Dependent Tiers

Basic Word Format

Each word on the %mor tier has the structure:

POS|stem

Where POS is a part-of-speech tag and stem is the root morpheme, separated by a pipe (|). Example:

*CHI: I wanted a toy.
%mor: pro|I v|want-PAST det|a n|toy.

Part-of-Speech Tags

POS tags may include subcategories separated by colons:

category:subcategory

Examples: n (noun), v (verb), v:aux (auxiliary verb), n:prop (proper noun), pro:dem (demonstrative pronoun).

Morphological Markers

Markers follow the stem and indicate morphological structure:

Prefixes

Morphological prefixes precede the POS tag and are separated by #:

prefix#POS|stem

Example: un#v|do for "undo".

Compound Words

Compound words use + to join constituent words under a shared POS:

POS|+word1+word2

Example: n|+ice+cream for "ice cream".

Clitics

Clitics are bound morphemes attached to a host word:

• Preclitics precede the main word and end with $: pro|I$ v|be&1S
• Postclitics follow the main word and begin with ~: v|do~neg|not for "don't"

Translations

English translations of non-English stems are marked with =. Multiple alternatives are separated by /:

n|perro=dog, v|aller=go/walk

Omitted Words

Words marked as omitted on the Orthography tier (with 0 prefix) appear on the %mor tier with 0 before the POS tag: 0det|the.

Terminators

Utterance terminators from the Orthography tier (., ?, !, and special terminators) are carried through to the end of the %mor line.

Entry Format

Each entry has the structure:

index|head|RELATION

• index — 1-based position of this word in the utterance
• head — position of the word this word depends on (0 = sentence root)
• RELATION — grammatical relation label

Example

A complete example showing Orthography, %mor, and %gra together:

*CHI: I wanted a toy.
%mor: pro|I v|want-PAST det|a n|toy.
%gra: 1|2|SUBJ 2|0|ROOT 3|4|DET 4|2|OBJ 5|2|PUNCT

Reading this dependency tree:

• 1|2|SUBJ — word 1 ("I") is the subject of word 2 ("wanted")
• 2|0|ROOT — word 2 ("wanted") is the root of the sentence
• 3|4|DET — word 3 ("a") is the determiner of word 4 ("toy")
• 4|2|OBJ — word 4 ("toy") is the object of word 2 ("wanted")
• 5|2|PUNCT — word 5 (".") is punctuation attached to the root

Common Relation Labels

Relation labels follow the Universal Dependencies framework. Common labels include: