Post on 08-Apr-2018
transcript
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 1/30
ON THE MENTAL LEXICON
HEINZ VATER
1. The subject and goals of psycholinguistics
1.0 Preliminary remarks
After a brief introduction to psycholinguistics, I will describe
the mental lexicon (ML) on the basis of the representations by
EMMOREY / FROMKIN (1988), LEVELT (1989), SCHWARZ (1992;
19962) and HANDKE (1995) and then show that some of their claims
concerning the organization of the ML and the structure of lexical
entries have to be modified due to more recent research. In the
beginning, I would like to describe the place of psycholinguistics
within the realm of linguistics. Linguistics covers a vast area of topics
and is divided into many subdisciplines including the nuclear areas(phonology, morphology, syntax and semantics), and interdisciplinary
areas as phonetics, computational ling., mathematical linguistics,
philosophical linguistics, textlinguistics, neuro-. and
psycholinguistics.
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 2/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
176
Fig. 1 subdisciplines of linguistics
Neurolinguistics
1.1 Areas of psycholinguistics
BIERWISCH (1987: 646) formulates three questions
psycholinguistics has to answer:
[1] What is knowledge of language?[2] How is knowledge of language acquired?[3] How is knowledge of language put to use?
[1] concerns the static ("representational") aspect of
psycholinguistics, i.e. primarly the question how language is stored in
Phonology
Syntax
Phonetics
SemanticsMorphology
Computationallinguistics
Psycho-linguistics
Languagephilosophy
Mathematicallingustics
Pragmatics
Sociolinguistics
Text linguistics
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 3/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
177
the human memory, whereas [2] and [3] concern the dynamic
("procedural") aspect. [2] characterizes the area of language
acquisition, and [3] the area of language processing (speech
production and speech comprehension). CHOMSKY (1986) mentions
language loss (e.g. in aphasia) as a fourth area of psycholinguistics;
this is also a section of neurolinguistics. Further areas concern:
– the relationship between language and perception (cf. MILLER
/
JOHNSON-LAIRD 1976);– the relationship between language and thought (WYGOTSKI 1964,
HÖRMANN 1967, ENGELKAMP 1974);– onomatopoesis and sound imitaion (cf. HÖRMANN 1967, XII);– the neuronal base of language functions (cf. NAUTA / FEIRTAG 19889
and GESCHWIND 19889).
Cognitive Linguistics (CL) has been defined (a) as a
subdiscipline of psycholinguistics concerning cognitive structures and
processes (complementary to affective processes; cf. the schema by
GRIMM / ENGELKAMP 1981: 18), (b) as linguistics under cognitive
aspect (cf. BIERWISCH 1987 and SCHWARZ 19962). I prefer viewing
CL as an interdisciplinary area (a), where linguistics meets cognitive
science, since I hold that there is no need for a term characterizing
linguistics as a cognitive discipline. "Cognition" can be defined,
according to SCHWARZ (19962: 36) as "the set of all mental structures
and processes comprising the total of human activities having to do
with knowledge“. Perception, language and thought are cognitive
systems belonging to the mental equipment of man. Fig. 2 gives a
survey of the cognitive areas.
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 4/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
178
Fig. 2 cognitive systems and processes related to them
Mind
Perception Language Thought
vis. acst. tact. olfct. gust. lang.re- lang.pro- lang.losssyst. syst. syst. syst. syst. present. cessing
vis. acst. tact. olfct. gust. Mental speech speechproc. proc. proc. proc. proc. Lexicon production recept. other
systems
language Memoryacquisition
SIS STM LTM
Abbreviations:
acst. = acoustic syst. = systemgust. = gustatory tact. = tactilelang. = language vis. = visualolfct. = olfactory LTM = long term memoryproc. = process SIS = sensory information storagerecept. = reception STM = short term memory
Concerning the memory, ATKINSON / SHIFFRIN (1968) dis-
tinguish three stages: a sensory information storage (SIS), where a
stimulus is stored only for a fragment of a second, a short term
memory, where elements are processed from ca. 250 msec up to a
couple of minutes, and a long term memory (LTM), being a
permanent store (cf. 1.3 for greater detail).
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 5/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
179
There are specific processes associated with every system; e.g.
speech production and reception processes are associated with the
linguistic system. The systematic part of thought comprises memory
(as a store of thoughts or thought structures), the mental lexicon
forming a part of it. A person’s memory is built up gradually, and it
can also be destroyed gradually. For FODOR (1983), perception and
language are input systems, whereas thought (incl. argumentation,
learning and problem solving) is considered to be a central system.
FODOR (1983: 2ff) gives the following description of input systems:
− Input systems are modality and domain specific. Domains are the areasof physical stimuli that are operated on in the specific modalities, e.g.light waves in the visual modality, acoustic waves in the auditorialmodality. Central systems are neutral in relation to modalities anddomains.
− Input systems function quickly and automatically, whereas centralsystems are slower and less automatic; thought processes can be exertedin a conscious way.
−
Input systems are "encapsulated" according to FODOR (1983:2), whereascentral systems have unrestricted access to data from memory as well asfrom input systems.
All this means that input systems are modular, central systems
are holistic (CARSTON 1988: 43). As for myself, I do not think that
language is automatic and reflex-like.1 Some phenomena speak
against it:
− Usually we have a choice as to how express a message. Thus, allsentences in (1-01) express the same state of affairs, but they do it in
different ways as to the emotional attitude expressed in relation to it:
1 "Fodor (1985) talks of natural language parsing and visual processing asreflex-like; they are reflex-like in their automaticity and encapsulation.However, this analogy breaks down with regard to the other main property of reflexes, that is their direct unmediated response to the appropriate stimulus".(CARSTON 1988:42).
