def is_np_structure(node):
# These tags are attested NP modifiers
# rule out NP-PRD as NP structure (0:88(15))
# rule out things already tagged explicitly as coordination by tag.py
if any(has_tags(kid, 'cC') for kid in node): return False
return node.tag.startswith('NP') and all(
(any(kid.tag.startswith(cat)
for cat in NominalCategories)) or kid.tag.startswith('ADJP')
or kid.tag.startswith('QP') or kid.tag.startswith('CD')
or kid.tag.startswith('DP') or kid.tag.startswith('DT') or
# 24:98(3)
kid.tag.startswith('PN') or kid.tag.startswith('CP') or kid.tag.
startswith('DNP') or kid.tag.startswith('ADVP') or kid.tag.startswith(
'AD') or kid.tag.startswith('IP') or kid.tag.startswith('JJ')
or # JJ is in here because ADJP < JJ may have been shrunk already
kid.tag.startswith('LCP') or
kid.tag.startswith('CLP') or # 0:57(12) reduced M
kid.tag.startswith('PP') or kid.tag in
("PU", "CC") or # CC for underspecified NP (27:24(3))
kid.tag.startswith('NP') or kid.tag.startswith('WHNP') or # 9:30(13)
kid.tag.startswith('FLR') or # ignore FLR
kid.tag.startswith('SP') or # ignore SP
has_tag(kid, 'p') for kid in node)
def is_np_structure(node):
# These tags are attested NP modifiers
# rule out NP-PRD as NP structure (0:88(15))
# rule out things already tagged explicitly as coordination by tag.py
if any(has_tags(kid, 'cC') for kid in node): return False
return node.tag.startswith('NP') and all(
(any(kid.tag.startswith(cat) for cat in NominalCategories)) or
kid.tag.startswith('ADJP') or
kid.tag.startswith('QP') or kid.tag.startswith('CD') or
kid.tag.startswith('DP') or kid.tag.startswith('DT') or
# 24:98(3)
kid.tag.startswith('PN') or
kid.tag.startswith('CP') or
kid.tag.startswith('DNP') or
kid.tag.startswith('ADVP') or kid.tag.startswith('AD') or
kid.tag.startswith('IP') or
kid.tag.startswith('JJ') or # JJ is in here because ADJP < JJ may have been shrunk already
kid.tag.startswith('LCP') or
kid.tag.startswith('CLP') or # 0:57(12) reduced M
kid.tag.startswith('PP') or
kid.tag in ("PU", "CC") or # CC for underspecified NP (27:24(3))
kid.tag.startswith('NP') or
kid.tag.startswith('WHNP') or # 9:30(13)
kid.tag.startswith('FLR') or # ignore FLR
kid.tag.startswith('SP') or # ignore SP
has_tag(kid, 'p') for kid in node)
def is_np_internal_structure(node):
# rule out things already tagged explicitly as coordination by tag.py
if any(has_tags(kid, 'cC') for kid in node): return False
return (node.tag.startswith('NP') and all(
has_tags(kid, 'nN') or any(
kid.tag.startswith(tag) for tag in NominalCategories) or kid.tag in
('PU', 'CC') or kid.tag.startswith('JJ') or kid.tag.startswith('CD')
or kid.tag.startswith('OD') or has_tag(kid, '&')
for kid in leaves(node)))
def is_np_internal_structure(node):
# rule out things already tagged explicitly as coordination by tag.py
if any(has_tags(kid, 'cC') for kid in node): return False
return (node.tag.startswith('NP') and
all(has_tags(kid, 'nN')
or any(kid.tag.startswith(tag) for tag in NominalCategories)
or kid.tag in ('PU', 'CC')
or kid.tag.startswith('JJ')
or kid.tag.startswith('CD')
or kid.tag.startswith('OD')
or has_tag(kid, '&') for kid in leaves(node)))
def is_topicalisation(node):
return has_tag(node[0], 't')
def is_apposition(node):
return any(has_tag(kid, 'A') for kid in node)
def is_modification(node):
return has_tag(node[0], 'm')
def is_coordination(node):
return has_tag(node[0], 'c') or has_tag(node[1], 'c')
def is_ucp(node):
# TODO: The test for node[1].tag != PU is to prevent a rare mis-analysis of punctuation in a UCP being
# identified as a conjunct. Need to investigate further
return has_tag(node[0], 'C') and node[1].tag != 'PU'
def is_prn(node):
return has_tag(node, 'p')
def is_head_initial(node):
fnpk = first_nonpunct_kid(node)
return (has_tag(fnpk, 'h') or has_tag(node[1], 'r')) if fnpk else False
def is_prn(node):
return has_tag(node, 'p')
def is_head_final(node):
lnpk = last_nonpunct_kid(node)
return (has_tag(lnpk, 'h') or has_tag(node[0], 'l')) if lnpk else False
def is_topicalisation_without_gap(node):
return has_tag(node[0], 'T')
def is_etc(node):
return node.count() > 1 and has_tag(node[1], '&')
def is_topicalisation(node):
return has_tag(node[0], 't')
def is_modification(node):
return has_tag(node[0], 'm')
def is_argument_cluster(node):
return all(has_tag(kid, '@') for kid in node)
def is_topicalisation_without_gap(node):
return has_tag(node[0], 'T')
def is_head_initial(node):
fnpk = first_nonpunct_kid(node)
return (has_tag(fnpk, 'h') or has_tag(node[1], 'r')) if fnpk else False
def is_etc(node):
return node.count() > 1 and has_tag(node[1], '&')
def is_adjunction(node):
return has_tag(node[0], 'a')
def is_head_final(node):
lnpk = last_nonpunct_kid(node)
return (has_tag(lnpk, 'h') or has_tag(node[0], 'l')) if lnpk else False
def is_partial_ucp(node):
return ((node[0].is_leaf() and
(node[0].tag.startswith('CC') or node[0].tag == 'PU')
and has_tag(node[1], 'C'))
and base_tag(node.tag) != base_tag(node[1].tag))
def is_adjunction(node):
return has_tag(node[0], 'a')
def is_apposition(node):
return any(has_tag(kid, 'A') for kid in node)
def is_partial_ucp(node):
return ((node[0].is_leaf() and (node[0].tag.startswith('CC') or node[0].tag == 'PU') and has_tag(node[1], 'C')) and
base_tag(node.tag) != base_tag(node[1].tag))
def is_argument_cluster(node):
return all(has_tag(kid, '@') for kid in node)
def is_coordination(node):
return has_tag(node[0], 'c') or has_tag(node[1], 'c')
def is_ucp(node):
# TODO: The test for node[1].tag != PU is to prevent a rare mis-analysis of punctuation in a UCP being
# identified as a conjunct. Need to investigate further
return has_tag(node[0], 'C') and node[1].tag != 'PU'
