Transcript 323-Morphology

323 Morphology
The Structure of Words
2. Basic Concepts
(Last updated 25 SE 06)
2.1 Lexemes and Word Forms
Words are not easy to define.
A preliminary definition is based on the English orthographic system.
The spaces used in orthography represent words (usually).
Most dictionaries list only one word of an inflected set:
E.g. sing, sang, sung, singing, sings.
The form ‘sing’ is always chosen as a dictionary entry.
The form is technically an infinitive.
In linguistics the term is lexeme represents the basic or dictionary form
of the word.
Lexemes are usually written in CAPS: SING
Lexemes are abstract representations, which presumably are listed in the
brain in a component called the lexicon.
Each inflected form of a lexeme is called a word-form.
E.g. ‘sing, sang, sung, singing, sings’ are each a word-form and
each one belongs to the lexeme SING.
The set of word-forms of a given lexeme is called a paradigm.
2.1 Lexemes and Word Forms
By convention in each language, the dictionary representation may be the
infinitive form of the verb as in Russian, the first person singular in Latin (which
has no infinitive), the third person singular in Arabic, or perhaps by some other
form. The entry form for nouns in normally the singular nominative case form
of the noun: Latin, Russian, English, Czech, German.
A lexeme family, or less formally a word family, is a set of lexemes that are
related. They should share some phonological properties and be related
semantically. The latter is easier said than determined.
E.g. print, printable, unprintable, printer, printability, reprint.
This list is not necessarily complete.
Complex lexemes are lexemes formed with an affix (a morpheme).
E.g. ‘able’, ‘un’, ‘er’, ‘ity’, ‘re’ in the above list.
Complex lexemes must each be listed separately in a dictionary as the
meaning may differ.
The various word-forms of a given lexeme do not change the
meaning of the lexeme.
Which affixes that occur with which basic lexeme is not predictable.
E.g. we find in English un-happy, un-ripe, but not *un-sad, *unred, *un-tall, and so forth.
2.1 Lexemes and Word Forms
Sometimes a lexeme with an affix occurs but the basic form does not
exist:
E.g. dis-gruntled but not *gruntled, in-cognito, but not *cognito,
un-gainly, but not*gainly.
Sometimes the expected affix does not occur but another affix does:
E.g. natural-ness in place *natural-ity.
Or the expected affix occurs with another meaning:
E.g. cook, cook-er (an instrument for cooking, not a person who
cooks, which is simply the noun ‘cook’.
Kinds of morphological relationship
inflection: the relationship between the word-forms of a lexeme.
E. g. mask, masks; sit, sat, sitting, sits; blue, bluer, bluest.
derivation: the relationship between lexemes of a lexical family.
E. g. singer, singer; write, writer; cookV, cookN, cooker.
Derivation usually implies forming one lexeme from another lexeme in the same
lexical family.
E.g. sing -> singer, write -> writer, cookV, cookN and cooker.
Word is used whenever the distinction between derivation and inflection is
uncertain. (no examples currently).
Compound *lexeme) refers to words that are made up of two or more lexemes:
doghouse, catfish, greenhouse, whiplash, tattletale, and so forth.
2.2 Morphemes
A morpheme is the smallest constituent with a function. I prefer this distinction to
‘smallest constituent with meaning. There are some forms that appears to be
constituents but have no discernable meaning, but have a function in terms of word
building:
E.g. doof-us, radi-us, cf. radi-al, radi-an.
Some inflectional morphemes have no true meaning, but they have a grammatical
function:
E.g. he, him; who, whom; they, them,
The suffix ‘-m’ marks the accusative (objective) Case. This is a syntactic relation and no
meaning can be associated with it.
The term function includes meaning.
To go one step further than H., the hierarchy for constituents is:
Sentence -> phrase -> word -> morpheme.
Phrases are very important constituents in syntax.
Some grammatical categories cannot be expressed in terms of morphemes. For
example, note the following partial inflection of the English verb sing and others similar
to it:
E.g. sing, sang, sung.