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 7/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
181
neutral as to speech production and comprehension). In online tests,
the test persons themselves can be involved, e.g. choosing a certain
type of utterance by pushing a button (cf. SCHWARZ 19962). In (04),
the testees have to choose one of the two options as an adequate
continuation:
(04) a If you are a pilot, landing planes ... is /are dangerous.b If you are close to an airport, landing planes ... is/are dangerous.
There are several ways to carry through offline tests, e.g. by
having the testees repeat some utterances or judge their grammat-
icality. As examples for slips of the tongue, (03)a concerns lexical
planning, indicating that there must be a semantic as well as a
morphological organization of the entries in the ML; the two entries
that got mixed up (vergammeln and verkommen) are closely related in
semantic respect, the first one expressing a deterioration of the outer
appearance, the second one a deterioration of outer appearance or of
inherent quality. At the same time, the two verbs are morphologically
related by being derivates with the same prefix ver -. (03)b, showing a
confusion in the phonological realization, is a case of metathesis.
1.3 The memory
The memory is, according to SCHWARZ (19962: 35) "the storing
of ontogenetically acquired information". The ability of the humanorganism to store experience in a permanent way is the precondition
for all complex behavioral activities (including thought processes).
The structure of the memory has been investigated especially by
ATKINSON / SHIFFRIN (1968). Their model contains three components:
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 8/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
182
Fig. 3 the memory
memorizing
sensory informa- short-term long-termtion storage (SIS) memory (STM) memory (LTM)
forgetting forgetting The sensory information storage is an iconic storage storing
visual or acoustic stimuli for a fraction of a second. WETTLER (1980:
16) describes an experiment, where visul stimuli, e.g. letters, were
offered to the testees in a chessboard pattern for 50 milliseconds.
Immediately after it, the same stiuli were shown in the same positions
as the stimuli presented before. The testees have to tell whether the
second stimulus is identical to the first one in the same position. The
testees were able to do this. However, if the task is modified slightly,
e.g. by replacing a small letter by a capital letter, the performance of
the testee does not exceed the level of accidental hit or miss.
According to MILLER (1956), seven elements (plus minus two)
can be stored in the STM simultaneously. They remain there as long
as you pay them conscious attention (cf. WETTLER 1990: 17). The
information stored in the LTM is not just a copy of the memory tracein the STM. Thus, the representation of Schimmel does, in STM, not
distinguish between "white horse" ("weißes Pferd" and "fungus"
(Schimmelpilz"). In the LTM, for each of these homonyms a special
trace is constructed; on the other hand, Schimmel1 and weißes Pferd
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 9/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
183
("white horse") have the same or similar representations: the LTM is
considered to be a semantic memory. According to ATKINSON/
SHIFFRIN (1968) stimuli are transferred from the SIS into the STM,
from there into LTM. They neglect the fact that the representation of a
word in the STM (our working memory) obviously depends on
knowledge (also concerning the social acceptability) stored in the
LTM, which can be exemplified by "refusal of perception": The time
span for the correct recognition of a word in brief visual presentation
is higher for words that are socially tabooed than for neutral words
(e.g. Arsch "ass" vs. Barsch "sole"). Inspite of this shortcoming, the
model by ATKINSON/ SHIFFRIN (1968) was successfully used for the
interpretation of many experiments (WETTLER 1980: 18).
The main differences between STM and LTM are the following:
– The STM has a limited capacity, the LTM has an unlimited one.– The STM encodes primarily physical and phonological characteristics,
the LTM semantic ones (cf. Schimmel).
– Whereas the synthesis of certain protein enzymes plays an important rolefor the functioning of the LTM (as shown by experiments with rats thathad to remember ways through a labyrinth), it does not play a role in thefunctioning of the STM.
According to SCHWARZ 19962: 87), concepts are the minimal
entries of storage in memory, storing knowledge of the world.2
2 According to SCHWARZ (19962: 80), clinical observations show that theHippocampus is responsable for long term storing of information. But on thewhole, there is no distinct relation between local damages of the brain and theoccurrence of disturbances of memory.
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 10/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
184
2. The mental lexicon (ML)
2.0 General remarks
The mental lexicon constitutes a central area of language
representation. SCHWARZ (19962: 35) holds that "linguistic
competence is a system of knowledge, encoded in memorial traces".
The human lexicon is, according to HANDKE (1995: 50) the central
module of a language processing system, interacting with the other components of the language processor and containing detailed
information concerning the words to be produced or comprehended.
The mechanical lexica contain a section of the total vocabulary of a
language (e.g. the weather-forecast lexica), whereas the mental
lexicon (unless it has not been damaged by a lesion of the brain)
contains the total vocabulary. CUTLER (1994: 81) emphasizes that a
lexicon is acquired:
"The contents of a lexicon are so patently language-specific that itgoes without saying that a lexicon cannot be inborn – it must beacquired, on the basis of linguistic experience."
As to the entries of the lexicon, there is agreement on the fact
"that the contents of a lexicon consist of sound-to-meaning mappings
in discrete chunks" CUTLER (1994: 82). But what is the size of these
"sound-to-meaning mappings"? Most linguists agree on words being
the main type of entries stored in the ML. The word as a linguistic unitis not easy to define.
A difference has to be made between inflectional words like
child vs. children and lexical words like CHILD. There is more or less
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 11/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
185
agreement on lexical words (lexemes) rather than inflectional words
being considered to be entries of the ML (cf. HANDKE 1995: 52). On
the other hand, it is not easy to tell what has to be considered to be a
lexeme.3 Homonyms are usually counted as two different lexemes.
There are unproblematic cases like Eng. hole "hollow space" and
whole "entire(ty)", which are differentiated in spelling or Ger. Bank 1
(Eng. "bank") and Bank 2 ("Eng. "bench") that are differentiated in
their inflection (Banken vs. Bänke).
