oFORGET, FORGETFUL, UNFORGETABLE = word family (à derivation)
Base: relative starting point for morphological analysis (always last step back)
·Read is the base of readable, readable is the base of readability
à only one step back only
Root: absolute base (never changes)
·Read is the root of readable and readability
à back to the root
Aufgabe: unforgetfulness: Root, Base, Affixes
·Root: forget
·Base: unforgetful/forgetfulness
·Affixes:
oPrefix: un-
oSuffix: -ful, -ness
Other Affixes:
·Circumfixes: e.g. GE-gang-EN
·Infixes: e.g. is-T-agala (arabic)
Allomorph: different realisations with the same meaning/function
·E.g. plural form in English: –s is pronounced in different ways ([s] in „cats“, [z] in „dogs“, [ez] in „houses“
àphonological allomorphs
·E.g. time in Spanish (Verb „gehen“):
oVa-mos = present
oFu-imos = past
oIr-emos = future
à va-, fu-, ir- = Morphological conditioned allomorphs
·E.g. plural forms: -s in „cats“, -en in „oxen“, 0 in „sheep“
Two operations:
·Concatenative: involves the combination of segmentable morphological material
oCats à cat + s
oHouses à house + s
·Non-concatenative: change in morphological form that cannot be described in terms of segmentable material
oWoman à women
oMouse à mice
Four classes of morphological operations:
·Affixation: cats à cat + s
= word with specific content + grammatical morpheme
·Compounding: school building à school + building
= two words with specific content
·Base modification: arabic: darasa (learn) à darrasa (teach = make somebody learn); murle: nyoon (lamb) à nyoo (lambs); hindi: mar (die) à ma:r (kill); english: import (N) à import (V)
·Reduplication
oComplete: malagasy: hafa (different) à hafa-hafa (somewhat different)
oPartial: ponpean: weak (confess) à weakak (be confessing)
oWith duplifix: somali:
§buug (book) à buug-ag (books)
§Fool (face) à fool-al (faces)
§Koob (cup) à koob-ab (cups)
à one might argue duplifixation is a special case of affixation!
·Conversion: hammer (V) à hammer (N); plant (V) à plant (N); etc.
à Affixation, Compounding, (Reduplication) = concatenative
à Base Modification, Reduplication, Conversion = non-concatenative
Exercise:
·Kitab à kutub = base modification
·Hortus à horto = affixation (standard cases in latin with the stam „hort“)
·Savali à savavali = partial reduplication
·[oef] à [oe] = base modification?
·Dobry à najlepszy = Suppletion
Two Models
·Morpheme-based model
oConcatenative operations as prototype of morphological operations
oAssumes that Morphology can be described in the same way as syntax
oMorphological rules can either be formalized as word-structure rules or as lexical entries
à Phrase structure rules
§Noun phrase English: NP à DET + ADJ + N
§Determiner: DET = the
§Adjective: ADJ = red
§Noun: N = house
à the red house
à Word structure rules
§Word form: WF à STEM (+ -ISFX)
§Stem: ST à (DPFX- +) Root (+ -DSFX)
§Root: ROOT = happy
§Derivational prefix: DPFX = un-
§Inflactional suffix: DSFX = -er
à un-happi-er
oLexical Entries are written as follows:
oAdvantages:
§Treats Concatenation as the most fundamental type of morphological operation and thus offers a natural explanation for the dominance of this pattern
§Can show the well-known model similarities between morphology and syntax in terms of linear ordering and hierarchical arrangement
à in syntax:
·We need more (intelligent leaders)
·We need (more intelligent) leaders
à in morphology:
·Un-(do-able)
·(un-do-)able
oDisadvantages:
§Difficulties to accommodate non-concatenative patterns like base modification and conversion à too restricted
·Word-based model
oNo segmentation of complex words into concatenative segments, but formulation of word-schemas that represent features common to morphologically related words
oMorphological rules are modeled as morphological correspondences which describe form-meaning correlations between different schemas and also specify the combinatorial potential of the relevant schemas
oà Word schema for singular nouns:
à Word schema for plural nouns:
à Morphological correspondence between SG nouns scheme and PL nouns schema:
à other example:
§Advantages:
úNon-concatenative patterns can be described naturally
úCan explain back-formations in elegant manner: baby + sitter à babysitter à to babysit; à babsitter gab es vor babysit
à the morpheme-based model just assumes that new words are formed by combining morphemes (e.g. baby + to sit à to babysit) and not by subtracting (e.g. babysitter à to babysit)
úCan explain cross-formation in an elegant manner: To attract à attraction ßà attractive; *to aggress à aggression ßà aggressive;
à verb attract exists, aggress doesn’t
à the morpheme-based model just assumes that morphologically complex words have to be derived from a root. If that root doesn’t exist, the very existence of the derived words cant be explained!
