TARSP Language Measures¶
We will describe here all Tarsp language measures, in the order of the identifiers assigned to them. However, a few language measures are special in that they systematically composed out of recurring macros. See TarspAnnotation for more on these language measures, which we will call composed language measures. We will discuss the most important macros that are used in the queries for such language measures first, in section Macros for Composed Language Measures.
Macros for Composed Language Measures¶
Composed language measures often consist of:
a macro for the mood of the clause (declarative, question, imperative)
macros for ‘zinsdelen’
macros for counts of complements and modifiers
We will discuss each of these.
Macros for clause mood¶
The mood of a clause is determined on the basis of a lot of different clues. Interpunctuation symbols also play a role: we especially do not allow interpunctuation symbols that contradict a mood (e.g. an imperative followed by a question mark). We therefore use several macros for interpunctions symbols:
topcontainsquestionsmark
topcontainsperiodmark
topcontainsexclamationmark
We give the definition of topcontainsquestionmark:
topcontainsquestionmark = """(ancestor::node[@cat="top" and node[%questionmark%]])"""
questionmark = """(@pt="let" and @word="?")"""
The other two are defined analogously.
The most problematic clausal category is sv1, since it can head declaratives, questions and imperatives (see Clauses in Alpino). Distinguishing these correctly is no trivial matter, and the queries we made are certainly not perfect.
Declaratives¶
Declarative clauses are identified by the macro declarative:
declarative = """(@cat="smain" or
(@cat="ssub" and not(%partofwhquestion%)) or
(@cat="sv1" and not(%basicimperative%) and not(%ynquery%) and not(%partofwhquestion%)) )"""
Explanation:
smain clauses are always declarative
ssub clauses are declarative if they are not the body of a (subordinate) wh-question: not (%partofwhquestion%)
sv1 clauses are declarative if they are neither imperatives (not(%basicimperative%)), nor yes-no-questions (not(%ynquery%)), nor the body of (main) wh-question (not(%partofwhquestion%))
For the definition of the macro basicimperative, see Imperatives, for the macro ynquery see Questions.
The macro partofwhquestion is defined as follows:
partofwhquestion = """((@cat="sv1" or @cat="ssub") and
@rel="body" and
parent::node[@cat="whq" or @cat="whsub" ]) """
Questions¶
Yes-no questions are identified by the macro ynquery:
ynquery = """@cat="sv1" and
(@rel="--" or @rel="dp") and
not(%topcontainsperiodmark%) and
not(%topcontainsexclamationmark%) and
node[@rel="hd" and @pt="ww" and @pvtijd !="conj" and
(@stype="ynquestion" or %topcontainsquestionmark% ) ] and
(node[@rel="su"] or %topcontainsquestionmark%) and
(not(%impmodfound%) or %topcontainsquestionmark% )
"""
Explanation:
Tarsp does not mark subordinate yes-no question (introduced by the subordinate conjunction of) in any special way, so we only deal with main clauses
The category must be sv1 and it must have one of the relations – or dp, so that it indeed is a main clause
There must not be an period or an exclamation mark at the end of the utterance
There must be a verb, but not in conjunctive form, and either Alpino has marked the utterance as a yes-no-question (via the attribute stype on the verb), or there is a question mark at the end of the sentence.
There must be a subject or a question mark. Yes-no questions normally contain a subject. Imperatives most often do not. So absence of a subject is seen as a indication for an imperative unless there is a question mark
Certain adverbs are often used with imperatives. Such adverbs should therefore not be present (not(%impmodfound%)), or there must be question mark.
Wh-questions are defined as follows:
Tarsp_whq = """((@cat="whq" and @rel="--") or (@cat="whsub") or (@cat="whrel" and @rel="--"))"""
covering both main clauses (whq) and subordinate clauses (whsub), but also clauses that have been analysed by Alpino as independent (relation=”–” ) relative clauses, which usually are actually wh-questions.
Imperatives¶
Imperatives are identified by the macro basicimperative:
basicimperative = """(%normalimp% or %bareverbimp% )"""
Here the macro normalimp is defined as follows:
normalimp = """(@cat="sv1" and
not(%topcontainsquestionmark%) and
(not(%topcontainsperiodmark%) or %impmodfound%) and
%imprelok% and
%impsubjectok% and
node[%wwimpok%]
)"""
Explanation:
The category must be sv1
the utterance must not contain a question mark
it does not contain a period unless an adverbial modifier typical for imperatives is present. The latter is checked by the macro impmodfound. Such an adverbial modifier can occur in the main clause but also in a subordinate nonfinite clause (macro nonfinvc). The adverbs typical for imperatives are maar and eens (macro basicimpmod):
impmodfound = """(%basicimpmod% or node[%nonfinvc% and %basicimpmod%])""" basicimpmod = """(node[@rel="mod" and (@lemma="maar" or @lemma="eens")])""" nonfinvc = """(@rel="vc" and %nonfincat%) """ nonfincat = """(@cat="inf" or @cat="ppart")"""
the relation of the node must be typical for imperative clauses: imprelok:
imprelok = """(@rel="--" or @rel="nucl")"""
There are restrictions on subjects in imperative clauses, expressed by the macro impsubjectok:
impsubjectok = """(not(%Ond%) or (node[%subject% and %impsubject%] and (%impmodfound% or %topcontainsexclamationmark%) ) )""" impsubject = """(@rel="su" and (@word="jij" or @word="u"))"""
Explanation:
there either is no subject, or
there is a subject but
it is jij or u (impsubject) and
either a imperative modifier is found (impmodfound) or the utterance contains an exclamation mark
there are restrictions on the verb in an imperative clause, expressed by the macro wwimpok:
wwimpok = """( %wwimpfeatok% and not(%modalframe%) and %zijnimpok% and %potentialimpverb% )"""
These restrictions concern:
conditions on the verbal properties (wwimpfeatok). The verb must be a finite present tense singular form (wwimpfin), or an infinitive (wwimpinf):
wwimpfeatok = """(%wwimpfin% or %wwimpinf%)""" wwimpfin = """(@rel="hd" and @pt="ww" and @pvtijd="tgw" and @pvagr="ev" and @wvorm="pv") """ wwimpinf = """(@rel="hd" and @pt="ww" and @wvorm="inf") """
the verb must not be a modal verb (not(%modalframe%)):
modalframe = """(contains(@frame,"modal"))"""
if the lemma of the verb is zijn, the form must be wees or weest:
zijnimpok = """(not(@lemma="zijn") or @word="wees" or @word="weest")"""
the verb must be able to form an imperative. This is defined by the macro potentialimpverb, which basically excludes a list of verbs that have no imperative:
potentialimpverb = """(not(contains(@lemma,"moeten") or contains(@lemma,"hoeven") or contains(@lemma,"zullen") or contains(@lemma,"kunnen") or contains(@lemma,"mogen") or @lemma="hebben" or contains(@lemma, "_hebben") or contains(@lemma, "weten") ) )"""
The macro bareverbimp is defined as follows:
bareverbimp = """(%wwimpok% and
%imprelok% and
parent::node[@cat="top"] and
not(%topcontainsquestionmark%) and
not(%topcontainsperiodmark%)) """
It is intended for single verb utterances:
the verb must be compatible with imperative mood (wwimpok)
the relation that the verb bears must be compatible with imperatives (imprelok)
the parent node must have cat equal to top
there must not be a question mark in the utterance
there must be no period in the utterance.
Remark We must find a way to exclude kijk and kijk eens as an imperative in most cases, since it should be analysed as a V.U.
Remark We must allow other case forms for jij en u as subjects in imperatives (but not the reduced variant je)
remark There surely is overlap betwen the condition on modal verbs and the conditions imposed by potentialimpverb.
Schlichting (p. 62)
“De kenmerken van de Gebiedende Wijs zijn in het kader van TARSP:
de zin moet altijd beginnen met een werkwoord; dat werkwoord staat meestal in de vorm van de stam van het werkwoord, een enkele maal vinden we Stam+t of de infinitief. (covered except Stam+t)
Er is altijd een niet-vragende intonatie (covered by disallowing a question mark)
Er is vaak een tweede of derde woord in de zin, bijvoorbeeld ‘maar’, ‘even’, ‘eens’. Covered
Een gebiedende wijs begint nooit met een modaal werkwoord, behalve met ‘laten’ (covered)
Macros for ‘Zinsdelen’¶
The ‘zinsdelen’ are B, Ond, VC and W and they correspond to some extent to the following language measures, for ‘zinsdelen’:
T007: B, adverbial modifier (see T007: B)
T063: Ond, subject (see T063: Ond)
T097: VC, object or complement (see T097: VC)
T120: W, verb / predicate (see T120: W)
But the correspondence is only perfect for B.
Remark For Ond the definitions are different, but it is not clear whether that cannot be avoided.
Remark For VC the definitions are also different but perhaps they should be identical.
Remark For W there is no immediately obvious correspondence, but perhaps there should be. It looks as if the necessary restrictions have been implemented independently twice.
Out of these language measures, only T120/W occurs in the form (‘profielkaart’).
Language measures such as WBVC, OndVC, OndW, OndWB, OndWBVC, and several others are defined by combinations of the definitions of recurring macros for ‘zinsdelen’ (and some others conditions). For such language measures, special macros for Ond, VC and W have been defined:
Ond: has a simple definition:
Ond = """node[%subject%]"""
where:
subject = """(@rel="su" and parent::node[(@cat="smain" or @cat="sv1" or @cat="ssub")])"""
Here we only count as subjects those nodes (full or index nodes) that have a finite clause node as parent. Overt subjects never occur in nonfinite clauses, and subject index nodes should not be counted inside nonfinite clauses. Subject index nodes do occur in finite clauses, e.g., as the “trace” of a wh-movement.
It differs from the definition of T063: Ond because T063 has to cover an additional case (existential er as a subject (though maybe that should be included here as well)
Tarsp_B: see T007: B
Tarsp_VC: is defined as follows:
Tarsp_VC = """((@rel="obj1" or (@rel="pc" and not(%Tarsp_pc_vc_exception%)) or @rel="predc" or @rel="ld" or @rel="obj2" or %Tarsp_finvc% or %Tarsp_vcvnw% or (@rel="svp" and @pt!="vz")) and not(%Tarsp_B%) )"""
So, it covers:
direct objects (obj1)
prepositional complements that are not Tarsp_pc_vc_exception (pc)
predicative complements (predc)
locative/directional complements (ld)
indirect objects (obj2)
finite verbal complements (macro Tarsp_finvc)
pronouns with relation vc, this is for sluiced subordinate clauses as in ik weet wat, effe kijken waar
separable particles of a verb that are not adpositions (svp)
and we exclude all cases that can be analysed as an adverbial modifier (Tarsp_B, see T007: B)
The macro Tarsp_pc_vc_exception is generated in the generatemacros module on the basis of a list of pairs (verb, adposition), e.g. (slaan, op) in which Alpino analyzes the adposition as the head of a prepositional complement (pc), but where it should be considered the head of a modifier. See Generation of macros for more details.
Tarsp_W: It is defined as:
Tarsp_W = """node[@rel="hd" and @pt="ww"]"""
Nonfinite verbs are not excluded by this definition but they are excluded by the mood conditions. For language measure T120, W, the macro Tarsp_coreW is used.
Macros for Counts of ‘zinsdelen’¶
Many composed language measures differ only in the number of complements or adverbial modifiers required. For example Tarsp_OndWB and Tarsp_OndWBB differ only in the number of adverbial modifiers that should be present (one, resp, two). This requires conditions on the number of such complements or modifiers. Note that one cannot simply include two expressions joined by and in an Xpath query (e.g. %Tarsp_B% and %Tarsp_B%) to cover the BB part of the language measure Tarsp_OndWBB. The reason is that in XPath such a query will then still match with a structure containing a single adverbial modifier. We therefore use the Xpath function count in composed language measures.
Adverbial modifiers and complements can occur not only under the node where the subject and verb occur but also inside a nonfinite verbal complement, so we gather both in a node set, union (|) these node sets and count the number of elements in the union. This leads to the following formulations:
Tarsp_B_X_count = """count(node[%Tarsp_B%] | node[%nonfinvc%]/node[%Tarsp_B%]) """
Tarsp_VC_X_count = """count(node[%Tarsp_VC%] | node[%nonfinvc%]/node[%Tarsp_VC%]) """
Definitions for Tarsp_B and Tarsp_VC were provided above.
We must also have counts for the overall number of “zinsdelen” in the clause. This is taken care of by the macro realcomplormodnodecount. It counts occurrences of realcomplormodnode in the current node or in a nonfinite complement:
realcomplormodnodecount = """count(%realcomplormodnode% | node[%nonfinvc%]/%realcomplormodnode%)"""
realcomplormodnode = """node[%realcomplormod%]"""
The macro realcomplormod is defined as follows:
realcomplormod = """(not(%particlesvp%) and not(%indexnode%) and not(%nonfinvc%) and not(@rel="hd"))"""
Here we exclude:
adpositional separable verb particles (particlesvp). These are considered to be part of the verb:
particlesvp = """(@rel="svp" and @pt="vz")"""
index nodes (indexnode). We do not want to include nodes that do not cover lexical material:
indexnode = """(@index and not (@cat or @pt or @pos))"""
nonfinite verbal complements, using the macro nonfinvc defined above
the head verb (not(@rel=”hd”)). We exclude this because there can be multiple verbs in a clause but they should always count as one ‘zinsdeel’, viz. predicate. The value that realcomplormodnodecount is compared with is therefore always one lower than one would expect, because the predicate is assumed to be always present, and is not counted by realcomplormodnodecount . For example, for OndWBB, with 4 ‘zinsdelen’, the value that realcomplormodnodecount must have is equal to 3, not 4.
Concrete Example: Tarsp_OndWBB¶
Given all of the above, it should be clear why the definition of Tarsp_OndWBB is as follows:
Tarsp_OndWBB = """(%declarative% and
%Ond% and
%Tarsp_W% and
%Tarsp_B_X_count% = 2 and
%realcomplormodnodecount% = 3)"""
Language Measures¶
T001: (Vr)WOnd+¶
Name: (Vr)WOnd+.
Category: Zinsconstructies
Subcat: Vragen
Level: Zc
Original: yes
In form: yes
Page: 60
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “Vraagwoordzin waarbij het vraagwoord is weggelaten. Het werkwoord +, meestal, het Onderwerp zijn gerealiseerd. Daarna kunnen nog één of meer Zinsdelen volgen.”
T003: 6+¶
Name: 6+.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Sz
Original: yes
In form: yes
Page: 56
Implementation: Python function
Query defined as:
sziplus6
Schlichting: “6 Zinsdelen of meer in een zin”
T004: Aan/uit¶
Name: Aan/uit.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 57;58
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “Soms wordt een zinsdeel aan het begin of aan het einde van de zin met andere woorden herhaald. Die herhaling noemen we ‘Aanloop’ aan het begin van een zin en ‘Uitloop’ aan het einde van de zin.”
T005: als¶
Name: als.
Category: Verbindingswoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 77
Implementation: Xpath
Query defined as:
//node[@lemma="als" and @pt="vg"]
Straightforward Xpath expression for the conjunction als.
Schlichting “‘als’ is het eerste verbindingswoord van de ondergeschikte zin. ‘als’ wordt hier ook gescoord wanneer de onderschikkende zonder hoofdzin gebruikt wordt.
Remark: Fully covered but it is unclear whether als as used in comparisons also should be included. All Schlichting’s examples concern conditional als. So maybe this query should be restricted somewhat.
T006: Avn¶
Name: Avn. (Aanwijzend Voornaamwoord als nomen)
Category: Voornaamwoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 79
Implementation: Xpath
Query defined as:
AVn = """(%coreavn% or %avnrel%)"""" coreavn = """(@pt="vnw" and @vwtype="aanw" and @lemma!="hier" and @lemma!="daar" and @lemma!="er" and @rel!="det" and (not(@positie) or @positie!="prenom") )""" avnrel = """(%diedatrel% and parent::node[@cat="rel" and @rel!="mod"])""" diedatrel = """(@pt="vnw" and @vwtype="betr" and @rel="rhd" and (@lemma="die" or @lemma="dat"))"""
The query for AVn consists of two subcases: the core case, and the case of die and * dat* incorrectly analysed as a relative pronoun in an independent relative clause.
The core case (coreavn):
which has pt equal to vnw and vwtype equal to aanw selects demonstrative pronouns, but
these include R-pronouns, so they are explicitly excluded
the relation must not be det (otherwise the pronouns are not used independently)
and if a position attribute is present it should not have the value prenom (otherwise it is not used independently)
The relative case (avnrel) covers the relative pronouns die and dat (diedatrel) in an independent relative clause (i.e. rel is not equal to mod).
Schlichting: “Aanwijzend Voornaamwoord: ‘die’, ‘dit’, ‘deze’, ‘dat’ zelfstandig gebruikt.” Fully covered.