Inflectional words as stateN and stateV have to be associated to
two different lexemes for the same reason. The question how to
differentiate between different lexemes has not been answered in a
satisfactory way. For this reason, the number of words (lexemes) an
average person has stored in the ML is a matter of disagreement.4
There are probably big differences between different speakers of the
same language; the temporal factor plays a role as well, since every
person learns new words or lexical expressions continually. Thus, we
should assume that a person’s ML has changed more or less after
reading a scholarly article or listening to a lecture.
3 A lexical word (lexeme), being an abstract unit can only be made audible (or visible) by using one of the inflectional words as its representative (e.g. thenom. sing. with nouns or the infinitive with verbs).
4 AITCHISON (1987:6ff) assumes that a speaker of English knows between50.000 and 250.000 words. What does it mean to know a word? "Knowing" isto be interpreted as comprehending; it concerns the passive vocabulary. Thereare no estimates concerning the active vocabulary – except that it forms asubset of the passive vocabulary.
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 12/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
186
Also morphemes (at least as constituents of regular words; cf.
2.2) have to be taken into account as storage entries, as well as idioms
or "phraseolexemes" (cf. TOPCZEWSKA 2004).5
The structure of the ML is a matter of discussion too. Most
models postulate modular lexical structures as well as (phonological,
morphological etc.) subcomponents of the ML. Disturbances concern-
ing specific areas of the lexicon are taken as evidence (cf. 1.3). JeanAITCHISON (1987) − cf. also AITCHISON (19984) − gives a declarative
representation of the lexicon. In describing the "discovery trips" ling-
uists have to carry out, she quotes a claim made by Sherlock Holmes:
"'From a drop of water', said the writer, 'a logician could infer thepossibility of an Atlantic or a Niagara without having seen or heard of one or the other. So all life is a great chain, the nature of which is knownwhenever we are shown a single link of it.'" (Arthur Conan Doyle, AStudy in Scarlet)
According to AITCHISON (1987), evidence relevant for the study
of the ML consists of four types:
– Word-retrieval problems of normal speakers,– word-retrieval problems of speakers with language deficits,– psycholinguistic experiments (association tests etc.),– results of theoretiscal linguistics.
EMMOREY / FROMKIN (1988: 145f) propose a model of the ML:
– The mental lexicon contains independent sublexica that specify different
categories of lexical information – e.g. phonological, semantic and (withliterate people) orthographic information. They are connected in anetwork (as shown by priming effects); in dyslexia, these connectionsare cut off.
5 FROMKIN / R ODMAN (19935: 124) call the lexicon "a mental storehouse of information about words and morphemes".
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 13/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
187
– The phonological units contain abstract phonemic representations. Inspeech production, the phonological chain must be transformed into asequence of commands to the motor system.
– Orthographic representations take regular and irregular spellings intoaccount.
− Morphologically related words are listed in one entry or in a cluster (cf.2.2).
– Lexical and grammatical morphemes are marked as such, as has beenshown by differences in the ways they are processed.
– In the lexical-semantic subcomponent words are grouped according tosemantic features.
Detailed information concerning the structure of the ML is
found in LEVELT (1989), in connection with language production.
HANDKE (1995: 69) adopts Levelt's structure of lexical entries:
Fig. 4 ACCESS UNIT (according to HANDKE 1995: 69)
Morphological Phonological Specification Specification Form
lexical specification Lexical Pointer
Syntactic Conceptual Specification Specification
Lemma
Unfortunately, most linguists dealing with the ML do not
answer the question whether linguistic concepts are stored separately
from cognitive concepts or together or as part of them (i.e. linguistic
concepts forming a subset of general concepts).
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 15/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
189
/s/, all other feature specifications being the same), whereas /ts/
behaves as a unitary entry in phonotactic respect (e.g. by having the
same combinatory potential as /t/ or /p/). It has been shown that in
slips of the tongue and the tip-of-the-tongue phenomenon, the
phonotactic structure (i.e. the skeletal tier of AP) is maintained,
whereas a replacement, omission or addition took place on the
segmental tier. Thus e.g., when I could not find the name of the
famous tenor Luciano Pavarotti, Ruggiero came to my mind as a first
try, a name that has the same syllable structure and stress pattern as
Luciano.
Facts from early child language indicate that phonological
representations stored in the ML are not necessarily identical to
"spelled out" articulated sound sequences. WODE (1993, 1994)
mentions cases where small children said [tat] instead of [ʃ tat] butrejected the pronunciation [tat] offered by grown up people insisting
on that they said [tat] rather than [tat]! This shows that their
articulatory organs were not ready for the cluster [ʃ tat] yet, although
they had stored it in the ML.6
2.2 Morphological structure of lexical entries
According to LEVELT (1989), the morphological structures of
simple or complex words are stored in the ML as shown in (07) −
6 JAKOBSON (1969: 23) reports a similar case, where a French speaking girl usedtorchon for garcon as well as for cochon, but protested when grown ups didnot make the difference, using cochon for a boy or garcon for a pig.
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 16/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
190
(09).7 In this representation, roots of words are considered to be cyclic
entities, which I doubt. I think that we need the traditional (cyclic) unit
"stem" to count for the structure of derivates like (09).
(07) WORD (08) WORD
R OOT R OOT
house R OOT AFFIX
R OOT AFFIX
nation al ity
(09) WORD
WORD WORD
WORD WORD WORD WORD
WORD JOINT R OOTN R OOTV JOINT R OOTN
Straß en bahn halt e stelle
For EMMOREY / FROMKIN (1988: 145f), it has not been decided
yet whether morphologically complex words are listed as stem +
affixes or as individual morphemes to be combined by rules. Recent
morphological research seems to give an answer to their question:
− CLAHSEN et al. (1991) found evidence for the so-called "dual route"model, according to which regular inflectional words like leb-t-e "lived"can be said to be combinations of (separately stored) morphemes (rootsand affixes), whereas irregular forms like war "was", tat "did" or wurde "became" are not decomposed but rather stored as whole words.