The past tense is marked by a change of the root vowel. The latter form marks two
distinct grammatical functions — the passive form of the verb and the perfect form of the
verb. Each form is a distinctive morpheme with a different function but phonologically the
same.
2.3 Affixes, Bases and Roots
Affixes are morphemes that are adjoined to the left of the base of a word or to the
right of the base of a word:
A prefix is an affix that is adjoined to the left of the base of a word.
E.g. ‘un-’ in un-happy, un-regulated; ‘re-’ re-do, re-heat, re-write, and so
forth.
A suffix is an affix that is adjoined to the right of the base of a word.
E.g. ‘s’ in book-s, cat-s; eat-s, smell-s; linguistic-s.
An infix is an affix that is inserted into the base of the word forming a noncontiguous base. There are no infixes in English. Infixes occur in the Semitic
language.
E. g. “ktb” is the base for book and read and words which refer to
book/read in some related sense. To form the noun in
Arabic, the infixes ‘I’ and ‘a’ are inserted into the base
between the firsts two consonants and the second two
consonants, respectively:
E.g. kitab.
A circumflex is an affix that occurs on both sides of the base. (H.)
E. g. (per H) German ge-les-en.
English dialects: a-walk-ing, a-read-ing..
The English “a-” is etymologically related to the German “ge-”.
Actually, this is not quite the case. In German the prefix ‘ge-’ is a morpheme and
allomorph in that it occurs in other constructions. The prefix ‘ge- must occur in construction
with certain suffixes. A circumflex is a morphological construct which contains a prefix and a
suffix in a noncontiguous string. Together the affixes in a circumflex represent one function.
Outside of the aforementioned dialect, there are no circumflexes in English.
2.3 Affixes, Bases and Roots
Stem and Root
A root is a morpheme that cannot be broken down into further
morphemes.
A base is a contiguous strings of one or more morphemes which can hold
lexical meaning.
In English the word dog, for example, is a root since it cannot be broken into further
morphological units:
E. g. ‘do’ is not a morpheme of dog, it is basically a verb. There is no
morpheme ‘og’ that has any kind of function.
Dog is also a base. It has lexical meaning.
The English word disgruntled consists of three morpheme dis-, gruntle,
and ed. ‘dis’ is a
prefix, and ed’ is an inflectional affix marking the past tense among other functions. The
morpheme gruntle is a root, since two affixes are adjoined to it. It is not a base, since it has
no lexical meaning (what does gruntle mean?) Once both affixes are adjoined to it, then
disgruntled, which is a base, is a lexical stem since it does have meaning.
Technically, the prefix ‘dis-’ is adjoined first to gruntle to form the base
‘disgruntle’.
Apparently this form has no lexical meaning and remains a base. Once the adjectival suffix ‘ed’ is added to disgruntle
then the base receives lexical meaning and is a stem.
English has several words usually considered compounds, where at least one member of the
compound doesn’t behave like a normal prefix or affix.
E. g. tele-graph. Although graph may have lexical meaning, tele- does not. It does not
occur in isolation. The form is borrowed from Greek where it
means ‘far’. It is
more like a root that cannot become a stem in its own
right, but it may
be adjoined to a stem to form a new stem. This
particular property makes it
look like an affix, or, why are affixes not
roots?
2.4 Formal Operations
Some words such as derive imply a process. A true process is a historical phenomenon and
does not imply a process in terms of how language is represented in the mind (the grammar of
a language). For some yet to be determined reason, H considers affixation and compounding to
be concatenative (the addition of morphemes on to a string; e.g. hope-less-ness. Certain kinds
of inflection and other constructions he considers to be non-concatenative; e.g. English sing,
sang, sung (there is no past tense morpheme here)
Another non-concatenative structure include word whose final consonant becomes voiced, final
consonant becomes palatalized, or gemination of a root consonant.
E.g. Albanian: armik [-q] (Sg.), armiq [-c] (Pl.).
Note: [c] is not a palatalized consonant. The form came about through
palatalization, which is not visible/hearable in the phone [c].
E.g. English: hoof [hƱf] (Sg.), hooves [hƱv-z] (Pl.).
E.g. Arabic causative verbs: darasa (noncausative), darrasa (causative).
Gemination is the doubling of a consonant.
Reduplication is the copying of a syllable or part of a syllable:
1. Prereduplication:
E.g. Ponapean: duhp (nonprogressive), du-duhp (progressive) ‘(be) diving.’
A weak syllable (no coda) is copied).
2. Postreduplication.
E.g. Mangap-Mbula: kuk (nonprogressive), kuk-uk )progressive) ‘(be) barking’
The rhyme of the syllable is copied.
3. Duplifixing is adding an affix and reduplicated part of the stem:
E.g. Somali: buug (Sg.), buug-ag (Pl.) ‘book(s;, fool (Sg.), fool-al (Pl.) .book’.
The vowel ‘a’ is like a suffix in that it is invariable. The consonants
‘g’ and ‘l’ are copied from the stem final consonant and placed after ‘a’.
2.4 Formal Operations
E.g. Tsutujil: saq (Sg.) ‘’white’, saq-soj ’whitish’.
‘s’ is reduplicated from the initial consonant of the stem, and ‘oj’ is a nonvariable suffix.
Subtraction is the omission of one or more final segments of the base.
E.g. Murle: nyoon (Sg.), nyoo (Pl.) ‘lamb’
Strong suppletion is replacing one form with another form (allomorph) that is phonologically unrelated
to it the replacee.
E.g. the forms of the English verb be: ‘is’, ‘are’, ‘was/were’.
Weak suppletion if replacing one form with another form (allomorph) which share some common
phonological forms, but not all phonological forms are common to both:
E.g. sing, sang (/i/, /æ/), foot, feet (/Ʊ/, /i/).
‘Base’ is redefined (H):
The base of a morphologically complex word is the element to which a morphological operation
applies.
This definition works a long as we assume a zero operation that may derive one form from
another forms is derived with no phonological change. We can say a base may be derived from
a root with a zero morphological operation.
E.g. the noun ‘push’ (he gave me a push) is derived from the verb ‘push’. The derivation is
a zero operation in that there is no overt sign marking this. We could represent this as:
[N[V PUSH]], probably in later chapters.
A morphological pattern refers to the various ways a particular grammatical or lexical feature can be
expressed. There are four morphological patterns of the past tense of the English verb:
E.g. the default suffix ’-ed’, the irregular suffix ‘-d’ (tell, told), ,the irregular suffix ‘-T (feel, felt),
and the vowel replacement system of strong verbs (sing, sang; drink, drank).
2.5 Morphemes and Allomorphs
A morpheme is a set of allomorphs. Most linguists would agree with this even if they are not familiar
with set theory. The problem is how to account the variation. For the past 45 years or so, the theory
of underlying representation. This theory states that there is an abstract (usually) form from which
the other allomorphs are derived. H refers to these as a ’fictitious underlying representation’
(p.27). H does not elaborate here. I agree with him.
The approach that I favour is set theory.To review Korean, there are two allomorphs (members) of
the set for the plural of nouns: {ul lul} (also written as {{ul} {lul}}. The standard to write morphemes
and allomorphs with hyphens to show that the morpheme or allomorph is an affix. I is not a
theoretical divergence. As I mentioned before one of the allomorphs of the plural morpheme is the
default. The nondefault allomorph must be marked with information indicating the contexts in which
the allomorph occurs. The default allomorph usually corresponds with the underlying form. The
default or underlying allomorph is normally determined, in part, at least, by is distribution. There are
fewer vowels than consonants in Korean. If -ul, which follows consonants, is the default, then the
selection of -lul has a more constrained condition. The rule writing form will be dealt with later.
In the Russian example on p. 27, H considers ZAMOK-I castles to be the underlying form for the
plural form. The suffix ‘-ok’ must be marked in its grammatical entry (the grammaticon) to indicate
that the vowel /o/ in the suffix /ok/ is deleted if the inflectional affix begins with a vowel. I, too, would
consider the allomorph /ok/ as the default. And I would marked the other allomorph with the same
information indicating that /k/ is chosen if the suffix begins with a vowel.
Lexical conditioning is the situation where on suppletive (weak or strong) allomorph is dependent on
a particular lexical item but not on a class. The English plural ‘-en’ occurs with only three nouns, one
of them nearly obsolete: ox-en, childr-en, and brethr-en. The suffix the result of a lexical property
called lexical conditioning.
2.6 Some Problems in Morpheme Analysis
H mentions a problem arising from suppletion. The plural allomorphs ‘-s’ and ‘-en’ in English are related
by suppletion. They share no exclusive phonological properties. H raises the question whether the two
suffixes are manifestations of the same morpheme. H leans toward this view. So do I.
My view is determined by the claim that all morphemes must have a form, a function and a sign. I will
illustrate with the progressive participle suffix ‘-ing’:
The program I am using to make graphics does not import unicode phonetics. I am using here ‘ñ’ for
engma, the nasal velar [ŋ].
[+Progressive] is the feature denoting the progressive aspect; the form is a suffix which is adjoined
to a noun host (base); and the sign is /ɩŋ/.
There are plural signs for nouns in English: /z/ and /ɩn/. These two allomorphs are strongly
suppletive. They are shown in the following grammeme (entry form for grammatical morphemes):
2.6 Some Problems in Morpheme Analysis
Grammeme: [+Pl]
[+Plural, Noun]
function
+Suffix
+Host
+Noun
form
{/in/, / {CHILD, BROTHER, OX} ___},
{/iz/, [default]}
sign
The two allomorphs here form a ‘natural’ set, since they share the same function. The fact
that they are the same form supports this claim. If they are in the same set, then they must
be a member of the [+Pl]. And if they are in the same set they must be allomorphs.
A morpheme may consist of two or more features. For example, the English verbal suffix ‘s’ marks agreement with a third person singular subject and it marks the present tense.
The suffix in the above figure contains two subfeatures [+Host] and [+Noun]. Agglutinating
languages do not do this, with some minor exceptions. This cumulative expression is
also called fusion.
A zero expression ‘ø’ means that there is no overt affix to mark a function. ‘ø’ has been the
topic of notable debates. Until very recently I was opposed to the notion of ‘ø’ until I started
learning set theory. Set theory permits empty sets often written as ‘ø’. A zero expression
grammeme is not entirely empty; the sign and the form are empty. It is now considered
better to consider the singular morphological operation for nouns as ‘ø’. It thus has the
following grammemical entry:
2.6 Some Problems in Morpheme Analysis
Grammeme: [-Pl, +Noun]
[-Plural, +Noun]
function
ø
form
ø
sign
Only the function is not empty; it merely has no form and no sign.
An empty morpheme is an affix that has no meaning, but has a function: it forms a base
to which certain meaningful affixes are adjoined. This occurs in English when nouns are
borrowed from Greek and Latin and retain their plural form. The singular ending occurs
in English an empty (ø) morph:
E.g. radi-us (Sg.), radi-i (Pl.); agend-a (Sg.), agend-ae (Pl.); phenomen-on (Sg.),
phenomen-a (Pl.).
The plural form is adjoined to the base, respectively: ‘I’, ‘ae’, ‘a’. In English the Sg. form
is morphologically null. The suffixes in the above three examples are stem-enders, an
empty morpheme required when there is no suffix adjoined to the word. This applies to
derivatives as well: radi-al, phenomen-al, and so forth. The grammemical entry for ‘-us’
is:
Grammeme: [stem extender]
stem extender when
there is no affix, 'us'
class
•
function
[+suffix]
form
us
sign
Go to Course Outline, Go to Chapter 1, Go to Chapter 3, Go to Exercise 2