T007: B¶
Name: B. (Bijwoordelijke Bepaling)
Category: Zinsdelen
Subcat:
Level: Zc
Original: yes
In form: no
Page: 42
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_B%]
The definition of this macro is as follows:
Tarsp_B = """(
((((@rel="mod" or @rel="ld" or @rel="predm") and
(not(@cat) or @cat!="conj") and
(not(@pt) or @pt!="tsw")
)or
(%predcB%)
) and
(../node[@pt="ww" and @rel="hd"])
) or
((@pt="vz" or @pt="bw" or %Rpronoun%) and (@rel="dp" or @rel="--" or @rel="nucl" or @rel="body") and %notonlyrealnode%) or
(@cat="pp" and (@rel="--" or @rel="dp") and %notonlyrealnode%) or
(@rel="pc" and ../node[@rel="hd" and %locverb%]) or
(@rel="cnj" and parent::node[@rel="mod" or @rel="ld" or @rel="predm"]) or
(@rel="mod" and @pt="bw" and parent::node[@cat="np"] ) or
%Tarsp_nonmodadvcp% or
%Tarsp_pc_vc_exception%
)"""
Explanation: Adverbial modifiers are:
Case 1:
- Nodes with grammatical relation mod, ld or predm
but if the node is a phrase, it should not have the value conj for the attribute cat. In a conjunction the whole conjunction is not considered an adverbial modifier. The individual conjuncts are, see below.
if the node is a word, it should not be an interjection
Nodes that meet the conditions of the macro predcB. This macro is defined below.
but only if they modify a verb
Case 2: adpositions, adverbs and R-pronouns that bear one of the relations dp, –, nucl or body, if they are not the only real node in the structure. The macro notonlyrealnode is defined below.
Case 3: adpositional phrases with grammatical relation dp or – if they are not the only realnode
Case 4: adpositional complements to a verbs that meets the requirements of the macro locverb. Verbs that have a locative interpretation often also have use where they are combined with a particular adposition (e.g. staan + op). This often leads to an ambiguity, and Alpino very often selects the pc analysis. However, very young children do not know these uses yet, but almost always intend the locative use. This part of the query corrects for Alpino’s disambiguation strategy.
Case 5: A conjunct if its parent node has a grammatical relation from mod, ld or predm
Case 6: adverbs that are a modifier inside an NP. Probably because Alpino analyses ambiguous cases by including the adverb in the NP.
Case 7: nodes that meet the requirements of the macro Tarsp_nonmodadvcp.
Case 8: nodes that meet the conditions of the macro tarsp_pc_vc_exception. See section Generation of macros
The macro predcB is defined as follows:
predcB = """(@rel="predc" and
(@pt="adj" or @pt="bw" or @cat="ap" or @cat="advp") and
../node[@rel="obj1"]
)"""
The idea is that predc nodes that do not occur with a copula are covered here.
The macro notonlyrealnode is defined as follows:
notonlyrealnode = """(parent::node[count(node[%realnode%])>1])"""
where realnode excludes interpunction signs, interjections, or nodes without any pos or category label:
realnode = “””((not(@pt) or (@pt!=”let” and @pt!=”tsw”)) and (not(@postag) or @postag!=”NA()”))”””
The macro locverb is defined as follows:
locverb = """(@lemma="staan" or @lemma="zitten" or @lemma="rijden" or
@lemma="vallen" or @lemma="doen" or @lemma="gaan" or
@lemma="komen" or @lemma="zijn" or %locmodalverb% )"""
locmodalverb = """ (@lemma="kunnen" or @lemma="moeten" or
@lemma="hoeven" or @lemma="willen" or
@lemma="mogen")"""
The macro Tarsp_nonmodadvcp is defined as follows:
Tarsp_nonmodadvcp = """(@cat="cp" and
(@rel="dp" or @rel="--") and
node[@pt="vg" and @conjtype="onder" and
fBas@lemma!="dat" and @lemma!="of" ] )"""
covering cp-nodes that could not be integrated in the whole structure (hence relation dp or –) and contain a subordinate conjunction other than dat or of.
Schlichting: “De bijwoordelijke bepaling zegt iets over de hele inhoud van de zin of iets over het werkwoord, een bijwoord of een bijvoeglijk naamwoord. Een zin kan meer dan één bijwoordelijke bepaling hebben.” Basically covered.
Remark The locverb macro very probably has overlap with or is subsumed by the generated macro Tarsp_pc_vc_exception
Remark Case 5 does not cover conjuncts under a node conj that meets the requirements of predcB, e.g ik vind haar mooi en aardig
Remark The macro predcB does not cover cases where the object is absent due to topic drop (e.g. vind ik mooi). We could include a condition which includes cases where surely non-copular verbs (such as vinden) are listed, as was done in ASTA.
T010: BBBv¶
Name: BBBv. (Bijwoord + Bijwoord + Bijvoeglijk woord, adverb + adverb + adjective)
Category: Woordgroepen
Subcat: Ov
Level: Wg
Original: yes
In form: no
Page: 70
Implementation: Xpath
Query defined as:
//node[@cat="ap" and node[@rel="mod" and (@cat="ap" or @cat="advp") and node[@rel="mod" or @rel="me" ] and node[@rel="hd" ] and count(.//node)=2] and node[@rel="hd" and @pt="adj"] and count(node) =2]
Schlichting: “Bijwoord + Bijwoord + Bijvoeglijk Woord. Deze constructie heeft een Bijvoeglijk woord als kern. Het is eigenlijk een BBv/B voorafgegaan door een bijwoord.
The first adverb is analysed as a modifier or measure expression of the second adverb. Together they form a phrase of category ap or advp that cooccurs with a head adj. Both phrases mentioned can contain only two nodes as children. This analysis contradicts the analysis by Schlichting, which suggest that the first adverb modifies the combination of the second adverb with the adjective. But an analysis that matches with the query given seems the most plausible analysis for the second example (hij is veel te hoog). As for the first example (een kindje is nog te klein voor), a more plausible analysis has nog as a sentential modifier, and that is how Alpino analyses it.
T011: Bbijzin¶
Name: Bbijzin. Bijwoordelijke Bijzin (adverbial subordinate clause)
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Sz
Original: yes
In form: yes
Page: 57
Implementation: Xpath
Query defined as:
//node[@rel="mod"and @cat="cp" and node[@rel="body" and node[@pt="ww" and @pvagr and @rel="hd"]]]
Schlichting: “Bijwoordelijke bijzin met verbindingswoord. Dit is dus een ondergeschikte zin die in de hoofdzin de functie van bijwoordelijke bepaling heeft. Het verbindingswoord is in deze Fase meestal als”,.
A straightforward query for a cp (“complementizer phrase”) with grammatical relation mod. The condition that the body must contain a finite verb is necessary because a cp can also have other phrases (e.g. np) as complement.
T012: BBv/B¶
Name: BBv/B.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 67
Implementation: Xpath with macros
Query defined as:
//node[(node[@rel="mod" and %adjoradv%] and node[@rel="hd" and %adjoradv%]) ]
This query is straightforward implementation of an adverb modifying an adjective or adverb functioning as a head. It makes use of the macro adjoradv (twice), which is defined as follows:
adjoradv = “””(@pt=”bw” or @pt=”adj” or (@pt=”vnw” and (@pos=”adj” or @pos=”adv”)))”””
The macro covers not only words with the value adj or bw for pt but also words with pt=vnw where the Alpino pos attribute has the value adj or adv. The latter is needed for a word such veel and for R-words such as daar.
Schlichting: “Bijwoord + Bijvoeglijk woord of Bijwoord. De eerste B is vaak een bijwoord van graad, of zegt in ieder geval iets van het tweede woord in deze woordgroep. Het tweede woord is meestal cde kern van de woordgroep.”
T013: BBvZn¶
Name: BBvZn.
Category: Woordgroepen
Subcat: Ov
Level: Wg
Original: yes
In form: no
Page: 70
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “Bijwoord + Bijvoeglijk woord + Zelfstandig naamwoord”
T014: BBX¶
Name: BBX.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 50
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_BBX%]
The macro Tarsp_BBX is defined as follows:
Tarsp_BBX = """((@cat="top" and
count(.//node[@pt!='let' and @pt!='vg'])=3 and
count(.//node[@special="er_loc" or @pt="bw" or
@pt="vz" or (@pt="adj" and @rel="dp" ) ])=2 and
.//node[@pt="n" or (@pt="vnw" and not(@special="er_loc")) ]
) or
((@cat="sv1" or @cat="smain") and not(%basicimperative%) and
not(%ynquery%) and not(%Ond%) and %Tarsp_W% and
count(.//node[%Tarsp_B%]) = 2 and %realcomplormodnodecount% = 2
)
)"""
Explanation: there is match if either cas 1 holds or Case2:
Case 1: the utterance contains:
anywhere in the stucture exactly 3 real nodes, and
anywhere in the structure 2 adverbs, adverbial pronouns, adpositions, or grammatically isolated (rel = dp) adjectives, and
anywhere in the structure a noun, or a pronoun that is not an R-pronoun
Case 2: there is a clause with category sv1 or smain
that is not an imperative clause, and
that is not yes-no question, and
that does not contain a subject, and
that does contain a head verb, and
that does contain 2 adverbial modifiers, and
that contains exactly 2 real complements or modifiers
Schlichting: “Bijwoordelijke bepaling + Bijwoordelijke bepaling + een ander zinsdeel (Ond of W of VC of B)”
T015: BepBvZn¶
Name: BepBvZn.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 71
Implementation: Xpath
Query defined as:
//node[@cat="np" and node[@rel="det" ] and node[@rel="mod" and @pt="adj"] and node[@rel="hd" and @pt="n"] ]
Straightforward implementation of noun phrase with a determiner, a modifying adjective and a head noun
Schlichting: “Bepaler + Bijvoeglijk woord + Zelfstandig naamwoord”
Remark: does not cover numerals, which also fall under “bijvoeglijk”. Probably best to make a macro for “bijvoeglijk woord”
Remark: restrict this to exactly 3 nodes under NP.
T016: BepZnBv¶
Name: BepZnBv.
Category: Woordgroepen
Subcat: Ov
Level: Wg
Original: yes
In form: no
Page: 70
Implementation: Xpath
Query defined as:
//node[(@cat="ap" or @cat="advp") and node[(@rel="mod" or @rel="me") and @cat="np"]]
Schlichting: “Bepaler + Zelfstandig naamwoord + Bijvoeglijk woord. Soms vindt men een Zn op de plaats van de Bv: een heleboel water”
The current query covers both adjectives and adverbs. Unclear whether that is necessary or desired.
Remark: The example een heleboel water is not covered by this query
Remark: Currently any NP is allowed, this should be restrcted to an NP with only two nodes (determiner or modifier + head)
T017: BezZn¶
Name: BezZn.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 68
Implementation: Xpath
Query defined as:
//node[@cat="np" and node[@pdtype="det" and @rel="det" and @positie="prenom" and @vwtype="bez" and @pt="vnw" ] and node[(@pt="n" or (@pt="adj" and @positie="nom")) and @rel="hd"]]
Straightforward implementation. Also covers substantivised adjectives as a head.
Schlichting: “Bezittelijke voornaamwoord + Zelfstandig naamwoord”
Remark: according to the query, the NP can consist of more than just the possessive pronoun and the noun. This should be restricted
T018: Bijvoeglijke Bijzin¶
Name: Bijvoeglijke Bijzin.
Category:
Subcat:
Level: Zc
Original: yes
In form: no
Page:
Implementation: Xpath
Query defined as:
//node[@rel="mod"and @cat="rel" and parent::node[node[@pt="n" and @rel="hd"]]]
We have assumed that this involves relative clauses.
Schlichting: Though “bijvoeglijke bijzin” is mentioned with T020 (p. 53-54) there is to our knowledge no separate description of them.
Remark: Should whrel be included here as well, perhaps with a condition that there is also a noun or subtantivised adjective as antecedent?
T019: Bijwoordelijke bepalingwoordgroep¶
Name: Bijwoordelijke bepalingwoordgroep.
Category: Uitbreiding Zinsdelen
Subcat:
Level: Zc
Original: yes
In form: no
Page: 75
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “Bijwoordelijke bepalingwoordgroep”
T020: Bijzin z Verb¶
Name: Bijzin z Verb.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 53;54
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “Bijzin zonder verbindingswoord. De bijzinnen kunnen Bijwoordelijke bijzinnen zijn, VC bijzinnen of Bijvoeglijke bijzinnen.”
T023: Bv/B¶
Name: Bv/B.
Category: Eenwoordzin
Subcat:
Level: Zc
Original: yes
In form: yes
Page: 40
Implementation: Xpath with macros
Query defined as:
//node[@cat="top" and node[@pt='adj' or @pt='bw' or @pt='vz'] and count(.//node[%realnode%])=1]
These are single word utterances consisting of an adjective or an adverb, and based on the examples that Schlichting gives (such as uit, op) adpositions should be included as well. (uit is parsed as an adverb by Alpino, but op as an adposition).
The condition to restrict it to single word utterances uses the macro realnode, which is defined as follows:
realnode = """((not(@pt) or (@pt!="let" and @pt!="tsw")) and (not(@postag) or @postag!="NA()"))"""
It excludes interpunction symbols, interjections and words not classified for DCOI part of speech by Alpino (i.e., where postag=NA()),
Schlichting: “Bijvoeglijke naamwoorden en Bijwoorden”
Remark: R-pronouns (such as daar, explictly mentioned by Schlichting) are not covered yet. Perhaps use the macro adjoradv here.
T024: Bv z e¶
Name: Bv z e.
Category: Woordstructuur
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 87;88
Implementation: Xpath
Query defined as: not implemented yet
It is not clear whether predicative adjectives should be covered by this as well. We assume not. Schlichting only discusses attributive uses and only gives examples of attributive uses. A query could be:
//node[@pt="adj" and @buiging="zonder" and
(@rel="mod" and
parent::node[@cat="np" and node[@rel="hd" and (@pt="n" or(@pt="adj" and @positie="nom"))]])]
covering attributive adjectives modifying a noun or a substantivised adjective. The phenomenon occurs only once in the VKLTarsp dataset (sample 6, utterance 25) but it has not been annotated as such. It does not occur in [Schlichting 2017].
Schlichting: “Bijvoeglijk naamwoord zonder ‘e’-uitgang”
T025: BvBepZn¶
Name: BvBepZn.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 69
Implementation: Xpath
Query defined as:
//node[@cat="np" and node[@rel="det"] and node[@rel="mod" and @pt="bw" and (not(@special) or @special!="er")] and count(node)=3]
Straightforward implementation, some remarks:
the conditions on the special attribute are intended to exclude adverbial pronouns
there is no condition on the head. Perhaps a condition allowing only nouns and substantivised adjectives should be added (or should substantivised infinitives be included as well?)
there is no specific condition on the adverb (e.g. no mention of nog), Schlichting’s criterion “als het gevolgd kan worden door een telwoord” is difficult to implement.
Schlichting: “Bijvoeglijk woord + bepaler + Zelfstandig naamwoord. Het bijvoeglijk woord is meestal ‘nog’. We beschouwen ‘nog’ alleen als bijvoeglijk woord als het gevolgd kan worden door een telwoord.”
T027: BvZn¶
Name: BvZn.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 66
Implementation: Xpath with macros
Query defined as:
//node[%BvZn%]
Schlichting: “Bijvoeglijk woord + Zelfstandig naamwoord. Onder een Bijvoeglijk woord (nota bene niet een ‘bijvoeglijk naamwoord’) verstaan we woorden die een eigenschap of een hoeveelheid aangeven:
Bijvoeglijke naamwoordem, bijv. ‘grote’, ‘andere’
Telwoorden, bijvoeglijk gebruikt, bijv. ‘twee’, ‘eerste’
Bijvoeglijke woorden die een hoeveelheid aangeven, bijvoorbeeld ‘meer’, ‘veel’. Ook ‘nog’ wordt tot deze groep van bijvoeglijke woorden gerekend, namelijk dan wanneer het gevolgd kan worden door een telwoord: ‘nog auto’ is dus BvZn”
The query is defined using the macro BvZn, which is defined as follows:
BvZn = """ (count(node)=2 and
( (%headn% and (%coreadjmod% or %corevnwmod% or %corebwmod%))
)
)
"""
in which four other macros are used:
headn: simple head noun:
headn = """ ( node[@rel="hd" and @pt="n"] )"""
coreadjmod: adjective or numerals as modifier (note that ordinal numerals are analysed as adjectives by Alpino):
coreadjmod = """ (node[@rel="mod" and ( @pt="adj" or @pt="tw")] and not(node[@rel="det"])) """
corevnwmod: indefinite pronouns acting as determiner, excluding genitives such as iemands, and geen:
corevnwmod = """ (node[@rel="det" and (@pt="tw" or (@pt="vnw" and @vwtype="onbep" and @naamval!="gen" and @lemma!="geen"))] )"""
corebwmod: basically any modifying adverb that is quantificational in nature (by the macro qbwlemma (so far only nog):
corebwmod = """ (node[@rel="mod" and @pt="bw" and %qbwlemma%]) """ qbwlemma = """ (@lemma="nog" ) """
T029: BWondBB¶
Name: BWondBB.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 53
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_BWOndBB%]
Schlichting: “Onderwerp + Werkwoord + drie Bijwoordelijke bepalingen.”
