ParlaMint-TEI.ana.rnc

default namespace = "http://www.tei-c.org/ns/1.0"
namespace a = "http://relaxng.org/ns/compatibility/annotations/1.0"

a:documentation [
  "RelaxNG XML schema for ParlaMint corpora: schema\x{a}" ~
  "  for linguistically annotatated corpus component file."
]
a:documentation [
  "The linguistic annotation markup covers sentences, tokens, lemmas,\x{a}" ~
  "  morphosyntactic tags, UD morphological features, UD dependency relations, and named\x{a}" ~
  "  entities."
]
include "ParlaMint-TEI.rnc" {
  
  ## Encoding description linguistically annotated corpus compoment.
  encodingDesc =
    element encodingDesc {
      global.atts,
      element projectDesc { paras },
      tagsDecl
      >> a:documentation [
           "Definition of listPrefix for MTed corpora."
         ],
      element listPrefixDef { prefixDef }?
    }
  
  ## The only element that contains analysed text is the segment. It can,
  ##       however, contain transcription commentary.
  seg = element seg { global.atts, (comment | pb | sentence)+ }
}

## A sentence.
sentence =
  element s {
    global.atts, (comment | pb | ner | phr | word | punct)+, syntax?
  }

## Multi-Word Unit
phr =
  element phr {
    global.atts,
    attribute function { xsd:token },
    attribute ana { xsd:anyURI }
    >> a:documentation [
         "Currently we only support phrases with semantic markers"
       ],
    attribute type { "sem" },
    mwe_tokens
  }

## Named entity element
ner =
  
  ## Standard name element
  (element name {
     global.atts,
     ref.att,
     attribute ana { xsd:anyURI }?,
     attribute type { "PER" | "LOC" | "ORG" | "MISC" }?,
     attribute subtype { xsd:token }?,
     ner_tokens
   }
   >> a:documentation [ "Additional content used by CZ:" ]
   | element date { ner.atts, temporal.atts?, ner_tokens }
   | element time { ner.atts, ner_tokens }
   | element num { ner.atts, ner_tokens }
   | element unit { ner.atts, ner_tokens }
   | element email { ner.atts, ner_tokens }
   | element ref { ner.atts, ner_tokens })

## Attributes for NEs.
ner.atts =
  global.atts,
  attribute ana { xsd:anyURI }

## Annotations that can appear in a MWE.
mwe_tokens = (word | punct)+

## Annotations that can appear in a NE.
ner_tokens = (word | punct | comment | pb | ner)+

## A series of tokens.
tokens = (word | punct)+

## Definition of word.
word =
  
  ## cf. https://clarin-eric.github.io/parla-clarin/#sec-ananorm
  element w {
    
    ## Simple word:
    ((word.atts, normalized-space.string)
     >> a:documentation [ "Word with several normalised words:" ]
     | (word-optional.atts,
        # This doesn't work: error: group of "string" or "data" element
        
        # ref name="normalized-space.string"/
        text,
        element w { word.atts, empty }+)
       >> a:documentation [
            "Normalised word with several surface words:"
          ]
     | (word.atts,
        element w { normalized-space.string }+,
        empty))
  }

## A punctuation element.
punct = element pc { token.atts, non-empty.string }

## Dependency syntax as a link group (inside a sentence).
syntax =
  element linkGrp {
    attribute targFunc { "head argument" },
    attribute type { "UD-SYN" },
    element link {
      attribute ana { xsd:anyURI },
      attribute target { anyURIs },
      empty
    }+
  }

## Attributes of token elements.
token.atts =
  id.att?,
  lang.att,
  attribute join { "right" }?,
  attribute norm { non-empty.string }?,
  attribute msd { non-empty.string },
  attribute ana { anyURIs }?,
  attribute pos { non-empty.string }?,
  attribute function {
    # This is different from TEI, where it is a name token (which can't contain comma!?)
    
    # ref name="normalized-space.string"/
    xsd:token
  }?

## Obligatory attributes of word.
word.atts =
  token.atts,
  attribute lemma { normalized-space.string }

## Optional attributes of word.
word-optional.atts =
  id.att?,
  lang.att,
  attribute join { "right" }?,
  
  ## UD PoS and morph.features, e.g. "UPosTag=NUM|NumForm=Digit":
  (attribute msd { non-empty.string }?),
  attribute ana { anyURIs }?,
  attribute lemma { normalized-space.string }?