This paper investigates the syntax of present active participles in Polish, morphologically similar forms that nevertheless split into agreeing and invariant types. Agreeing participles show full φ-agreement and appear in adjectival environments, while invariant participles lack agreement and function as adverbial, converb-like modifiers. Despite this contrast, both types retain core verbal properties, including argument structure and aspectual interpretation.

I argue that this split does not reflect a categorial distinction between verbs and adjectives. Instead, both participle types are derived from the same verbal structure and the difference follows from the presence or absence of an additional φ-probe and from structural position. Agreement is licensed by syntactic Agree, not by recategorization to adjective. Adopting a feature-driven minimalist approach, I propose that non-categorial features such as φ can provide labels under the Labeling Algorithm, making an adjectival head unnecessary. Present active participles will therefore be defined as “centaur” constructions, showing properties of multiple categories, which however does not entail syntactic category change.

This study investigates linguistic patterns of gender coding in Polish job titles used in online recruitment advertisements, with the aim of determining the extent to which these patterns reflect gender-inclusive or gender-exclusive linguistic practices. Drawing on a large corpus of contemporary job postings, the analysis examines the distribution of gendered forms in occupational titles across a range of sectors and professional domains.

The findings demonstrate an overall predominance of masculine forms across the corpus, while inclusive strategies constitute a secondary pattern, most notably realised through masculine–feminine pairings and, more marginally, abbreviated gender markers. However, item-level statistical analyses reveal substantial internal differentiation, indicating that patterns of gender coding tend to be aligned with the gender structure of the Polish labour market. Inclusive practices appear to be weakest in traditionally male-dominated technical and manual occupations, particularly where established feminine equivalents are lacking, and stronger in feminised sectors where comparable word-formation constraints do not occur. Lexical distributional analysis further reveals a Zipfian pattern across all coding categories, with a small number of high-frequency, semantically generic job titles accounting for a disproportionate share of tokens.

Overall, the results demonstrate that gender coding in Polish recruitment discourse is neither random nor chaotic, but systematically shaped by both linguistic factors and the extralinguistic realities of occupational gender stratification.