Straigthforward implementation using of the macro Tarsp_BWOndBB, which is defined as follows:
Tarsp_BWOndBB = """(%declarative% and %Ond% and %Tarsp_W% and %Tarsp_B_X_count% = 3 and %realcomplormodnodecount% = 4)"""
See section Macros for Composed Language Measures for details.
T030: BX¶
Name: BX.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 46;47
Implementation: Xpath with macros
Query defined as:
//node[(@cat="top" and not(.//node[%pv%]) and not(.//node[@lemma="niet"]) and not(.//node[@rel="dlink"]) and ( (count(.//node[@cat="du" and count(node[%realnode%])=2 and node[%singlewordbw%] and node[(@pt or %corephrase%) ]])=1 ) or (node[%bxnp1%]) or (node[@cat="du" and count(node[%realnode%])=1 and node[%bxnp1%]]) or (node[%bxnp2%]) or (node[@cat="du" and count(node[%realnode%])=1 and node[%bxnp2%]]) )) or %Tarsp_bnonfin% ]
Explanation: Alpino cannot deal very well with such structures. If there are additional interjections or interpunction signs, Alpino often creates additional structure (with nodes with category du), which makes formulating the query not easy. Furthermore, the X-part of BX can be a full constituent consisting of multiple words, so simple conditions on the number of words will not work here. Two cases are distinguished:
Case 1: a match is found with a node with cat = top that does not contain a finite verb (pv) nor the word niet (niet is covered by a different language measure), and that does not contain at any depth a node with relation dlink: if such a node is present a full phrase or clause is present as well. If these conditions are met, several subcases are distinguished one of which must evaluate to True:
there must be exactly one node anywhere in the structure with cat = du, containing exactly 2 realnodes (macro realnode), one of which is a single adverb (macro singlewordbw, see below)* and the other is a single word or a corephrase. (example: daar auto), The macro corephrase is defined as follows:
corephrase = """(@cat="np" or @cat="pp" or @cat="advp" or @cat="ap")"""
there is a child node that meets the requirements of the macro bxnp1:
bxnp1 = """(@cat="np" and count(node)=2 and node[@rel="hd" and @pt="ww"] and node[@rel="mod" and @pt])"""
i.e. a NP that contains a verb as a head and a word that modifies it, and nothing else (example: even pakken .).
there is a node with cat = du that contains exactly one realnode that meets the requirements of the macro bxnp1
there is a child node that meets the requirements of the macro bxnp2, i.e. a noun phrase consisting of a head and a single word modifier, and nothing else (example: die hier):
bxnp2 = """(@cat="np" and count(node)=2 and node[@rel="hd"] and node[@rel="mod" and %singlewordbw%])"""
there is a node with cat = du that contains exactly one realnode that meets the requirements of the macro bxnp2
Case 2: the node meets the requirements of the macro Tarsp_bnonfin:
Tarsp_bnonfin = """((@cat="inf" or @cat="ppart") and @rel="vc" and parent::node[@cat="smain" and count(node)=1] and node[%Tarsp_B%] and node[@pt="ww" and @rel="hd"] and count(node[%realcomplormod%])=1 )"""
This covers cases where a nonfinite verbal complement is the only child of an smain category and contains only a head verb and an adverbial modifier (defined by macro Tarsp_B, see T007: B). Such structures can only arise after correction through the smallclauses module. (See Small clauses).
The macro singlewordbw is defined as follows:
singlewordbw = """ (@pt="bw" or %Rpronoun% or %adjadv%)"""
adjadv = """(@pt='adj' and (@lemma='wel'))"""
The macro Rpronoun simply lists all R-pronouns in a disjunction, and the adjadv macro is intended for words that are adverbs but analysed by Alpino (in some cases) as an adjective. So far, we only encountered this for wel (or was the adjoradv macro intended here?.
- Schlichting: “Bijwoordelijke bepaling + een ander Zinsdeel. Dit andere Zinsdeel kan zijn een W, een VC of een B. BW, BVC en BB staan namelijk niet apart op de profielkaart. In de praktijk worden hier ook zinnen gescoord met een B en een ander zinsdeel waarvan men niet weet of het een onderwerp of een VC is. De zin ‘die ook’ bijvoorbeeld, kan betekenen ‘die moet ook’ of ‘die moet je ook daar doen’. Zulke twijfelgevallen kunnen we hier dus ook scoren.”
Basically covered. if noun or pronoun precedes the adverb SASTA analyses the utterance as a OndB, not as a BX. That corresponded best to the actual practice in the annotated data received.
T031: C¶
Name: C .
Category: Zinsdelen
Subcat:
Level: Zc
Original: yes
In form: no
Page: 43;44
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “Complement. Een aanvulling van het gezegde. Het Complement kan zijn:
Naamwoordelijk deel van het gezegde. Het naamwoordelijk gezegde bestaat uit een koppelwerkwoord en (meestal) een zelfstandig naamwoord of bijvoeglijk naamwoord. Voorbeelden Zij is mijn zusje, Ben je moe, Dat is van hem
Bepaling van gesteldheid. Deze zegt iets van het werkwoord en van het lijdend voorwerp of het onderwerp. De bepaling van gesteldheid wordt ook wel ‘dubbelverbonden bepaling’ genoemd. Voorbeelden: Die ga ik groen kleuren, Dat vind ik lekker, hij heeft het kapot gemaakt, ik kan dat alleen
Bij de analyse worden Voorwerpen en Complementen alle twee ‘VC’ genoemd. “
T032: de¶
Name: de.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 66
Implementation: Xpath with macros
Query defined as:
//node[@cat="np" and node[@rel="det" and @lemma="de"] and count(node)=2]
Schlichting: “Bepaald lidwoord + enkelvoudig mannelijk of vrouwelijk Zelfstandig naamwoord of meervoudig Zelfstandig naamwoord. De constructies ‘in de mand’ en ‘naar het ziekenhuis’ worden gescoord bij VzBepZn èn bij ‘de’.” Covered.
T033: die/dezeZn¶
Name: die/dezeZn.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 68
Implementation: Xpath with macros
Query defined as:
//node[@cat="np" and node[@pt="n" and @rel="hd"] and node[@pt="vnw" and @vwtype="aanw" and @rel="det" and (@lemma="die" or @lemma="deze")] and count(node)=2]
The query is self-explanatory.
Schlichting: “Aanwijzend Voornaamwoord + Zelfstandig naamwoord in enkelvoud of meervoud”. Covered.
T034: dit/datZn¶
Name: dit/datZn.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 71
Implementation: Xpath with macros
Query defined as:
//node[@cat="np" and node[@pt="n" and @rel="hd"] and node[@pt="vnw" and @vwtype="aanw" and @rel="det" and (@lemma="dit" or @lemma="dat")] and count(node)=2]
The query is self-explanatory.
Schlichting: “Aanwijzend Voornaamwoord + onzijdig Zelfstandig naamwoord. Ook onjuist gebruik mag gescoord worden.” Covered.
T035: een¶
Name: een.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 64
Implementation: Xpath with macros
Query defined as:
//node[node[@lemma="een" and @pt="lid"] and node[@pt="n" and @rel="hd"] and count(node)=2 ]
The query is self-explanatory.
Schlichting: “Onbepaald lidwoord + Zelfstandig naamwoord. Ook het onbepaalde lidwoord + Bijvoeglijke woord, gebruikt als Zelfstandig naamwoord, wordt hier gescoord, bijvoorbeeld ‘een andere’.”
Remark Substantivised adjectives are not covered yet.
T036: en¶
Name: en.
Category: Verbindingswoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 77
Implementation: Xpath with macros
Query defined as:
//node[@pt="vg" and @lemma="en" and @rel="dlink"]
This query is self-explanatory. The condition on rel excludes en in conjunctions.
Schlichting: “Het verbindingswoord ‘en’ wordt in deze Fase gebruik om een zinsdeel of een zin te introduceren. In Fase VI wordt ‘en’ gebruikt om twee hoofdzinnen met elkaar te verbinden.”
T037: er¶
Name: er.
Category: Voornaamwoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 81
Implementation: Xpath
Query defined as:
//node[@lemma="er"]
The query is self-explanatory. Note that the condition formulated in terms of the atribute lemma ensures that also variants of er such as Er, d’r and even der can be dealt with.
Schlichting: “Dit ‘er’ heeft slechts in enkele gevallen het karakter van een voornaamwoord. Het kan als zinsdeel de functie hebben van een bijwoordelijke bepaling: ‘zij komt er wel’, ‘ik heb er nog drie’, of van een onderwerp, bv. In: ‘er stond een plant voor het raam’. Zo’n zin kan dan twee onderwerpen hebben (‘er’ is dan een zgn. plaatsonderwerp, het andere onderwerp is dan het getalsonderwerp, dat wil zeggen dat dat onderwerp in het meervoud kan komen te staan: ‘er staat een kind’, ‘er staan twee kinderen’)”
T038: geen X¶
Name: geen X.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 72
Implementation: Xpath
Query defined as:
//node[@cat="np" and node[@pt="vnw" and @rel="det" and @lemma="geen"] and node[@pt="n" and @rel="hd"]]
The query is self-explanatory.
Schlichting: “‘geen’ meestal gevolgd door een zelfstandig naamwoord”
Remark perhaps the number of nodes should be restricted to 2? Though geen N meer should also be accepted
T039: hè¶
Name: hè.
Category: Zinsconstructies
Subcat: Vragen
Level: Zc
Original: yes
In form: yes
Page: 59;60
Implementation: Xpath
Query defined as:
//node[ @lemma="hè"]
The query is self-explanatory, but it is perhaps too broad. Occurrences of hè at the beginning of an utterance will also be scored now.
Schlichting: “De naklank ‘hè’ in de betekenis van ‘vind je niet?’ of ‘dat vind je toch ook?’ maakt de voorafgaande uiting vragend.”
T040: hem¶
Name: hem.
Category: Voornaamwoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 82
Implementation: Xpath
Query defined as:
//node[@lemma="hem" and @pt="vnw"]
The query is self-explanatory. It is crucial that the formulation is in terms of lemma (and not word), otherwise we will miss cases where the word is written with an initial capital (Hem) and reduced forms (‘m).
Schlichting: “Het persoonlijk voornaamwoord ‘hem’, gebruikt als voorwerp.”. Covered.
T041: het¶
Name: het .
Category: Voornaamwoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 80
Implementation: Xpath
Query defined as:
//node[@lemma="het" and @pt='vnw' ]
The query is self-explanatory. The condition that the pt is equal to vnw excludes occurrences of het as an article (because then pt = lid). The check on the lemma form allows also reduced forms (‘t) and occurrences where the word contains capital characters (e.g. Het).
Schlichting: “Persoonlijk voornaamwoord ‘het’, onder andere als lijdend voorwerp en als onderwerp.” Covered.
T042: hetZn¶
Name: hetZn.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 71
Implementation: Xpath
Query defined as:
//node[@cat="np" and node[@pt="lid" and @rel="det" and @lemma="het"] and node[@rel="hd" and @pt="n"]]
The query is self-explanatory. * Remark We shoud add a condition so that only these two elements are allowed inside the NP.
Schlichting: “Bepaald lidwoord + onzijdig Zelfstandig naamwoord. Ook onjuist gebruik mag gescoord worden. De constructie ‘naar het strand’ wordt gescoord bij VzbepZn èn bij hetZn.”
T043: hij¶
Name: hij.
Category: Voornaamwoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 80
Implementation: Xpath with macros
Query defined as:
//node[@lemma="hij" and @pt="vnw"]
The query is self-explanatory. The condition formulated in ters of lemma allows reduced forms (ie, even tie and -ie) and occurrences with capital letters (Hij)
Schlichting: “‘hij’ en ‘ie’ worden op de profielkaart beide bij ‘hij’ gescoord. ‘ie’ komt vaker voor dan ‘hij’ in deze Fase.”
T044: Hww i¶
Name: Hww i.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 65
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_Hwwi%]
The macro Tarsp_Hwwi is defined as follows:
Tarsp_Hwwi = """(( @pt="ww" and @rel="hd" and @wvorm="pv" and
%Tarsp_hww% and
((parent::node[node[@cat="inf"and @rel="vc"]]) or
(parent::node[node[@pt="ww" and @rel="vc" and @wvorm="inf"]])
)
) or
%robusthwwi%
)"""
Explanation:
The core case can be described as follows:
We look for a finite verb acting as a head
that is an auxiliary verb (macro Tarsp_hww)
and cooccurs with a verbal complement (rel = vc) that is either an infinitval phrase (cat = inf) or a bare infinitive (wvorm = inf)
There is also a second case, for robustness (i.e. if Alpino misanalysed), with the macro robusthwwi, defined as follows:
robusthwwi = """ (@cat="top" and
.//node[@pt="ww" and %Tarsp_hww% and @wvorm="pv" and (@rel="--" or @rel="dp")] and
.//node[@pt="ww" and @wvorm="inf" and (@rel="--" or @rel="dp")]
)"""
This macro searches for a misanalysed (rel equal to – or dp) finite auxiliary verb anywhere in the structure, in combination with a misanalysed infinitive verb anywhere in the structure.
Schlichting: “Hulpwerkwoord gevolgd door een infinitief oftewel het hele werkwoord. De volgende werkwoorden rekenen we tot de hulpwerkwoorden; enkele worden verschillende malen genoemd:
Hebben, zijn, zullen (hulpww van tijd|)
Gaan, komen, zijn, blijven, zitten, liggen lopen, staan (aspect)
Worden, zijn (lijdende vorm)
Kunnen, zullen, mogen, moeten, willen, laten, hoeven, (be)horen (modaliteit)
Doen en laten (causaliteit)
Doen (omschrijving)
De hulpwerkwoorden die gevolgd worden door een infinitief worden niet gescoord bij HwwZ, Stam of Stam+t”.
Remark: the auxiliary verbs that take te plus infinitive are not covered yet
Remark: should we count zitten, staan, liggen, etc. as HwwZ when not followed by infinitive or te + infinitive?
T045: Hww Vd¶
Name: Hww Vd.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 69
Implementation: Xpath with macros
Query defined as:
//node[(@pt="ww" and @rel="hd" and (@lemma="hebben" or @lemma="worden" or @lemma="zijn") and parent::node[node[@rel="vc" and (@cat="ppart" or (@pt="ww" and @wvorm="vd"))] ]) or %robusthwwvd%]
The core case of the query searches for one of the verbs hebben, worden or zijn in combination with a verbal complement that is either a participial phrase (ppart) or a bare participle (wvorm = ppart).
There is also a robust case (macro robusthwwvd):
robusthwwvd = """ (@cat="top" and
.//node[@pt="ww" and %vdhwws% and @wvorm="pv" and (@rel="--" or @rel="dp")] and
.//node[@pt="ww" and @wvorm="vd" and (@rel="--" or @rel="dp")]
)"""
This macro searches for a misanalysed (rel equal to – or dp) finite auxiliary verb that takes a participial clause (macro vdhwws) anywhere in the structure, in combination with a misanalysed participial verb anywhere in the structure. The macro vdhwwws is defined as follows:
vdhwws = """(@lemma="hebben" or @lemma="zijn" or @lemma="worden")"""
Schlichting: “Hulpwerkwoord gevolgd door het Voltooid deelwoord. Hulpwerkwoord + Voltooid deelwoord komt voor bij de voltooide tijd (‘jij hebt gemaakt’). Het Voltooid deelwoord wordt apart bij de Woordstructuur gescoord. ook voltooid verleden tijd (‘jij had gemaakt’) kan hier gescoord worden. dan wordt eveneens bij Verleden tijd en bij Voltooid deelwoord gescoord. Het hulpwerkwoord wordt niet gescoord bij Woordstructuur.” Fully covered.
Remark The macro vdhwws should also be used in the core case of this query.
T046: HwwZ¶
Name: HwwZ.
