn + tumnēn = talatt itmān
sabaʕt itlāt + suds = ʕašart irbāʕ
tusʕēn + tusʕ = sudsēn
xamast ixmās + subʕ = tamant isbāʕ
subʕēn + xumsēn = 2435

All words and sums are fractions. Words and sums' numerators and denominators ≤ 10 and there is no x1.

We know that:

  • Ø-u-Ø-Ø = 1root
  • Ø-u-Ø-ēn = 2root

Ø-u-Ø-Ø+Ø-u-Ø-ēnt-m-n=1+2t-m-n=t-l-tt-m-n

∴ t-l-t = 3

Observe that Ø-a-a-t always occurs with i-Ø-ā-n, and all the remaining fractions have them:

∴ 1a1a1t i22ā2n = 1-1-12-2-2

... when 1-1-1 > 2

12 / 30

IOL 2003, Problem 2: Arabic Arithmetic

tumn + tumnēn = talatt itmān
sabaʕt itlāt + suds = ʕašart irbāʕ
tusʕēn + tusʕ = sudsēn
xamast ixmās + subʕ = tamant isbāʕ
subʕēn + xumsēn = 2435

All words and sums are fractions. Words and sums' numerators and denominators ≤ 10 and there is no x1.

x-m-sx-m-s+1s-b-ʕ=t-m-ns-b-ʕs-b-ʕ+1=t-m-n

We know that s-b-ʕ ∈ {5, 7}, so t-m-n ∈ {6, 8}

But we also know that 6 must be either t-s-ʕ or s-d-s. So t-m-n cannot be 6.

∴ t-m-n = 8; s-b-ʕ = 7

Because {x-m-s, s-b-ʕ} = {5, 7}
∴ x-m-s = 5

13 / 30

IOL 2003, Problem 2: Arabic Arithmetic

tumn + tumnēn = talatt itmān
sabaʕt itlāt + suds = ʕašart irbāʕ
tusʕēn + tusʕ = sudsēn
xamast ixmās + subʕ = tamant isbāʕ
subʕēn + xumsēn = 2435

All words and sums are fractions. Words' and sums' numerators and denominators ≤ 10 and there is no x1.

s-b-ʕt-l-t+1s-d-s=ʕ-š-rr-b-ʕ73+1s-d-s=ʕ-š-rr-b-ʕ

Because s-d-s ∈ {4, 6}

73+1s-d-s=73+14=3112, or=73+16=52=104

Given the constraints and integer uniqueness, we get

  • s-d-s = 6
  • t-s-ʕ = 9
  • ʕ-š-r = 10
  • r-b-ʕ = 4
14 / 30

IOL 2003, Problem 2: Arabic Arithmetic

tumn + tumnēn = talatt itmān
sabaʕt itlāt + suds = ʕašart irbāʕ
tusʕēn + tusʕ = sudsēn
xamast ixmās + subʕ = tamant isbāʕ
subʕēn + xumsēn = 2435

All words and sums are fractions. Words' and sums' numerators and denominators ≤ 10 and there is no x1.

And we're done!

Roots

  1. t-m-n = 8
  2. t-l-t = 3
  3. s-b-ʕ = 7
  4. s-d-s = 6
  5. ʕ-š-r = 10
  6. r-b-ʕ = 4
  7. t-s-ʕ = 9
  8. x-m-s = 5

Transfixes

  1. Ø-u-Ø-Ø=1root
  2. i-Ø-ā-Ø=2root
  3. 1a1a1t 2u22ēn=1-1-12-2-2
15 / 30

Derivation and inflection

  • Derivation = morphological processes that change the core meaning of a word
    • read → re-read (affixation)
    • dɔ̀ drink → tɔ drink:caus ‘cause to drink’ (base modification)
  • Inflection = no change in the core meaning, change in grammatical features
    • read → (he) read-s (affixation)
    • paːrʃt̪ bard:nom:sg:indf ‘a bard’ → vaːrʃt̪ bard:gen:pl:indf ‘of bards’ (base modification)

Universal!!! 🤩:

  • Across spoken languages, derivational morphemes are (generally) closer to the root than inflectional ones.
    • read-er-s read-nmlz-pl
16 / 30

UR & SR

  • Often, the same morphemes (same meaning) have different forms, changing some sounds:
    • Buch ~ Büch-er book-pl ‘books’ (German Umlaut)
    • kami ‘paper’ ~ ori-gami fold-paper ‘folding paper’ (Japanese Rendaku)
  • Not random but predictable from sound and word structure, or from the specific words themselves.
    • German Umlaut (simplified): ü [yː] occurs in the morpheme when the word is plural, u [uː] elsewhere
    • Japanese Rendaku (simplified): g [g] at the start of the second morpheme in a compound, when preceded by a vowel; k [k] elsewhere
17 / 30

