Corpus Gesproken Nederlands

registration: 2007/10/24 06:53:47, last modified: 2016/03/02 09:13:34
title
  • Corpus Gesproken Nederlands
title.abbreviation
  • CGN
title.alternative
  • Spoken Dutch Corpus (version 1.0)
creator
  • researcher
  • developer
  • annotator F. Van Eynde, Katholieke Universiteit Leuven (chairman of the corpus annotation group)
  • annotator W. Goedertier, ELIS, Rijks Universiteit Gent (project manager)
  • annotator N. Oostdijk, CLS, Katholieke Universiteit Nijmegen (project manager)
  • annotator H. Baayen, Max Planck Instituut, Nijmegen (chairman of the corpus design group)
  • annotator L.W.J. Boves, Katholieke Universiteit Nijmegen (chairman of the signal processing group)
subject
  • spoken language corpus
subject.linguisticField
  • phonetics
  • text_and_corpus_linguistics
subject.monoMultilingual
  • monolingual
subject.resourceSubject
  • corpus speech corpus - monolingual
  • glossary speech corpus - monolingual
description
  • The Spoken Dutch Corpus is a collection of approximately 900 hours of Standard Dutch from Flemish and Dutch speakers. All recordings have been aligned with an orthographic transcription and each word has been given a POS tag and a lemma. Part of the data has been enriched with syntactic, prosodic and/or phonetic information.
description.price *This metadata is only as a guide
description.language
  • English
  • German
description.inputDevice
description.inputEnvironment
description.speakingStyle
  • read_speech
  • spontaneous_speech
description.speechMode
  • monologue
  • dialogue
  • meeting_speech
  • lecture_speech
description.samplingRate
description.additionalData
  • analysis_parameter orthographic transcripts
  • analysis_parameter lemmatization and POS information
publisher
contributor
  • sponsor Dutch Government
  • sponsor Netherlands Organization for Scientific Research (NWO) (http://www.nwo.nl/nwohome.nsf/pages/index)
  • sponsor Flemish Government
contributor.attribute.author.motherTongue
contributor.attribute.author.dialect
contributor.attribute.author.level
contributor.attribute.author.age
contributor.attribute.author.gender
contributor.attribute.speaker.motherTongue
  • native 1334 Flemish and 2901 Dutch
contributor.attribute.speaker.dialect
  • standard_dialect contemporary standard Dutch as spoken in Flanders and the Netherlands
contributor.attribute.speaker.level
  • professional
  • amateur
  • educational level: 2697 high, 672 medium, 74 low
contributor.attribute.speaker.age
  • adult
contributor.attribute.speaker.gender
  • male
  • female
contributor.attribute.speaker.number
  • number_of_male_speakers 2498
  • number_of_female_speakers 1737
  • number_of_total_speakers 4235
date.created
  • 1998-06-01 2004-03-01
date.issued
  • 2004-03-01
date.modified
type
  • Sound
  • Text
type.discourseType
type.linguisticType
  • primary_text
  • lexicon
type.purpose
  • analysis To construct a database of contemporary standard Dutch
type.style
  • speech lecture
  • speech dialogue
  • speech about 2 million words monologue and about 7 million words with two or more speakers
  • speech dialogue
  • speech speech-interviews with teachers of Dutch
  • speech speech-spontaneous telephone dialogues
  • speech speech-simulated business negotations
  • speech speech-discussions
  • speech speech-lessons recorded in the classroom
  • speech speech-live sports broadcast
  • speech speech-news and commentaries broadcast
  • speech speech-ceremonious speeches
  • speech lecture
  • speech reading
type.form
  • unfixed
type.sentence
  • short
  • long
type.annotation
type.annotationSample
  • <<Orthographic transcritions>>

    (ort format)
    [Codes]:
    *v: foreign (= non-Dutch) word
    *d: dialect
    *a: incomplete word
    *u: slip of the tongue or onomatope
    *z: word with dialectal pronunciation
    *x: word difficult to hear
    ggg: a non-speech sound produced by the speaker
    xxx: one or more incomprehensible words or partial words
    Xxx: an incomprehensible word that is clearly a titel or proper name

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/text/ort.htm
    (pri format)
    [Tags]:
    <text>: text.
    <au>: annotation unit.
    <w>: word
    <l>: the punctuation. Possible values : ".", "..." or "?".
    <mu>: mark-up unit with COMMENT or BACKGROUND information
    <m>: marker within a <mu></mu>.
    [Attributes]:
    s: speaker ID. Possible values in <au>: Nxxxxx, Vxxxxx or UNKOWN (x = a digit). Possible values in <mu>: COMMENT or BACKGROUND.
    id: identification code. The information is as follows: <sample ID>.<annotation unit rank number>.<word/marker/punctuation sign-rank number>
    marked: translates the *-codes in the original orthographic transcription as optional attribute of the <w> element. Possible values: foreign, dialect, incomplete, mispr, regionalpr and uncertain (corresponding to *v, *d, *a, *u, *z and *x respectively).

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/xml/pri.htm
    (skp format)
    [Tags]:
    <ttext>: time aligned text
    <tau>: time aligned annotation unit
    <tw>: time aligned word
    <tmu>: time aligned mark-up unit with COMMENT or BACKGROUND information.
    <tm>: time aligned marker within a <tmu></tmu>.
    [Attributes]:
    ref: The reference code
    s: speaker ID
    w: the orthographic transcription of the word.
    m: the orthographic transcription of a marker.
    tb: time begin (in seconds) of a time aligned annotation unit.
    te: time end (in seconds) of a time aligned annotation unit.
    tt: type of a time span.
    tq: quality of the time span. Values: (1) man (manually): time markers have been inserted manually. (2) auto (automatically): time markers have been inserted automatically and have not been manually verified. (3) auto_unrel (automatically, unreliable): time markers that have been generated automatically and which are known to be unreliable.

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/xml/skp.htm
    "Protocol for Orthographic Transcription"
    http://ww2.tst.inl.nl/images/stories/docs/cgndocs/ort_prot_en.pdf
    <<Part-of-speech tagging, lemmatizing and lexicon link-up>>

    (plk format)
    [Tags and attributes for lines with a sentence boundary marker]:
    <au>, <mu>, s(speakerID), tb(time begin)
    [For tab-separated 8-column lines]:
    column1: word form (token) as it occurs in the orthogrfaphic transcription
    column2: part-of-speech tag
    column3: lemma of the token. The underscore ("_") indicates that a lemma is absent.
    column4: lexicon-ID
    column5: lexicon-ID of the lemma associated with the word form
    column6: multi-word lemma (when different from column 3)
    column7: lexicon-ID of the multi-word lemma
    column8: References to the different parts of the multi-word expression by means of the rank number of the word in the sentence.

    Thus, part of data unit looks like as follows:

    <au id="3" s="N01265" tb="175.824">
    ja TSW() ja 141336 45366
    Partij SPEC(deeleigen) _ 0 0 Partij_Van_De_Arbeid 610975 2,3,4,5
    Van SPEC(deeleigen) _ 0 0 Partij_Van_De_Arbeid 610975 2,3,4,5
    De SPEC(deeleigen) _ 0 0 Partij_Van_De_Arbeid 610975 2,3,4,5
    Arbeid SPEC(deeleigen) _ 0 0 Partij_Van_De_Arbeid 610975 2,3,4,5

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/text/plk.htm
    (tag format)
    [Tags]:
    <ptext>: text with part-of-speech tagging, lemmatisation and lexicon link-up.
    <pau>: annoattion unit.
    <pw>: word
    <pl>: punctuation mark
    <pmu>: mark-up unit with comprise COMMENT or BACKGROUND information
    <pm>: marker within a <pmu></pmu>.
    [Attributes]:
    ref: (same)
    s: spreaker ID
    w: word form as it occurs in the orthographic transcription
    pos: part-of-speech tag
    lem: lemma
    wid: Lexicon ID
    lid: Lexicon ID of the lemma of the word form
    nlid: Lexicon ID of the multi-word lemma
    pq: quality of the POS tag. Possible values: (1) man, (2) auto (automatically generated).
    marked: (same)

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/xml/tag.htm
    "PART OF SPEECH TAGGING AND LEMMATIZING OF THE CORPUS GESPROKEN NEDERLANDS (SPOKEN DUTCH CORPUS)"
    http://ww2.tst.inl.nl/images/stories/docs/cgndocs/tg_prot_en.pdf
    <<Lexicon link-up>>

    (lxk format)
    [Tags]:
    <ltext>: text with lexicon link-up
    <lau>: annotation unit
    <lw>: word
    <lmu>: mark-up unit with COMMENT or BACKGROUND information
    <lm>: marker within a <lmu></lmu>
    <lkop>: link-up unit within a <lw></lw>
    [Attributes]:
    ref: (same)
    s: speaker identification
    w: word form
    klem: lemma
    nlid: lexicon ID of the single or multi-word lemma
    ksize: the number of parts in the multi-word expression
    kparts: references to the individual parts of the multi-word expression: <annotation unit, rank number>.<word rank number>

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/xml/lxk.htm
    (lex format)
    Comprises 14 columns:
    column1: Id-Nummer
    column2: Orthografie
    (diacritics representation): "&" + capital/small letter + diacritic representation + ";")
    "a" + "grave"
    "c" + "acute"
    "e" + "circ"
    "i" + "uml"
    "n" + "cedil"
    "o" + "tilde"
    "u" + "ring"
    "A"
    "C"
    "E"
    "I"
    "N"
    "O"
    "U"
    column3: LEXICON
    (values):
    ADJ = adjective
    BW = adverb
    LID = article
    N = substantief
    SPEC(afgebr) = code that is used almost exclusively in the lexicon for parts of contracted multi-word expressions (eg 'in- en uitvoer'); in the corpus it has also been used for all incomplete words
    SPEC(deeleigen) = code for part of a multi-word proper name
    SPEC(meta) = code for a word mention
    SPEC(onverst) = code for an incomprehensible utterance
    SPEC(vreemd) = code for an utterance in a foreign language or a foreign word
    TSW = interjection
    TW = numeral
    VG = conjunction
    VNW = pronoun
    VZ = preposition
    WW = verb
    column4: Orthographic representation of the lemma
    column5: Rank number (Id = 'identification') to indicate which word forms belong to one and the same paradigm
    column6: Syntactic complementation patterns per word form
    column7: Status of the word form:
    B = Belgian
    INF = informal
    *d = dialect
    *u = mispronunctiation
    *v = foreign word
    *x = incomprehensible word
    *z = word pronounced with a heavy regional accent
    column8: Canonical pronunciation, representation of (standard) Dutch pronunctiation
    column9: Canonical pronunciation, representation of (standard) Flemish pronunciation
    column10: Formal Flemish pronunciation
    column11: Canonical pronunciation, representation of the word form incl. syllable boundaries and main stress
    column12: Hierarchical morphological segmentation of the lemma:
    (part-of-speech codes):
    N = noun
    A = adjective
    Q = numeral
    V = verb
    D = article
    O = pronoun
    B = adverb
    P = preposition
    C = conjunction
    I = interjection
    X = rest category
    . = affix
    x = part of discontinuous affix
    column13: Orthographical statsu:
    C = correct spelling of a corpus type
    I = incorrect spelling of a corpus type
    O = non-validated spelling of a corpus type
    V = validated spelling of a corpus type
    column14: definition

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/lexicon/README_cgnlex7.htm
    For CGN Multi-word lexicon:
    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/lexicon/README_cgnmlex7.htm
    "Linguistic Protocol for Lexicological Linking"
    http://ww2.tst.inl.nl/images/stories/docs/cgndocs/lxk_prot_en.pdf
    <<Phonetic Transcription>>

    (fon format)
    Class Example CGN symbol
    Plosives put p
    bad b
    tak t
    dak d
    kat k
    goal g
    Fricatives fiets f
    vat v
    sap s
    zat z
    sjaal S
    ravage Z
    licht x
    regen G
    geheel h

    The rest is omitted.

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/text/fon.htm
    <<Word segmentation>>

    (bpt format)
    [Tags]:
    <ftext>: text with a broad phonetic transcription, word segmentation and phone segmentation
    <fau>: annotation unit
    <fw>: word
    <fmu>: mark-up unit with COMMENT or BACKGROUND information
    <tm>: marker within a <fmu></fmu>.
    <fl>: punctuation mark.
    [Attributes]:
    ref: (same)
    s: speaker ID
    w: orthographic transcription of the word
    fon: phonetic transcription of the word
    left/right: left/righ boundary of the word. Values: SEP:normally separated, SHARE-P(x):shared plosive x, SHARE-NP(x):shared non-plosive x, INSERT(x):inserted phoneme x, SHARE-W(x):shared full word (in CGN version 1.0 only "da's")
    marked: (same)

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/xml/bpt.htm
    "Protocol for manual verification of automatically generated word segmentations"
    http://ww2.tst.inl.nl/images/stories/docs/cgndocs/wrd_prot_en.pdf
    <<Syntax>>

    (syn format)
    Each .syn file contains a three-line header(sample number, empty comment line, and NeGra format version). The fourth line: field names, the fifth: #BOS, followed by 7 clumns:
    column1: "word" word form for a terminal node and node number for a non-terminal node
    column2: "tag" POS tag for a terminal node and node label (syntactic category) for a non-terminal
    column3: "morph" official POS tag
    column4: "edge" syntactic function
    column5: "parent" the first filed where the mother node is described
    column6: "secedge" double syntactic function in its original position
    column7: "comment" function of the constituent in its current position is given

    Thus, part of a sentence will look like this:

    %% sample fn123456
    %%
    #FORMAT 3
    ...
    %% word tag morph edge parent secedge comment
    #BOS 8 ...
    welke VNW11 U521b DET 500
    films N2 T107 HD 500
    hebben WW2 T302 HD 501
    zij VNW1 U501u SU 501
    ? LET T007 -- 0
    #500 NP -- WHD 502 OBJ1 501
    (tig format)
    [Tags]:
    <subcorpus>: sample with a syntactic annotation
    <s>: sentence
    <graph>: graphic representation of the syntactic annotation
    <terminals>: list of terminal nodes, end nodes <t>.
    <nonterminals>: list of non-terminal nodes <nt>.
    <edge>: syntactic function
    <secedge>: syntactic function
    <nt>: non-terminal node
    <t>: terminal node
    [Attributes]:
    root: ID of the mother node <s></s>.
    id: unique node identification
    word: word form as it occurs in the orthographic transcription
    pos: POS tag of the terminal node
    morhp: POS tag corresponding to the POS tag from the attributepos.
    cat: syntactic category of a non-terminal node.
    label: syntactic function
    idref: reference to the id of the daughter node

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/xml/tig.htm
    "CGN Syntactic Annotation"
    http://ww2.tst.inl.nl/images/stories/docs/cgndocs/syn_prot_en.pdf
    <<Prosodic annotation>>

    (pix format)
    [Tags]:
    <prtext>: text with a prosodic annotation.
    <prau>: annotation unit
    <prw>: word
    <prl>: punctuation mark
    <prmu>: mark-up unit with COMMENT or BACKGROUND information
    <prm>: a marker within a <prmu></prmu>
    [Attributes]:
    ref: (same)
    s: speaker identification
    w: word form as it occurs in the orthographic transcription
    annot: prosodic annotation
    nprom: number of prominent syllables
    nlength: number of lengthened sounds
    nweakb: number of weak boundaries
    nstrongb: number of strong boundaries
    tbeg: time marker of the beginning of <prau></prau>
    tend: time marker of the end of <prau></prau>.
    leftb/rightb: nature of the left/right boundary. Values: auto, none, weak, strong

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/xml/prx.htm
    (pro format)
    [Labels]:
    :strong break.(e.g. ja dat weet ik maar wanneer)
    : weak break.(e.g. jan en ook piet)
    ^: vowel part of a prominente syllable(e.g. ^i^k ben thuis)
    %: unusual lengthening of a sound without giving prominence to the syllable(e.g. %ja%)

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/text/pro.htm
    "Protocol for Prosodic Annotation"
    http://ww2.tst.inl.nl/images/stories/docs/cgndocs/pr_prot_en.pdf
type.sample
  • Transcritions:

    (pri format)

    <text id="fn123456">
    <mu id="fn123456.1" s="COMMENT">
    <m id="fn123456.1.1"> De </m>
    <m id="fn123456.1.2"> televisie </m>
    <m id="fn123456.1.3"> staat </m>
    <m id="fn123456.1.4"> aan </m>
    <m id="fn123456.1.5"> op </m>
    <m id="fn123456.1.6"> de </m>
    <m id="fn123456.1.7"> achtergrond. </m>
    </mu>
    <au id="fn123456.2" s="N01168">
    <w id="fn123456.2.1"> maar </w>
    <w id="fn123456.2.2"> zij </w>
    <w id="fn123456.2.3"> gaat </w>
    <w id="fn123456.2.4"> uh </w>
    <w id="fn123456.2.5"> drankjes </w>
    <w id="fn123456.2.6"> verkopen </w>
    <l id="fn123456.2.7"> . </l>
    </au>
    <au id="fn123456.3" s="N01167">
    <w id="fn123456.3.1"> gratis </w>
    <w id="fn123456.3.2"> verkopen </w>
    <l id="fn123456.3.3"> ? </l>
    </au>

    The rest is omitted.

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/xml/pri.htm
    (skp format)

    <ttext ref="fn123456">
    <tmu ref="fn123456.1" s="COMMENT" tb="0.000" te="599.523" tt="eq" tq="man">
    <tm ref="fn123456.1.1" tb="0.000" te="599.523" tt="in" tq="man" m="De"/>
    <tm ref="fn123456.1.2" tb="0.000" te="599.523" tt="in" tq="man" m="televisie"/>
    <tm ref="fn123456.1.3" tb="0.000" te="599.523" tt="in" tq="man" m="staat"/>
    <tm ref="fn123456.1.4" tb="0.000" te="599.523" tt="in" tq="man" m="aan"/>
    <tm ref="fn123456.1.5" tb="0.000" te="599.523" tt="in" tq="man" m="op"/>
    <tm ref="fn123456.1.6" tb="0.000" te="599.523" tt="in" tq="man" m="de"/>
    <tm ref="fn123456.1.7" tb="0.000" te="599.523" tt="in" tq="man" m="achtergrond."/>
    </tmu>
    <tau ref="fn123456.2" s="N01168" tb="0.251" te="2.250" tt="eq" tq="man">
    <tw ref="fn123456.2.1" tb="0.251" te="2.250" tt="in" tq="man" w="maar"/>
    <tw ref="fn123456.2.2" tb="0.251" te="2.250" tt="in" tq="man" w="zij"/>
    <tw ref="fn123456.2.3" tb="0.251" te="2.250" tt="in" tq="man" w="gaat"/>
    <tw ref="fn123456.2.4" tb="0.251" te="2.250" tt="in" tq="man" w="uh"/>
    <tw ref="fn123456.2.5" tb="0.251" te="2.250" tt="in" tq="man" w="drankjes"/>
    <tw ref="fn123456.2.6" tb="0.251" te="2.250" tt="in" tq="man" w="verkopen"/>
    </tau>

    The rest is omitted.

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/xml/skp.htm
    Part-of-speech tagging, lemmatizing and lexicon link-up

    (plk format)

    <au id="1" s="N01036" tb="000.000">
    ga WW(pv,tgw,ev) gaan 93037 30559
    je VNW(pers,...) je 620014 135108
    nou BW() nou 620167 135232
    met VZ(init) met 620087 135170
    de LID(bep,...) de 619612 134796
    trein N(soort,ev,...) trein 317006 104897
    naar VZ(init) naar 620133 135200
    Loon SPEC(deeleigen) _ 0 0 Loon_Op_Zand 608839 8,9,10
    Op SPEC(deeleigen) _ 0 0 Loon_Op_Zand 608839 8,9,10
    Zand SPEC(deeleigen) _ 0 0 Loon_Op_Zand 608839 8,9,10
    of VG(neven) of 620170 135234
    met VZ(init) met 620087 135170
    de LID(bep,...) de 619612 134796
    bus N(soort,ev,...) bus 54520|4521 16763|6764
    ? LET() ? 0 0
    <mu id="2" s="BACKGROUND" tb="152.867">
    inschenken SPEC(achter) _ 0 0
    water. SPEC(achter) _ 0 0

    The rest is omitted.

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/text/plk.htm
    |4521" lid="16763|6764" nlid="16763|6764#1" pq="man"/>
    <pl ref="fn123456.1.15" w="?" pos="LET()" lem="?"
    wid="0" lid="0" nlid="0#1" pq="man"/>
    </pau>

    The rest is omitted.

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/xml/tag.htm
    |6764"
    ksize="1" kparts="265.14"/> </lw>
    <ll ref="fn123456.1.15" w="?"/>
    </lau>

    The rest is omitted.

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/xml/lxk.htm
    Word segmentation

    (bpt format)

    <ftext ref="fn123456">
    <fau ref="fn123456.1" s="N01161">
    <fw ref="fn123456.1.1" w="ja" fon="ja"
    left="SEP" right="SEP" fq="auto"
    times="67.905 67.978 68.112"/>
    <fw ref="fn123456.1.2" w="da" fon="dAs"
    left="SEP" right="SHARE-W(da's)" fq="auto"
    times="68.112 68.143 68.205 68.267"/>
    <fw ref="fn123456.1.3" w="'s" fon="dAs"
    left="SHARE-W(da's)" right="SEP" fq="auto"
    times="68.112 68.143 68.205 68.267"/>
    <fw ref="fn123456.1.4" w="waar" fon="war"
    left="SEP" right="SEP" fq="auto"
    times="68.267 68.319 68.371 68.423"/>
    <fl ref="fn123456.1.5" w="."/>
    </fau>

    The rest is omitted.

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/xml/bpt.htm
    Syntax

    (syn format)

    %% sample fn123456
    %%
    #FORMAT 3
    ...
    %% word tag morph edge parent secedge comment
    #BOS 8 ...
    welke VNW11 U521b DET 500
    films N2 T107 HD 500
    hebben WW2 T302 HD 501
    zij VNW1 U501u SU 501
    ? LET T007 -- 0
    #500 NP -- WHD 502 OBJ1 501
    #501 SV1 -- BODY 502
    #502 WHQ -- -- 0
    #EOS
    ...
    %% 432 sentences (2530 tokens, 926 phrases)

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/text/syn.htm
    (tig format)

    <s id="fn123456.1">
    <graph root="fn123456.1.506">
    <terminals>
    <t id="fn123456.1.1" word="in" pos="VZ1" morph="T701"/>
    <t id="fn123456.1.2" word="de" pos="LID" morph="T602"/>
    <t id="fn123456.1.3" word="Amsterdam" pos="SPEC" morph="T005"/>
    <t id="fn123456.1.4" word="Arena" pos="SPEC" morph="T005"/>
    <t id="fn123456.1.5" word="is" pos="WW1" morph="T301"/>
    <t id="fn123456.1.6" word="Sensation" pos="N5" morph="T110"/>
    <t id="fn123456.1.7" word="de" pos="LID" morph="T602"/>
    <t id="fn123456.1.8" word="grootste" pos="ADJ3" morph="T208"/>
    <t id="fn123456.1.9" word="houseparty" pos="N1" morph="T101"/>
    <t id="fn123456.1.10" word="ter" pos="VZ3" morph="T703"/>
    <t id="fn123456.1.11" word="wereld" pos="N1" morph="T101"/>
    <t id="fn123456.1.12" word="gehouden" pos="WW7" morph="T320"/>
    <t id="fn123456.1.13" word="." pos="LET" morph="T007"/>
    </terminals>
    <nonterminals>
    <nt id="fn123456.1.500" cat="MWU">
    <edge label="MWP" idref="fn123456.1.3"/>
    <edge label="MWP" idref="fn123456.1.4"/>
    </nt>
    <nt id="fn123456.1.501" cat="PP">
    <edge label="HD" idref="fn123456.1.10"/>
    <edge label="OBJ1" idref="fn123456.1.11"/>
    </nt>
    <nt id="fn123456.1.502" cat="NP">
    <edge label="DET" idref="fn123456.1.2"/>
    <edge label="HD" idref="fn123456.1.500"/>
    </nt>
    <nt id="fn123456.1.503" cat="NP">
    <edge label="DET" idref="fn123456.1.7"/>
    <edge label="MOD" idref="fn123456.1.8"/>
    <edge label="HD" idref="fn123456.1.9"/>
    <edge label="MOD" idref="fn123456.1.501"/>
    </nt>
    <nt id="fn123456.1.504" cat="PP">
    <edge label="HD" idref="fn123456.1.1"/>
    <edge label="OBJ1" idref="fn123456.1.502"/>
    </nt>
    <nt id="fn123456.1.505" cat="NP">
    <edge label="HD" idref="fn123456.1.6"/>
    <edge label="APPOS" idref="fn123456.1.503"/>
    </nt>
    <nt id="fn123456.1.506" cat="SMAIN">
    <edge label="HD" idref="fn123456.1.5"/>
    <edge label="VC" idref="fn123456.1.12"/>
    <edge label="MOD" idref="fn123456.1.504"/>
    <edge label="SU" idref="fn123456.1.505"/>
    </nt>
    </nonterminals>

    The rest is omitted.

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/xml/tig.htm
    Prosodic annotation

    (prx format)

    <prtext ref="fn123456">
    <prau ref="fn123456.1" s="N02008">
    <prw ref="fn123456.1.1" w="je" annot="je"
    nprom="0" nlength="0" nweakb="0" nstrongb="0"
    tbeg="157.056" tend="160.988" leftb="auto" rightb="none"/>
    <prw ref="fn123456.1.2" w="kunt" annot="kunt"
    nprom="0" nlength="0" nweakb="0" nstrongb="0"
    tbeg="157.056" tend="160.988" leftb="none" rightb="none"/>
    <prw ref="fn123456.1.3" w="ook" annot="ook"
    nprom="0" nlength="0" nweakb="0" nstrongb="0"
    tbeg="157.056" tend="160.988" leftb="none" rightb="none"/>
    <prw ref="fn123456.1.4" w="ha" annot="ha"
    nprom="0" nlength="0" nweakb="0" nstrongb="0"
    tbeg="157.056" tend="160.988" leftb="none" rightb="none"/>
    <prw ref="fn123456.1.5" w="tegen" annot="tegen"
    nprom="0" nlength="0" nweakb="0" nstrongb="0"
    tbeg="157.056" tend="160.988" leftb="none" rightb="none"/>
    <prw ref="fn123456.1.6" w="haar" annot="haar"
    nprom="0" nlength="0" nweakb="0" nstrongb="0"
    tbeg="157.056" tend="160.988" leftb="none" rightb="none"/>
    <prw ref="fn123456.1.7" w="zeggen" annot="z^e^ggen"
    nprom="1" nlength="0" nweakb="0" nstrongb="0"
    tbeg="157.056" tend="160.988" leftb="none" rightb="none"/>

    The rest is omitted.

    http://lands.let.kun.nl/cgn/doc_English/topics/version_1.0/formats/xml/prx.htm
format.extent
  • 9 million words (about 900 hours of speech)(some 3.3 million words :Flanders, well over 5.6 million: The Netherlands) on 33 DVDs (32 audio containing 12,780 wav files and one annotation)
format.medium
  • DVD
format.encoding
  • ASCII
format.markup
  • XML (orthgraphic transcription, POS tagging, lemmatisation, lexicon link-up, phonetic transcription, word segmentation, lexicon, syntactic annotation, prosodic annotation)
  • TextGrid format generated by the PRAAT software (orthgraphic transcription, phonetic transcription, word segmentation, prosodic annotation)
format.functionality
  • "COREX" contained in the corpus package
  • (POS tagging) a POS tagging system to speed up and facilitate the annotation process, developed by combination of a TnT tagger, a Brill tagger, a maximum entropy tagger and a memory-based tagger.
  • (POS tagging verification) a tag selection program for checking and correcting the output to view the tagger output and correct the output by selecting the correct tag in a menu and have it automatically replace the erroneous tag, developed at the University of Nijmegen
  • (transcription) "Praat" for the manual verification of the word segmentation. Praat makes it possible to see and play the speech signal and the transcription tiers and edit segmentation
  • (http://www.fon.hum.uva.nl/praat/)
  • (lemmatisation) a lemmatiser to facilitate the process, developed at the University of Tilburg
  • (syntax) "Annotate"(@nnotate) to facilitate and speed up the annotation process
  • (http://www.coli.uni-saarland.de/projects/sfb378/negra-corpus/annotate.html)
  • "MBDTagLem" The Memory-based Dutch Tagger-Lemmatizer for tagginh Dutch transcribed spontaneous speech and written text (POS and lemma)
  • "TIGERSearch 2.1" A specialized search engine for retrieving information from a database of graph structures (treebank). (for syntactic annotation and viewing tig files)
  • http://www.ims.uni-stuttgart.de/projekte/TIGER/TIGERSearch/
identifier
identifier.doi
identifier.islrn
source
language
  • Dutch
language.area
  • Europe
relation
  • isReferencedBy CGN final evaluation report (version 1.0)
  • isReferencedBy Online Course Spoken Dutch Corpus
  • isReferencedBy CGN promotional CD(http://ww2.tst.inl.nl/images/stories/docs/Engels/cgndemo_en.pdf)
  • isReferencedBy C-003642: COREA-coreferentiecorpus
relation.utilization *This metadata is automatically extracted
  • The accuracy of <usage>the part-of-speech tagger TnT trained</usage> on CGN is reported to be 96.2% (Oostdijk et al., 2002).
  • <usage>The tagger</usage> requires an input file with one word per line and <usage>was trained</usage> on a preliminary version of the CGN corpus (Internal Release 6), and is a purely statistical tagger, based on mono-, bi-, and trigram tag frequencies.
  • <usage>The acoustic models needed for this alignment are estimated</usage> based on the dictated speech in CGN (a CGN component for which no prosodic annotations are available).
  • In this paper the accuracy of <usage>prosodic models</usage> based on the CGN database was evaluated and compared to the accuracy of manual annotations.
  • A first conclusion is that the available amount of manual prosodic annotations in the CGN database is sufficient <usage>for the development of the proposed baseline prosodic models</usage>, a 5 times smaller training database would also suffice.
  • Note that in order <usage>to generate those automatic prosodic transcriptions</usage>, only an orthographic transcription (available in CGN) is needed.
  • A later update of the CGNcorpus may <usage>provide new automatic phonemic transcriptions and segmentations</usage> if such ‘matched' 8kHz model provides substantially better results.
  • Our analysis leads us to conclude that in principle CGN data are appropriate <usage>for developing a model of the (wh-) questions that people will use in interactive QA</usage>.
  • <usage>All speech data and the orthographic transcriptions of the data were taken</usage> from the sixth release of the Spoken Dutch Corpus (Oostdijk, 2000).
  • We <usage>compare the results obtained from the Spoken Dutch Corpus with a similar analysis of a large random collection of FAQs from the internet</usage>.
coverage.temporal
  • 1998/06/01-2004/03/01
coverage.spacial
rights