Category: Woordstructuur
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 84-85
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_HwwZ%]
The macro Tarsp_HwwZ is defined as follows:
Tarsp_HwwZ = """(@pt="ww" and @rel="hd" and @wvorm="pv" and
((%Tarsp_hww% or @lemma = "hebben") and
not(%Tarsp_vc_sibling%)
) or
(@lemma="zijn" and not(%Tarsp_vc_sibling%) and %Tarsp_ld_sibling% )
)"""
The query searches for finite forms of a head verb and distinguishes two subcases:
the finite form is a form of one of the auxiliary verbs that take an infinitival complement, or of the verb hebben: in this case no vc complement should be present, which is achieved by negting the Tarsp_vc_sibling macro, defined as follows:
Tarsp_vc_sibling = """parent::node[ node[@rel="vc"]]"""
the finite form is a form of the verb zijn: in this case a vc complement must be absent but an ld sibling must be present. This has been done to exclude regular copular uses of the verb zijn (when combined with an adjective (phrase) or noun (phrase) predicate. The presence of the ld sibling is ensured by the Tarsp_ld_sibling macro:
Tarsp_ld_sibling = """parent::node[ node[@rel="ld"]]"""
- Schlichting: “Hulpwerkwoord Zelfstandig gebruikt. De eerste persoonsvorm die door de kinderen na het koppelwerkwoord geleerd wordt is de tegenwoordige tijd enkelvoud van het hulpwerkwoord, ‘zelfstandig’ gebruikt, ‘Zelfstandig’ betekent in dit verband dat het hulpwerkwoord niet gevolgd wordt door een hoofdwerkwoord. Tot deze ‘zelfstandig’ gebruikte hulpwerkwoorden rekenen we: blijven, doen, gaan, hebben, hoeven, horen, komen, kunnen, laten, liggen, lopen, moeten, mogen, staan, willen, worden, zijn, zitten, zullen”. The list provided here is identical to the list provided under T044 (see T044: Hww i), except that behoren is not explicitly mentioned, plus the auxiliaries hebben, zijn en worden.” Covered.
“Wanneer het hulpwerkwoord wordt gevolgd door een onbepaalde wijs (bijvoorbeeld; ‘ik wil fietsen’) of door een onvoltooid deelwoord (bijvoorbeeld: ‘ze heeft geroepen’) wordt het hier niet gescoord (zie Hww i, Fase III en Hww Vd, Fase IV).” Covered
“Wanneer het ‘zelfstandig’gebruikte hulpwerkwoord in het meervoud staat (bijvoorbeeld: ‘nou gaan we’) wordt het niet hier gescoord maar bij het Meervoud van de Tegenwoordige Tijd.” Not covered
“Wanneer het ‘zelfstandig’ gebruikte hulpwerkwoord in de verleden tijd staat (bijvoorbeedl: ‘dat deed jij’) wordt het niet hier gescoord maar bij de Verleden tijd.” Not Covered
“De hulpwerkwoorden die hier gescoord worden, behoeven niet gescoord te worden bij Stam of Stam + t.”See under Stam and Stam + t
Remark the uncovered cases can be improved by adding the following condition: @pvagr!=”mv” and @pvtijd!=”verl”
T047: ik¶
Name: ik.
Category: Voornaamwoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 79
Implementation: Xpath with macros
Query defined as:
//node[@lemma="ik" ]
The query is self-explanatory. The condition formulated in terms of lemma (instead of word) also allows variants in case (Ik), reduced variants (‘k, k) en emphasised variants (ikke)
Schlichting: “Het persoonlijk voornaamwoord ‘ik’ wordt meestal in deze Fase ontwikkeld” (It involves Stage III).
T048: Into¶
Name: Into.
Category: Zinsconstructies
Subcat: Vragen
Level: Zc
Original: yes
In form: yes
Page: 59
Implementation: Xpath with macros
Query defined as: not implemented yet
Though it is trivial to implement this by looking for the presence of a question mark or exclamation mark, the use of these annotations by annotators appears to be qufaantekeninite inconsistent.
Schlichting: “In Fase II stellen de kinderen alleen vragen door een vraagintonatie te gebruiken. Bij het uitschrijven van het sample wordt die intonatie aangegeven door een vraagteken. Bij het analyseren van een uiting die alleen vraagintonatie heeft maar niet de vorm van de vraag analyseren we de uiting eerst als Mededelende Zin; de aantekening intonatie komt er extra bij. Vraag-Intonatie is dus een extra scoring “
T049: Inv¶
Name: Inv.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 50
Implementation: Xpath with macros
Query defined as:
//node[(((@cat="sv1" and not(parent::node[@cat="whq"])) or (@cat="smain" ) ) and node[@pt="ww" and @rel="hd" and (not(@stype) or @stype!="imparative") ] and node[@rel="su" and number(@end)>../node[@rel="hd"]/@end] ) or %robustinversion%]
The core case of this query looks for
sv1 nodes not dominated by a whq node or smain node
that are not imperatives
in which the finite verb precedes the subject.
The robust case has been implemented by the macro robustinversion, defined as follows:
robustinversion = """(@cat="top" and
.//node[@pt="ww" and @wvorm="pv" and @rel!="hd" and
(not(@stype) or @stype!="imparative")] and
.//node[@pt="vnw" and @naamval="nomin" and @rel!="su" and
@end>..//node[@pt="ww" and @wvorm="pv"]/@end])"""
It searches for wrongly analyzed structures (cf. the conditions on the grammatical relations) that contain:
anywhere a non-head finite verb that is not marked for stype = imparative (Alpino indeed uses this misspelling of imperative as value)
anywhere a non-subject nominative pronoun
where the pronoun precedes the verb
- Schlichting:
“Hiermee wordt bedoeld de inversie of omkering van persoonsvorm en onderwerp. In de eerste constructie met onderwerp en persoonsvorm die zij leren, gebruiken de kinderen de volgorde onderwerp - persoonsvorm. heel spoedig daarna leren ze de inversievolgorde”
Inversie hoeft in principe niet elke keer dat het voorkomt gescoord te worden. Het is van belang dat men bij elke leerling die in de Grammaticale Ontwikkelingsfase III of hoger valt controleert dat het kind twee- of driemaal de inversievolgorde gebruikt. Fouten tegen de regel van inversie worden apart genoteerd, bijvoorbeeld achterop de profielkaart. Het is waarschijnlijk dat in Zuid-Nederland deze Inversie wat later wordt geleerd dan in Noord-Nederland”
Remark It is probably best to restrict Inversion to declarative clauses only. We probably can best use the macro declarative for this. Currently we exclude wh-questions and imperatives, but not yes-no questions. That will also make the exclusion of imperatives better and consistent with other language measures.
T050: jij¶
Name: jij.
Category: Voornaamwoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 79
Implementation: Xpath
Query defined as:
//node[@lemma="jij" or (@lemma="je" and (@vwtype="pr" or @vwtype="pers") and (@rel="su" ))]
The reduced form je does not have jij as its lemma, probably because it is also a reduced form of jou and jouw. Its lemma is je. The values of vwtype can be pers (personal pronoun, for jij) or pr (personal or reflexive pronoun, for je).
Schlichting: “We vinden overwegend ‘jij’ in deze Fase; een enkele maal ‘je’.” (it concerns Stage III).
Remark use of jij as a single word utterance, or as a single word as dp, sat, or nucl is not covered now.
T051: jou¶
Name: jou.
Category: Voornaamwoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 82
Implementation: Xpath
Query defined as:
//node[(@lemma="je" or @lemma="jou") and (@vwtype="pr" or @vwtype="pers") and (@rel="obj1" or @rel="obj2")]
The query is self-explanatory.
The lemma differs for jou and je, see also T050: jij
The vwtype for je and jou differs, so both options must be allowed
The condition on the rel attribute is to exclude the possessive pronoun je
Remark It should be checked whether both a condition on vwtype and a condition on rel are needed
Remark: the query does not account for cases where jou is the only word in an utterance, or the only word in a dp, sat or nucl node. Extension of the query is needed. Or when it is a predc (als ik jou was).
Schlichting: “Het persoonlijk voornaamwoord ‘jou’ wordt zowel gebruikt als voorwerp als in bepalingen met een voorzetsel.”
T052: Kop¶
Name: Kop.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 47;48
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_Kop%]
The macro Tarsp_Kop is defined as follows:
Tarsp_Kop = """( @pt="ww" and @rel="hd" and not(%Tarsp_hww%) and
((%Tarsp_predc_sibling% and not(%Tarsp_obj1_sibling%)) or
(@lemma="zijn" and not(%Tarsp_vc_sibling%)
and not(%Tarsp_ld_sibling%) and
not(%Tarsp_onlymodR_sibling%))
)
)"""
Explanation: a node is a copular verb if it is a verb acting as a head but not an auxiliary verb (Tarsp_hww), and
either it has a predc sibling (tarsp_predc_sibling) but no direct object (obj1) sibling (Tarsp_obj1_sibling), or
the lemma equals zijn and it has no vc sibling (Tarsp_vc_sibling), no ld sibling, and no Rpronoun as modifier if no predicate is present (Tarsp_onlymodR_sibling)
The sibling macros are defined as follows:
Tarsp_vc_sibling = """parent::node[ node[@rel="vc"]]"""
Tarsp_predc_sibling = """parent::node[ node[@rel="predc"]]"""
Tarsp_obj1_sibling = """parent::node[ node[@rel="obj1"]]"""
Tarsp_ld_sibling = """parent::node[ node[@rel="ld"]]"""
Tarsp_onlymodR_sibling = """(parent::node[node[@rel="mod" and %Rpronoun%] and not(node[@rel="predc"])])"""
Remark These definition could be simplified by replacing parent::node[node by ../node
The definition of zijn as a copula does not require the presence of a predicate, it rather excludes complements and modifiers that indicate a different usage of zijn (e.g., as an auxiliary verb, or as an independent verb of location). This has been done because in incomplete sentences or misanalysed sentences containing zijn, the most probable interpretation of zijn is in most cases as a copula (with a left-out or unpronounced predicate).
The query does not list the copulas explicitly because there are many more copulas than traditionally assumed and also many more than [Schlichting 2005:47] mentions for child language (e.g. gaan, aanvoelen, raken, etc).
Schlichting: “Koppelwerkwoord. We onderscheiden twee soorten gezegdes:”
“Werkwoordelijke, bijvoorbeeld ‘zij werkt ‘s morgens’”
“Naamwoordelijke gezegdes, bijvoorbeeld ‘hij is tuinman’ en ‘jij bent moe’.”
“Naamwoordelijke gezegdes bestaan uit een koppelwerkwoord en een naamwoordelijk deel van het gezegde. Het koppelwerkwoord heeft niet veel betekenis van zichzelf. het vormt de koppeling tussen het onderwerp en het naamwoordelijk deel van het gezegde. Het meest gebruikte koppelwerkwoord, ook in de kindertaal, is ‘zijn’. Andere voorkomende koppelwerkwoorden zijn ‘worden’, ‘blijven’ en ‘lijken’.
“Aanvankelijk wordt alleen de vorm ‘is’ nog gebruikt. Het Koppelwerkwoord wordt geanalyseerd als speciaal Zinsdeel ‘W-Kop’en wordt apart bij de Zinsdelen in Fase II gescoord.”
Note In Sasta a copular verb is annotated by the code Kop, not by W-Kop. Since it is also a verb, two codes are used: W,Kop
T053: maar¶
Name: maar.
Category: Verbindingswoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 77
Implementation: Xpath with macros
Query defined as:
//node[@lemma="maar" and ((@pt="vg") or (@pt="bw" and parent::node[@cat="smain"] and @begin=parent::node/@begin)) ]
The first part of this query is self-explanatory, the second part has been included because Alpino often analyses an sentence-initial maar as an adverb instead of conjunction
Schlichting: “Het Verbindinsgwoord ‘maar’. Dit wordt introducerend gebruikt en om twee hoofdzinnen te verbinden.”
T054: Sv1¶
Name: Sv1.
Category:
Subcat:
Level: Zc
Original: no
In form: no
Page:
Implementation: Xpath with macros
Query defined as:
//node[@cat="sv1" and (@rel="--" or @rel="nucl")]
This is not a language measure of Tarsp, but it has been defined to identify verb-initial clauses (for development purposes)
T055: Mededelende Zin¶
Name: Mededelende Zin.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: no
Page: 44;45
Implementation: none
Query defined as: not implemented
This is not really a language measure, but rather a class of language measures that apply to declarative sentences.
T056: mij¶
Name: mij.
Category: Voornaamwoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 81
Implementation: Xpath with macros
Query defined as:
//node[(@lemma="me" or @lemma="mij") and (@vwtype="pr" or @vwtype="pers")]
The lemma differs for mij and me, see also T050: jij
The vwtype for mij and me differs, so both options must be allowed
Schlichting: “Persoonlijk voornaamwoord ‘mij’, ‘me’, gebruikt als voorwerp en na een voorzetsel”
T057: MvTT¶
Name: MvTT.
Category: Woordstructuur
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 87
Implementation: Xpath with macros
Query defined as:
//node[@pt="ww" and @pvagr="mv" and @pvtijd="tgw" ]
The query is self-explanatory.
Schlichting: “Meervoud van de tegenwoordige Tijd. Het gaat hier om congruentie tussen Onderwerp en Werkwoord in het meervoud van de 1e, 2e en 3e persoon van de tegenwoordige tijd van het werkwoord, ook van het hulpwerkwoord. We bedoelen hier constructies als ‘ze komen’, ‘wij spelen’. Het moet duidelijk zijn dat het werkwoord geen infinitief is. Ook de congruentie tussen persoonsvorm en naamwoordelijk deel van het gezegde wordt hier gescoord”
Remark Though Schlichting refers to “congruentie”(agreement), the query can be formulated in terms of properties of the finite verb alone.
T058: MvZn¶
Name: MvZn.
Category: Woordstructuur
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 83;84
Implementation: Xpath with macros
Query defined as:
//node[@pt = "n" and @getal="mv" ]
The query is self-explanatory.
Schlichting: “Meervoudig Zelfstandig naamwoord. Zowel meervoudsvormen op ‘-en’ als op ‘-s’ komen voor. Wanneer kinderen een meervoud van de zelfstandige naamwoorden gebruiken betekent dit nog niet altijd dat zij de meervoudsregel begrepen hebben. Het woord ‘schoenen’ bijvoorbeeld kan heel goed alleen in het meervoud gebruikt worden door het kind. Daarom is het van belang dat men aantekent op de profielkaart welke meervoudsvormen men gevonden heeft. Zodra er een meervoudsvorm is gevonden van een woord dat waarschijnlijk ook in het enkelvoud is aangeboden, is het extra aantekenen niet meer nodig.”
Remark The plural forms are currently not put into the form, and it is not tracked whether a singular form has already occurred.
T059: Nabep¶
Name: Nabep.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 73
Implementation: Xpath
Query defined as:
// node[@rel="mod" and (not(@lemma) or (@lemma!="ook" and @lemma!="dan")) and parent::node[@cat="np" and node[@rel="hd" and @pt!="ww"]] and @begin >= ../node[@rel="hd"]/@end]
Explanation:
the node must be a modifier, and either a phrase, or if it is a single word it should not be ook or dan
it must modify a head that is not a verb inside an NP
the node must follow the head
Schlichting: “In ‘de grote plant’ zijn ‘de’ en ‘grote’ bepalende woorden bij ‘plant’, in ‘de plant met de grote bladeren’is ‘met de grote bladeren’een bepaling bij ‘plant, maar deze bepaling staat na het kernwoord. Dit noemen we een ‘nabepaling’.”
Remark Alpino (incorrectly) analyses many verbless utterances as involving a nabepaling (e.g., ladder hier). We try to reanalyse these in the module Small clauses.
T060: Nevens¶
Name: Nevens.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Sz
Original: yes
In form: yes
Page: 56
Implementation: Xpath
Query defined as:
//node[node[@rel="cnj" and (@cat="smain" or @cat="sv1")] and node[@rel="crd" ]]
Explanation:
the node must contain an smain or sv1 with grammatical relation cnj, and
it must contain a coordinator
Schlichting: “Nevenschikkende zin met Verbindingswoord, dus twee hoofdzinnen verbonden door ‘en’, ‘maar’, ‘want’, ‘of’, of ‘dus’.” Covered.
T061: Nevenschikkende¶
Name: Nevenschikkende.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: no
Page: 54
Implementation: Xpath
Query defined as: not implemented yet
This is actually not a language measure but subclass of “samengestelde zinnen”.
T062: Ombep¶
Name: Ombep.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 74
Implementation: Xpath with macros
Query defined as: not implemented yet
Alpino does not analyse the only example given (geen … meer) as single constituent. So far we did not encounter any examples. It does not occur in [Schlichting 2017].
Schlichting: “Omsluitende bepaling: een gedeelte van de bepaling staat vóór het kernwoord, een gedeelte erachter.”
T063: Ond¶
Name: Ond.