·Disadvantages:
oIs non-restrictive and allows morphological processes that do not exist in natural languages
oHas no good explanation for the dominance of concatenation
What is the lexicon?
·Language user’s mental dictionary, which contains all the information necessary to create and understand words
What is inside of the lexicon?
·Information not predictable from general rules
·But disagreement whether the lexicon also contains predictable information
Theoretical approaches
·Morpheme lexicon
·Strict Word-form lexicon
·Moderate word-form lexicon
Morpheme lexicon
·Basic unit:individual morphemes, which are combined into complex words. Treats morphology like syntax: language users do not store every possible sentence, same for words
·Advantage:
oSmall number of stems and affixes allows to create a massive amount of inflected word-forms (economical)
·Disadvantages:
oCant model
§non-concatenative morphology
úMutter à Mütter
§cumulative expressions
úHort-us (garden-NOM.SG.M)
úhort-orum (garden-GEN.PL.M)
à not only one, but two meanings (Case & Number)
§zero expressions
úOli-n (i was)
úOli-t (you where)
úOli (he/she was)
§empty morphs
úAbsolutive: Sew
úGenitive: Sew-re-n
úDative: Sew-re-z
úSubessive: Sew-re-k
§non-compositional (additional) meaning
úe.g. read- + -er (someone who reads OR lecturer) à possible additional meanings are excluded
oCan’t explain unproductive patterns
§e.g. action nouns like arriv-al, deni-al, refus-al etc. are stable and not growing anymore and so are not always built anew from e.g. arrive and –al!
Strict word-form lexicon
·Basic unit: entire word-forms, both simple and complex. Basic unit of analysis is the word (or word-form)
·Advantage:
oCan handle non-concatenative morphology
oCan handle unproductive patterns
·Disadvantage:
oNot elegant/economical: even predictable forms are stored in the lexicon
oSpeakers of agglutinative languages (e.g. turkish = language with extremly morpheme-rich word-forms) don’t memorize thousands of possible word-forms
oPhonological (e.g. rules of intonation (e.g. „there must be an ‘s’ when between to vocals“) and morphological patterns (e.g. „ge-kauft“ but „be-sprochen“) in certain languages that make reference to morphemic structure
Moderate morpheme-based lexicon
·Basic unit: word-forms are primary, but morphological patterns are allowed as a second type of lexical entry. Some lexical words are listed, others are created on the fly
·Word-schemas are created for different cases: e.g. „always if you have X, there are the suffixes z“. Or for example:
ê this listing gives the schema:
·Challenge:
oWhat is stored (direct route) and what is derived (decomposition)?
oProcess of looking up a word in the mental lexicon = “lexical acces”
oDecomposition: eg. In- + sane à insane; direct route: eg. Insane à insane
§Frequency (+ direct) à sane – insane vs. Helpful - unhelpful
§Segmentability (+ decomposition) à man – men (built non-concenatively) vs. Dog – dogs (built concatenatively)
oà this gives us a three-dimensional continuum, in which every word-form can be posited
à a word form is most likely to be stored, if
§allomorphy is high
§frequency is high
§segmentability is low
Flow of language
·There is not one lexicon shared by all speakers of a language. Rather, each speaker has his own lexicon, depending on education, profession, origin, etc.
·Dictionaries contain the basic vocabulary of a language that consists of actually existing and common words that are expected to be shared by most speakers. These are called actual words
·Additional, there are possible words that could theoretically exist, but are rarely or never used (e.g. „bagelize“ from the noun „bagel“)
Inflection and Derivation
·Inflection = Relationship between the word-forms of a lexeme (usually listed in the dictionary)
oE.g. Latin: insula, insulam, insulae, . à predictable forms
oAlways expresses certain functions, which are referred to as inflectional values
ovalues can be classified into groups of values (inflectional categories) if they
§express the same semantics (gleiche Bedeutung, z.B. number)
§are mutually exclusive (schliessen sich gegenseitig aus, z.B. Singular und Plural)
oE.g. read, readable, unreadable, reader, readability, reread, . à not predictable
Inflectional categories
·On nouns, pronouns:
oNumber, case, gender, person
·On Verbs
oNumber, person
oonly on verbs:
§tense (present, future, past, .)
úspecifies the temporal location of an event
§aspect (perfective, imperfective, habitual, .)
úspecifies the internal temporal structure of an event (e.g. he was cooking vs. He cooked)
§mood (indicative, subjunctive, imperative, .)
úspecifies the speaker’s subjective attitude to the event (e.g. he cooks vs. He might cook vs. He should cook)
à Interferences
§Tense, aspect and mood are not always to keep apart because there are strong interferences between individual categories: e.g. value “perfective” and “present” are not compatible, because a perfective event has necessarily been completed and hence must belong to the past
§Because of these interferences, the three categories are often subsumed under the umbrella term „TAM“ = time, aspect, mood
·Is the distinction between inflection and derivation
ogradual (continuum approach) or
oabsolute (dichotomy approach)
à before answering this question, one must think about the different parameters that set inflection and derivation apart from each other
Differences between inflection and derivation:
·Relevance to syntax:
oinflection is relevant to the syntax
oderivation is not relevant to the syntax
à Problem: tense, aspect and mood are not obviously relevant for syntax, TAM-value of a verb does usually not depend on the presence/absence of other words in the same sentence
·Obligatoryness:
oinflectional features are obligatorily expressed on all applicable word-forms (eg. Latin noun: case is obligatory)
oDerivational meanings are not obligatory expressed
·Limitations:
oinflectional values can be applied to their base without arbitrary limitations
oderivational formations may be limited in an arbitrary way
à Problem: certain verb forms may have defective paradigms, they may simply lack certain inflectional forms = gaps in paradigms: E.g. english modals lack an infinitve and a gerund (*to can, *canning, *to may, *maying, etc.)
·Same concept as base
ocanonical inflected word-forms express the same concept as the base (e.g insula/insulae à both refer to an island)
ocanonical lexemes express a new concept (e.g. schön/Schönheit à different concepts)
à Problem: sometimes there are meaning splits in inflected form (eg. Brother – brothers/brethren, which demonstrates that inflected word forms can become associated with idiosyncratic meaning
·Abstractness:
oinflectional values express a relatively abstract meaning
oderivational meanings are relatively concrete
à Problem: derivational meaning can be highly abstract (eg. -hood), while inflectional meaning can be relatively concrete (eg. dual forms)
·Meaning compositionality:
ocanonical inflected word-forms have compositional meaning
ocanonical derived lexemes have non-compositional meaning
à Problem: some derivational affixes have an entirely compositional meaning (e.g. German: -in)
·Position relative to base
ocanonical inflection is expressed at the periphery of words
ocanonical derivation is expressed close to the root
à Problem: there are cases of inflection affixes being closer to the root than derivational suffixes (e.g. ver-schön-er-n)
·Base Allomorphy
oInflection induces less base allomorphy
oderivation induces more base allomorphy
à Problem: while the tendency may be correct, inflection may induce base allomorphy as well (e.g. sterblich (derivation), starb (inflection), stürbe (inflection))
·Word-class change:
ocanonical inflection does not change the word-class of the base
oderivational affixes may change the word-class of the base
à Problem: there are instances of derivation in which the base of a derivational processes still retains some morphosyntactic autonomy (e.g. mojeho bratr-ow-e dzeci = lit. My brotherly children à ‘mojeho’ agrees not with dzeci, but with bratrowe, and bratrowe agrees with dzeci)
·Cumulative ecpressions:
oinflectional values may be expressed cumulatively
oderivational meaning are not expressed cumulatively
à Problem: some derivational affixes may express more than one meaning at a time (e.g. Dutch ‘-ster’ = female + agent)
·Iteration:
oinflectional values cannot be iterated
oderivational meaning can sometimes be iterated
à Problem: while iterative derivation is attested, it is far too rare to be helpful as a diagnostic between inflection and derivation
·Dichotomy approach: First three parameters are the most important
·Continuum approach: No arbitrary line between ‘more important’ and ‘less important’! A morpheme may be more inflecional with regard to one property and more derivational wih regard to another aspect
·Other Linguists: trichotomic view of inflection and postulate a distinction between inherent and contextual inflection
oInherent inflection comprises features that convey information that is usually independent of the syntactic context (z.B. „das Mädchen“ ist Singular und sachlich)
oContextual inflection comprises features that convey information that is demanded by the syntactic context (z.B. „das Mädchen“ ist Akkusativ)
Inherent inflection often shares properties with derivation, whereas contextual inflection tends not to do so. Inherent inflection
·is more prone to develop idiosynscratic meanings than contextual inflection (- same concept as base)
·tends to be closer to the root than contextual inflection (position relative to base)
·is more likely to induce base allomorphy than contextual inflection (+ base allomorphy)
Architecture of grammar
·Dichotomy approach: derivation is always presyntactic, inflection is always post-syntactic (split-component-hypothesis à derivation happens first)
·Continuum approach: derivation and inflectoin are part of one and the same grammatical component (single-component-hypothesis à both happens at the same time)
What is Productivity?
·A property of morphological patterns and determines the likelihood that an extant morphological pattern is used to derive a new form
·Productive patterns have a high likelihood to derive new forms, whereas non-productive patterns have a low likelihood
why worry about it?
·Linguists interested in speaker’s competence (knowledge of rules), not performance (use of rules in discourse)
·Some linguists argue that
oproductivity is a diachronic phenomenon that does not matter for synchronic description (= nicht nachvollziehbar/regelbasiert)
à but if productivity wouldn’t matter from a synchronic perspective, how would speakers know that the past tense form of the verb ‘to mide’ is ‘mided’ rather than ‘mid?
à productivity is an integral part of a speaker’s knowledge about the morphology of a language and morphological patterns cannot be adequately described without reference to it!
·Some linguists argue that
oProductivity can be reduced to selectional restrictions. So differences in productivity are just differences in selectional restrictions in terms of phonology, morphology, semantics, etc.
à but there are certain morphemes that are not restricted in any obvious way (e.g. the English ‘-let’), but are not productive at all!
Productivity and creativity
·Some linguists draw a sharp line between productivity and creativity: productive formations are coined unconsciously and look natural/do not strike as new, whereas creative formations are coined consciously and look new/noticeable.
oProblem:
§what does consciousness/intentionality refer to here and how can we measure it?
§Many neologisms are equally remarkable and noticeable
oSolution:
§creativity is the violation of language norms, the use of nonproductive patterns to form a new word
§Creativity is thus not a concept that is distinct from productivity, but a concept that depends on (un)productivity
§Productivity is not a dichotomic notion (productive/unproductive), but a gradient scale on which morphological patterns are arranged into more/less productive patterns
Productivity and restrictions on word-formation rules
·Many morphological patterns are subject to some kind of selectional restrictions. E.g. nominalizer ‘–ity’ does not attach to adjectives ending in ‘-ish’, ‘y’, or ‘ful’: its domain of application is restricted
oDefinition of productivity with regard to domains: morphemes are productive if they can be routinely used to create new words within a domain à a morpheme with highly restrictive domain can still be productive
oDefinition of productivity irrespective of domains but in relation to the entire lexicon à a morpheme with highly restrictive domain can never be productive
Problem:
·Even if one assumes that productivity is related to domain restrictions, morphological patterns can be largely unrestricted and still differ in terms of their productivity
·Has something to do with the way that words are stored in the lexicon
Processing restrictions:
·when an affix often occurs in words that are decomposed, it has a high probability of being stored as a separate morpheme and hence a high memory strength, others have a low memory strength
à high memory strength correlates with high productivity
·if a complex word is less frequent than it’s base, it will be decomposed and contribute to the memory strength of the affix
à (e.g. modernity > modern + -ity)
·if a complex word is more frequent than it’s base, it will be stored and not contribute to the memory strenght of the affix
à (e.g. security > secure + -ity)
·So Productivity can be used as parsing ratio (for each affix: the proportion of words that are estimated to be decomposed à more words estimated decomposed = stronger productivity)
Synonymy blocking
·Different morphological patterns can restrict each other through synonymy blocking
à e.g in English the suffix ‘-ity’ is highly productive, but there is a blocking of the adverb ‘*goodly’ because it already exists an adverb ‘well’
·synonymy blockling related to the frequency of the blocking form
à highy productivity can be outdone by synonymy blocking!
