Item and Pattern Morphology

James P. Blevins

University of Cambridge
[email protected]

Quantitative Measures in Morphology & Morphological

Development, ..
Outline
 Morphological analysis
Types of phenomena
Conceptions of analysis

 Models of grammatical analysis

Morphemic models
Item and pattern models
Stem-based implication
 Implicational analysis
System factorization
Fractured noun declensions in German
Modelling implicational structure
 Summary and implications
 Morphological analysis

Items and patterns

Why are item/pattern models relevant to quantitative measures?

Because they provide appropriate items to count, i.e., words.

Why are these models relevant to morphological development?

Because they incorporate a speaker-oriented perspective, which
addresses tasks that speakers face in language acquisition and use.
 Morphological analysis

Items and patterns

Why are item/pattern models relevant to quantitative measures?

Because they provide appropriate items to count, i.e., words.

Why are these models relevant to morphological development?

Because they incorporate a speaker-oriented perspective, which
addresses tasks that speakers face in language acquisition and use.
 Morphological analysis

Items and patterns

Why are item/pattern models relevant to quantitative measures?

Because they provide appropriate items to count, i.e., words.

Why are these models relevant to morphological development?

Because they incorporate a speaker-oriented perspective, which
addresses tasks that speakers face in language acquisition and use.
 Morphological analysis

Items and patterns

Why are item/pattern models relevant to quantitative measures?

Because they provide appropriate items to count, i.e., words.

Why are these models relevant to morphological development?

Because they incorporate a speaker-oriented perspective, which
addresses tasks that speakers face in language acquisition and use.
 Morphological analysis Types of phenomena

Speaker-oriented analysis
Morphological analysis models the morphological information
that speakers exploit in the acquisition and use of a language.
Speakers need to be able to produce and interpret the forms of a
language, including forms that they may not have encountered.
In languages with rich inectional morphology, this is the
‘paradigm cell lling problem’ (Ackerman et al. ).
The diculty of this task does not appear to dier signicantly
across languages (Malouf & Ackerman a,b), presumably
reecting adaptive pressures on language acquisition and use.
 Morphological analysis Types of phenomena

Speaker-oriented analysis
Morphological analysis models the morphological information
that speakers exploit in the acquisition and use of a language.
Speakers need to be able to produce and interpret the forms of a
language, including forms that they may not have encountered.
In languages with rich inectional morphology, this is the
‘paradigm cell lling problem’ (Ackerman et al. ).
The diculty of this task does not appear to dier signicantly
across languages (Malouf & Ackerman a,b), presumably
reecting adaptive pressures on language acquisition and use.
 Morphological analysis Types of phenomena

Speaker-oriented analysis
Morphological analysis models the morphological information
that speakers exploit in the acquisition and use of a language.
Speakers need to be able to produce and interpret the forms of a
language, including forms that they may not have encountered.
In languages with rich inectional morphology, this is the
‘paradigm cell lling problem’ (Ackerman et al. ).
The diculty of this task does not appear to dier signicantly
across languages (Malouf & Ackerman a,b), presumably
reecting adaptive pressures on language acquisition and use.
 Morphological analysis Types of phenomena

Speaker-oriented analysis
Morphological analysis models the morphological information
that speakers exploit in the acquisition and use of a language.
Speakers need to be able to produce and interpret the forms of a
language, including forms that they may not have encountered.
In languages with rich inectional morphology, this is the
‘paradigm cell lling problem’ (Ackerman et al. ).
The diculty of this task does not appear to dier signicantly
across languages (Malouf & Ackerman a,b), presumably
reecting adaptive pressures on language acquisition and use.
 Morphological analysis Types of phenomena

Theoretical lexicography

Many other familiar properties of morphological descriptions fall

under what might be called ‘theoretical lexicography’:
Speakers do  need to be able to associate a single meaning
with individual formatives, to assign invariant segmentations to
forms, to identify a unique set of principal parts, to assign forms to
a xed number of inection classes, etc.

These properties of languages exhibit far greater variation,

presumably reecting the lack of adaptive pressures.
 Morphological analysis Types of phenomena

Theoretical lexicography

Many other familiar properties of morphological descriptions fall

under what might be called ‘theoretical lexicography’:
Speakers do  need to be able to associate a single meaning
with individual formatives, to assign invariant segmentations to
forms, to identify a unique set of principal parts, to assign forms to
a xed number of inection classes, etc.

These properties of languages exhibit far greater variation,

presumably reecting the lack of adaptive pressures.
 Morphological analysis Types of phenomena

Theoretical lexicography

Many other familiar properties of morphological descriptions fall

under what might be called ‘theoretical lexicography’:
Speakers do  need to be able to associate a single meaning
with individual formatives, to assign invariant segmentations to
forms, to identify a unique set of principal parts, to assign forms to
a xed number of inection classes, etc.

These properties of languages exhibit far greater variation,

presumably reecting the lack of adaptive pressures.
 Morphological analysis Conceptions of analysis

Atomistic (Neo-Bloomeldian) approaches

Language acquisition involves learning form-meaning mappings.

Language use involves the interpretation of forms and the
realization of meanings in isolation from other forms/meanings.
The lexicon is a redundancy-free repository of meaningful units.
 Morphological analysis Conceptions of analysis

Atomistic (Neo-Bloomeldian) approaches

Language acquisition involves learning form-meaning mappings.

Language use involves the interpretation of forms and the
realization of meanings in isolation from other forms/meanings.
The lexicon is a redundancy-free repository of meaningful units.
 Morphological analysis Conceptions of analysis

Atomistic (Neo-Bloomeldian) approaches

Language acquisition involves learning form-meaning mappings.

Language use involves the interpretation of forms and the
realization of meanings in isolation from other forms/meanings.
The lexicon is a redundancy-free repository of meaningful units.
 Morphological analysis Conceptions of analysis

Neo-Neogrammarian models

Language acquisition involves learning (i) form inventories, the

(ii) distribution of forms in paradigms and in syntactic contexts
and (iii) the functions and meanings associated with forms.
Language use involves highly aggressive prediction (possibly to
minimize/reduce the amount of real-time processing of input).
The lexicon is a statistical inferencing engine that provides an
analogical base for extending encountered patterns to new items.
 Morphological analysis Conceptions of analysis

Neo-Neogrammarian models

Language acquisition involves learning (i) form inventories, the

(ii) distribution of forms in paradigms and in syntactic contexts
and (iii) the functions and meanings associated with forms.
Language use involves highly aggressive prediction (possibly to
minimize/reduce the amount of real-time processing of input).
The lexicon is a statistical inferencing engine that provides an
analogical base for extending encountered patterns to new items.
 Morphological analysis Conceptions of analysis

Neo-Neogrammarian models

Language acquisition involves learning (i) form inventories, the

(ii) distribution of forms in paradigms and in syntactic contexts
and (iii) the functions and meanings associated with forms.
Language use involves highly aggressive prediction (possibly to
minimize/reduce the amount of real-time processing of input).
The lexicon is a statistical inferencing engine that provides an
analogical base for extending encountered patterns to new items.
 Morphological analysis Conceptions of analysis

Implicational morphology

Implicational approaches model speaker-oriented analysis.

‘Word and Paradigm’ models are pure implicational approaches.
Words and paradigms merely sanction reliable implications.
 Morphological analysis Conceptions of analysis

Implicational morphology

Implicational approaches model speaker-oriented analysis.

‘Word and Paradigm’ models are pure implicational approaches.
Words and paradigms merely sanction reliable implications.
 Morphological analysis Conceptions of analysis

Implicational morphology

Implicational approaches model speaker-oriented analysis.

‘Word and Paradigm’ models are pure implicational approaches.
Words and paradigms merely sanction reliable implications.
 Models of grammatical analysis Morphemic models

Item and arrangement

The essence of IA is to talk simply of things and the arrangements

in which those things occur (Hockett : )

The grammar, or grammatical system, of a language is () 

    , and ()   
        
. (Hockett : )
 Models of grammatical analysis Morphemic models

Item and arrangement

The essence of IA is to talk simply of things and the arrangements

in which those things occur (Hockett : )

The grammar, or grammatical system, of a language is () 

    , and ()   
        
. (Hockett : )
 Models of grammatical analysis Morphemic models

Problems of morphemic analysis

One motive for the post-Bloomeldian model consisted, that is to

say, in a genuinely factual assertion about language: namely, that
there is some sort of matching between minimal ‘sames’ of ‘form’
(morphs) and ‘meaning’ (morphemes). Qua factual assertion this
has subsequently proved false: for certain languages, such as Latin,
the correspondence which was envisaged apparently does not exist
… One is bound to suspect, in the light of such a conclusion, that
the model is in some sense wrong. (Matthews : )
 Models of grammatical analysis Morphemic models

Item and process

Morphophonemes, morphs, phones, and acoustic phones are

   or   , not
elements in a language. (Hockett : )

Some of the phonemic material in a derived form may be, not part
of any underlying form, but rather a  or  of
the process. (Hockett : )
 Models of grammatical analysis Morphemic models

Item and process

Morphophonemes, morphs, phones, and acoustic phones are

   or   , not
elements in a language. (Hockett : )

Some of the phonemic material in a derived form may be, not part
of any underlying form, but rather a  or  of
the process. (Hockett : )
 Models of grammatical analysis Item and pattern models

‘WP’ models

Quite apart from minor variants of IP or IA, or models that might

be invented tomorrow, there is one model which is clearly distinct
from either IA or IP, and which is older and more respectable than
either. This is the    (WP) model, the traditional
framework for the discussion of Latin, Greek, Sanskrit, and a good
many more modern familiar languages. … (Hockett : )

An inectional system is factored into a set of exemplary patterns

and a set of diagnostic principal parts for non-exemplary items.
 Models of grammatical analysis Item and pattern models

‘WP’ models

Quite apart from minor variants of IP or IA, or models that might

be invented tomorrow, there is one model which is clearly distinct
from either IA or IP, and which is older and more respectable than
either. This is the    (WP) model, the traditional
framework for the discussion of Latin, Greek, Sanskrit, and a good
many more modern familiar languages. … (Hockett : )

An inectional system is factored into a set of exemplary patterns

and a set of diagnostic principal parts for non-exemplary items.
 Models of grammatical analysis Item and pattern models

A (slight) mischaracterization

‘Item and arrangement’ and ‘item and process’ both refer to units
(‘items’) and modes of combination (‘arrangements’/‘processes’).

But ‘word and paradigm’ refers to a  unit (‘word’) and

a  network of units (inectional ‘paradigm’).

The third model should be ‘item and pattern’, where comparison

of the item against the pattern sanctions the deduction of forms.
 Models of grammatical analysis Item and pattern models

A (slight) mischaracterization

‘Item and arrangement’ and ‘item and process’ both refer to units
(‘items’) and modes of combination (‘arrangements’/‘processes’).

But ‘word and paradigm’ refers to a  unit (‘word’) and

a  network of units (inectional ‘paradigm’).

The third model should be ‘item and pattern’, where comparison

of the item against the pattern sanctions the deduction of forms.
 Models of grammatical analysis Item and pattern models

A (slight) mischaracterization

‘Item and arrangement’ and ‘item and process’ both refer to units
(‘items’) and modes of combination (‘arrangements’/‘processes’).

But ‘word and paradigm’ refers to a  unit (‘word’) and

a  network of units (inectional ‘paradigm’).

The third model should be ‘item and pattern’, where comparison

of the item against the pattern sanctions the deduction of forms.
 Models of grammatical analysis Item and pattern models

Why words?

Words are often of greater predictive value than sub-word units.

They are  because they are more consistently
demarcated in the speech stream than sub-word units.
They are more unambiguously  than sub-word units.

The interpretation of a morphological unit is just not the sum of

the interpretation of its parts:  may be distinctive.
 Models of grammatical analysis Item and pattern models

Why words?

Words are often of greater predictive value than sub-word units.

They are  because they are more consistently
demarcated in the speech stream than sub-word units.
They are more unambiguously  than sub-word units.

The interpretation of a morphological unit is just not the sum of

the interpretation of its parts:  may be distinctive.
 Models of grammatical analysis Item and pattern models

Why words?

Words are often of greater predictive value than sub-word units.

They are  because they are more consistently
demarcated in the speech stream than sub-word units.
They are more unambiguously  than sub-word units.

The interpretation of a morphological unit is just not the sum of

the interpretation of its parts:  may be distinctive.
 Models of grammatical analysis Item and pattern models

Why words?

Words are often of greater predictive value than sub-word units.

They are  because they are more consistently
demarcated in the speech stream than sub-word units.
They are more unambiguously  than sub-word units.

The interpretation of a morphological unit is just not the sum of

the interpretation of its parts:  may be distinctive.
 Models of grammatical analysis Item and pattern models

Why words?

The word is a more stable and solid focus of grammatical relations

than the component morpheme by itself. Put another way,
grammatical statements are abstractions, but they are more
protably abstracted from words as wholes than from individual
morphemes. (Robins : ).
 Models of grammatical analysis Item and pattern models

Why paradigms?

Paradigms exhibit the most reliable patterns of interpredictability

because they are dened over a closed, uniform feature space:
Notions like ‘morphological gap’, ‘suppletion’, and even
‘syncretism’ are mainly applied to inectional paradigms.
Conversely, inected forms are rarely taken to be ‘established’.
 Models of grammatical analysis Item and pattern models

Why paradigms?

Paradigms exhibit the most reliable patterns of interpredictability

because they are dened over a closed, uniform feature space:
Notions like ‘morphological gap’, ‘suppletion’, and even
‘syncretism’ are mainly applied to inectional paradigms.
Conversely, inected forms are rarely taken to be ‘established’.
 Models of grammatical analysis Item and pattern models

Why paradigms?

Paradigms exhibit the most reliable patterns of interpredictability

because they are dened over a closed, uniform feature space:
Notions like ‘morphological gap’, ‘suppletion’, and even
‘syncretism’ are mainly applied to inectional paradigms.
Conversely, inected forms are rarely taken to be ‘established’.
 Models of grammatical analysis Item and pattern models

The scope of implication

Yet ‘word and paradigm’ is a specic instance of an ‘item and
pattern’ model and debates about the morphological status of
words and paradigms, though important, are subsidiary:
In periphrastic constructions (Ackerman & Stump ), units
larger than the word may be an equally “stable and solid focus of
grammatical relations” (cf. Robins ).
A perfectly agglutinative system would also sanction reliable
predictions between ‘units of meaning’ and ‘units of form’.
Derivational paradigms and other types of morphological families
also exhibit (a typically weaker) implicational structure.
 Models of grammatical analysis Item and pattern models

The scope of implication

Yet ‘word and paradigm’ is a specic instance of an ‘item and
pattern’ model and debates about the morphological status of
words and paradigms, though important, are subsidiary:
In periphrastic constructions (Ackerman & Stump ), units
larger than the word may be an equally “stable and solid focus of
grammatical relations” (cf. Robins ).
A perfectly agglutinative system would also sanction reliable
predictions between ‘units of meaning’ and ‘units of form’.
Derivational paradigms and other types of morphological families
also exhibit (a typically weaker) implicational structure.
 Models of grammatical analysis Item and pattern models

The scope of implication

Yet ‘word and paradigm’ is a specic instance of an ‘item and
pattern’ model and debates about the morphological status of
words and paradigms, though important, are subsidiary:
In periphrastic constructions (Ackerman & Stump ), units
larger than the word may be an equally “stable and solid focus of
grammatical relations” (cf. Robins ).
A perfectly agglutinative system would also sanction reliable
predictions between ‘units of meaning’ and ‘units of form’.
Derivational paradigms and other types of morphological families
also exhibit (a typically weaker) implicational structure.
 Models of grammatical analysis Item and pattern models

The scope of implication

Yet ‘word and paradigm’ is a specic instance of an ‘item and
pattern’ model and debates about the morphological status of
words and paradigms, though important, are subsidiary:
In periphrastic constructions (Ackerman & Stump ), units
larger than the word may be an equally “stable and solid focus of
grammatical relations” (cf. Robins ).
A perfectly agglutinative system would also sanction reliable
predictions between ‘units of meaning’ and ‘units of form’.
Derivational paradigms and other types of morphological families
also exhibit (a typically weaker) implicational structure.
 Models of grammatical analysis Stem-based implication

Stem and pattern systems?

Sub-word units consisting of roots and formatives are often

informative in ways that roots in isolation are not, leading to
analyses based on abstract stems (Arono ), stem sets
(Anderson ) or stem spaces (Bonami & Boyé ).
Yet stem-based approaches create a pair of new challenges:
The demarcation of stems raises new diculties of segmentation.
The interpretation and predictive value of stems often depends on
their distribution within full word forms.
 Models of grammatical analysis Stem-based implication

Stem and pattern systems?