Category: Zinsdelen
Subcat:
Level: Zc
Original: yes
In form: no
Page: 41
Implementation: Xpath with macros
Query defined as:
//node[%FullOnd%]
Here, FullOnd is defined as:
FullOnd = """(%subject% or %erx%)"""
where subject (see Macros for ‘Zinsdelen’) and erx are defined as:
subject = """(@rel="su" and parent::node[(@cat="smain" or @cat="sv1" or @cat="ssub")])"""
erx = """((@rel="mod" and @lemma="er" and ../node[@rel="su" and @begin>=../node[@rel="mod" and @lemma="er"]/@end]) or
(@rel="mod" and @lemma="er" and ../node[@rel="su" and not(@pt) and not(@cat)])
)
"""
The erx macro is to ensure that so-called expletive er also counts a subject. We implemented this in the following manner: er is considered (also) expletive (and thus must count as a subject):
if it precedes the subject (as in er kwam iemand binnen), or
if there is an empty subject, to cover cases such as wie zwom er. Note that in wie heeft er gezwommen* this er is considered a subject because of the empty subject of the participial clause, which is perhaps not what we want.
Schlichting: “Dit is de persoon of de zaak die de handeling van het werkwoord uitvoert. Wanneer het onderwerp van een zin in het meervoud staat, staat de persoonsvorm van die zin ook in het meervoud.”
T064: OndB¶
Name: OndB.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 45
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_OndB%]
Straightforward implementation:
Tarsp_OndB = """(%Ond% and
node[%Tarsp_Basic_B%] and
count(node) = 2)"""
See Small clauses for details.
Schlichting: “Onderwerp + Bijwoordelijke Bepaling”
T065: OndBVC¶
Name: OndBVC.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 51
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_OndBVC%]
Straightforward implementation:
Tarsp_OndBVC = """(%Ond% and
node[%Tarsp_Basic_B%] and
node[%Tarsp_Basic_VC%] and
count(node) = 3) """
See Small clauses for details.
Schlichting: “Onderwerp + Bijwoordelijke Bepaling + Voorwerp of Complement”
T066: Onderbr¶
Name: Onderbr.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 58
Implementation: Xpath with macros
Query defined as: not implemented yet
This is not really a language measure. It is rather an indication of how sentences containing an interruption by another sentence or phrase should be analysed.
Schlichting: “Soms wordt de zin onderbroken met een korte zin of een Zinsdeel. Beide zinnen worden ook apart geanalyseerd”
T067: Onderschikkend: B¶
Name: Onderschikkend: B.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: no
Page: 42;55;56
Implementation: Xpath with macros
Query defined as:
Tarsp_B = """( ((((@rel="mod" or @rel="ld" or @rel="predm") and (not(@cat) or @cat!="conj") and (not(@pt) or @pt!="tsw") )or (%predcB%) ) and (../node[@pt="ww" and @rel="hd"]) ) or ((@pt="vz" or @pt="bw" or %Rpronoun%) and (@rel="dp" or @rel="--" or @rel="nucl" or @rel="body") and %notonlyrealnode%) or (@cat="pp" and (@rel="--" or @rel="dp") and %notonlyrealnode%) or (@rel="pc" and ../node[@rel="hd" and %locverb%]) or (@rel="cnj" and parent::node[@rel="mod" or @rel="ld" or @rel="predm"]) or (@rel="mod" and @pt="bw" and parent::node[@cat="np"] ) or %Tarsp_nonmodadvcp% or %Tarsp_pc_vc_exception% ) """
The query distinguishes several subcases as alternatives (joined by or):
Any node with one of the grammatical relations mod, ld or predm provided:
the category, if present, is not equal to conj (the individual conjuncts do count, see below.
if the node is for a word, the word should not be an interjection
it has a head verb as sibling
Any adverbial predicate that has a head verb as sibling. Adverbial predicates are defined by the macro predcB:
predcB = """(@rel="predc" and (@pt="adj" or @pt="bw" or @cat="ap" or @cat="advp") and ../node[@rel="obj1"] )"""
so these are predc nodes with a direct object as sibling
Remark We now miss cases where the object is absent due to Topic drop (e.g. vind ik goed). Maybe replace it by disallowing real copulas.
Any adpositon, adverb or Rpronoun not in a core grammatical relation but with any of the relations dp, –, nucl, body provided that they are not the only realnode present, where a real node is any node that is not a interjection, interpunction sign or of unknown part of speech tag:
realnode = """((not(@pt) or (@pt!="let" and @pt!="tsw")) and (not(@postag) or @postag!="NA()"))""" notonlyrealnode = """(parent::node[count(node[%realnode%])>1])"""
Any PP with any of the grammatical relations – or dp if it is not the only real node.
Any adpositional complement to a locative verb. Alpino analyses locative and directional complements to a locative verb often as an adpositional complement. That is often correct but not for very young children that Tarsp addresses. This part of the query corrects for that.
Any conjunct (relation cnj) in a conjunction that bears the grammatical relation mod, ld or predm.
Any adverb inside an NP with relation mod. This is especially for adverbs such as ook, nog, niet, alleen, but also for adverbs inside NPs with a substantivised verb as a head (e.g. even pakken).
Adverbial clauses that occur independently (e.g. als mama thuiskomt), as defined by the macro Tarsp_nonmodadvcp:
Tarsp_nonmodadvcp = """(@cat="cp" and (@rel="dp" or @rel="--") and node[@pt="vg" and @conjtype="onder" and @lemma!="dat" and @lemma!="of" ] )"""
remark I am not sure that this should be kept. Maybe move it to Bbijzin
- Adpositional complements that should have been analysed as adverbial modifiers. Achieved by the macro Tarsp_pc_vc_exception, explained elsewhere (see section Generation of macros).
Remark There is overlap with the condition on locative verbs above. It has to be found out how big the overlap is, and whether this can be simplified.
Schlichting: “De bijwoordelijke bepaling zegt iets over de hele inhoud van de zin of iets over het werkwoord, een bijwoord of bijvoeglijk naamwoord. Een zin kan meer dan één bijwoordelijke bepaling hebben”
T068: Onderschikkend: VC¶
Name: Onderschikkend: VC.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: no
Page: 43-44;55
Implementation: Xpath with macros
Query defined as:
//node[%complement% and parent::node[(@cat="smain" or @cat="sv1" or @cat="ssub" or @cat="inf" or @cat="ppart") ]]
This is not a language measure that occurs in the form. The macro complement is defined as follows:
complement = """((@rel="obj1" or @rel="obj2") or
(@rel="predc" and not(%predcB%)) or
(@rel="pobj1" and not(%pobj1B%) ) or
%verbalcomplement%
)"""
verbalcomplement = """(@rel="vc" and (@cat="cp" or @cat="whq" or @cat="whsub"))"""
predcB = """(@rel="predc" and
(@pt="adj" or @pt="bw" or @cat="ap" or @cat="advp") and
../node[@rel="obj1"]
)"""
pobj1B = """(@rel="pc" and ../node[@rel="hd" and %locverb%])"""
Remark The differences with Tarsp_VC (see Macros for ‘Zinsdelen’) should be studied, and we should probably make them equal. T068 has not been maintained systematically, so very likely Tarsp_VC is the best one.
Schlichting: “Lijdend Voorwerp, Meewerkend Voorwerp en Voorzetselvoorwerp worden alle drie afgekort ot V. Dit betekent dat ze na de analyse alle drie onder V gescoord worden. Een Complement is een aanvulling van het gezegde. Het complement kan zijn: (1) naamwoordelijk deel van het gezegde; (2) bepaling van gesteldheid”
Remark Schlichting’s example ‘Ik kan dat alleen’ is not a complement, in my view, but a modifier (secondary predicate).
Bij de analyse worden Voorwerpen (V) en Complementen (C) alle twee ‘VC’ genoemd. Dit doen we omdat het, vooral bij de tweewoordzinnen, niet altijd mogelijk is de Voorwerpen en de Complementen te onderscheiden. Bovendien wordt hierdoor het aantal constructies beperkt.”
T069: Onderschikkende¶
Name: Onderschikkende.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: no
Page: 54;55
Implementation:
Query not implemented
This is mot really a query, but rather a class of clauses: subordinate clauses.
Schlichting: “Bijzinnen horen meestal bij een hoofdzin. Men kan een bijzin herkennen aan de plaats waar de persoonsvorm staat, dat is namelijk helemaal of bijna helemaal achteraan in de zin.”
T070: Onderwerpswoordgroep¶
Name: Onderwerpswoordgroep.
Category: Uitbreiding Zinsdelen
Subcat:
Level: Zc
Original: yes
In form: no
Page: 75
Implementation: Xpath
Query defined as:
//node[@rel="su" and @cat]
Schlichting: ” Bij het behandelen van de Woordgroepen hebben we gezien dat een zinsdeel uitgebreid kan zijn tot een woordgroep.”
Remark We currently do not underline these phrases, nor do we circle the utterance id in the form.
T071: OndVC¶
Name: OndVC.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 46
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_OndVC%]
Straightforward implementation:
Tarsp_OndVC = """(%Ond% and
node[%Tarsp_Basic_VC%] and
count(node) = 2) """
where:
Tarsp_BasicVCW = """(node[@pt="ww" and @rel="hd"] and
node[%Tarsp_Basic_VC%] and
count(%fillednode%)=2)"""
See section Small clauses for details.
Schlichting: “Onderwerp + voorwerp of Complement”
T072: OndW¶
Name: OndW.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 46
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_OndW%]
Straightforward implementation:
Tarsp_OndW = """(%declarative% and
%Ond% and
(%Tarsp_W% or node[%Tarsp_onlyWinVC%]) and
%realcomplormodnodecount% = 1 )"""
See section Macros for Composed Language Measures for details.
The only special part here is the macro Tarsp_onlyWinVC, defined as follows:
Tarsp_onlyWinVC = """(@rel="vc" and node[@rel="hd" and @pt="ww" and %realcomplormodnodecount% = 0])"""
This covers cases in which the predicate is a nonfite verb. See Small clauses for clarification.
Schlichting: “Onderwerp + Werkwoord”
T073: OndWB¶
Name: OndWB.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 49
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_OndWB%]
Straightforward implementation:
Tarsp_OndWB = """(%declarative% and
%Ond% and
%Tarsp_W% and
%Tarsp_B_X% and
%realcomplormodnodecount% = 2 )"""
See section Macros for Composed Language Measures for details.
Schlichting: “Onderwerp + Werkwoord + Bijwoordelijke Bepaling”
T074: OndWBB¶
Name: OndWBB.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 52
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_OndWBB%]
Straightforward implementation:
Tarsp_OndWBB = """(%declarative% and
%Ond% and
%Tarsp_W% and
%Tarsp_B_X_count% = 2 and
%realcomplormodnodecount% = 3)"""
See section Macros for Composed Language Measures for details.
Schlichting: “Onderwerp + Werkwoord + Bijwoordelijke Bepaling + Bijwoordelijke Bepaling”
T075: OndWBVC¶
Name: OndWBVC.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 52
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_OndWBVC%]
Straightforward implementation:
Tarsp_OndWBVC = """(%declarative% and
%Ond% and
%Tarsp_W% and
%Tarsp_B_X% and
%Tarsp_VC_X% and
%realcomplormodnodecount% = 3 )"""
See section Macros for Composed Language Measures for details.
Schlichting: “Onderwerp + Werkwoord + Bijwoordelijke Bepaling + lijdend voorwerp of voorzetselvoorwerp of Complement”
Remark Should indirect objects not be included (we did, Schlichting does not mention it, though she states that it hardly occurs on p. 53 in Stage V)?
T076: OndWVC¶
Name: OndWVC.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 48
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_OndWVC%]
Straightforward implementation:
Tarsp_OndWVC = """(%declarative% and
%Ond% and
%Tarsp_W% and
%Tarsp_VC_X% and
%realcomplormodnodecount% = 2 )
"""
See section Macros for Composed Language Measures for details.
Schlichting: “Onderwerp + Werkwoord + lijdend voorwerp of voorzetselvoorwerp of Complement””
T077: OndWVCVC(X)¶
Name: OndWVCVC(X).
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 53
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_OndWVCVCX%]
Straightforward implementation:
Tarsp_OndWVCVCX = """(%declarative% and
%Ond% and
%Tarsp_W% and
%Tarsp_VC_X_count% = 2 and
%realcomplormodnodecount% <= 4 )"""
See section Macros for Composed Language Measures for details.
Schlichting: “Deze constructie bestaat uit vier of vijf Zinsdelen: Onderwerp,+ Werkwoord + twee VC’s (+ een ander Zinsdeel). De VC’s zijn meestal lijdend voorwerp, bepaling van gesteldheid, of naamwoordelijk deel van het gezegde.”
T078: Ov2¶
Name: Ov2.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 48
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: ” “
T079: Ov3¶
Name: Ov3.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 51
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_Ov3%] Tarsp_Ov3 = """(%declarative% and not(%Tarsp_OndWVC%) and not(%Tarsp_OndWB%) and not(%Tarsp_BBX%)and not(%Tarsp_WBVC%) and not(%Tarsp_OndB%) and not(%Tarsp_OndVC%) and %realcomplormodnodecount% = 2) """
The definition is straightforward, and mostly negative:
declarative clauses
with two complements or modifiers
but not any of OndWVC, OndWB, BBX, WBVC, OndB, or OndVC
Schlichting: “behalve de genoemde zeven constructirs in Fase III van de Mededelende zin vinden we nog enkele constructies die bestaan uit drie zinsdelen, die minder voorkomen.”
Remark OndBVC is missing here
T080: Ov7¶
Name: Ov7.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 58
Implementation: Xpath
Query defined as: not implemented yet
Schlichting: ” “
T081: OvWg7¶
Name: OvWg7.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 74
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: ” “
T082: OvVerb6¶
Name: OvVerb6.
Category: Verbindingswoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 78
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: ” “
T083: Ov4¶
Name: Ov4.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 52
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_Ov4%]
The macro Tarsp_Ov4 is defined as follows:
Tarsp_Ov4 = """(%declarative% and
%realcomplormodnodecount% = 3 and
not(%Tarsp_OndWBVC%) and
not(%Tarsp_OndWBB%) and
not(%Tarsp_OndWVCVCX%))"""
The definition is straightforward:
declarative clauses
with three complements or modifiers (and a verb as the fourth ‘zinsdeel’)
but not any of OndWBVC, OndWBB, or OndWVCVCX.
Schlichting: “Overige constructies mededelende zin Fase IV, bestaande uit vier Zinsdelen”
T084: Ov5¶
Name: Ov5.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 53
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_Ov5%]
The macro Tarsp_Ov5 is defined as follows:
Tarsp_Ov5 = """(%declarative% and %realcomplormodnodecount% = 4 and
not(%Tarsp_VCWOndBB%) and
not(%Tarsp_OndWVCVCX%) and
not(%Tarsp_BWOndBB%) )"""
The definition is straightforward:
declarative clauses
with four complements or modifiers (and a verb as the fifth ‘zinsdeel’)
but not any of VCWOndBB, OndWVCVCX, or BWOndBB.
Schlichting: “overige constructies met vijf zinsdelen.”
T085: Overige¶
Name: Overige.
Category: Voornaamwoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 82
Implementation: Xpath with macros
Query defined as:
//node[@lemma="alles" or @lemma="niets" or @lemma="niks" or (@lemma="wat" and @vwtype= "onbep") or @lemma="u" or @lemma="ons" or @lemma="zelf"]
Schlichting: “zelf, niets/niks, alles, ons, wat, u”
Remark We probably should exclude ons as a possessive pronoun here.
T086: SamZn¶
Name: SamZn.
Category: Woordstructuur
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 86
Implementation: Xpath with macros
Query defined as:
getcompounds
For words that do not occur in the Alpino lexicon, Alpino has a rule system to determine whether a string is a compound. So it correctly analyzes:
ponyautootje (as a compound with the parts pony and auto, lemma = pony_auto)
The Alpino lexicon contains words that are compounds and many of these are explicitly marked as a compound. The lemma of such words contains an underscore between the parts of the compound. Examples:
teddybeer ( lemma = teddy_beer)
Alpino leaves any affixes between the compounds out puts the lemmas of each part in the lemma:
aardbeienijs (lemma = aardbei_ijs)
zonnepanelen (lemma = zon_paneel)
grotemensenfeestje (lemma = groot_mens_feest)
but there are various inconsistencies (as acknowledged by Gertjan van Noord), eg. the analysis of kinderbadje:
kinderbadje (lemma = kinder_bad)
In this example not the lemma (glijden) but the actual form has been used:
glijbaan (lemma = glij_baan)
If the first part is a diminutive, various strategies are followed by Alpino inconsistently:
ijsjesverkoop (lemma = ijsje_verkoop)
worstjeswinkel (lemma = worst_DIM_winkel)
grapjespiet (lemma = grap_DIM_piet)
New versions of Alpino may treat these words differently.
However, not all words that are a compound are marked as such in the Alpino lexicon (e.g. tafelkleed). Therefore we also use a list of compounds derived from CELEX.
The language measure T086 has been implemented by the Python function getcompounds from the compounds module, which has been described here.
Schlichting: ” “
T087: Stam¶
Name: Stam.
