Morphology

Question 1

What is Morphology

Answer 1

The study of the internal structure of words (and the mental process of word formation). Aronoff & Fudeman.

Question 2

What is the traditional Chomskyan take on Morphology?

Answer 2

Assume that words are composed of morphemes.

A morpheme is:
‘a minimal unit of lexical form’ (Bloomfield 1933)
‘the smallest linguistic piece with grammatical function’ (Aronoff & Fudeman 2011)

Question 3

Why are morphemes and words different?

Answer 3

A word is the smallest meaningful element which can occur in isolation, whereas morphemes have meaning but don’t necessarily occur in isolation.

Question 4

Main types of word formation

Answer 4

Affixation, Truncation, Blends/Portmanteaux

Question 5

Affixation

Answer 5

Prefix, suffix or infix. (for some affixes this is fixed and for others it can vary).

Saturative affixation is when the marker spreads to both elements of the phrase being marked, e.g ‘your guys’ phone’.

Question 6

Truncation

Answer 6

Shortening.

Main patten in English is to truncate right side e.g Robert > Rob

This said, truncation of right side also happens e.g mushroom > shroom
telephone > phone

Or both happens:
influenza > flu

Acronymy:
Self-Contained Underwater Breathing Apparatus > Scuba

TWO TYPES:
Templatic: a certain section (according to a template) of the word is kept and the rest discarded.
Subtractive: A certain section of the word is removed and discarded.
(Examples in Lecture 1 notes)

Question 7

Blends / Portmanteaux

Answer 7

2 parts: brunch
3 parts: frappalattecino
infixing: ricockulous

Question 8

What is a morph? Aronoff & Fudeman

Answer 8

The phonological realisation of a morpheme

Question 9

What are allomorphs? Aronoff & Fudeman

Answer 9

variants of a morpheme (e.g the three different english plural markers

Question 10

What is a stem? Aronoff & Fudeman

Answer 10

A base unit to which other morphological units are attached. Can be a simple (unanalysable form) or complex (made of more than one simple morphological unit) form.

Question 11

What is a root? Aronoff & Fudeman

Answer 11

A simple stem.

Question 12

Aronoff & Fudeman - why is morphology a distinct component of languages and grammars?

Answer 12

Aronoff 1994 shows that some elements of morphology, e.g arbitrary classes called declensions in Latin which determine the form taken in different environments, have little or nothing to do with syntax or phonology. So while morphology interacts with other aspects of language probably more so than other aspects, it should be considered a distinct component of language.

Question 13

2 approaches to Morphological Analysis - Aronoff & Fudeman

Answer 13

Analytic - breaking words down (American structural linguistics of the first half of the twentieth century - dealing with unfamiliar languages). Methods independent of the structures we are examining - preconceived notions might interfere with an objective, scientific analysis.

Synthetic - (often - perhaps unfairly - more associated with theory than methodology). Assumes that you know the pieces which make up a word, and asks how a speaker constructs a grammatically complex word.

Question 14

Eugene Nida’s (1949, revised edition 1965) textbook Morphology sets out some basic analytic principles.

Answer 14

Principle 1
Forms with the same meaning and the same sound shape in all their occurrences are instances of the same morpheme. (If they have same form but different meanings they are different morphemes).
First step is to look for these forms with same shape and meaning.
Type-token distinction. Need multiple tokens (on diff stems?!) with same form and meaning to really confirm one TYPE of morpheme.
Principle 2
Forms with the same meaning but different sound shapes may be instances of the same morpheme if their distributions do not overlap.
English plural marker for example.
Principle 3
Not all morphemes are segmental.
Some depend on other morphemes for their realisation (can’t be pronounced alone).
Take for example ‘ablaut’ vowel alternations in English vowels: e.g run/ran, speak/spoke, eat/ate.
Not a segment but a CONTRAST
Another example would be breath/breathe, cloth/clothe, house/house, where there is a contrast in the voicing feature.
Principle 4
A morpheme may have zero as one of its allomorphs provided it has a non-zero allomorph.
So ‘fish’ shows a zero plural marker allomorph in English, but the same could not be posited for ‘fish’ in Japanese which is also singular or plural, because there is no non-zero plural marker with which to contrast the zero marker.

Question 15

Historically what is morphological typology thought to be linked to?