−
PENKE
/ K
RAUSE(2000) elicitated 20 irregular noun plurals and 49participles (23 regular ones, 26 irreg. ones) from 2 German subjects who
were diagnosed with William’s syndrome (WS). Eight unimpairedchildren served as controls. Subjects had to transform (i) a given1.pers.sg. present form into a participle or (ii) a singular noun into its
7 WUNDERLICH/ FABRI (1995) offer different ways of morphological structuring.
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 17/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
191
plural form. The authors found that with regard to the default – s-plural,WS children show lower error scores compared to the controls (17% vs.30%); in contrast, the distribution of errors for the irregular – er - and – n-plurals show the reverse pattern (32% WS, 17% controls). Regular participle formation turned out to be intact (correctness scores being thesame as those for the controls). In contrast to the English data byCLAHSEN / SONNENSTUHL (1999), both WS and control children achieverelatively low correctness scores for irregular participles (62% WS, 66%control). A clear difference, however, emerges when frequency is takeninto consideration: 78% of the control’s errors occur with infrequentirregular participles; in WS children, 45% of errors result inoverregularizations of frequent irregular participles. 8
− K OSTIKAS-TSELEPIS (2000) reports on a research project in which 22Aromunian/Greek bilinguals were tested. Presented with Aromuniansimple words, complex and ill-formed (prefixed and suffixed) nonwords,the testees accepted 99% of the well-formed novel words prefixed withpalju- ("old/bad") as compared to only 36.66% of the novel wordssuffixed with the ornative suffix – osu. This means that "a decompositionprocess for affixed words is … operant, however to different degrees."
− In their study of the mental representation of German compounds, ISEL / GUNTER / FRIEDERICI (2000) offered their 24 native German testees 160
compounds (40 transparent-transparent, 40 opaque, 40 transp.-opaqueand 40 opaque-transp.). The testees responded by pressing one of twobuttons of a response box. Left constituents were activated only for transp.-transp. and opaque-transp. compounds. The authors conclude thatleft constituents are not activated on-line and not processed before theprocessing of the head has been completed. The findings propose ahierarchical model with the transparency of the head being the crucialfactor in determining the mode of lexical access. Decomposition onlytakes place when the head is transparent (cf. FRIEDERICI et al. 2000).
− Neuro-anatomic procedures like the positron-emission tomography(PET) have been applied successfully to the functioning of the ML, e.g.in the study of verb morphology in English by JAEGER et al. (1996). Theauthors found that subjects produced the past tense forms of regular verbs significantly faster than irregular and that error rates for irregular
8 The papers by K OSTIKAS-TSELEPIS and ISEL/GUNTER /FRIEDERICI, are researchreports printed in the documentation Second Intern. Conference on the Mental
Lexicon. Montréal, 2000 (cf. references).
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 18/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
192
verbs were considerably higher than for regular or nonce past tenseforms. Past-tense forms of regular verbs were computed on-line:Activation of left dorsolateral prefrontal cortex, which had been shownin other studies to be involved in on-line behavior of intentional novelbehavior, occurred only in the regular and nonce conditions, which canbe interpreted such that this cortical area is involved in the computationof the regular rule. The hypothesis is that irregular past tense forms arecomputed by activating some aspect of lexical memory; the activation of the left middle temporal gyrus (involved with auditory memory traces)occurred only in the irregular task.
2.3 Syntactic structure of lexical units
In speech production, an important step (after conceptualization
and lexical insertion) is the construction of a syntactic structure, into
which lexical entries have to be inserted (cf. LEVELT 1989). The
syntactic constructions have to take the syntactic properties of the
lexical items into consideration. Since TESNIÈRE (1959), “valency”
has been known to be the most relevant of these syntactic properties.
It concerns the number and quality of linguistic elements a certain
lexical entry can be combined with; valency is associated especially
with verbs; but nouns and adjectives can have valency too. According
to ÁGEL (1995), we have to distinguish between a valency potential
associated with a lexical entry and valency realization depending on
the valency potential of a lexical item and on the syntactic
construction that has been chosen (cf. also VATER 2003).
The verb to paint requires two complements, one of them
designating the agent, the other one the theme. In an active
construction, both complements are obligatory, the agent being
realized as the subject of the sentence, the theme as its object: John
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 19/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
193
painted the wall. In a passive construction, only the theme is
obligatory (as the subject of the sentence), whereas the agent can be
realized by a PP (as a prepositional object) or be omitted: The wall
was painted (by John). A verb can have several variants, each having
its own valency, often also a specific meaning, like the German verb
gehen “to go”. Since each of the variants can be inserted into different
syntactic frames, a considerable amount of syntactic realizations
arises:
(10) a Paul geht zum Bahnhof. 'P. is walking to the station'b Paul geht nach Amerika. 'P. is going (emigrating") to A.'c Paul geht zum Theater. 'P. is going to the theatre
(i.e. becoming an actor)'d Paul geht (zum 31.12.). 'P. is leaving (by the 31.12.)'e Das Paket geht nach München. 'The parcel is sent to Munich'f Die Uhr geht. 'The watch works'g Der Roman von Walser geht ausgezeichnet (wie warme Semmeln).
'The novel by W. sells extremely well'
h Wie geht es? – Es geht. 'How is it going?' − 'Not (too) bad'
Each of the examples illustrates a different variant (with
decreasing semantic similarity): Whereas gehen in (10)a means 'to
walk' (i.e. 'on foot'), in b and c it still means moving but not 'by means
of one’s feet'; variant a allows dropping of the directional adverbial,
which is not possible with b or c; a − d express intentional movement
as opposed to e (a parcel is not being moved by its own intention).