Category: Woordstructuur
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 86
Implementation: Xpath with macros
Query defined as:
//node[@pt="ww" and @pvtijd="tgw" and not(%Tarsp_kijkVU%) and not(%Tarsp_hww% or @lemma = "hebben" or @lemma = "worden" or @lemma = "zijn" ) and @pvagr="ev" ]
Schlichting: “Stam van het zelfstandig werkwoord (uitgezonderd die in de gebiedende Wijs)”
the stem of the verb is identical to the present tense singular form first person: in DCOI features: @pvtijd=”tgw” and @pvagr=”ev”
we must exclude auxiliary verbs that are scored under Hww i (p. 65), Hww Vd (p. 69) or Hww Z (p. 85)
we must exclude the use of kijk as a V.U. We do so though the macro Tarsp_kijkVU. it basically allows any occurrence of kijk that does not have a PP complement or modifier, nor a vc complement:
Tarsp_kijkVU = """(@pt="ww" and @lemma="kijken" and @wvorm="pv" and @pvagr="ev" and @pvtijd="tgw" and not(../node[%Tarsp_pporvc%]))""" Tarsp_pporvc = """ (((@rel="pc" or @rel="mod" or @rel="ld") and @cat="pp") or @rel="vc")"""
T088: Stam+t¶
Name: Stam+t.
Category: Woordstructuur
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 86
Implementation: Xpath with macros
Query defined as:
//node[@pt="ww" and @pvagr="met-t" and not(%Tarsp_hww% or @lemma = "hebben" or @lemma = "worden" or @lemma = "zijn" ) ]
Schlichting: “Stam van het zelfstandige werkwoord, gevolgd door een toegevoegde, hoorbare ‘t’.”
The definition is straightforward:
in D-Coi the value met-t is only applied in case there really is a ‘t’ in addition to the stem (as in hij loopt, hij doodt, but not in hij verlaat haar). The ‘t’ in doodt is “toegevoegd”, but not “hoorbaar”, SASTA counts it as as Stam + t. The samples support this: VKLtarsp 01 (vindt), VKLTarsp 07 (houdt), TD22 (vindt, 4x) are all marked as Stam + t
Auxiliary verbs that are scored under Hww i (p. 65), Hww Vd (p. 69) or Hww Z (p. 85) should not be marked for Stam + t
T089: Sz2+¶
Name: Sz2+.
Category:
Subcat:
Level: Zc
Original: yes
In form: no
Page:
Implementation: Xpath with macros
Query defined as:
//node[@cat="top" and count(.//node[@cat="smain" or @cat="cp" or @cat="whsub" or @cat="rel" or @cat="whrel" or @cat="whq" or @cat="whsub" or ( @cat="sv1" and not(parent::node[@cat="whq"])) ] ) >= 2]
Unclear whether this measure exists at all.
Schlichting: ” “
T090: Sz3+¶
Name: Sz3+.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Sz
Original: yes
In form: yes
Page: 57
Implementation: Xpath with macros
Query defined as:
//node[@cat="top" and count(.//node[@cat="smain" or @cat="cp" or @cat="whsub" or @cat="rel" or @cat="whrel" or @cat="whq" or @cat="whsub" or ( @cat="sv1" and not(parent::node[@cat="whq"])) ] ) >= 3]
The query counts the number of clauses in the whole tree. If 3 or more, a match is found. A node with cat*=*sv1 can be a clause of a clause body. It is a clause body if dominated by a node with category whq, so sv1 nodes are counted in only if they are not dominated by a node with category whq.
Schlichting: “Samengestelde zin die bestaat uit drie of meer zinnen. Dit kunnen drie hoofdzinnen zijn of een combinatie van hoofd- en bijzinnen. Na drie hoofdzinnen begint er een nieuwe uiting”
Remark We better define a macro for clauses and clause bodies and use that in this query.
T091: V (lijdend)¶
Name: V (lijdend).
Category: Zinsdelen
Subcat:
Level: Zc
Original: yes
In form: no
Page: 42
Implementation: not implemented
Query defined as: not implemented
Not implemented because they are subsumed under VC
Schlichting: “Het lijdend voorwerp van een zin kan meestal gevonden worden door een vraag te stellen met ‘wat’ betreffende het onderwerp en het werkwoord van de zin. […] Het lijdend voorwerp van de bedrijvende vorm wordt het onderwerp van dezelfde zin in de lijdende vorm.”
T092: V (meewerkend)¶
Name: V (meewerkend).
Category: Zinsdelen
Subcat:
Level: Zc
Original: yes
In form: no
Page: 42;43
Implementation: not implemented
Query defined as: not implemented
Not implemented because they are subsumed under VC
Schlichting: “Het meewerkend voorwerp kan bijna altijd voorafgeaan worden door ‘aan’ of ‘voor’. Dit ‘voor’kan altijd vervangen worden door ‘ten behoeve van’. Het meewerkend voorwerp komt niet zo vaak voor. Het wordt gebruikt bij werkwoorden als ‘geven’ en ‘kopen’ en ‘schrijven’ of ‘mededelen’.”
T093: V (Voorzetselvoorwerp)¶
Name: V (Voorzetselvoorwerp).
Category: Zinsdelen
Subcat:
Level: Zc
Original: yes
In form: no
Page: 43
Implementation: not implemented
Query defined as: not implemented
Not implemented because they are subsumed under VC
Schlichting: “Wanneer een werkwoord gebruikt wordt met een vast voorzetsel is het zinsdeel dat begint met dat voorzetsel het voorzetselvoorwerp. Ook een werkwoordelijke uitdrukking kan gebruikt worden met een voorzetselvoorwerp.”
T094: V.U. Divers¶
Name: V.U. Divers.
Category: V.U.
Subcat:
Level: Zc
Original: yes
In form: yes
Page: 37
Implementation: Xpath with macros
Query defined as:
//node[ (@lemma!="ja" and @lemma!="nee" and @word!="xxx" and @lemma != "mama" and @word!="xx" and ((@pt="tsw" and @lemma!="hè") or (@pt="tsw" and @lemma="hè" and @rel="tag" and number(@end)<=number(../node[@rel="nucl"]/@begin)) or ((@lemma="au" or @lemma="hoepla" or @lemma="dag" or @lemma="kijk" or @lemma="hap" or @lemma="aai" ) and (@rel="--" or @rel="sat" or @rel="tag") ) ) ) or %Tarsp_kijkVU% or %Tarsp_hehe% ]
The query distinguishes three subcases:
- Case 1: interjections or otherwise specifically mentioned extragrammatical words
It excludes ‘ja’and ‘nee’ (these fall under T150) and the CHAT code ‘xxx’ and the (wrong) CHAT code ‘xx, as well as the lemma ‘mama’
the word must be a ‘tsw’ but not ‘hè’, or
the word must be a ‘tsw’ with lemma= ‘hè’, provided that this hè occurs as a tag to the left of a nucleus
Case 2: the word ‘kijk’ as a VU. This is achieved by the macro Tarsp_kijkVU, defined in T087: Stam.
Case 3: the combination of words hè, hè, defined by the macro Tarsp_hehe, which basically searches for the bigram hè hè:
Tarsp_hehe = """ (@lemma="hè" and @end = ancestor::node[@cat="top"]/descendant::node[@lemma="hè"]/@begin)"""
Schlichting: “Diverse Sociale Uitdrukkingen. hieronder vallen de uitingen die in het sociale verkeer veel gebruikt worden zoals ‘dag’, ‘hallo’, ‘goedemorgen’, ‘au’, ‘hoepla’ etc. Ook ‘hoor’(‘die is van mij, hoor’), ‘kijk’ (‘kijk, dat is hem’) en ‘hè’ in de betekenis van ‘wat zeg je’of ‘hè, hè’ worden hier gescoord. Verder vinden we in deze categorie de klanknabootsingen als ‘boem’ en ‘tingeling’”
Remarks There is also a macro for kijkeens (Tarsp_kijkeens), but I am not sure it is used anywhere now
T095: V.U. Soc. AangP¶
Name: V.U. Soc. AangP.
Category: V.U.
Subcat:
Level: Zc
Original: yes
In form: yes
Page: 37;38
Implementation: Xpath with macros
Query defined as:
//node[@pt="n" and @word!="hè" and @word!="xxx" and @word!="xx" and @rel="tag" ]
The query selects nouns that bear the grammatical relation tag. It excludes hè (apparently sometimes analysed as a noun) and (wrong) CHAT codes.
Schlichting: “Aangesproken persoon, bijvoorbeeld ‘mama’, ‘Peter’”
T096: V.U. Ster¶
Name: V.U. Ster.
Category: V.U.
Subcat:
Level: Zc
Original: yes
In form: yes
Page: 38;39
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “Hieronder verstaan we uitingen die in hun geheel worden geleerd, zoals:
Spreekwoorden en uitdrukkingen, bijvoorbeeld ‘bezint eer ge begint’ en regels uit versjes of liedjes, bijv. ‘alle eendjes zwemmen in het water’
Ook gewone spreektaalzinnen die het kind in zijn geheel leert scoren we hier en verder niet meer. Nemen we bijvoorbeeld de uiting ‘weet ik niet’. Wanneer we deze uiting vinden in een taalsample waarin verder geen ‘ik’ wordt gebruikt, of geen inversie van onderwerp en persoonsvorm voorkomt, dan is de zin waarschijnlijk als Stereotiepe Uitdrukking geleerd.
Opsommingen, bijv. ‘een, twee, drie, vier, vijf.’
Zelfherhalingen. Men scoort hier wanneer een kind een uiting van zichzelf onmiddellijk herhaalt met precies dezelfde woorden.
T097: VC¶
Name: VC.
Category:
Subcat:
Level: Zc
Original: yes
In form: no
Page: 42-44
Implementation: Xpath with macros
Query defined as:
//node[%complement% and parent::node[(@cat="smain" or @cat="sv1" or @cat="ssub" or @cat="inf" or @cat="ppart") ]]
Here the macro complement is defined as follows:
complement = """((@rel="obj1" or @rel="obj2") or
(@rel="predc" and not(%predcB%)) or
(@rel="pobj1" and not(%pobj1B%) ) or
%verbalcomplement%
)"""
So, it covers:
direct objects (obj1)
indirect objects (obj2)
predicative complements (predc), at least if they are not predicates that are considered adverbial modifiers in Tarsp (predcB)
prepositional complements (pc) if they are not considere adverbial complements in Tarsp(pobj1B)
and verbals complements (verbal complement).
The latter is defined as follows:
verbalcomplement = """(@rel="vc" and (@cat="cp" or @cat="whq" or @cat="whsub"))"""
So it excludes nonfinite verbal complements such as inf, teinf, ppart, etc.
Predicates that cooccur with a transitive verb are considered modifiers in Tarsp. We exclude them by checking for the presence of a direct object:
predcB = """(@rel="predc" and
(@pt="adj" or @pt="bw" or @cat="ap" or @cat="advp") and
../node[@rel="obj1"]
)"""
Remark that fails if the direct object is left out (e.g. by topic drop).
Prepositional complements should be considered modifiers if they cooccur with a verb that can also take locative or directional complements. This has been implemented by pob1B:
pobj1B = """(@rel="pc" and ../node[@rel="hd" and %locverb%])"""
locverb = """(@lemma="staan" or @lemma="zitten" or @lemma="rijden" or
@lemma="vallen" or @lemma="doen" or @lemma="gaan" or
@lemma="komen" or @lemma="zijn" or %locmodalverb% )"""
locmodalverb = """ (@lemma="kunnen" or @lemma="moeten" or
@lemma="hoeven" or @lemma="willen" or
@lemma="mogen")"""
Remark This language measure is not in the form and has not been kept up to date. In particular, it should be investigated to what extent it differs from the macro Tarsp_VC used in the composed language measures. See Macros for Composed Language Measures.
Schlichting discusses several cases of “Voorwerpen” and “Complementen” separately:
“Lijdend Voorwerp. Het lijdend voorwerp kan meestal gevonden worden door een vraag te stellen met ‘wat’ betreffende het onderwerp en het werkwoord van de zin. […] Het lijdend voorwerp van bedrijvende vorm wordt het onderwerp van dezelfde zin in de lijdende vorm.” Covered
“Meewerkend Voorwerp. Het meewerkend voorwerp kan bijna altijd voorafgegaan worden door ‘aan’ of ‘voor’. Dit ‘voor’ kan altijd vervangen worden door ‘ten behoeve van’. Het meewerkend voorwerp komt niet zo veel voor. Het wordt gebruikt bij werkwoorden als ‘geven’ en ‘kopen’, en ‘schrijven’ of ‘mededelen’. Covered
“Voorzetselvoorwerp. Wanneer een werkwoord gebruikt wordt met een vast voorzetsel is het zinsdeel dat begin met dat voorzetsel het voorzetselvoorwerp. Ook een werkwoordelijke uitdrukking kan gebruikt worden met een voorzetselvoorwerp (een hekel hebben aan).” Covered.
- “Complement. Een aanvulling van het gezegde. het complement kan zijn:
naamwoordelijk deel van het gezegde. Het naamwoordelijk gezegde bestaat uit een koppelwerkwoord en (meestal) een zelfstandig naamwoord of bijvoeglijk naamwooord.” (maar ook een voorbeeld met een PP wordt gegeven). Covered
Bepaling van gesteldheid. Deze zegt iets van het werkwoord en van het onderwerp of lijdend voorwerp. de bepaling van gesteldheid wordt ook wel ‘dubbelverbonden bepaling’ genoemd.” This covers primary predicates (which Alpino consideres predc) but also secondary predicates (in Alpino: predm, e.g in ik kan dat alleen). Mostly covered.
Remark we do not treat secondary predicates such as ik kan dat alleen as complements, but as adverbial modifiers.
T098: Vcbijzin¶
Name: Vcbijzin.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Sz
Original: yes
In form: yes
Page: 56;57
Implementation: Xpath with macros
Query defined as:
//node[((@rel="vc" or @rel="su") and (@cat="cp" or @cat="whsub")) or %directerede_vcbijzin%]
Alpino does not have clauses with grammatical relation obj1 or obj2: instead clause have the grammatical relation vc.
A clause acting as an adpositional complement also has the grammatical relation vc, so is also covered.
The macro directerede_vcbijzin is used as weel, but I do not understand or remember why. (was it intended for ASTA?). It is defined as follows:
directerede_vcbijzin = """( %clausecat% and not(@rel="cnj") and
(preceding-sibling::node[%metasmain%] or
following-sibling::node[%metasv1%]))"""
and it uses several other macros:
clausecat = """(@cat="smain" or @cat="whq" or %baresv1% )"""
metaverb = """(@lemma="zeggen" or @lemma="denken" or
@lemma="vinden" or @lemma="vragen" or
@lemma="schreeuwen" or @lemma="fluisteren" )"""
metasmain = """(@cat="smain" and not(@rel="cnj") and node[@rel="hd" and %metaverb%]) """
metasv1 = """(@cat="sv1" and not(@rel="cnj") and node[@rel="hd" and %metaverb%]) """
baresv1 = """( @cat="sv1" and not(parent::node[(@cat="whq" or @cat="whrel")]))"""
Schlichting: “Ondergeschikte zin die in de hoofdzin de functie heeft van een lijdend voorwerp, een voorzetselvoorwerp of meewerkend voorwerp of een complementsbijzin: deze laatste heeft de functie van een naamwoordelijk deel van het gezegde of van een bepaling van gesteldheid.”
Remark A clause with grammatical relation predc does occur, but is currently not covered.
Remark Nodes with category cp are not necessarily clauses. That is only so if they also contain a ssub node. This should be adapted. It is especially important for cp with grammatical relation predc.
T099: VCW¶
Name: VCW.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 46
Implementation: Xpath with macros
Query defined as: not implemented yet
For infinitival constructons, this can be dealt with by using the smallclauses module (See Small clauses).
For finite constructions, we can cover is + Noun or NP, but what else can occur has to be investigated. There are many examples in the Auris data.
Schlichting: “Voorwerp of Complement + Werkwoord”
T100: VCWOndBB¶
Name: VCWOndBB.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 52
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_VCWOndBB%]
Straightforward implementation:
Tarsp_VCWOndBB = """(%declarative% and
%Ond% and
%Tarsp_W% and
%Tarsp_B_X_count% = 2 and
%Tarsp_VC_X% and
%realcomplormodnodecount% = 4)"""
Schlichting: “Onderwerp + Werkwoord + VC + twee Bijwoordelijke bepalingen”
T101: VCWoordgroep¶
Name: VCWoordgroep.
Category: Uitbreiding Zinsdelen
Subcat:
Level: Zc
Original: yes
In form: no
Page: 75
Implementation: Xpath
Query defined as:
//node[@cat and ((@rel="obj1" or @rel="obj2" or @rel="pc" or @rel="predc" or @rel="ld" or (@rel="vc" and (@cat="cp" or @cat="whq" or @cat="whsub")) ) and parent::node[(@cat="smain" or @cat="sv1" or @cat="ssub" or @cat="inf" or @cat="ppart") ])]
The query selects nodes for phrases (@cat) and with a grammatical relatoin for objects and complements that occur in a clause or clause body.