Sub-word units consisting of roots and formatives are often

informative in ways that roots in isolation are not, leading to
analyses based on abstract stems (Arono ), stem sets
(Anderson ) or stem spaces (Bonami & Boyé ).
Yet stem-based approaches create a pair of new challenges:
The demarcation of stems raises new diculties of segmentation.
The interpretation and predictive value of stems often depends on
their distribution within full word forms.
 Models of grammatical analysis Stem-based implication

Stem and pattern systems?

Sub-word units consisting of roots and formatives are often

informative in ways that roots in isolation are not, leading to
analyses based on abstract stems (Arono ), stem sets
(Anderson ) or stem spaces (Bonami & Boyé ).
Yet stem-based approaches create a pair of new challenges:
The demarcation of stems raises new diculties of segmentation.
The interpretation and predictive value of stems often depends on
their distribution within full word forms.
 Models of grammatical analysis Stem-based implication

Stem and pattern systems?

Sub-word units consisting of roots and formatives are often

informative in ways that roots in isolation are not, leading to
analyses based on abstract stems (Arono ), stem sets
(Anderson ) or stem spaces (Bonami & Boyé ).
Yet stem-based approaches create a pair of new challenges:
The demarcation of stems raises new diculties of segmentation.
The interpretation and predictive value of stems often depends on
their distribution within full word forms.
 Models of grammatical analysis Stem-based implication

Problems of segmentation I

The ‘recalcitrance’ of English children:

 child + ren ∼ childr + en — “each of the points of division has
advantages and disadvantages” (Harris  ).
 child + r + en ∼ child + ablaut + en ∼ children — “this is one of the
cases in which all of our preferential criteria … fail and nothing
remains but a resort to convenience” (Hockett : ).
 Models of grammatical analysis Stem-based implication

Problems of segmentation I

The ‘recalcitrance’ of English children:

 child + ren ∼ childr + en — “each of the points of division has
advantages and disadvantages” (Harris  ).
 child + r + en ∼ child + ablaut + en ∼ children — “this is one of the
cases in which all of our preferential criteria … fail and nothing
remains but a resort to convenience” (Hockett : ).
 Models of grammatical analysis Stem-based implication

Problems of segmentation I

The ‘recalcitrance’ of English children:

 child + ren ∼ childr + en — “each of the points of division has
advantages and disadvantages” (Harris  ).
 child + r + en ∼ child + ablaut + en ∼ children — “this is one of the
cases in which all of our preferential criteria … fail and nothing
remains but a resort to convenience” (Hockett : ).
 Models of grammatical analysis Stem-based implication

Problems of segmentation II

The order of morphemes is xed [in Spanish]: (derivational

prex(es)) + lexical stem + theme vowel + tense marker (sometimes
including an empty morph) + person marker. Some forms,
however, have fused in the course of history and a neat
segmentation is not always possible. The preterit is the most
dicult paradigm to analyse, since the theme vowel is sometimes
indistinguishable, and segmenting the second and third person
plural markers in the regular way, /-is, -n/, leaves an awkward
residue that occurs nowhere else in the system. (Green : )
 Models of grammatical analysis Stem-based implication

Parasitism
Sg Pl Sg Pl
Nom kuotam kuotamazh kuorta kuortozh
Gen kuotama kuotamii kerta kuortoi
Dat kuotamaa kuotamazhta kertaa kuortozhta
Erg kuotamuo kuotamazh kertuo kuortozh
All kuotamaga kuotamazhka kertaga kuortozhka
Ins kuotamaca kuotamazhca kertaca kuortozhca
Lat kuotamagh kuotamegh kertagh kuortuojegh
Ins kuotamal kuotamel kertal kuortuojel
hen () head ()
Ingush noun declensions (Nichols )
 Models of grammatical analysis Stem-based implication

Predictive syncretism in Ingush

In the singular, a genitive form X predicts (and is predicted by):

a dative X + a, an allative X + ga and an instrumental X + ca.

In the plural, an ergative form X predicts (and is predicted by):

a dative X + ta, an allative X + ka and an instrumental X + ca.
 Models of grammatical analysis Stem-based implication

Predictive syncretism in Ingush

In the singular, a genitive form X predicts (and is predicted by):

a dative X + a, an allative X + ga and an instrumental X + ca.

In the plural, an ergative form X predicts (and is predicted by):

a dative X + ta, an allative X + ka and an instrumental X + ca.
 Models of grammatical analysis Stem-based implication

Predictive syncretism in Ingush

In the singular, a genitive form X predicts (and is predicted by):

a dative X + a, an allative X + ga and an instrumental X + ca.

In the plural, an ergative form X predicts (and is predicted by):

a dative X + ta, an allative X + ka and an instrumental X + ca.
 Models of grammatical analysis Stem-based implication

Predictive syncretism in Ingush

In the singular, a genitive form X predicts (and is predicted by):

a dative X + a, an allative X + ga and an instrumental X + ca.

In the plural, an ergative form X predicts (and is predicted by):

a dative X + ta, an allative X + ka and an instrumental X + ca.
 Models of grammatical analysis Stem-based implication

Parasitism
Sg Pl Sg Pl
Nom kuotam kuotamazh kuorta kuortozh
Gen kuotama kuotamii kerta kuortoi
Dat kuotamaa kuotamazhta kertaa kuortozhta
Erg kuotamuo kuotamazh kertuo kuortozh
All kuotamaga kuotamazhka kertaga kuortozhka
Ins kuotamaca kuotamazhca kertaca kuortozhca
Lat kuotamagh kuotamegh kertagh kuortuojegh
Ins kuotamal kuotamel kertal kuortuojel
hen () head ()
Ingush noun declensions (Nichols )
 Models of grammatical analysis Stem-based implication

An abstract stem in Daghestanian?

Two dierent opinions can be found in the literature: (a) these
markers [e.g. -zh JPB] are markers of the ergative case and all
oblique cases are formed from the ergative; (b) these markers are
markers of the oblique stem (of the singular or plural) and the
ergative has no special marker and coincides with the oblique stem
of the appropriate number. The rst point of view is unsatisfactory:
it does not take account of the semantics of the oblique cases
(ergative meaning is not a component here), nor of the data from
other Daghestanian languages, where the ergative frequently has a
special morphological marker like other oblique cases …(Kibrik
: )
 Models of grammatical analysis Stem-based implication

Oblique stems in Ingush?

It is straightforward to state the fact that the  of genitive

singulars and ergative plurals predict the form of the
corresponding datives, allatives and instrumentals.
Expressing these patterns using oblique stems just creates the
problem of annotating underspecied stem entries in such a way
that they can be ‘selected’ as the base for the oblique case forms.
 Models of grammatical analysis Stem-based implication

Oblique stems in Ingush?

It is straightforward to state the fact that the  of genitive

singulars and ergative plurals predict the form of the
corresponding datives, allatives and instrumentals.
Expressing these patterns using oblique stems just creates the
problem of annotating underspecied stem entries in such a way
that they can be ‘selected’ as the base for the oblique case forms.
 Models of grammatical analysis Stem-based implication

Layered parasitism in Estonian

Sing Plu
Nominative lukk lukud
Genitive luku lukkude
Partitive lukku lukkusid
Illa2/Part2 lukku lukke
Illative lukusse lukkudesse
Inessive lukus lukkudes
Elative lukust lukkudest
Allative lukule lukkudele
Adessive lukul lukkudel
Ablative lukult lukkudelt
Translative lukuks lukkudeks
Terminative lukuni lukkudeni
Essive lukuna lukkudena
Abessive lukuta lukkudeta
Comitative lukuga lukkudega
 Models of grammatical analysis Stem-based implication

Layered structure of case forms

lukk → lukku → lukkude → lukkudesse
Nominative singular lukk → partitive singular lukku → genitive
plural lukkude → illative plural lukkudesse.
Stem  lukk → Stem  lukku → Stem  lukkude → lukkudesse.

Stems are identiable from word forms that they underlie.

But the stems cannot be assigned properties in isolation that
determine their distribution in the paradigm of an item.
‘Indexing’ each of these stems in ways that denes their
distribution gives rise to a diacritic model of morphology.
 Models of grammatical analysis Stem-based implication

Layered structure of case forms

lukk → lukku → lukkude → lukkudesse
Nominative singular lukk → partitive singular lukku → genitive
plural lukkude → illative plural lukkudesse.
Stem  lukk → Stem  lukku → Stem  lukkude → lukkudesse.

Stems are identiable from word forms that they underlie.

But the stems cannot be assigned properties in isolation that
determine their distribution in the paradigm of an item.
‘Indexing’ each of these stems in ways that denes their
distribution gives rise to a diacritic model of morphology.
 Models of grammatical analysis Stem-based implication

Layered structure of case forms

lukk → lukku → lukkude → lukkudesse
Nominative singular lukk → partitive singular lukku → genitive
plural lukkude → illative plural lukkudesse.
Stem  lukk → Stem  lukku → Stem  lukkude → lukkudesse.

Stems are identiable from word forms that they underlie.

But the stems cannot be assigned properties in isolation that
determine their distribution in the paradigm of an item.
‘Indexing’ each of these stems in ways that denes their
distribution gives rise to a diacritic model of morphology.
 Models of grammatical analysis Stem-based implication

Layered structure of case forms

lukk → lukku → lukkude → lukkudesse
Nominative singular lukk → partitive singular lukku → genitive
plural lukkude → illative plural lukkudesse.
Stem  lukk → Stem  lukku → Stem  lukkude → lukkudesse.

Stems are identiable from word forms that they underlie.

But the stems cannot be assigned properties in isolation that
determine their distribution in the paradigm of an item.
‘Indexing’ each of these stems in ways that denes their
distribution gives rise to a diacritic model of morphology.
 Models of grammatical analysis Stem-based implication

Layered structure of case forms

lukk → lukku → lukkude → lukkudesse
Nominative singular lukk → partitive singular lukku → genitive
plural lukkude → illative plural lukkudesse.
Stem  lukk → Stem  lukku → Stem  lukkude → lukkudesse.

Stems are identiable from word forms that they underlie.

But the stems cannot be assigned properties in isolation that
determine their distribution in the paradigm of an item.
‘Indexing’ each of these stems in ways that denes their
distribution gives rise to a diacritic model of morphology.
 Models of grammatical analysis Stem-based implication

Layered structure of case forms

lukk → lukku → lukkude → lukkudesse
Nominative singular lukk → partitive singular lukku → genitive
plural lukkude → illative plural lukkudesse.
Stem  lukk → Stem  lukku → Stem  lukkude → lukkudesse.

Stems are identiable from word forms that they underlie.

But the stems cannot be assigned properties in isolation that
determine their distribution in the paradigm of an item.
‘Indexing’ each of these stems in ways that denes their
distribution gives rise to a diacritic model of morphology.
 Models of grammatical analysis Stem-based implication

Implicational structure of grammatical case forms

The paradigm of  is predictable from partitive singular lukku.

The grammatical forms are directly predictable via

Identity: ‘short’ illative singular lukku,
Truncation: nominative singular lukk,
Shortening: genitive singular luku,
Suxation: genitive plurals lukkude and partitive plural lukkusid.
 Models of grammatical analysis Stem-based implication

Implicational structure of grammatical case forms

The paradigm of  is predictable from partitive singular lukku.

The grammatical forms are directly predictable via

Identity: ‘short’ illative singular lukku,
Truncation: nominative singular lukk,
Shortening: genitive singular luku,
Suxation: genitive plurals lukkude and partitive plural lukkusid.
 Models of grammatical analysis Stem-based implication

Implicational structure of grammatical case forms

The paradigm of  is predictable from partitive singular lukku.

The grammatical forms are directly predictable via

Identity: ‘short’ illative singular lukku,
Truncation: nominative singular lukk,
Shortening: genitive singular luku,
Suxation: genitive plurals lukkude and partitive plural lukkusid.
 Models of grammatical analysis Stem-based implication

Implicational structure of grammatical case forms

The paradigm of  is predictable from partitive singular lukku.

The grammatical forms are directly predictable via

Identity: ‘short’ illative singular lukku,
Truncation: nominative singular lukk,
Shortening: genitive singular luku,
Suxation: genitive plurals lukkude and partitive plural lukkusid.
 Models of grammatical analysis Stem-based implication

Implicational structure of grammatical case forms

The paradigm of  is predictable from partitive singular lukku.

The grammatical forms are directly predictable via

Identity: ‘short’ illative singular lukku,
Truncation: nominative singular lukk,
Shortening: genitive singular luku,
Suxation: genitive plurals lukkude and partitive plural lukkusid.
 Models of grammatical analysis Stem-based implication

Implicational structure of grammatical case forms

The paradigm of  is predictable from partitive singular lukku.

The grammatical forms are directly predictable via

Identity: ‘short’ illative singular lukku,
Truncation: nominative singular lukk,
Shortening: genitive singular luku,
Suxation: genitive plurals lukkude and partitive plural lukkusid.
 Models of grammatical analysis Stem-based implication

Implicational structure of semantic case forms

The endings of the Illative through Comitative are invariant.

Hence the signicant variation lies in stem selection.
Semantic forms are based on the corresponding genitive:
Singular forms are based on genitive singular luke,
(The nominative singular is also based on the genitive singular),
Plural forms are based on genitive plural lukkude.

But since the case endings are invariant, implications are

 and semantic case forms imply their genitive bases.
 Models of grammatical analysis Stem-based implication

Implicational structure of semantic case forms

The endings of the Illative through Comitative are invariant.

Hence the signicant variation lies in stem selection.
Semantic forms are based on the corresponding genitive:
Singular forms are based on genitive singular luke,
(The nominative singular is also based on the genitive singular),
Plural forms are based on genitive plural lukkude.

But since the case endings are invariant, implications are

 and semantic case forms imply their genitive bases.
 Models of grammatical analysis Stem-based implication

Implicational structure of semantic case forms

The endings of the Illative through Comitative are invariant.

Hence the signicant variation lies in stem selection.
Semantic forms are based on the corresponding genitive:
Singular forms are based on genitive singular luke,
(The nominative singular is also based on the genitive singular),
Plural forms are based on genitive plural lukkude.

But since the case endings are invariant, implications are

 and semantic case forms imply their genitive bases.
 Models of grammatical analysis Stem-based implication

Implicational structure of semantic case forms

The endings of the Illative through Comitative are invariant.

Hence the signicant variation lies in stem selection.
Semantic forms are based on the corresponding genitive:
Singular forms are based on genitive singular luke,
(The nominative singular is also based on the genitive singular),
Plural forms are based on genitive plural lukkude.

But since the case endings are invariant, implications are

 and semantic case forms imply their genitive bases.
 Models of grammatical analysis Stem-based implication

Implicational structure of semantic case forms

The endings of the Illative through Comitative are invariant.

Hence the signicant variation lies in stem selection.
Semantic forms are based on the corresponding genitive:
Singular forms are based on genitive singular luke,
(The nominative singular is also based on the genitive singular),
Plural forms are based on genitive plural lukkude.

But since the case endings are invariant, implications are

 and semantic case forms imply their genitive bases.
 Models of grammatical analysis Stem-based implication

Implicational structure of semantic case forms

The endings of the Illative through Comitative are invariant.

Hence the signicant variation lies in stem selection.
Semantic forms are based on the corresponding genitive:
Singular forms are based on genitive singular luke,
(The nominative singular is also based on the genitive singular),
Plural forms are based on genitive plural lukkude.

But since the case endings are invariant, implications are

 and semantic case forms imply their genitive bases.
 Implicational analysis System factorization

Domains of implication

An implicational analysis captures the fact that the stem variation

exhibited by Ingush and Estonian does not signal semantic
properties but instead sanctions deductions about other forms.
In Estonian, grammatical case forms, singular semantic forms
and plural semantic forms each comprise interpredictable cohort
sets. The same factorization applies to more familiar systems.
Implicational analyses oer revealing factorizations of other
systems by identifying domains over which implications apply.
 Implicational analysis System factorization

Domains of implication

An implicational analysis captures the fact that the stem variation

exhibited by Ingush and Estonian does not signal semantic
properties but instead sanctions deductions about other forms.
In Estonian, grammatical case forms, singular semantic forms
and plural semantic forms each comprise interpredictable cohort
sets. The same factorization applies to more familiar systems.
Implicational analyses oer revealing factorizations of other
systems by identifying domains over which implications apply.
 Implicational analysis System factorization

Domains of implication

An implicational analysis captures the fact that the stem variation

exhibited by Ingush and Estonian does not signal semantic
properties but instead sanctions deductions about other forms.
In Estonian, grammatical case forms, singular semantic forms
and plural semantic forms each comprise interpredictable cohort
sets. The same factorization applies to more familiar systems.
Implicational analyses oer revealing factorizations of other
systems by identifying domains over which implications apply.
 Implicational analysis Fractured noun declensions in German

How many declension classes in German?

In reality, the choice of plural formation depends largely on gender

and/or inection class as manifested also in the expression of the
four German cases in the singular. Thus if a masculine has the
sux -en in the Gen.Sg., it must also have it for the plural, e.g. der
Fürst ‘prince, sovereign’, Gen.Sg. des Fürst-en implies the plural
Fürst-en. (Laaha ea : )
 Implicational analysis Fractured noun declensions in German

Singular patterns

Masc Neut Fem

S1 S2 S1 S3
Nom Pegel Prinz Segel Regel
Acc Pegel Prinzen Segel Regel
Dat Pegel Prinzen Segel Regel
Gen Pegels Prinzen Segels Regel
‘level’ ‘bear’ ‘sail’ ‘rule’
Singular declensional patterns in German (cf. Duden (: ))
 Implicational analysis Fractured noun declensions in German

Plural patterns

P1 P2 P3 P4 P5
Ending -s -(e)n -e -e -er Ø Ø
Stem –uml –uml –uml +uml +uml –uml +uml

N/A/G Uhus Prinzen Hunde Bünde Münder Balken Gärten

Dat Uhus Prinzen Hunden Bünden Mündern Balken Gärten
(Masc) ‘owl’ ‘bear’ ‘dog’ ‘waistcoat’ ‘mouth’ ‘beam’ ‘garden’
N/A/G Autos Ohren Jahre Flöße Länder Muster Klöster
Dat Autos Ohren Jahren Flößen Ländern Mustern Klöstern
(Neut) ‘cars’ ‘eyes’ ‘years’ ‘rafts’ ‘countries’ ‘patterns’ ‘cloisters’
N/A/G Bars Regeln — Hände — — Töchter
Dat Bars Regeln — Händen — — Töchtern
(Fem) ‘bars’ ‘rules’ — ‘hands’ — — ‘daughters’
Plural declensional patterns in German (cf. Duden (: ))
 Implicational analysis Fractured noun declensions in German

Declensional patterns

S1 S2 S3
uml Masc Neut Masc Fem
P1 –   — 
P2 –    
P3 –   — —
P3 +  () — 
P4 +   — —
P5 –   — —
P5 +  () — 
Combinations of singular and plural patterns in German
 Implicational analysis Fractured noun declensions in German

Structure of German noun declensions

Singular forms comprise a cohort set, in which the genitive is the

most informative form. Plural forms also comprise a cohort set.
There is a weaker implicational relation between singular and
plural forms, conditioned by gender and phonotactics:
S → P (‘weak’ masculine singulars are weak in the plural).
S → ¬P (no feminine plurals in -er).
S (‘strong’ masculine/neuter) is compatible with all plurals.
 Implicational analysis Fractured noun declensions in German

Structure of German noun declensions

Singular forms comprise a cohort set, in which the genitive is the

most informative form. Plural forms also comprise a cohort set.
There is a weaker implicational relation between singular and
plural forms, conditioned by gender and phonotactics:
S → P (‘weak’ masculine singulars are weak in the plural).
S → ¬P (no feminine plurals in -er).
S (‘strong’ masculine/neuter) is compatible with all plurals.
 Implicational analysis Fractured noun declensions in German

Structure of German noun declensions

Singular forms comprise a cohort set, in which the genitive is the

most informative form. Plural forms also comprise a cohort set.
There is a weaker implicational relation between singular and
plural forms, conditioned by gender and phonotactics:
S → P (‘weak’ masculine singulars are weak in the plural).
S → ¬P (no feminine plurals in -er).
S (‘strong’ masculine/neuter) is compatible with all plurals.
 Implicational analysis Fractured noun declensions in German

Structure of German noun declensions

Singular forms comprise a cohort set, in which the genitive is the

most informative form. Plural forms also comprise a cohort set.
There is a weaker implicational relation between singular and
plural forms, conditioned by gender and phonotactics:
S → P (‘weak’ masculine singulars are weak in the plural).
S → ¬P (no feminine plurals in -er).
S (‘strong’ masculine/neuter) is compatible with all plurals.
 Implicational analysis Fractured noun declensions in German

Structure of German noun declensions

Singular forms comprise a cohort set, in which the genitive is the

most informative form. Plural forms also comprise a cohort set.
There is a weaker implicational relation between singular and
plural forms, conditioned by gender and phonotactics:
S → P (‘weak’ masculine singulars are weak in the plural).
S → ¬P (no feminine plurals in -er).
S (‘strong’ masculine/neuter) is compatible with all plurals.
 Implicational analysis Modelling implicational structure

Measuring predictive value

Let paradigm cells be random variables that take realization

‘outcomes’ as their values (e.g., for [Gen Sg] the value ‘Xs’).
The uncertainty associated with the realization of a cell C can be
dened in terms of the  (Shannon ) of the cell, H(C):

H(C) = − p(x) log p(x)

∑

x∈RC

RC here represents the set of realization outcomes for C, x an

outcome in R, and p(x) the probability that C is realized by x.
 Implicational analysis Modelling implicational structure

Measuring predictive value

Let paradigm cells be random variables that take realization

‘outcomes’ as their values (e.g., for [Gen Sg] the value ‘Xs’).
The uncertainty associated with the realization of a cell C can be
dened in terms of the  (Shannon ) of the cell, H(C):

H(C) = − p(x) log p(x)

∑

x∈RC

RC here represents the set of realization outcomes for C, x an

outcome in R, and p(x) the probability that C is realized by x.
 Implicational analysis Modelling implicational structure