In f , g and h no movement is involved: in f, geht means 'funct-
ions', in g, 'is sold'. In h, gehen occurs in impersonal constructions: the
first sentence means 'how is it going?' and the second one 'more or less
well' (Fr. comme ci, comme ça; Pol. tak sobie).
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 20/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
194
The only psycholinguistic study concerning syntactic inform-
ation in lexical entries of the ML I could find is BROWN TESOLIN / DE
ALMEIDA (2000).9 The task for the 25 participants was to provide a
noun that would make two sentences serving as a frame conceptually
plausible (as e.g. door in The person closed the ____ and The ____
closed as well as nouns that were impossible in that frame (as e.g.
moon). Data from this experiment showed equal priming effects for
both plausible and implausible conditions. The authors do not offer an
interpretation of their results (maybe it is offered in ALMEIDA 2004,
which was not accessible for me).
2.4 Semantic structure of lexical units
According to SCHWARZ (19962: 35), experiments show that the
LM is subdivided semantically into subsystems equivalent to semantic
fields as shown by the loss of sections of the vocabulary (like loss of the field of plant names) with people suffering from certain types of
aphasia. What does the semantic structure of lexical entries look like?
Basing herself on the types of evidence mentioned in 2.0, AITCHISON
(1987: 72) holds that words are not composed of semantic primitives,
but are rather connected "in a gigantic multi-dimensional cobweb".10
9 Paper printed in Second Intern. Conference on the Mental Lexicon, Montréal.10 A holistic hypothesis is also forwarded by LÖBACH (2000). Other
psycholinguists analyze words semantically in (necessary and sufficient)features or in prototypes, like SCHWARZ (19962:88), dividing the concept of DOG into prototype features (including categorical information like IS AN
ANIMAL), HAS FOUR LEGS, HAS A TAIL and CAN BARK . According to SCHWARZ (19962:85) it is not clear yet, "ob das mentale Lexikon nur sprachliches ...Wissen oder auch allgemeines Weltwissen beinhaltet" ("whether the mentallexicon contains only ling. knowledge or also general knowledge").
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 21/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
195
Concepts are not stored in isolation but in narrow relations with each
other (cf. SCHWARZ 19962: 67). This is also true of linguistic concepts
as entries of the ML. The cobweb is only partially similar to a
semantic field as described by BAUMGÄRTNER (1967), where lexical
entries are connected with each other by superordinates or
subordinated semantic features (i.e. by hyponomy relations) or on the
basis of their similarity (as co-hyponymes). AITCHISON (1987: 74f)
assumes four different types of semantic relations:
(11) Relations among entries of the Mental Lexicon
– SUPERORDINATION: orange − citrus fruit − fruit – COORDINATION: orange − mandarine − lemon – COLLOCATION: orange − to peel – SYNONYMY: oculist − eye doctor
Association tests as well as slips of the tongue and tip of the
tongue (TOT) cases show that the concept ORANGE is in a narrow
relationship with superordinate nouns (like citrus fruit and fruit )11 as
well as with co-hyponyms like mandarine and lemon, synononyms
(like Ger. Apfelsine and Orange) and frequently co-occuring
predicational concepts like KNIFE, TO PEEL, HAS A PEEL, IS JUICY etc.
Unfortunately SCHWARZ (19962) employs (like PIAGET 1928) the term"concept" for entities of thought as well as for entities of language (stored inthe memory), whereas BIERWISCH (1983) postulates a difference (but does notgive any details as to the different types of storage).
11 STEINITZ (1968) shows, however, that not all kinds of superordinate nouns areequally natural, e.g. as anaphoric replacements: Auf der Straße lag ein Pudel.
Der Hund / *das Säugetier / das Tier bellte mich an (On the street, a poodle
was lying. The dog / *the mammal / the animal barked at me).
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 22/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
196
A fifth relationship should be added: ANTONYMY (cf. beginning
– end ; warm – cold ; rise – fall).12
LÖBACH (2000) shows, in a very interesting study that the
assumption of "innate ideas" as suggested by Fodor and other
representatives of Generative Grammar has to be rejected. Children
form their first concepts on the basis of percepts before deriving other
concepts out of these basic concepts. The author thinks that theseconcepts are holistic as to their semantics rather than composed of
features or prototypes.
Klaudia GROTE (2000) investigated whether the language
modality (visual-gestural vs. auditory-vocal) affects the architecture
and nature of the semantic lexicon and whether structural diversities
of the lexicon influence memory processes.13 Signs and spoken words
are produced and perceived in a completely different modality. Signlanguages take full advantage of the parallel capacity of the visual
system expressing information both linearly and simultaneously. The
first study involved examining whether deaf and hearing subjects
show different reaction times in a verification task where they had to
judge the presence or absence of a semantic relation between two
signs and two spoken words, respectively. The second verification
task investigated whether the structure of the semantic lexicon wasprojected into the visual semantic system: Subjects had to judge line- 12 I noticed that antonymy is involved in slips of the tongue (produced by myself
as well as by other persons) and in TOT phenomena. AITCHISON (1987)mentions antonymy but does not include it into her scheme.
13 Paper printed in Second intern. conference on the mental lexicon. Montréal.
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 23/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
197
drawings instead of signs which were (1) not related, (2) associatively
related to studied target, (3) associatively related + using a classifier
which referred to target item. False recognition rate the best for (1),
followed by (2), the worst for (3). A group of 20 native DGS-signers
and 20 native DLS-speakers was compared. The author's hypothesis −
strength of syntagmatic and paradigmatic relations within lexical
organization of DGS and DLS vary with different language modality
− was supported. Temporal and structural differences between signed
and spoken languages have an effect on implicit or explicit activation
of semantically related words and lead to different memory errors.