Schlichting: “<no text>”
Remark It is not obvious to me that the condition on being contained in a clause is really needed. A simple test on the VKLtarsp data suggests that it is not.
T102: Verb Ov¶
Name: Verb Ov.
Category:
Subcat:
Level: Zc
Original: yes
In form: yes
Page: 78
Implementation: Xpath with macros
Query defined as:
//node[((@pt="vg" and (@lemma="dat" or @lemma="of" or @lemma="toen") ) or (@pt="vnw" and @vwtype="vb") ) and parent::node[(@cat="whsub" or @cat="ssub" or @cat="rel" or @cat="cp" or @cat="whrel") and @rel!="nucl"] ]
The query selects
the conjunctions dat, of, toen, and
wh-pronouns
provided that they are dominated by a clause that does not bear the grammatical relation ‘nucl’
Schlichting: “Overige onderschikkende verbindingswoorden: ‘waar’, ‘dat’, ‘toen’, ‘of’, ‘wat’, ‘hoe’.”
T103: Vergr trap¶
Name: Vergr trap.
Category: Woordstructuur
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 88
Implementation: Xpath with macros
Query defined as: not implemented yet, but it could be //node[@graad=”comp”]
In the VKLTarsp dataset this only occurs with meer, mostly in a combination with a negative word (niet meer, nergens meer, geen … meer), but also in nog meer kindjes. Should these be included? No annotations for this measure occur in the Auris TD and DTD data. The language measure is absent in [Schlichting 2017].
Schlichting “Vergrotende trap van het bijvoeglijk naamwoord.”
T104: Verkl¶
Name: Verkl.
Category: Woordstructuur
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 83
Implementation: Xpath with macros
Query defined as:
//node[@pt="n" and @graad="dim"]
The query is self-explanatory.
Schlichting: “Verkleinwoord van zelfstandige naamwoorden. Sommige woorden hebben als meeste gebruikte of enig gebruikte vorm het verkleinwoord, bijvoorbeeld ‘meisje’, ‘kopje’, ‘koekje’. Deze worden niet gescoord bij verkleinwoord.”
Remark: meisje is correctly not counted (because Alpino does not analyse it as a diminutive), but koekje en kopje are incorrectly counted.
T105: Verl Tijd¶
Name: Verl Tijd.
Category: Woordstructuur
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 87
Implementation: Xpath with macros
Query defined as:
//node[ @pvtijd="verl" and @pt="ww" and @wvorm="pv"]
Schlichting: “De Verleden Tijd van Koppelwerkwoord, Hulpwerkwoord en soms het onregelmatige zelfstandige werkwoord. Zowel enkelvoud als meervoud worden hier gescoord. de verleden tijd van de regelmatige werkwoorden wordt in Fase VI nog niet gebruikt.”
Remark We have no exceptions for regular verbs. We interpret Schlichting’s text as follows: Past tenses of regular verbs should be counted, but they occur only rarely.
T106: Vo/bij¶
Name: Vo/bij.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 71
Implementation: Python function
Query defined as:
voslashbij
Schlichting: “Voornaamwoordelijk bijwoord, gesplitst. Het gesplitste voornaamwoordelijk bijwoord behoeft niet gescoord te worden bij Vobij, kolom Voornaamwoorden in Fase IV, alleen hier bij de Woordgroepen in Fase V.”
Remark Schlichting only gives examples with nonadjacent Rpronouns and adpositions. We agreed with Rob Zwitserlood that adjacent R-pronoun + adposition, even if written separately, is to be counted under Vobij. However, the notion “adjacent” is not so easy to define in inflated tree structures. This has been done in a python function adjacent in the module treebankfunctions, and for this reason this query must also be defined as a python function.
T107: Vobij¶
Name: Vobij.
Category: Voornaamwoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 80
Implementation: Python function
Query defined as:
vobij
Schlichting: “Voornaamwoordelijk bijwoord. Het voornaamwoordelijk bijwoord is een combinatie van ‘er’, ‘daar’, ‘hier’, ‘waar’ met een voorzetsel (bijv. ‘aan’, ‘bij’, ‘voor’) of een bijwoord (‘af’, ‘heen’, ‘toe’)”
Remark Separately wiritten but adjacen R-pronoun + adposition cases are also considered Vobij
Remark Alpino considers words such as ‘af’, ‘heen’, and ‘toe’ as postpositions, not as adverbs.
T108: Volt dw¶
Name: Volt dw.
Category: Woordstructuur
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 85;86
Implementation: Xpath with macros
Query defined as:
//node[@pt="ww" and @wvorm="vd"]
The query is self-explanatory
Schlichting: “Voltooid deelwoord. Bij het vroege gebruik van het Voltooid deelwoord wordt ‘ge’ nog wel eens weggelaten. Bij woorden als ‘vallen’ of ‘maakt’ zal men alleen uit de situatie kunnen opmaken of het een voltooid deelwoord betreft of een andere vorm. Ook zonder het voorvoegsel ‘ge’ mogen de voltooide deelwoorden gescoord worden.”
Remark Wrongly formed forms without ge can only be dealt with if they are dealt with by the deregularise module.
T109: Voltd fg¶
Name: Voltd fg.
Category: Woordstructuur
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 88
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “Voltooid deelwoord, fout gevormd. In deze Fase (VII) beginnen de kinderen de regel voor het maken van een voltooid deelwoord door te krijgen. Dit blijkt uit het gebruik van fout gevormde voltooide deelwoorden. Natuurlijk komt deze constructie nooit als leerdoel in een behandelingsplan.”
Remark Can be dealt with via the module deregularise.
T110: Vr¶
Name: Vr.
Category:
Subcat:
Level: Zc
Original: yes
In form: no
Page:
Implementation: Xpath
Query defined as:
//node[ @cat="whq"]
This language measure actually does not exist. It has been included and implemented because the code Vr occurs in the Schlichting appendix (example 20, p. 91, and example 22, p. 92). Example 20 should have been annotated as Vr(XY), and example 22 as Vr4.
T111: Vr(XY)¶
Name: Vr(XY).
Category: Zinsconstructies
Subcat: Vragen
Level: Zc
Original: yes
In form: yes
Page: 60
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_VrXY%]
Straightforward implementation:
Tarsp_VrXY = """(%Tarsp_whq% and
node[%Tarsp_whqhead%] and
node[%whqbody% and %realcomplormodnodecount% <= 1])"""
where Tarsp_whqhead and whqbody cover both wh-questions (main and subordinate) and independent relatives:
Tarsp_whqhead = """(@rel="whd" or @rel="rhd") """
whqbody = """((@cat="sv1" or @cat="ssub") and @rel="body")"""
See section Macros for Composed Language Measures for details.
Schlichting
“Vraagwoord (+ een of twee Zinsdelen).”Covered
Het Vraagwoord zonder meer, of gevolgd door één Zinsdeel komt niet vaak voor. We vinden meestal Vr + X + Y.” Not currently covered
“‘Wat is dat?’ is bij sommige kinderen als Vaste Uitdrukking geleerd. We scoren deze dan bij Stereotiepe Uitdrukkingen”. Not covered
T112: Vr4¶
Name: Vr4.
Category: Zinsconstructies
Subcat: Vragen
Level: Zc
Original: yes
In form: yes
Page: 61
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_Vr4%]
Schlichting “Vraagwoord + drie Zinsdelen.” Covered.
Straightforward implementation:
Tarsp_Vr4 = """(%Tarsp_whq% and
node[%Tarsp_whqhead%] and
node[%whqbody% and %realcomplormodnodecount% = 2])"""
For Tarsp_whqhead and whqbody, see T111: Vr(XY)
See section Macros for Composed Language Measures for details.
T113: Vr5+¶
Name: Vr5+.
Category: Zinsconstructies
Subcat: Vragen
Level: Zc
Original: yes
In form: yes
Page: 61;62
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_Vr5plus%]
Straightforward implementation:
Tarsp_Vr5plus = """(%Tarsp_whq% and
node[%Tarsp_whqhead%] and
node[%whqbody% and %realcomplormodnodecount% > 2])"""
For Tarsp_whqhead and whqbody, see T111: Vr(XY)
See section Macros for Composed Language Measures for details.
Schlichting “Vraagzinnen die beginnen met een vraagwoord en behalve dat Vraagwoord nog vier of meer Zinsdelen hebben.” Covered.
Remark There is also an older implementation as a Python function in the module SZiplus
T114: VzB¶
Name: VzB.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 72
Implementation: Xpath with macros
Query defined as:
//node[node[@rel="hd" and @pt="vz"] and node[@rel="obj1" and ( (@pt="vz" or @pt="bw") or (%Rpronoun% and @begin=../node[@rel="hd"]/@end) ) ] ]
The query selects a node that contains
an adposition as head, and
- a direct object, which can be
an adposition, or
an adverb, or
an Rpronoun, provided that it follows the adposition (so van hier is ok, hier van not)
Schlichting: “Voorzetsel + Bijwoord”
T115: VzBepBvZn¶
Name: VzBepBvZn.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 74
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “Voorzetsel + Bepaler + Bijvoeglijk woord + Zelfstandig naamwoord.”
T116: VzBepZn¶
Name: VzBepZn.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 66;67
Implementation: Xpath with macros
Query defined as:
//node[@cat="pp" and node[ @rel="obj1" and node[@rel="det"] ]]
The query searches for a PP that contains a direct object that contains a determiner.
Schlichting: “Voorzetsel + Bepaler + Zelfstandig naamwoord”
Remark Maybe we must restrict the content of the object to determiner + Noun and nothing else.
T117: VzN¶
Name: VzN.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 65
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_VzN%]
The macro Tarsp_VzN is defined as follows:
Tarsp_VzN = """(%vzn1xpath% or %vzn2xpath% ) """
vzn1xpath = """(@cat="pp" and
(node[@pt="vz"] and
node[(@pt="n" or @pt="vnw") and
not (%Rpronoun%) and @rel="obj1"] and
not(node[@pt="vz" and @vztype="fin"])
))"""
vzn2xpath = """(node[@lemma="in" and @rel="mwp"] and
node[@lemma="deze" and @rel="mwp"])"""
The macro distinguishes two subcases:
a PP with a preposition but no postposition and a direct object that is a noun or a pronoun but no R-pronoun.
the combination in deze, which has an idiosyncratic parse in Alpiono
Schlichting: “Voorzetsel + Zelfstandig naamwoord of Voornaamwoord. In Fase III vinden we hoofdzakelijk Vz + Zn”
T118: VzZnAz¶
Name: VzZnAz.
Category: Woordgroepen
Subcat: Ov
Level: Wg
Original: yes
In form: no
Page: 70
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “Voorzetsel + Zelfstandig naamwoord + achterzetsel”
T119: S1W¶
Name: S1W.
Category: Eenwoordzin
Subcat:
Level: Zc
Original: yes
In form: yes
Page: 40;41
Implementation: Xpath with macros
Query defined as:
//node[@cat="top" and node[@pt='ww'] and count(node)=1]
Schlichting: “Werkwoord. De kinderen gebruiken inde éénwoordfase een aantal werkwoorden in de onbepaalde wijs.”
Remark Schlichting does not distinguish a code for a verb as a single word utterance, and a verb or predicate in a full sentence. We introduced a different code for a verb as a single word utterance.
T120: W¶
Name: W.
Category: Zinsdelen
Subcat:
Level: Zc
Original: yes
In form: yes
Page: 41;42
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_coreW%]
The macro Tarsp_coreW is defined as follows:
Tarsp_coreW = """ ( @pt="ww" and
(@wvorm="pv" or parent::node[@rel!="vc"] or %Tarsp_BarenonfinW%) and
not(%Tarsp_kijkVU%) and
not((@lemma="zijn" or @lemma="worden") and parent::node[node[@rel="vc"]]) )"""
The query selects:
nodes with pt equal to ww* (verbs)
- that are
finite (@wvorm equal to pv), or
whose parent node does not bear the grammatical relation vc (because then a higher verb should count as W), or
or whose parent node does bear the grammtical relation vc but that parent is the only node under an smain node. This is for utterance that contain an infiitive or past particple only (e.g. boekje lezen). See the module Small clauses. This is achieved by the macro Tarsp_BarenonfinW:
Tarsp_BarenonfinW = """parent::node[@rel="vc" and parent::node[@cat="smain" and count(node)=1]]"""
that is not a VU kijk, by the macro tarsp_kijkVU, defined under @@, and
that is not a form of ‘zijn’ or ‘worden’ with a vc-sibling. I think that this is to exclude zijn as a W in e.g. hij kan gekomen zijn. In hij is gekomen, is does count because it is finite.
Schlichting: “Werkwoord”
Remark I do not really understand the last condition concerning zijn and * worden*, should hebben not be added here as well?
T121: W(X)¶
Name: W(X).
Category: Zinsconstructies
Subcat: Gebiedende Wijs
Level: Zc
Original: yes
In form: yes
Page: 62
Implementation: Xpath with macros
Query defined as:
wx
It has been implemented by means of a Python function wx.
In a later stage a query with macros was defined for it, but the Python function has not been replaced yet.
The query has a straightforward implementation:
wx = """(%basicimperative% and
%realcomplormodnodecount% <= 1)"""
See section Macros for Composed Language Measures for details.
Remark Experiment with replacing the wx function by the query with macros
Schlichting “De Gebiedende Wijszinnen bestaan in deze Fase uit één of twee zinsdelen.” Covered
T122: PV-loos¶
Name: PV-loos.
Category:
Subcat:
Level: Zc
Original: no
In form: no
Page:
Implementation: Xpath with macros
Query defined as:
//node[@cat="top" and not(.//node[@pt="ww" and @pvagr])]
This is not a language measure but has been added for development purposes.
T123: waarschijnlijk fout geanalyseerde nevenschikking¶
Name: waarschijnlijk fout geanalyseerde nevenschikking.
Category:
Subcat:
Level: Zc
Original: no
In form: no
Page:
Implementation: Xpath with macros
Query defined as: not implemented yet
This is not a language measure but has been added for development purposes.
T124: want¶
Name: want.
Category: Verbindingswoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 77;78
Implementation: Xpath with macros
Query defined as:
//node[@lemma="want" and @pt="vg"]
The query is self-explanatory.
Schlichting: “‘want’ verbindt twee hoofdzinnen. Soms wordt het introducerend gebruikt.”
T125: WBVC¶
Name: WBVC.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 51
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_WBVC%]
Straightforward implementation:
Tarsp_WBVC = """(%declarative% and
%Tarsp_W% and
%Tarsp_B_X% and
%Tarsp_VC_X% and
%realcomplormodnodecount% = 2 )"""
See section Macros for Composed Language Measures for details.
Schlichting “Werkwoord + Bijwoordelijke bepaling + Voorwerp of Complement.” Covered
T126: Wdeel¶
Name: Wdeel.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 68
Implementation: Xpath with macros
Query defined as:
//node[@pt="ww" and contains(@lemma,"_") and @rel ="hd" and parent::node[node[@rel="svp"]]]
The query searches for a head verb that contains an underscore in its lemma and that has an svp node as a sibling.
Schlichting: “Werkwoord + Scheidbaar deel”
T127: Werkwoordswoordgroep¶
Name: Werkwoordswoordgroep.
Category: Uitbreiding Zinsdelen
Subcat:
Level: Zc
Original: yes
In form: no
Page: 75
Implementation: Xpath with macros
Query defined as:
//node[@pt="ww" and @rel="hd" and parent::node[node[ @rel="vc" and (@cat="inf" or @cat="ppart")]]]
Schlichting: “<no text>”
Remark The given query finds the first verb of the verbal group. Note that Alpino (correctly) does not consider verbal groups to be phrases. if we want to identify all elements of the verbal group, a different query will have to be formulated, one that yields multiple nodes as single result.
T128: wij¶
Name: wij.
Category: Voornaamwoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 81
Implementation: Xpath with macros
Query defined as:
//node[@lemma="wij" or @lemma="we" ]
The query is self-explanatory.
Schlichting: “Persoonlijk voornaamwoord ‘wij’, ‘we’.”
T129: Wond(X)¶
Name: Wond(X).
Category: Zinsconstructies
Subcat: Vragen
Level: Zc
Original: yes
In form: yes
Page: 60
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_WOndX%]
Straightforward implementation:
Tarsp_WOndX = """(%ynquery% and
not(%basicimperative%) and
%topcontainsquestionmark% and
%Tarsp_W% and
%Ond% and
%realcomplormodnodecount% <= 2)"""
See section Macros for Composed Language Measures for details.
Schlichting “Werkwoord + Onderwerp, meestal gevolgd door een derde zinsdeel.” Covered.
T130: WOnd4¶
Name: WOnd4.