Answer 15

The stages of language formation. (Max Muller)

Question 16

What is an analytic language?

Answer 16

The distinction between analytic and synthetic languages is not binary but rather a spectrum. In the ideal analytic language (an isolating language), there is a one-one correspondence between words and morphemes, meaning that morphemes are not bound. An example of an analytic language is Vietnamese (however it is not completely isolating as a small group of affixes are attested). Each morpheme is represented by a word, rather than being bound to another word. Another language might bind the demonstrative to the noun (as Swedish binds determiners) or the totality marker to the quantifier, but the apparent lack of binding is characteristic of analytic languages.

Question 17

What is a synthetic language?

Answer 17

At the other end of the spectrum, a synthetic language is one in which words tend to consist of several morphemes bound together. Take for example the sentence below in French:

‘J’allais recommencer’
J(e) - all - ais re - commenc - er
1st singular nominative pronoun - stem of aller ‘to go’ - 1st singular imperfect tense - re ‘again’ - stem of commencer ‘begin’ - infinitive marker

In this example, the word ‘recommencer’ consists of 3 morphemes bound together, making French a synthetic language given the tendency for words to consist of more than one morpheme, as opposed to Vietnamese where the majority of words consist of a single morpheme.

Question 18

What is a concatenative language?

Answer 18

Concatenative languages are a subset of synthetic languages in which morphemes are bound to one another sequentially, but the forms remain separate. French is an example of a concatenative language, take for example the sentence: ‘c’est fini’ where the past participle ‘fini’ consists of the stem ‘fin-’ and perfective past marker ‘-i’. This is, two distinct morphemes attached in sequence, without alteration to their original forms. This is consistent with the formation of English equivalent ‘It’s finished’ (finish+ed), however one might note that the similar construction ‘it’s done’, does not express the perfect past in stem+suffix form, rather the form of the stem is altered to indicate the past participle, an example of the alternative: templatic morphology:

Question 19

What is templatic morphology?

Answer 19

A templatic language is one for which morphemes are not bound sequentially, but rather the root, which carries the basic meaning, is altered according to different (a set of fixed) templates to express additional meaning. For example, in Arabic, some nouns are pluralised with a suffix like in English, but others are formed as a so-called ‘broken plural’, where the triconsonantal stem is arranged using different vowels to form the plural. Take for example ‘kitāb’ (‘book’), which is transformed to ‘kutub’ (‘books’), where the differing vowel template has the same pluralising property as the English ‘-s’ suffix. This shows alteration of the stem rather than sequential addition of morphemes.

Question 20

What is an agglutinative language?

Answer 20

In an agglutinative language, a single morpheme indicates one piece of grammatical information. For example, present tense, or indicative mood. In Wishram, a dialect of Chinook (a North American Indian language), the word ačimluda (‘He will give it to you’) is composed of the elements a- ‘future,’ -č- ‘he,’ -i- ‘him,’ -m- ‘thee,’ -l- ‘to,’ -ud- ‘give,’ and -a ‘future.’ This is in contrast with ‘il commencera’ in French for example, where the suffix ‘-era’ on the verb is indicative of 1st person, singular, future tense & indicative mood. In Wishram, as it is agglutinative, a single morpheme could not convey all these separate pieces of information.

Question 21

What is a fusional language?

Answer 21

In contrast, a fusional language, as described with the French example above, is a language for which a single morpheme can encode multiple different items (pieces) of grammatical information. Another example is German adjectival pronoun declension, for which the suffix ‘-er’ can conveys masculine, singular & nominative, e.g. in gruen-er (Mann). Unlike agglutinative languages, here a single morpheme conveys multiple different units of information.

Question 22

What is a polysynthetic language?

Answer 22

According to Payne (1997), a polysynthetic language is distinguished by a synthetic language by its capacity to use a single word to express a whole sentence (subject and predicate). Taking the French example of synthetic morphology ‘J’allais recommencer’, it is clear that bound morphemes are used, however the full subject-predicate formation is expressed with 3 (with contraction for liaison) words. See in contrast the example below from Chukchi, a language spoken in Siberia.

Təmeyŋəlevtpəγtərkən.
t-ə-meyŋ-ə-levt-pəγt-ə-rkən
1.SG.SUBJ-great-head-hurt-PRES.1
‘I have a fierce headache.’