Productivity and analogy
·Analogical extension: sting – stung à fling – X ? àhow can such a restricted pattern still be productive?
·analogical extension = application of a productive word-based rule (Bild oben)
à diese Regel ist strikt, lässt aber trotzdem neue Formen zu, wenn die Form passt
·Productivity is tied to words in the lexicon! The number of words that instantiate (realisieren) a certain pattern determines the strength of the pattern
·Number of actual words formed according to a certain pattern= type frequency
oProblem: type-frequency not the same as productivity: e.g. English ‘-ment’ is extremely frequent but not productive
oExtremely good dictionary needed
·Number of possible words that can be formed by a certain pattern
oProblem: restrictions on possible formations are often too elusive (schwer fassbar, flüchtig) to be captured: e.g. random restrictions on English en- / em-)
oDifficult question: what actually is a possible word?
·Ratio of actual words to possible words formed with a pattern
oProblem: builds on the problematic concept of the “possible word”
·Number of neologisms (new word forms) with a certain pattern within a fixed time period (diachronic productivity)
oAdvantage: reasonable proxy for productivity
oBut extremely good dictionary/corpus needed
·Category-conditioned productivity: the ratio of hapax legomena with a given pattern to the total token frequency of words with that pattern
ohapax legomenon = a word that occurs only once within a context, either in the written record of an entire language, in the works of an author, or in a single text
oExtremely good dictionary/corpus needed
oOnly diachronic study is possible
·Hapax-conditioned productivity: the ratio of hapax legomena with a given pattern to the total number of all hapax legomena
oAdvantages: reasonable proxy for productivity, can measure productivity at any point in time
oBut extremely good dictionary/corpus needed
Compound = lexeme consisting of two or more base lexemes
·N + N (z.B. Hausdienst)
·A + N (z.B. Hardware)
·V + N (z.B. Laufjunge)
·N + V (z.B. babysit)
·N + A (z.B. vogelfrei)
·A + A (z.B. bittersüss)
Heads
·Semantic head:
oSemantically more important member of the compound
oDetermines what the compound “is about”
·Formal head
oMorphosyntactically more important member of the compound
oLocus of morphosyntactic marking/morphosyntactic properties
Side of formal head
·Right-headed compounds
oFormal Head = the right member of the compound (e.g. English light year à light years)
·Left-headed compounds
oFormal Head = left member of the compound (e.g. Spanish año luz àaños luz)
·Endocentric compounds
oSemantic Head = member of the compound
§English: “lipstick” = Lippenstift
§Spanish: “pez espada” =Schwertfisch
§Swedish: “spraklärare” = Sprachlehrer)
·Exocentric compounds
oSemantic Head is outside of the compound
§German: “Taugenichts” = Jemand, der für nichts gut ist
oTwo semantic heads with equal status denoting separate referents
§English: “bitters-weet”
§Hindi “mata|pita” = Mother and father
·Appositional compounds
oTwo semantic heads with equal status denoting the same referent
§English “player-coatch”
§Spanish “poeta-pintor” = poet who is also a painter
§English “bitter-sweet”
Typological differences between languages, e.g.:
·Unrestricted compounding
oIn German: Eintopf à Fleischeintopf à Rindfleischeintopf à Hochlandrindfleischeintopf
·Restricted compounding
oIn French: potage à *boeuf potage
·Frequent N-V compounding
oIn Alutor: “I washed my hands” is said as literally “I hand-washed” (= (noun) incorporation) à N-V-Compounding doesn’t exist in e.g. English or German
Morphological trees
·Hierarchical structure in compounds can be illustrated with morphological trees
Hierarchical structure in derivation
·Derived lexemes can also be modeled with morphological trees
·Derived lexemes can also display an internal hierarchical structure
Excursus: scope
·English “You must not come” vs. German “du musst nicht kommen” à substantial difference in necessity meaning (it is necessary for you not to come vs. it is not necessary for you to come)
Parallels between Syntax and morphology
·Morphology can – at least to some extent – be described with the same methods like syntax
·Question: how similar are morphology and syntax?