The question whether we have to assume two lexicons in our
mind, a mental lexicon containing linguistic units, and another lexicon
containing non-linguistic concepts, has not been investigated
thoroughly, although two different positions concerning therelationship between semantric and conceptual structure have been
advocated: JACKENDOFF (1983) holds that "semantic structure is
conceptual structure", whereas BIERWISCH (1983) thinks that both are
related but basically independent of each other. In any case, there is
evidence of the fact that, in language comprehension, the recipient
needs linguistic knowledge as well as knowledge of the world both
being in close interaction with each other. EMMOREY / FROMKIN
(1988: 146) assume that the mental lexicon does not contain
knowledge of the world:
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 24/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
198
"The fact that cases of agnosia show a dissociation between lexicalknowledge and nonlinguistic knowledge suggests that lexicalknowledge should be represented separately from nonlinguisticknowledge (Damasio 1985, Newcombe 1985)." 14
This would confirm the approach advocated by BIERWISCH
(1983). SCHWARZ (1992) holds that there is an "amodal" level in our
mental representation where concepts are stored without being
specified as perceptional, linguistic or purely notional. But this is a
completely theoretical assumption, not based on empirical evidence.
Also in this area, further research is necessary.
3. Conclusions
The overview of problems and results of studies concerning the
ML given by EMMOREY / FROMKIN (1988: 144ff) has to be
complemented (and partially modified) in several respects if we take
results of more recent research into account:
– The mental lexicon of a language contains information about words,morphemes and/or idioms.
– The information concerns the phonological form of a lexical entry(including phonological constraints), its morphological structure, itsmeaning (semantic representation), its combinatory potential (syntax), itsorthographic representation (which I disregarded in this article).
14 According to ARNOLD / EYSENCK / MEILI (1991: 37) agnosia is a malfunctionof recognition in spite of intact functions of the perceptual organs, intelligenceand consciousness. A perceived object is not recognized, i.e. is not related toremembrance.
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 25/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
199
– Based on these kinds of information, the speaker decides whether aphonic or graphic chain contains words really occurring in his languageor potential words.15
− Recent research projects concerning morphology in language processing,language acquisition or language loss support the dual storagehypothesis, according to which decomposition takes place in transparentregular words, whereas irregular and/or non-transparent words seem tobe stored as wholes (cf. CLAHSEN et al. 1991 and PENKE / K RAUSE 2000, where also problems concerning the dual storage are mentioned).
− Concerning the syntax of lexical entries, the differentiation into valencypotential (in the ML) and valency realization seems to be fruitful (cf.AGEL 1995, VATER 2003). In this area, there has not been done verymuch research; thus, a lot of work is still to do.
− Concerning the semantics of lexical entries, there is agreement thatentries are stored not in isolation but in close relations which each other.There is disagreement as to the nature of these relationships: whereassome linguists stick to semantic features (ordered within a network),others (like SCHWARZ 19962) prefer prototypes or "cobwebs" relatingentries by relations of coordination, subordination, co-occurrence etc.(AITCHISON 1987 and 19984); still others like LÖBACH (2000) consider lexical entries to be holistic in semantic respect.
Heinz Vater Universität zu Köln
heinzvater@yahoo.de
15 EMMOREY / FROMKIN (1988:144f) speak about "non-occurring nonsenseforms". This is too general. Potential words like the derivate stealer ("one whosteals") or the compound clearhead can be predicted because of their morphological regularities. Nonsense words may be potential words (formedaccording to morphological rules) like Ringelnatz' Tintenschwein "inkpig", or perversions ('Verballhornungen') of consisting words like Robert GERN-HARDT’s Schleistgoß (from Schloßgeist "castle ghost") or "half legal" wordslike MORGENSTERN’s Gingganz as opposed to monomorphematic forms likeMORGENSTERN’s Greule which are not likely to be used as words.
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 26/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
200
ReferencesÁgel, Vilmos (1995). Valenzrealisierung, Grammatik und Valenz. Zeitschr.
f. germanist. Linguistik 234:2−32.Ágel, Vilmos (2000). Valenztheorie. Tübingen: Narr (= narr studienbücher ).Aitchison, Jean (19984). The Articulate Mammal. An Introduction to
Psycholinguistics. London: Routledge.Aitchison, Jean (1987). Words in the Mind: An Introduction to the Mental
Lexicon. Oxford: Blackwell.Almeida, Roberto de (2000). Accessing arguments and conceptual fillers of
verbs. Poster presented at the Second International Conference on the
Mental Lexicon, Montreal, Canada.Almeida, Roberto de (2004). The effect of context on the processing of type-
shifting verbs. Brain and Language 90: 249−261.Anderson, J.R. (1983). The Architecture of Cognition. Cambr./MA: Harvard
University Press.Arnold, Wilhelm / Eysenck, Hans Jürgen / Meili, Richard (19918): Lexikon
der Psychologie. Freiburg, Basel, Wien.Atkinson, Richard C. / Shiffrin, Richard M. (1968). Human memory: A
proposed system and its control processes. In: Spence, K.W./ Spence,J.T. (eds.) (1968). Advances in the Psychology of Learning and
Motivation Research and Theory, Vol. II. New York: Academic Press,89-195.
Baumgärtner, Klaus (1967). Die Struktur des Bedeutungsfeldes. In: Moser,Hugo (ed.) (1967). Satz und Wort im heutigen Deutsch. Düsseldorf:Schwann (= Sprache der Gegenwart 1), 165-167.