It is evident that simply through bound morphology, speakers of this language can express full subject-predicate constructions which correspond to multi-word (in this case 5) sentences in English or other non-polysynthetic languages.

Question 23

How to write templatic stems?

Answer 23

√K-T-B

Question 24

Can a word be non-concatenative but not templatic?

Answer 24

Yes, for example the irregular past tense form ‘went’ is not concatenative, but it follows no template so cannot be templatic.

Question 25

Difference between inflectional and derivational morphology?

Answer 25

Inflection only refers to morphological change communicating grammatical meaning
(derivational changes semantic meaning or WORD CLASS) basically makes a new word.

Question 26

Where is irregular morphology retained?

Answer 26

On more frequent words. This can be thought of as evidence for morphological rules rather than analogy.

Question 27

What is the value of morphological typology?

Answer 27

The value of typology is that it may make predictions regarding important characteristics of the individuals represented by the types.
(you would not typologise motor vehicles based on the colour)

Question 28

Leading views on morphological structure:

Answer 28

Word based (Word and paradigm; Matthews 1972, Blevins 2017) 
Lexeme based (Item & Process; Anderson 1992, Aronoff 1994) 
Morpheme based (Item & Arrangement; Halle and Marantz 1993) 
> we study morpheme based views

Question 29

Evidence for morphological decomposition and storage:

Answer 29

New base forms - e.g agentive: daughter because i dot (Terri and Rose, January 2015), singulars: homo sapien, ME crevis>crayfish-> shellfish, starfish, cuttlefish
New morphemes - e.g i- (mac), -a/oholic, -burger, mc-

Question 30

Psycholinguistic evidence for morphemes.

Answer 30

Overregularisation: went > goed > went (in late 2 years of acquisition) (Berko Gleason 1958 wug test shows it’s rules). Problem here is that the reverse also applies - blunk - simply analogy?
ERPs: non-words have greater N400 than real words, however non-words formed of 2 unproductive morphemes have similar ERPs to real words. This could just be a frequency effect though! Morphological reality despite no word reality. (McKinnon et al. 2003.)
Psycholinguistic Priming: Rastle et al. 2004 - PRIMING a) semantically transparent morphological relationship with the target cleaner/clean b) an apparent morphological relationship but no semantic relationship with the target corner/corn NOT PRIMING c) a non-morphological form relationship with the target brothel/broth (these tests sometimes fail)
Synesthesia: color-letter/phoneme binding. One for first C of each regular morpheme.

Question 31

Morphological Speech Errors

Answer 31

Exchange: Slicely thinned (Fromkin)
Feature shift: have to went (for had to go)
Faulty access: have teachen, concludement
Janssen and Humphreys 2002: if a meaningful phoneme unit can move independently in error it must be a production unit (some independent representation).
Inflectional morphemes (e.g. -ed, -ing, -s) are much more error prone than derivational morphemes (e.g. -er, -ness, -able, -ion) (Garrett 1980, Humphreys, 2002).
However, an experiment showed that more affix errors occurred on morpho complex forms than pseudo morphs, showing that derivational morphemes also have a production representation.

Question 32

Difference between a phoneme and a morpheme

Answer 32

Phonemes are meaning distinguishing not meaningful.

Question 33

phonological word vs lexeme vs grammatical/morphosyntactic word

Answer 33

phonological words behave as a unit for certain kinds of phonological processes especially stress or accent. A lexeme is a unit of the lexicon with 1 meaning despite grammatical inflection. Words with different grammatical inflextions are considered different morphosyntactic words.

Question 34

Grammatical categories used for linguistic analysis, and the morphological (not syntactic or semantic) contrasts we draw in ENGLISH

Answer 34

gender: masculine, feminine, neuter
case: common case - possessive, object fro possessives
number: singular - plural
person: 1st 2nd 3rd
tense: past, non-past
aspect: progressive - non progressive, perfect - non perfect
voice: active - passive (- mediopassive).
comparison: absolute - comparative - superlative

Question 35

What’s the efficiency hypothesis?

Answer 35

The idea that we only have enough memory space to store rules rather than varying forms. There is no evidence for this though we might well have enough space.

Question 36

What is the Chomskyan view of Morphological Hierarchy?

Answer 36

morphological structure of words is hierarchically organised, as with words in syntax.

Question 37

analyses of the derivational process

Answer 37

A general pattern of derivation is affix + stem where the affix assigns a different lexical category to the stem. This could be analysed as a function where the output has no internal structure simply a new meaning, or according to the hierarchical view, the derivational history is preserved in the structure.
Tree: [N[adj Sad]-ness]

Question 38

Reasons to assume the hierarchical structure:

Answer 38

ambiguity of words such as unlockable can be represented.

recursive derivation

Question 39

How do we know the hierarchical structure?

e.g how do we know that white board eraser is right headed?

Answer 39

Semantics helps: white board is obviously a board, white board eraser is obviously an eraser.

Also: stress contours:

Chomsky, Halle & Lukoff 1956
Compound stress contours can be generated from hierarchical structure via a simple set of recursive rules.

Start assuming all constituents are primary.
Work inward from the most nested element.
Mark stress of first constituent of that element as primary.
Demote remainder by 1 degree.

(Then because it’s recursive you’d go to next nested level)

Question 40

What are selectional restrictions?

Answer 40

Basic principle: if an affix with a selectional restriction forms a legitimate word, it must be attaching to a constituent that satisfies its selectional restrictions
Semantic: e.g un- can only attach to words which DON’T have negative connotations.
Syntactic: e.g -ness only attaches to adjectives
Morphological: e.g -ion only attaches to latinate bases (decision vs choosion)
Phonological: e.g -er cannot attach to words of more than 2 syllables

Question 41

How does morphological headedness explain patterns in inflection on compounds?

Answer 41

Feature percolation: 
When there’s a constituent head
(endocentric):
 percolation to head (snowmen)
 When there’s no head (exocentric):
 no percolation (Maple Leafs, flied out)
 When there are two heads (dvandva):
 saturation (your guys’s, women writers, blew
dried)

Question 42

What are bracketing Paradoxes?

Answer 42

Morphologically compex expressions which can be assigned two different hierarchies. The phonological hierarchy based on phonological (stress contours created by addition of clas I & II affixes), morphological and syntactic selectional restrictions as opposed to the semantic bracketing.

Question 43

evidence for semantic features

Answer 43

capture generalisations about natural classes & subsets
Classifiers - Thai uses different numbers to count [+human] items
some verbs require [+human] or at least [+animate] subjects - like ‘admire’.

Question 44

using semantic features to model semantic change

Answer 44

Loss / Addition of semantic features

Question 45

using semantic features to model acquisition of meaning

Answer 45

Clark 1973 - much development can be characterized as semantic feature addition to initial incomplete structures which produced overgeneralisation errors.

The Gavagai problem (Quine) refers to the fact that given a word and a referent, how is one to know whether it refers to the individual, a certain class of individuals based on a (which) property of the individual, a single property of the individual, a part of the individual?

There are some semantic priors which children seem to follow.

• Whole object bias (not property or part)
• Object category bias (not individual)
• Shape bias (Imai, Gentner & Uchida 1994) is the basis for determining which objects are alike
• Form over color bias (Baldwin 1989)

The idea is that humans break down meaning into properties and feature sets.

Question 46

Evidence for Morphological Features

Answer 46

Feature based errors are MOVEMENT (like metathesis) of COPY (like leading list for reading list)

Morphological speech errors e.g have to went for had to go (metathesis of past tense feature)

Semantic substitution errors:

• insertion of opposite words
• substitutions tend to preserve gramatical category
• tend to preserve grammatical gender (Vigliocco, Vinson, Indefrey, Levelt & Hellwig 2004).

Verb errors (copy) 
Ashenfelter & Eberhard 2007 hypothesised that (for contexts where words sharing features were competing for insertion) in speech errors replacement of the simpler form by the more complex form should be more common and their data supported this. This is because feature copy from the less complex form would not affect the feature matrix of the more complex form, however feature copy from the more complex form would add additional features and cause realisation of the more complex form. 

Lexical Access
Burke et al 1991 - in TOT states alternative words retrieved are of same grammatical class, number and verb tense as the target word.

Caramazza and Miozzo 1997 - gender retrievable without access of lexeme in TOTs.

Question 47

strong and weak version of features theory

Answer 47

strong: morphemes are composed of features
weak: morphemes are categorised by features.