UR & SR: German

  • Sometimes it is useful to think of a word and its morphemes as having an ‘‘original’’ form (underlying representation; UR), and the derived form (surface representation; SR)

Consider the German examples below:

Singular

  1. [taːk] ‘day:sg’
  2. [los] ‘lot:sg’
  3. [ʔanʁuːf] ‘phone.call:sg’
  4. [ʔaʁm] arm:sg’
  5. [hʊnt] ‘dog:sg’
  6. [boːt] ‘boat:sg’

Plural

  1. [taːg-ə] ‘day:pl’
  2. [loz-ə] ‘lot:pl’
  3. [ʔanʁuːf-ə] ‘phone.call:pl’
  4. [ʔaʁm-ə] arm:pl’
  5. [hʊnd-ə] ‘dog:pl’
  6. [boːt-ə] ‘boat:pl’

What are the URs (original form)? What is the rule for deriving the SRs?

18 / 30

UR & SR: German

  • If the forms of the root are different, the UR is the ones with a final voiced obstruent [g, d, z], if any. (Why choose the voiced variant?)
    • [taːg], [loz], [hund], [ʔaʁm]
  • Otherwise, the root's UR is the same as its SR.
    • [ʔanʁuːf], [boːt]
  • The rule is:

[+consonantal−sonorant]→[−voice]/__#

where

  • [+consonantal−sonorant] = obstruent consonants
  • [−voice] = voiceless

‘‘All obstruent consonants become voiceless at the end of the word (before pause, #).’’

19 / 30

UR & SR: Indonesian

Isolate the prefix common to the words below. What is its UR? What are the rules to derive the SR? (No need for fancy notation.)

Verb stems

  1. bat͡ʃa
  2. ɲaɲi
  3. antuk
  4. ŋat͡ʃo
  5. t͡ʃut͡ʃi
  6. lempar
  7. pukul
  8. masak
  9. gambar
  10. tulis
  11. d͡ʒawab
  12. isi

Prefixed forms

  1. məmbat͡ʃa
  2. məɲaɲi
  3. məŋantuk
  4. məŋat͡ʃo
  5. meɲt͡ʃut͡ʃi
  6. məlempar
  7. məmukul
  8. məmasak
  9. məŋgambar
  10. mənulis
  11. məɲd͡ʒawab
  12. məŋisi
20 / 30

UR & SR: Indonesian

The UR is məŋ. The rules are:

  • Change the place of articulation of ŋ to be the same as the following consonant

ŋ→[αplace]/__[+consonantalαplace]

  • Delete the nasal before a sonorant (n, ŋ, m, l)

[+nasal]→∅/__[+consonantal+sonorant]

  • Delete p, t, k after the nasal

[+consonantal−voice−delayed release]→∅/[+nasal]__

... where + = morpheme boundaries.

Do these rules have to be ordered? Why?

21 / 30

UR & SR: Indonesian

Why not choose other forms as URs, e.g. mən, məm, mə or məɲ?

22 / 30

UR & SR: Indonesian

Why not choose other forms as URs, e.g. mən, məm, mə or məɲ?

Because of cases like these:

  • antuk → məŋantuk
  • isi → məŋisi

You may have another set of rules. These are fine as long as your rules account for 100% of the data, but there are probably more rules/they are more complicated.

All unpredictable things in the UR:

Posit a UR that has all unpredictable information. In this case, we cannot predict what nasals will occur between vowels, so we put [ŋ] in the UR.

23 / 30

IOL 2004, Problem 5 Chuvash verbs

  1. aman
  2. aptra
  3. cĕt
  4. çit
  5. čühen
  6. hupăn
  7. kaç
  8. kăvakar
  9. kuç
  10. puçtarăn
  11. shăn
  12. taptan
  13. tupăn
  14. uçăn
  15. ük
  16. vĕre
  17. vĕren
  18. vitĕn
  1. amant
  3. çĕter
  5. čühe
  8. kăvakart
  9. kuçar
  10. puçtar
  11. shănt
  12. tapta
  13. tup
  14. uç
  15. üker
  16. vĕret
  17. vĕrent
  18. vit
  19. kĕrt
  20. pytar

Instructions:

  • Fill in the blanks
  • Indicate which blanks cannot be filled with certainty (i.e. more than one possible forms).
24 / 30

IOL 2004, Problem 5 Chuvash verbs

  1. aman
  2. aptra
  3. cĕt
  4. çit
  5. čühen
  6. hupăn
  7. kaç
  8. kăvakar
  9. kuç
  10. puçtarăn
  11. shăn
  12. taptan
  13. tupăn
  14. uçăn
  15. ük
  16. vĕre
  17. vĕren
  18. vitĕn
  1. amant
  3. çĕter
  5. čühe
  8. kăvakart
  9. kuçar
  10. puçtar
  11. shănt
  12. tapta
  13. tup
  14. uç
  15. üker
  16. vĕret
  17. vĕrent
  18. vit
  19. kĕrt
  20. pytar

The key is just finding the right URs for the stems!