Bierwisch, Manfred (1983). Semantische und konzeptuelle Repräsentationlexikalischer Einheiten. In: R ůžička, R./ Motsch, W. (eds.) (1983).Untersuchungen zur Semantik . Berlin: Akademie Vl. (= studia gramm. XXII), 61-99.
Bierwisch, Manfred, (1987). Linguistik als kognitive Wissenschaft.Zeitschrift für Germanistik 8, 645-667.
Blumstein, Sheila E. (1988). Neurolinguistics. An Overview of Language-Brain Relations in Aphasia. In: Newmeyer 1988, III:210–236.
Bosshardt, Hans-Georg (ed.) (1986). Perspektiven auf Sprache.
Interdiszipinäre Beiträge z. Gedenken an Hans Hörmann. Berlin: deGruyter.
Bühler, Karl (1934, 19652). Sprachtheorie. Stuttgart: Fischer.Caplan, David (1988). The biological basis for language. In: Newmeyer
1988, III: 237–255.Carston, Robyn (1988). Language and Cognition. In: Newmeyer (1988), III:
38-68.
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 27/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
201
Chomsky, Noam (1986). Knowledge of Language. Its Nature, Origin, and
Use. New York: Praeger.Chomsky, Noam (1988). Language and Problems of Knowledge. The
Managua Lectures. Cambridge, Mass: MIT.Clahsen, Harald / Eisenbeiß, Sonja / Vainikka, A. (1991). The Seeds of
Structure. A Syntactic Analysis of the Acquisition of Case Marking . (=Arbeiten des Sonderforschungsbereichs 282), Nr.2.
Clahsen, Harald / Sonnenstuhl, Ingrid (1999). Lexical entries and the rules of language: A multidisciplinary study of German inflection. Behavioral
and Brain Sciences 22: 991−1060.Cutler, Anne (1994). Segmentation problems, rhythmic solutions. In:
Gleitman, L. / Landau, B. (eds.), 81–104.Damasio, Antonio R. (1985): Prosopagnosia. In: Trends in Neuroscience
8:132–135.Emmorey, Karen D. / Fromkin, Victoria A. (1988). The mental lexicon. In:
Newmeyer, F. (1988), III: 124–149.Engelkamp, Johannes (1974). Psycholinguistik . München: Fink (= UTB
297).Fanselow, Gisbert / Felix, Sascha, (1987, 19933). Sprachtheorie. 2 Bde.
Tübingen: Francke (= UTB 1441/1442).Flores d'Arcais, Giovanni (1986). Konzeptuelle Strukturen und das mentale
Lexikon. In: Bosshardt (1986):130-148.
Fodor, Jerry A. (1983). The Modularity of Mind. Cambridge/MA: MITPress.Friederici, Angela (1984). Neuropsychologie der Sprache. Stuttgart, Berlin,
Köln: Kohlhammer.Friederici, Angela / Mecklinger, Axel / Schröger, Erich (2000). Working on
working memory. Leipzig: Leipziger Universitäts-Verlag.Fromkin, Victoria (1971). The non-anomalous nature of anomalous
utterances. Language 47: 27–52. Reprinted in: Fromkin, Victoria (ed.)(1973): 215–242.
Fromkin, Victoria / Rodman, Robert (1974, 19935). An Introduction to
Language. New York: Holt, Rinehart & Winston.Fromkin, Victoria (ed.) (1973). Speech Errors as Linguistic Evidence. The
Hague: Mouton.Garman, M (1990). Psycholinguistics. Cambridge: Cambr. Univ. Press.Garnham, Alan (1985). Psycholinguistics. Central Topics. London:
Methuen.Geschwind, Norman (19889). Die Großhirnrinde. In: Gehirn und
Nervensystem, 112-120.
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 28/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
202
Gleitman, Lila / Landau, Barbara (eds.) (1994). The Acquisition of the
Lexicon. Cambridge, MA: The MIT Press.Görz, Günther (ed.) (1993). Einführung in die künstliche Intelligenz. Bonn:
Addison-Wesley.Grimm, Hannelore / Engelkamp, Johannes (1981). Sprachpsychologie.
Handbuch und Lexikon der Psycholinguistik . Berlin: Schmidt.Handke, Jürgen (1995). The Structure of the Lexicon. Berlin, New York:
Mouton de Gruyter.Hörmann, Hans (1967). Psychologie der Sprache. Berlin, Heidelberg, New
York: Springer (verbesserter Nachdruck 1970).Jackendoff, Ray (1983). Semantics and Cognition. Cambridge, Mass.: MIT
Press.Jaeger, Jeri J. / Lockwood, Alan / Kemmerer, David / Van Valin, Robert /
Murphy, Brian / Khalak, Hanif (1996). A positron emission tomographicstudy of regular and irregular verb morphology in English. Language 72:451−497.
Jakobson, Roman (1969). Kindersprache, Aphasie und allgemeine
Lautgesetze. Frankfurt: Suhrkamp (= edition suhrkamp 330).Kleiber, George (1993). Prototypensemantik: eine Einführung. Tübingen:
Narr.Lang, Ewald (1994). Semantische vs. konzeptuelle Struktur: Unterscheidung
und Überschneidung. In: Schwarz, Monika (ed.) (1994). Kognitive
Semantik / Cognitive Semantics. Tübingen: Narr (= TBL 395), 25–40.Lenneberg, Eric (1967). Biological Foundations of Language. New York:John Wiley and Sons.
Leuninger, Helen (1989). Neurolinguistik. Probleme, Paradigmen,
Perspektiven. Opladen: Westdeutscher Verlag.Levelt, Willem J.M. (1989). Speaking: From Intention to Articulation.
