3.3. Polish Specifications

3.3.2.1. Table of categories

3.3.3.1. Notes

• The attribute "Human" is added to express derogativity from the IPIC. The Polish derogatives are a class of plural forms of nouns which are [−Human] in the nominative but [+Human] in the accusative.
• Gender is reduced to the traditional set of "masculine, feminine, neutral". Masculine-human (m1), masculine-animate (m2) and masculine-object (m3) are expressed as a combination of features "Gender", "Animate" and "Human", which are grouped for convenience.
• The gerund, e.g., bieganie, is considered a Type of Nouns. Strictly speaking, gerunds are a subtype of common nouns, but are treated as a type parallel to both common and proper nouns for convenience.
• Aspect and Negation are added to characterize gerunds.

3.3.4.1. Notes

• NB! Gerund as a VForm (imiesłów przysłówkowy) is not to be confused with the gerund as a Noun Type (gerundium).
• No voice for Polish verbs as all verbal forms are active (adjectival/attributive participles are treated as adjectives)
• The agglutination phenomenon in Polish is similar to Czech clitic_s for pronouns, but has a wider scope and can be found in more parts of speech. It is encoded as a more general "Clitic(y/n/a/d)" attribute and is specified, e.g., for the indicative VForm with Tense=pa(s)t, corresponding to "praet" flexeme in the IPIC to differentiate between forms like gniótł (clitic="n") and gniotł- (clitic="d"), where the latter not only demands a clitic but also has different form. The value "(a)gglutinant" indicates the clitic itself, e.g., -em in gniotłem. Values "y" and "n" are left to enable showing that a graphical word, i.e., delimited by white spaces, is a combination of a (d)emanding (or free) segment and an (a)gglutinant in case the word segmentation should be revised in the future.
• The Vocalicity of (a)gglutinated forms like -em vs -m is mapped on the Definiteness attribute with its values "(f)ull-art" and "(s)hort-art" respectively, meaning "full form" and "short form". The terms are very artificial, but this category is used due to the similarity of the phenomenon.

3.3.5.1. Notes

• Aspect and Voice are used for further differentiation of adjectival participles.
• As compared to the IPIC: ordinal numerals were extracted from adjectives and moved to numerals; pronominal adjectives were moved to pronouns.
• Post-prepositional adjectives like (po) polsku are treated as adverbs.
• The IPIC flexeme winien and predicatives like rad are treated as short adjectives—Definiteness="short-art".

3.3.6.1. Notes

• The IPIC tagset does not have pronoun types, those were defined manually.
• Accentability of Polish pronouns like jego, niego vs go, -ń is mapped on the Definiteness attribute with its values "full-art" and "short-art" respectively. Even though the terms are artificial, this attribute was used due to the similarity of the phenomenon, in order to avoid multiplication of attributes.
• Prepositionality is encoded as Clitic with values "y(es)" for nią, niego etc., "n(o)" for ją, go etc., "a(gglutinant)" for -ń. Cf. the Clitic value "bound" for Slovene pronouns like zate which refers to the whole cluster, formally a combination of a preposition and a pronoun. This coding can be used for similar phenomena in Polish, e.g dlań (for him), given the word segmentation is revised towards a more trraditional one.