Measuring predictive value

Let paradigm cells be random variables that take realization

‘outcomes’ as their values (e.g., for [Gen Sg] the value ‘Xs’).
The uncertainty associated with the realization of a cell C can be
dened in terms of the  (Shannon ) of the cell, H(C):

H(C) = − p(x) log p(x)

∑

x∈RC

RC here represents the set of realization outcomes for C, x an

outcome in R, and p(x) the probability that C is realized by x.
 Implicational analysis Modelling implicational structure

Morphological information

The diagnostic value of an individual cell correlates with the

degree to which it reduces uncertainty about other cells.
Uncertainty reduction can be expressed in terms of 
, H(C |C ), which measures the amount of uncertainty
that remains about C given knowledge of C .
  measures uncertainty reduction:

H(C |C )
M(C |C ) =  −
H(C )
 Implicational analysis Modelling implicational structure

Morphological information

The diagnostic value of an individual cell correlates with the

degree to which it reduces uncertainty about other cells.
Uncertainty reduction can be expressed in terms of 
, H(C |C ), which measures the amount of uncertainty
that remains about C given knowledge of C .
  measures uncertainty reduction:

H(C |C )
M(C |C ) =  −
H(C )
 Implicational analysis Modelling implicational structure

Morphological information

The diagnostic value of an individual cell correlates with the

degree to which it reduces uncertainty about other cells.
Uncertainty reduction can be expressed in terms of 
, H(C |C ), which measures the amount of uncertainty
that remains about C given knowledge of C .
  measures uncertainty reduction:

H(C |C )
M(C |C ) =  −
H(C )
 Implicational analysis Modelling implicational structure

Paradigm uncertainty

The cumulative uncertainty associated with a paradigm P

depends directly on the uncertainty of its cells C , C . . . , Cn .
On a traditional model, cells are generally assumed to be
interdependent, so that the entropy of a paradigm, H(P), will
correspond to the   of its cells, H(C , C . . . , Cn ).
The information that C expresses about a paradigm P is then

H(P|C)
M(P|C) =  −
H(P)
 Implicational analysis Modelling implicational structure

Paradigm uncertainty

The cumulative uncertainty associated with a paradigm P

depends directly on the uncertainty of its cells C , C . . . , Cn .
On a traditional model, cells are generally assumed to be
interdependent, so that the entropy of a paradigm, H(P), will
correspond to the   of its cells, H(C , C . . . , Cn ).
The information that C expresses about a paradigm P is then

H(P|C)
M(P|C) =  −
H(P)
 Implicational analysis Modelling implicational structure

Paradigm uncertainty

The cumulative uncertainty associated with a paradigm P

depends directly on the uncertainty of its cells C , C . . . , Cn .
On a traditional model, cells are generally assumed to be
interdependent, so that the entropy of a paradigm, H(P), will
correspond to the   of its cells, H(C , C . . . , Cn ).
The information that C expresses about a paradigm P is then

H(P|C)
M(P|C) =  −
H(P)
 Summary and implications

Principal parts

The implicational relations between cells invoked by traditional

grammars can be modelled by morphological information.
A cell or set of cells is  if its morphological information
value approaches .
Cells are highly nondiagnostic when their value approaches .
 Summary and implications

Principal parts

The implicational relations between cells invoked by traditional

grammars can be modelled by morphological information.
A cell or set of cells is  if its morphological information
value approaches .
Cells are highly nondiagnostic when their value approaches .
 Summary and implications

Principal parts

The implicational relations between cells invoked by traditional

grammars can be modelled by morphological information.
A cell or set of cells is  if its morphological information
value approaches .
Cells are highly nondiagnostic when their value approaches .
 Summary and implications

Principal parts

This provides one solution to the traditional problem of

justifying principal parts (cf. also Finkel & Stump , )

One objection to the Priscianic model … was that the choice of

leading form was inherently arbitrary: the theory creates a problem
which it is then unable, or only partly able, to resolve (Matthews
: ).
 Summary and implications

Principal parts

This provides one solution to the traditional problem of

justifying principal parts (cf. also Finkel & Stump , )

One objection to the Priscianic model … was that the choice of

leading form was inherently arbitrary: the theory creates a problem
which it is then unable, or only partly able, to resolve (Matthews
: ).
 Summary and implications

Summary

Word and paradigm approaches are item and pattern models.

The patterns provide a base for analogical deduction.
Words are the most informative items in a morphological system.
Inectional paradigms exhibit reliable implicational patterns.

Implicational analyses of morphological systems model a type of

information that speakers need for language acquisition and use.
 Summary and implications

Summary

Word and paradigm approaches are item and pattern models.

The patterns provide a base for analogical deduction.
Words are the most informative items in a morphological system.
Inectional paradigms exhibit reliable implicational patterns.

Implicational analyses of morphological systems model a type of

information that speakers need for language acquisition and use.
 Summary and implications

Summary

Word and paradigm approaches are item and pattern models.

The patterns provide a base for analogical deduction.
Words are the most informative items in a morphological system.
Inectional paradigms exhibit reliable implicational patterns.

Implicational analyses of morphological systems model a type of

information that speakers need for language acquisition and use.
 Summary and implications

Summary

Word and paradigm approaches are item and pattern models.

The patterns provide a base for analogical deduction.
Words are the most informative items in a morphological system.
Inectional paradigms exhibit reliable implicational patterns.

Implicational analyses of morphological systems model a type of

information that speakers need for language acquisition and use.
 Summary and implications

Summary

Word and paradigm approaches are item and pattern models.

The patterns provide a base for analogical deduction.
Words are the most informative items in a morphological system.
Inectional paradigms exhibit reliable implicational patterns.

Implicational analyses of morphological systems model a type of

information that speakers need for language acquisition and use.