Sometimes the UR is on the left, sometimes on the right. (Hax: The UR is just the shorter one.)

25 / 30

IOL 2004, Problem 5 Chuvash verbs

  1. aman
  2. aptra
  3. cĕt
  4. çit
  5. čühen
  6. hupăn
  7. kaç
  8. kăvakar
  9. kuç
  10. puçtarăn
  11. shăn
  12. taptan
  13. tupăn
  14. uçăn
  15. ük
  16. vĕre
  17. vĕren
  18. vitĕn
  1. amant
  3. çĕter
  5. čühe
  8. kăvakart
  9. kuçar
  10. puçtar
  11. shănt
  12. tapta
  13. tup
  14. uç
  15. üker
  16. vĕret
  17. vĕrent
  18. vit
  19. kĕrt
  20. pytar

The key is just finding the right URs for the stems!

Sometimes the UR is on the left, sometimes on the right. (Hax: The UR is just the shorter one.)

Choosing the wrong UR will result in unpredictability.

26 / 30

IOL 2004, Problem 5 Chuvash verbs

  1. aman
  2. aptra
  3. cĕt
  4. çit
  5. čühen
  6. hupăn
  7. kaç
  8. kăvakar
  9. kuç
  10. puçtarăn
  11. shăn
  12. taptan
  13. tupăn
  14. uçăn
  15. ük
  16. vĕre
  17. vĕren
  18. vitĕn
  1. amant
  3. çĕter
  5. čühe
  8. kăvakart
  9. kuçar
  10. puçtar
  11. shănt
  12. tapta
  13. tup
  14. uç
  15. üker
  16. vĕret
  17. vĕrent
  18. vit
  19. kĕrt
  20. pytar

For URs on the left side, there are two suffixes:

  • -Vr (e or a followed by r), used when the stem ends with an obstruent.
    • The choice of the vowel depends on vowel harmony.

Chuvash vowels

  • -t, used when the stem ends with a sonorant (vowels, n, r).
27 / 30

IOL 2004, Problem 5 Chuvash verbs

  1. aman
  2. aptra
  3. cĕt
  4. çit
  5. čühen
  6. hupăn
  7. kaç
  8. kăvakar
  9. kuç
  10. puçtarăn
  11. shăn
  12. taptan
  13. tupăn
  14. uçăn
  15. ük
  16. vĕre
  17. vĕren
  18. vitĕn
  1. amant
  3. çĕter
  5. čühe
  8. kăvakart
  9. kuçar
  10. puçtar
  11. shănt
  12. tapta
  13. tup
  14. uç
  15. üker
  16. vĕret
  17. vĕrent
  18. vit
  19. kĕrt
  20. pytar

For URs on the right side, there is one suffix:

  • -(V̆)n (n, might be preceded by ĕ or ă): The vowel is harmonic and is inserted when the stem ends with a consonant.
28 / 30

IOL 2004, Problem 5 Chuvash verbs

  1. aman
  2. aptra
  3. cĕt
  4. çit
  5. čühen
  6. hupăn
  7. kaç
  8. kăvakar
  9. kuç
  10. puçtarăn
  11. shăn
  12. taptan
  13. tupăn
  14. uçăn
  15. ük
  16. vĕre
  17. vĕren
  18. vitĕn
  19. ker, kertĕn
  20. pyt, pytarăn
  1. amant
  2. aptrat
  3. çĕter
  4. çiter
  5. čühe
  6. hup, hupănt
  7. kaçar
  8. kăvakart
  9. kuçar
  10. puçtar
  11. shănt
  12. tapta
  13. tup
  14. uç
  15. üker
  16. vĕret
  17. vĕrent
  18. vit
  19. kĕrt
  20. pytar

We can now fill in the blanks. Some blanks are ambiguous because the data are compatible with the UR on either side.

29 / 30

Notes on how to answer

When you write your answers, you can use formulae, made-up notations (a → b), technical terms (URs, obstruents, sonorants, etc.) and abbreviations (C for consonants, V for vowels, etc.). But you must define them in the answers.

Next module is grammatical features. Have fun!

30 / 30

Me

Ponrawee Prasertsom [pʰonráwiː pràsɤ̀ːtsǒm]

  • PhD student, Centre for Language Evolution, University of Edinburgh
2 / 30
Paused

Help

Keyboard shortcuts

↑, ←, Pg Up, k Go to previous slide
↓, →, Pg Dn, Space, j Go to next slide
Home Go to first slide
End Go to last slide
Number + Return Go to specific slide
b / m / f Toggle blackout / mirrored / fullscreen mode
c Clone slideshow
p Toggle presenter mode
t Restart the presentation timer
?, h Toggle this help
Esc Back to slideshow