Cambridge, MA: MIT Press.Löbach, Brigitte (2000). Semantikerwerb. Ein Beitrag zu einer empiristisch-
naturalistischen Bedeutungstheorie. Tübingen: Niemeyer (= LA 423).Meibauer, Jörg (1995). Neugebildete er -Derivate im Spracherwerb.
Ergebnisse einer Langzeitstudie. Sprache und Kognition 14: 138-160.Meibauer, Jörg / Rothweiler, Monika (eds.) (1999). Das Lexikon im
Spracherwerb. Tübingen: Francke (= UTB Mittl. Reihe 2039).Miller, George A. (1956). The magical number seven, plus or minus two:Some limits on our capacity for processing information. Psychological
Review 63:81-97.Miller, George / Johnson-Laird, Philip (1976). Language and Perception.
Cambridge, Mass: Harvard Univ. Press.
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 29/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
203
Nauta, Walle J. H. / Feirtag, Michael (19889). Die Architektur des Gehirns.In: Gehirn und Nervensystem, 19889: 88-98.
Newcombe, Nora (1985). Methods for the study of spatial representation. In:Cohen, R. (ed.). The development of spatial cognition. Hillsdale, N.J.:Lawrence Erlbaum, 277−300.
Newmeyer, Frederick J. (ed.) (1988). Linguistics: The Cambridge Survey.Vol. III: Psychological and Biological Aspects. Cambridge, Mass. et al.:CUP.
Penke, Martina / Krause, Marion (2000). German noun plurals − a challengeto the Dual-Mechanism model. Brain and Language 81: 303−311.
Pinker, Steven (1994). The Language Instinct . New York: Morrow.Ramers, Karl Heinz / Vater, Heinz / Wode, Henning (eds.) (1994).Universale phonologische Strukturen und Prozesse. Tübingen: Niemeyer (= LA 310).
Rickheit, Gert / Strohner, Hans (1993). Grundlagen der kognitiven
Sprachverarbeitung. Tbg.: Francke (UTB 1735).Rosch, Eleanor (1977). Human Categorization. In: Warren, N. (ed.) (1977).
Studies in Cross-Cultural Psychology, Vol.1. London: Academic Press,1-49.
Schnelle, Helmut (ed.) (1981, 19902). Sprache und Gehirn. Frankfurt a.M.:Suhrkamp (= stw 343).
Schwarz, Monika (1992). Kognitive Semantiktheorie und
neuropsychologische Realität. Repräsentationale und prozeduraleAspekte der semantischen Kompetenz. Tübingen: Niemeyer (= LA 273).
Schwarz, Monika (19962). Einführung in die Kognitive Linguistik . Tübingen:Francke (= UTB 1636).
Second international conference on the mental lexicon. Montréal, Canada,Oct. 18–20, 2000.
Sonnenstuhl, Ingrid / Eisenbeiss, Sonja / Clahsen, Harald (1999).Morphological priming in the German mental lexicon. Cognition 72 (3):203−326.
Steinitz, Renate (1968). Nominale Proformen. In: Kallmeyer, Werner et al.(eds.) (1974). Lektürekolleg zur Textlinguistik , Bd. 2: Reader .Frankfurt/M.: Athenäum Fischer, 246-265.
Stephany, Ursula (1994). Phonologische Gesetzmäßigkeiten der Kindersprache aus synchroner und diachroner Sicht. In: Ramers, K. H. /Vater, H. / Wode, H. (eds.), 205-222.
Tesnière, Lucien (1959). Eléments de syntaxe structurale. Paris:Klincksieck.
8/7/2019 Vater4.1
http://slidepdf.com/reader/full/vater41 30/30
Studi Linguistici e Filologici Online 4.1
Dipartimento di Linguistica – Università di Pisa
www.humnet.unipi.it/slifo
Topczewska, Urszula (2004). Phraseolexeme in Paulusbriefen und ihre
Wiedergabe im Deutschen und im Polnischen anhand ausgewählter
Bibelübersetzungen. Trier: WVT (= Fokus 28).Vater, Heinz (1994). Linguistik – Natur- oder Geisteswissenschaft? In:
Roggausch, W. (ed.). Germanistentreffen Bundesrepublik Deutschland –
Polen. Tagungsbeiträge. Bonn: DAAD, 219-232.Vater, Heinz (1994, 20024). Einführung in die Sprachwissenschaft .
München: Fink. (= UTB 1799).Vater, Heinz (2003). Valency and diathesis. In: Cuyckens, Hubert / Berg,
Thomas / Dirven, René / Panther, Klaus-Uwe (eds). Motivation inLanguages: Studies in Honor of Günter Radden. Amsterdam: J.
Benjamins, 99–122.Vater, Heinz (2005). Referenz-Linguistik . München: Fink (= UTB 2685).Wettler, Manfred (1980). Sprache, Gedächtnis, Verstehen. Berlin/New
York: de Gruyter.Wiese, Richard (1983). Psycholinguistische Aspekte der Sprachproduktion.
Hamburg: Buske.Wiese, Richard (1987). Versprecher als Fenster zur Sprachstruktur. Studium
Linguistik 21:45-55.Williams, Edwin (1994). Remarks on lexical knowledge. In: Gleitman, L. /
Landau, B. (eds.) (1994):7–34.Wode, Henning (1993). Psycholinguistik. Eine Einführung in die Lehr- und
Lernbarkeit von Sprachen. München: Hueber.
Wode, Henning (1994). Perzeption, Produktion und die Lernbarkeit vonSprachen. In: Ramers, K. H. / Vater, H. / Wode, H. (eds.), 169-187.
Wunderlich, Dieter / Fabri, Roy (1995). Minimalist morphology: Anapproach to inflection. Zeitschrift für Sprachwissenschaft 14: 236−294.
Wygotski, L. S. (1964). Denken und Sprechen. Berlin: Akademie-Verlag.