Category: Zinsconstructies
Subcat: Vragen
Level: Zc
Original: yes
In form: yes
Page: 61
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_WOnd4%]
Straightforward implementation:
Tarsp_WOnd4 = """(%ynquery% and
not(%basicimperative%) and
%topcontainsquestionmark% and
%Tarsp_W% and
%Ond% and
%realcomplormodnodecount% = 3)"""
See section Macros for Composed Language Measures for details.
Schlichting “Werkwoord + Onderwerp + twee Zinsdelen.” Covered.
T131: WOnd5+¶
Name: WOnd5+.
Category: Zinsconstructies
Subcat: Vragen
Level: Zc
Original: yes
In form: yes
Page: 61
Implementation: Xpath with macros
Query defined as:
//node[%Tarsp_WOnd5plus%]
Straightforward implementation:
Tarsp_WOnd5plus = """(%ynquery% and
not(%basicimperative%) and
%topcontainsquestionmark% and
%Tarsp_W% and
%Ond% and
%realcomplormodnodecount% =4)"""
See section Macros for Composed Language Measures for details.
Schlichting “Vraagzinnen die beginnen met een Werkwoord en een Onderwerp en daarbij nog drie of meer Zinsdelen hebben.” Covered.
Remark Maybe = 4 should be >= 4
T132: woordgroep(onderstrepen)¶
Name: woordgroep(onderstrepen).
Category: Zinsdelen
Subcat:
Level: Zc
Original: yes
In form: no
Page: 63-64
Implementation: Xpath with macros
Query defined as:
//node[(@cat="ap" or @cat="advp" or @cat="np" or @cat="pp" or node[@rel="vc" and (@cat="inf" or @cat="ppart")]) and parent::node[count(node[@cat or @pt!="let"])>1]]
Schlichting: “Elk zinsdeel kan, zoals gezegd, ook uit meerdere woorden bestaan. […] Bij de analyse worden de woordgroepen onderstreept, het nummer van de uiting wordt omcirkeld.”
Remark We can identify the word groups, but they are not undelined in the annotation form, and the utterance id is currently not surrounded by a circle.
T133: WVz¶
Name: WVz.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 72;73
Implementation: Xpath with macros
Query defined as:
//node[node[@pt="ww" and @rel ="hd"] and node[@rel="pc" and not(%Tarsp_B%)]]
The query search for a verb that has a pc as sibling if that pc node is not an adverbial modifier.
Schlichting: “Werkwoord + Voorzetselvoorwerp waarbij het vast voorzetsel gerealiseerd is”
T134: WW¶
Name: WW.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 73
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “Werkwoord + Werkwoord. Deze constructie omvat een hulpwerkwoord + twee infinitieven. Ook werkwoordelijke uitdrukkingen zoals ‘nodig hebben’, ‘het koud hebben’worden hier gescoord”
T135: WXY¶
Name: WXY.
Category: Zinsconstructies
Subcat: Gebiedende Wijs
Level: Zc
Original: yes
In form: yes
Page: 63
Implementation: Xpath with macros
Query defined as:
//node[%wxy%]
Straightforward implementation:
wxy = """(%basicimperative% and
%realcomplormodnodecount% = 2)"""
See section Macros for Composed Language Measures for details.
Schlichting “Werkwoord + twee Zinsdelen”. Covered
T136: WXYZ¶
Name: WXYZ.
Category: Zinsconstructies
Subcat: Gebiedende Wijs
Level: Zc
Original: yes
In form: yes
Page: 63
Implementation: Xpath with macros
Query defined as:
//node[%wxyz%]
Straightforward implementation:
wxyz = """(%basicimperative% and
%realcomplormodnodecount% = 3)"""
See section Macros for Composed Language Measures for details.
Schlichting “Werkwoord + drie Zinsdelen”. Covered
T137: WXYZ5*¶
Name: WXYZ5*.
Category: Zinsconstructies
Subcat: Gebiedende Wijs
Level: Zc
Original: yes
In form: yes
Page: 63
Implementation: Xpath with macros
Query defined as:
//node[%wxyz5%]
Straightforward implementation:
wxyz5 = """(%basicimperative% and
%realcomplormodnodecount% = 4)"""
See section Macros for Composed Language Measures for details.
Remark The label should be chaned to WXY5, the varinat with the asterisk should be an alternative.
Schlichting “Werkwoord + vier Zinsdelen”. Covered
T138: X en X (en X)¶
Name: X en X (en X).
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 72
Implementation: Python function
Query defined as:
xenx
Schlichting: “Dit zijn twee woorden van dezelfde woordsoort, verbonden door ‘en’, die samen één Zinsdeel vormen, soms worden drie elementen verbonden. Ook woordgroepen kunnen op een dergelijke wijze verbonden worden. Behalve door ‘en’ kunnen de woorden of woordgroepen ook verbonden worden door ‘of’.”
T139: X(W)deel¶
Name: X(W)deel.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 48;49
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “De X kan hier elk willekeurig Zinsdeel zijn behalve de W. De X is dus een Ond, een VC, of een B. (W)deel is het scheidbare deel van een samengesteld werkwoord. In ‘Ik doe de jas uit’ is ‘uit’ het scheidbare deel van het samengestelde werkwoord ‘uitdoen’. In de (W)deelconstructies zien we wel het scheidbare deel maar niet het werkwoord. daarom staat de W tussen haakjes. “
Remark Dubious analysis. Most of these, if not all, can also be analysed as OndB or as BX
T140: Xneg¶
Name: Xneg.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: yes
Page: 49;50
Implementation: Xpath with macros
Query defined as:
//node[(@rel='--' or @rel="nucl") and count(node) = 2 and node[ @lemma!="niet"] and (node[@lemma="niet" ] or node[@cat="advp" and node[@lemma="niet"]] ) ]
The query searches for a node:
with grammatical relation – or nucl
containing exactly two nodes
one of which is not equal to the word niet
and the other one is node for the word niet or and adverbial phrase containing the word niet.
Schlichting: “Deze constructie bestaat uit een Zinsdeel dat verder niet benoemd wordt + een negatief, ontkennend woord, meestal ‘niet’. Het eerst Zinsdeel, dat verder niet benoemd wordt, noemen we ‘X’. het kan een Onderwerp zijn, een Werkwoord, een VC of een Bijwoordelijke bepaling. […] Wanneer ‘niet’ voorkomt in constructiees van meer dan twee Zinsdelen wordt ‘niet’weer gewoon als B geanalyseerd.”
T141: XY(W)deel¶
Name: XY(W)deel.
Category: Zinsconstructies
Subcat: Mededelende Zin
Level: Zc
Original: yes
In form: no
Page: 51
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “Deze constructie bestaat uit twee Zinsdelen + het ‘deeltje’ van een scheidbaar samengesteld werkwoord (e.g. Marije strijkbout vast”
T142: ze¶
Name: ze.
Category: Voornaamwoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 80;81
Implementation: Xpath with macros
Query defined as:
//node[ @getal="mv" and (@persoon="3p" or @persoon="3") and @vwtype="pers" and @pdtype="pron" and @pt="vnw"]
The query is self-explanatory. With plural ze, the attribute persoon can only have the values 3 or 3p*.
Schlichting: “Persoonlijk voornaamwoord ‘ze’, meervoud mannelijk en vrouwelijk. ‘ze’ komt veel vaker voor dan ‘zij’. In het begin wordt ‘ze’ vooral in de functie van onderwerp gebruikt.”
T143: zelf¶
Name: zelf.
Category: Voornaamwoorden
Subcat: Ov
Level: VVW
Original: yes
In form: no
Page:
Implementation: Xpath with macros
Query defined as: not implemented yet
zelf is mentioned on p.82 under Overige, together with niets and niks. It should be covered there.
Schlichting: ” “
T144: zij¶
Name: zij.
Category: Voornaamwoorden
Subcat:
Level: VVW
Original: yes
In form: yes
Page: 82
Implementation: Xpath with macros
Query defined as:
//node[@pt="vnw" and @vwtype="pers" and @getal="ev" and @genus="fem" and @pdtype="pron"]
The query allows zij and ze, but also zijzelf (though I believe that is a misspelling for zij zelf)
Schlichting: “Persoonlijk voornaamwoord derde persoon enkelvoud, vrouwelijk”
T145: Zn¶
Name: Zn.
Category: Eenwoordzin
Subcat:
Level: Zc
Original: yes
In form: yes
Page: 40
Implementation: Xpath with macros
Query defined as:
//node[@cat="top" and (node[@pt="n" or (@pt="tw" and @numtype="hoofd") or (@lemma="paar" and @pt="lid") ] or node[@cat="du" and node[@pt="n" or (@pt="tw" and @numtype="hoofd") or (@lemma="paar" and @pt="lid")] ] ) and count(.//node[%realnode%])=1 ]
The query search for a top node that contains only one real node and has as a child:
a nominal node
a du node that contains a nominal node
where a nominal node is a node:
that is a noun, or
that is a substantively used numeral, or
that has as lemma paar with pt lid.
Schlichting: “Hiertoe worden gerekend de zelfstandige naamwoorden, de eigennnamen, en de getalsnamen vwanneer ze zelfstandig gebruikt worden, dus bijvoorbeeld: ‘boek’, ‘oma’, ‘Jantien’, ‘twee’”
Remark I do not know why paar with pt lid was added, there are no such cases. paar is never analysed as an article in the TARSP samples, Lassy Klein or CGN.
T147: ZnZn¶
Name: ZnZn.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: yes
Page: 69
Implementation: Xpath with macros
Query defined as:
//node[@cat="top" and .//node[node[@pt="n" and @rel="hd"] and node[@pt="n" and not(@rel="hd")] ] or node[@cat="du" and count(node[@rel="dp" and @pt="n"])>=2] ]
The query searches for a top node:
that contains a node (anywhere in the structure) which contains a head noun and a non-head noun, or
that contains a du node as a child which as two n children.
Schlichting: “Zelfstandig naamwoord + Zelfstandig naamwoord. Deze combinatie van twee Zelfstandige naamwoorden kan een bezitsrelatie uitdrukken, bijvoorbeeld ‘oma’s tas’ of de ‘tas van oma’, maar kan ook een andere combinatie van twee Zn’s tot een woordgroep vormen, bijvoorbeeld ‘stukje koek’. “
Remark The query searches for a top node, but it should not do that but search for the first noun of the two nouns. It covers stukje kaas or opa baardje but not vaders huis, however, Alpino analyses opa as a pronoun in the construction opa’s huis (idem for oma, papa and probably more words). It does not correctly analyse Jans huis, Piets huis or tantes huis.
T148: Znx¶
Name: Znx.
Category: Woordgroepen
Subcat:
Level: Wg
Original: yes
In form: no
Page: 40
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “Hiertoe worden gerekend de zelfstandige naamwoorden, de eigennnamen, en de getalsnamen wanneer ze zelfstandig gebruikt worden, dus bijvoorbeeld: ‘boek’, ‘oma’, ‘Jantien’, ‘twee’”
T150: V.U. Nee/ja¶
Name: V.U. Nee/ja.
Category: V.U.
Subcat:
Level: Zc
Original: yes
In form: yes
Page: 38
Implementation: Xpath
Query defined as:
//node[@lemma="ja" or @lemma="nee"]
The implementation is self-explanatory.
Schlichting: “Nee/ja”
T151: V.U. Totaal¶
Name: V.U. Totaal.
Category: Aggregate
Subcat:
Level:
Original: yes
In form: yes
Page: 39
Implementation: Python function
Query defined as:
vutotaal
Schlichting: “Hier noteren we het totaal aantal Sociale en Stereotype Uitdrukkingen”
T152: G Totaal¶
Name: G Totaal.
Category: Aggregate
Subcat:
Level:
Original: yes
In form: yes
Page: 20
Implementation: Python function
Query defined as:
gtotaal
Schlichting: “We gaan uit van het Totaal aantal gescoorde uitingen. Hiervan trekken we af:
het aantal Afvallers: A totaal
het aantal vaste Uitdrukking: V.U. totaal
Dit levert het aantal Analyse-eenheden op, dat is het aantal uitingen dat grammaticaal geanalyseerd is, oftewel G totaal.”
T153: G.O Fase¶
Name: G.O Fase. (Grammaticale Ontwikkelingsfase)
Category: Aggregate
Subcat:
Level:
Original: yes
In form: yes
Page: 20
Implementation: Python function
Query defined as:
gofase
Schlichting: Eerst wordt G totaal berekend. Zie hiervoor T152: G Totaal. “We berekenen hoeveel Zinsconstructies per Fase zijn gescoord. We letten hiervoor uitsluitend op de vakken van de Zinsconstructies: Mededelende zin, Vraag, en Gebiedende wijs. Per Fase noteren we het aantal Zinsconstructies in de eerste kolom bóven de Fase-aanduiding (niet meegeteld worden Intonatie en Koppelwerkwoord in Fase II, en ‘hè’ en Inversie in Fase III). Bij een bepaald percentage Zinsconstructies kunnen we zeggen dat het kind zich in die Fase aan het ontwikkelen is. De regel is dat 5% van de Analyse-eenheden van het taalsample in een Fase bij de Zinsconstructies gescoord moet zijn om het kind in die Fase te kunnen plaatsen.
T154: PFII¶
Name: PFII.
Category: Aggregate
Subcat:
Level:
Original: yes
In form: yes
Page: 23
Implementation: Python function
Query defined as:
pf2
Schlichting: “Profielscore voor Fase II”
T155: PFIII¶
Name: PFIII.
Category: Aggregate
Subcat:
Level:
Original: yes
In form: yes
Page: 23
Implementation: Python function
Query defined as:
pf3
Schlichting: “Profielscore voor Fase III”
T156: OvZnBv4¶
Name: OvZnBv4.
Category: Woordgroepen
Subcat:
Level: WG
Original: yes
In form: yes
Page: 69
Implementation: Xpath with macros
Query defined as: not implemented yet
Schlichting: “Overige Zn- en Bv-woordgroepen Fase IV”
T157: Screening¶
Name: Screening.
Category: Aggregate
Subcat:
Level:
Original: no
In form: no
Page: 94-97
Implementation: Python function
Query defined as:
tarsp_screeningSchlichting: “De TARSP-Screening us een methode om de Grammatical Ontwikkelingsfase van de spontane taal vast te stellen, zonder een analyse te maken.”
T158: PFIV¶
Name: PFIV.
Category: Aggregate
Subcat:
Level:
Original: yes
In form: yes
Page: 23
Implementation: Python function
Query defined as:
pf4
Schlichting: “Profielscore voor Fase IV”
T159: PFV¶
Name: PFV.
Category: Aggregate
Subcat:
Level:
Original: yes
In form: yes
Page: 23
Implementation: Python function
Query defined as:
pf5
Schlichting: “Profielscore voor Fase V”
T160: PFVI¶
Name: PFVI.
Category: Aggregate
Subcat:
Level:
Original: yes
In form: yes
Page: 23
Implementation: Python function
Query defined as:
pf6
Schlichting: “Profielscore voor Fase VI”
T161: PFVII¶
Name: PFVII.
Category: Aggregate
Subcat:
Level:
Original: yes
In form: yes
Page: 23
Implementation: Python function
Query defined as:
pf7
Schlichting: “Profielscore voor Fase VII”
T162: PF¶
Name: PF.
Category: Aggregate
Subcat:
Level:
Original: yes
In form: yes
Page: 23
Implementation: Python function
Query defined as:
pf
Schlichting: ” Profielscore. Een tweede mogelijkheid om twee samples meer in detail te vergelijken is de Profielscore (PF). De Profielscore is het totaal aantal constructies waarbij een of meermalen gescoord is. Bij het vergelijken van twee taalsamples door middel van de Profielscore moeten de samples eenzelfde aantal Analyse-eenheden bevatten. De Fase I constructies worden voor de Profielscore niet meegeteld”. Ook de constructies met twee ** tellen niet mee (Bijzin zonder verbindingswoord en de Vraagwoordconstructie zonder vraagwoord: ((Vr)WOnd+). OndVC telt men, indien deze niet in het taalsample voorkomt, wel mee als tenminste bij OndWVC of OndWBVC gescoord is. XNeg mag meegeteld worden indien een langere uiting met ‘niet’ gescoord is. Hetzelfde principe geldt voor OndB, VCW en BX. De vraagintonatie ‘Into’ mag meegeteld worden indien ergens anders in de vraagkolom gescoord is. In het meest linkse vak op de profielkaart staat het aantal constructies vermeld dat in een Fase ontwikkeld wordt. Hier kan men het gescoorde aantal constructies per Fase noteren. De Profielscore wordt genoteerd naast de G.O. Fase, bij PF. De maximale Profielscore is 100.
T165: Formulier¶
Name: Formulier.
Category: Forms
Subcat:
Level:
Original: yes
In form: yes
Page: 31-36
Implementation: Python function
Query defined as:
mktarspformSchlichting: “Profielkaart”