O

object

The object of a sentence is the noun phrase that appears after the verb in a declarative English sentence. For example, in The cat ate the pizza, the pizza is the object. In The pizza was eaten by the cat, there is no object. Object noun phrases can be arbitrarily long and complex. For example, in He ate a pizza with lots of pepperoni, pineapple, capsicum, mushrooms, anchovies, olives, and vegemite, the object is a pizza with lots of pepperoni, pineapple, capsicum, mushrooms, anchovies, olives, and vegemite. [No, I do not have shares in a pizza company.]

See also ditransitive, transitive, and intransitive.

ordinal

A form of number word that indicates rank rather than value. Thus "one, two, three, four, five, six, seven" are cardinal numbers, whose corresponding ordinal numbers are "first, second, third, fourth, fifth, sixth, seventh".

output probability

= lexical generation probability, but used in the context of a Hidden Markov Model.

 

P

parse tree

A parse tree is a way of representing the output of a parser, particularly with a context-free grammar. Each phrasal constituent found during parsing becomes a branch node of the parse tree. The words of the sentence become the leaves of the parse tree. As there can be more than one parse for a single sentence, so there can be more than one parse. Example, for the sentence "He ate the pizza", with the respect to the grammar with rules

S → NP VP, NP → PRO, NP → ART N, VP → V NP,

and lexicon

("ate" V) ("he" PRO) ("pizza" N) ("the" ART)

the parse tree is

Note that this graphical representation of the parse tree is unsuitable for further computer processing, so the parse tree is normally represented in some other way internally in NLP systems. For example, in a Prolog-like notation, the tree above could be represented as:

s(np(pro("He")),
  vp(v("ate"),
     np(art("the"), n("pizza")))).

parser

A parser is an algorithm (or a program that implements that algorithm) that takes a grammar, a lexicon, and a string of words, decides whether the string of words can be derived from the grammar and lexicon (i.e. is a sentence with respect to the grammar and lexicon.

If so, it produces as output some kind of representation of the way (or ways) in which the sentence can be derived from the grammar and lexicon. A common way of doing this is to output (a) parse tree(s).

part-of-speech tagging

The process of going through a corpus of sentences and labelling each word in each sentence with its part of speech. A tagged corpus is a corpus that has been so labelled. A tag is one of the labels. Large-scale corpora might use tag-sets with around 35-40 different tags (for English). See Allen Fig. 7.3 p. 196 for an example of a tag-set.

part of speech, POS

Synonymous with lexical category: the role, like noun, verb, adjective, adverb, pronoun, preposition, etc. that a word is either playing in a particular sentence (e.g. like is acting as a preposition in I like pizza) or that it can play in some sentence: e.g. like can act as a verb, noun, adjective, adverb, preposition, and conjunction. (It can also act as a "filled pause", as do um, er, and uh - e.g. >He's, like, a pizza chef in this, like, fast food joint downtown.

 

 

participle

Participles come in two varieties (in English) - present participles and past participles. (Often abbreviated to PRESPART and PASTPART or something equivalent, like ING and EN). Present participles are variants on verbs; they end in "-ing", as in "setting", "being", "eating", "hiring". Past participles end in "-ed", "-en", or a few other possibilities, as in "set" (past participle the same as the infinitive form of the verb), "been", "eaten", "hired", "flown" (from "fly").

Participles are used in constructing tensed forms of verbs, as in "he is eating", "you are hired", and also as though they were adjectives in phrases like "a flying horse" and "a hired man".

In some cases, present participles have become accepted as nouns representing an instance of the action that the underlying verb describes, as with "meeting".

PRESPART and PASTPART are lexical grammatical categories.

particle

A particle is usually a word that "normally" functions as a preposition, but can also modify the sense of a noun. Not all prepositions can be particles. An example of a word functioning as a particle is "up", in "The mugger beat up his victim". Here "beat up" functions as a unit that determines the action being described. A telltale sign of a particle is that it can often be separated from the verb, as in "The mugger beat the victim up". Sometimes it can be non-trivial for an NLP system to tell whether a word is being used as a particle or as a preposition. For example, in "Eat up your dinner", "up" is definitely a particle, but in "He eats up the street", "up" is a preposition, but it takes real-world knowledge to be sure of this, as the alternative possibility, that the person being referred to is eating the street, is syntactically reasonable (though not pragmatically reasonable, unless perhaps "he" refers to a bug-eyed asphalt-eating alien.)

See also phrasal verb.

passive voice

Both active and passive voice are described in the article on active voice.

past

See modal operators - tense and tense.

PASTPART

An abbreviation for Past Participle, particularly in grammar rules.

past perfect

See tense.

patient

= object.

person

Person is a feature of English noun phrases that is principally of significance with pronouns and related forms. The possible values of person are first person signifying the speaker (possibly with his/her companions), second person signifying the person addressed (possibly with his/her companions), and third person signifying anybody else, i.e. not speaker or person addressed or companion of either.

Below is a table of the forms of pronouns, etc. in English, classified by person and syntactic case:

case	first person	second person	third person
nominative	I/we	thou/you/ye	he/she/it/they
accusative	me/us	thee/you/ye	him/her/it/them
possessive adjective	mine/ours	thine/yours	his/hers/its/theirs
possessive	my/our	thy/your	his/her/its/their
reflexive	myself/ourselves	thyself/yourself yourselves	himself/herself itself/themselves

phone

A low-level classification of linguistic sounds - phones are the acoustic patterns that are significant and distinguishable in some human language. Particular languages may group together several phones and regard them as equivalent. For example, in English, the L-sounds at the beginning and end of the word "loyal", termed "light L" and "dark L" by linguists, are distinct in some languages. Light L and dark L are termed allophones of L in English. Similarly, the L and R sounds in English are regarded as equivalent in some other languages.

phoneme

Start by reading about phones. Phonemes are the groups of phones (i.e. allophones) regarded as linguistically equivalent by speakers of a particular language. Thus native English speakers hear light L and dark L as the same sound, namely the phoneme L, unless trained to do otherwise. One or more phonemes make up a morpheme.

phonetics

The study of acoustic signals from a linguistic viewpoint, that is, how acoustic signals are classified into phones.

phonology

The study of phones, and how they are grouped together in particular human languages to form phonemes.

phrasal category

A kind of non-terminal symbol of a grammar - a non-terminal determines a phrasal category if it cannot appear in a lexical insertion rule, that is, a rule of the form X → w, where w is a word. Examples include NP, VP, PP, ADJP, ADVP and S . Non-examples include N, V, ADJ, PREP, INTERJ, ADV (see lexical category).

 

Contrast with lexical category.

phrasal verb

A phrasal verb is one whose meaning is completed by the use of a particle. Different particles can give rise to different meanings. The verb "take" participates in a number of phrasal verb constructs - for example:

take in	deceive	He was taken in by the swindler
take in	help, esp. with housing	The homeless refugees were taken in by the Sisters of Mercy
take up