Vous êtes sur la page 1sur 14

11/10/2010

Lexical Morphology
Ling 4/510 Morphology Lynn Santelmann, Ph.D. 2010

Morphology: The Dark Years


Both Structuralism and early Generative Grammar ignored morphology Morphology was dealt with either as part of syntax or as part of phonology.

Lexicalist Morphology
Lexicalist Morphology: rescued morphology from other components of grammar Posited that derivational morphology (and possibly inflectional morphology) should be handled in the lexicon Note most of the citations are from the 1970s, and a few from the 1980s. What does that tell you about this theory?

11/10/2010

Early Generative Accounts


In early Generative Syntax, sentence that were related in meaning, but had different order of elements were derived from one another. Transformations were used to derive sentences Example: Passive NP V+ tense NP
1 2 3 4 The dog bit the child 4 be 3 2+en by 1 The child was bitten by the dog

This rule specifies both the order of elements and the morphemes needed

Extending transformations to morphology


If sentences can be derived via transformations, why not words? Example: transformations were used to describe the differences between verbs and their related nominal forms (a.ka. Nominalizations)
a. Robin is devoted to her mother. b. Robin's devotion to her mother.

Write a rule to derive (b) from (a)

Arguments against lexical transformations


Nominalizations changes meaning and violate constraints on transformations
transformations should not change meaning

A general rule cannot be determined; some verbs have two nominalizations that do not mean the same thing, e.g.,
Commit (v.) committal, commitment Affect (v.) affect (n.), affection

11/10/2010

Arguments against transformationally deriving nominals


Some verbs do not have corresponding noun forms.
Robin amused the children with stories. *Robin's amusement of the children with stories.

Syntactic transformations are supposed to apply to all members of a word class. How can you have a syntactic transformation that only applies to some verbs?

More evidence for separating morphology from syntax


Rules of syntax generally do not refer to phonological structure or make reference to the output of phonological rules. Morphological rules do, e.g.,
Infixes that insert into the onset of a syllable German diminutive -chen that cannot attach to bases ending in [x] or [ng].

The result?
Deriving nominals should be handled in a different area of grammar. What is that area? Morphology! If you have a morphology for nominals, it makes sense to extend it to all word formation.

11/10/2010

Lexical Morphology
The goal of morphology, according to Aronoff (1976) is the "enumeration of the class of possible words of a language" (p. 170), or to tell us what sort of words can be formed.

Lexical Morphology
Lexical Morphology primarily studies:
Potential word forms Rules and constraints on word formation

Parallel to thinking in Syntax, the morphology "overgenerates" words (creating potential words); non-words are ruled out by constraints.

Important Constructs in Lexical Morphology


Full Entry Theory Stratal Ordering Heads of Words & Feature Percolation Constraints on Word Formation Rules (WFRs)

11/10/2010

The Full Entry Theory


Full entry theory: theory of the lexicon which states that every word that 'exists' is listed in the lexicon. What does it mean for a word to exist?
Words that have been created, and used by an individual are stored in their lexicon. Lexicons are individualized

Whats stored in the Full Entry Lexicon?


In some versions of this theory, the lexicon contains:
all words that exist stems affixes

Advantages to a full entry lexicon


It can explain why a speaker can know two words without knowing the derivational relationship between them.
Are repair and reparations derivational related?

It makes sense for non-productive morphology/non-compositional forms

11/10/2010

Problems with the Full Entry Lexicon


Whats really stored?
Are all novel compounds stored once theyre made? What about Grammar Function Changing Morphemes?
Some GFC morphemes are rarely if ever used. Are forms

What about paradigms?


Some versions of the Full Entry lexicon store all forms in a paradigm. Really? That could lead to 1000s of forms to store for 1 verb/noun

Updated Lexicalist Views of the Lexicon


Aronoff & Anshen (1998): Morphology deals with the internal structure of words and the creation of words The lexicon stores existing words. This includes:
Non-compositional forms Affixes Idioms

Updated Lexicalist Views of the Lexicon (Aronoff & Anshen, 1998)


Examples of words that would be stored in the lexicon:
bamboozle (no internal structure) hornswoggle horn is recognizable, but swoggle is not hoodwink recognizable parts, but not compositional, so stored.

Contrast with: rigidification


Rigidification is clearly created from parts, and compositional in meaning. It will not be listed in the hearer/speakers

11/10/2010

Stratal (Level) Ordering


Recall: Languages appear to have more than one type of affix:
Neutral # (a.k.a. Primary, Level I) Non-neutral + (a.k.a. Secondary, Level II)

Non-neutral affixes:
Change stress pattern Change pronunciation of base Show obligatory phonetically conditioned allomorphy

Level Ordering
One way to account for differences between the effects of affixes is to order affixation:
Level 1 affixes added Then phonological rules (e.g., stress) applied Level 2 affixes added

Level Ordering
Predictions from Level Ordering
Level 1 affixes are always closer to the base than level 2 affixes Level 2 affixes should always appear further from the base than Level 1 affixes.

Recall: This distinction between affixes holds for other languages as well.

11/10/2010

Level Ordering, More Complex


Allen (1978) argued that compounding is carried out in a 3rd stratum (level), after Level 2. This predicts that there derivational morphemes must be added before compounding.
*[street-music]al, *[war-hero]ic, etc.

What about inflectional affixes? Added after compounding

Levels of Morphology
Root Non-neutral affixes (+) Phonological rules Neutral Affixes (#) Compounding Inflection Level 2 Level 3 Level 4 Level 1

Problems with Stratal/Level ordering


Affixes appear to be added to bases that don't exist,
e.g., enliven can you liven something? can you enlive something? So is en- or -en added first?

Orderings of affixes are found that are not predicted by the level order hypothesis.
For example, level 2 suffixes are found inside level 1 suffixes, e.g., governmental (#ment+al), or derviational suffixes outside compounds, e.g., nonsensical.

11/10/2010

Problems with level ordering


Most English affixes show evidence of being on more than one stratum. It accounts for a relatively small portion of the data.

Feature Percolation
Another major construct of Lexicalist Morphology is: Feature Percolation The features of the head of a word determine the feature of the whole word.

Right-hand Heads
Head: the element in a compound which
determines the gender and declension/conjugation class carries inflectional elements that apply to the whole compound, denotes the superordinate of the whole compound (i.e. major meaning)

11/10/2010

Whats the head?


In compounds: the right hand element :
It determines the type/class of the element The semantic information associated with the right-hand element is associated with the compound as a whole.

Example: laundry basket


Is a type of basket (hyponym) Is a noun

Whats the head?


In derived words, the head is usually the derivational suffix. With inflectional affixes, it's less clear.

Heads and Features


The features of the head percolate up the structure The features of the daughter nodes are available to the mother Examples with plurals
Coffee mug has the plural: coffee mugs Mailman has the plural: mailmen

10

11/10/2010

Feature Percolation
If the features from both parts are in agreement, they will both percolate up to the mother node. If the features conflict, then the mother node will take the features from the head.

Example of Feature Percolation: German Compounds


Der Landesmusikdirektor = state music director It consists of 3 parts:
das Land(es) die Musik der Direktor

German compounds
Der Landesmusikdirektor The head = direktor:
Determines gender (der) Determines word class (noun) Type of meaning (a type of control)

11

11/10/2010

Landesmusikdirektor

Your turn: Draw a tree, indicating gender features percolation


Arznei+mittel+ausgaben+begrenzungs+gesetz medicine distribution control law
die Arznei -medicine das Mittel - remedy die Ausgabe - distribution/dispersement die Begrenzung - restrictions das Gesetz - law

Assume theyre added in order of occurrence


[[[[[arznei]+mittel]+ausgaben]+begrenzungs]+gesetz]

English compounds
Lest you think that German has cornered the market on long compounds, let's look at: Neoorthodox existentialist theology
Whats the head? Whats the structure? What features percolate from the head?

12

11/10/2010

Feature Percolation and Derivation


The notion of feature percolation was generalized from endocentric compounds to all word formation. With derivational affixes, the derivational affix contains the features that percolate up.

Feature Percolation & Derivation

Issues with feature percolation


Does feature percolation apply to inflectional morphology? How do you recognize the head? English: Right-hand Head Rule
All other things being equal, the head in an English word is the right-hand element in that word.

13

11/10/2010

Issues with feature percolation


Feature percolation works well for endocentric compounds and derivational suffixes. It works less well for exocentric compounds and derivational prefixes. It's also not clear how to generalize this to other languages.

Constraints
Another major feature of Lexicalist Morphology was constraints on output, e.g., Generalized Lexicalist Hypothesis (Lapointe, 1981: 22):
Syntactic rules are not allowed to refer to, and hence cannot directly modify, the internal morphological structure of words.

The Adjacency Condition (Allen, 1978: 155):


No rule of word formation can involve X and Y, unless Y is uniquely contained in the cycle adjacent to X.

Conclusions
Lexicalist Morphology needs to explain:
Status of the lexicon Issues with ordering

Contributions of Lexical Morphology:


Made a case for a separate morphology, especially for derivational morphology Started the search for universals of word formation

14

Vous aimerez peut-être aussi