oTheory 1: compounds and derived lexemes both have internal structure/formal heads, so word-internal structure is governed by syntactic principles!
oTheory 2: the formal differences between morphology and syntax are substantial, so word-internal structure is independent of syntactic principle!
·How strong are the parallels between syntax and compounds/ derivation? Four criterias:
olocus of marking (full parallels)
ogovernment (no parallels)
oagreements (no parallels)
oVisualization of Locus of Marking, Government and Agreement:
oSummary: how strong are the parallels?
Two relations between linguistic units:
·Syntagmatic: words standing next to each other, e.g. “ich – sehe – den - Hund
·Paradigmatic: units in the syntagmatic relation that can be replaced by other units, e.g. ich – schlage/beisse/rufe – den – Hund/Mann/Mäuserich
à morphology can be analyzed from both points of view:
Inflection classes = set of paradigms with the same inflectional pattern
·e.g. o- and u-declension in Latin (see picture) à these both are inflection classes, because every Latin noun ending with ‘-us’ follows one of the two classes
·can be due to
ophonological criteria (e.g. classes due to amount of syllables)
omorphological criteria (e.g. classes due to specific base)
osemantic criteria (e.g. classes due to transitivity/intransitivity)
·inflection class and gender
oinflection classes sometimes correspond with gender, but there are exceptions! E.g. la man-o (Italian), el agu-a (Spanish)
oNew inflectional classes generally tend to absorb loanwords and neologisms and also attract members from less productive classes
Paradigmatic relations
·We’ve already seen: Relations between the members of a lexeme family with morphological correspondence schemes
·Also relations between members of a paradigm
·Inflectional class shift
oinflectional class shift can be described in terms of morphological correspondences between paradigmsà shift to another class because of same forms
inheritance hierarchies = inflectional classes grouped into hierarchies of rule-schemes
·abstraction over two paradigms (slides à V=Vowel, Z = any segment including zero)
ê
·Default rule
oAllow lower-level nodes to override specifications inherited from higher-level nodes. The relevant forms are marked by an exclamation mark in brackets <(!)>
Other important concepts
·Prisianic forms = forms in a paradigm that are derived from another form in that paradigm
·Syncretism = formal identity of functionally distinct forms (e.g. er spielt, ihr spielt; wir spielen, sie spielen)
oSystematic = show up systematical in all paradigms
oIf syncretic forms can be described in terms of the same inflectional values, they are said to form a natural syncretism (e.g. wir spielen, sie spielen: beide Plural)
·Defectiveness = unmotivated lack of word-forms for a lexeme
oE.g. im Spanischen gibt es für das Verb “abolir” im Präsens nur die 1. und 2. Person Plural (= Lücken im Paradigma ‘Personangabe im Präsens)
·Deponency = expression of an inflectional value with the “wrong” marker
oE.g. hortor (Latin) ist marked as passive word, but has a active meaning
·Periphrasis = expression of an inflectional value by means of a multi-word phrase
oLexical: lack of word-forms for certain values in specific lexemes
§E.g. active, *activer, *activest (à active, more active, most active)
oParadigmatic: lack of word-forms for certain values in an entire word-class (here passive + perfect/pluperfect)
oCategorical: lack of word-forms in an entire word-class for a specific grammatical category
Marke aktuell halten ⟶ Erhaltungswerbung An veränderte Umweltbedingungen anpassen Erschließung neuer Zielgruppen Degenerationsphase Kommunikation spielt untergeordnete Rolle Relaunch Andere Anwendungsbereiche Reaktion auf aktuelle Probleme/ Skandale Wiederherstellung des Vertrauens…
...[weiter lesen]