Adjective
1. Adjective in the classifications of parts of speech; basic semantic, syntactic and morphological properties
Adjectives are one of the basic (semantically independent) parts of speech and have featured in the most important Polish grammars since Statorius Stojeński. As a separate functional class of lexemes, they are distinguished on the basis of semantic, syntactic or morphological (inflectional) criteria, and usually a combination thereof (in various configurations and proportions and to a varying degree of systematization). The latter approach not only corresponds to the intuitive (i.e. by default based on a centuries-long grammatical tradition) understanding of adjectives or the specific problems related to the strict homogeneity of the criteria, but also the very trans-criterion nature of such sets. This is the case since lexemes enter into relations with both the external world (semantic-logical relations) and with each other within a specific text (syntactic relations). Identifying the mutual relations between the semantic, combinatorial, and formal properties of expressions is one of the central tenets of linguistics. On the one hand, syntactic properties are not initially independent of semantic ones (regarding the actual meaning), while inflectional properties are derived directly from syntactic ones, at least when it comes to lexemes that take only (disregarding the degree) syntactically dependent values of grammatical categories, as adjectives do. Thus, in the cognitive order, they are the exponents of such syntactic properties, even though, somewhat paradoxically, it is the morphological criterion – as the most tangible one – that tends the most towards emancipation. On the other hand, however, the form and combinatorial properties of expressions may modify their content, their reference to reality. The recognized extent of such an influence depends on the assumptions adopted for the purpose of linguistic description. For instance, the relationship between nouns and their adjectival derivatives in the classic semantic syntax approach (Karolak 2002) is viewed, as a rule, as equivalent (nouns and adjectives refer to the same predicates at the semantic level and differ only by the manner of content formalization). By contrast, in the approach based on the Saussurean (formal-semantic) unit of language, the smallest formal differences are assumed to have the potential to reflect differences in meaning, which requires an in-depth analysis every single time (cf. in the context of the description of adjectives, e.g. Danielewiczowa 2018); see also the early position of Adam Heinz (1957) who assumed that a conversion of a noun into an adjective in phrases such as dom ojca → ojcowski introduces a “qualitative moment” to meaning, erasing to a certain extent the explicitness of “the content of attitude” emphasized in the relevant dependent cases (see also e.g. Grzegorczykowa 1982). This operation serves as a starting point for further potential semantic modifications as in [iście] ojcowska opieka. As a result of differences in approach, even a tiny modification to the classification criteria, their hierarchy, or even the very understanding of what their names convey, may produce discrepancies in the results. Eventually, therefore, one needs to conclude that the concept of the adjective, as is the case of other parts of speech inherited from the linguistic tradition, is either purely
indicative or fully relativized to fit a specific model of grammar.
Even though the development of the grammar studies, which aim at exactness and the separation of classes, has been characterised by a gradual abandonment of purely semantic observations and a turn towards formal markers, the intersectionality of criteria still remains at the very core of most classifications. The selection of only one, atomistically treated criterion, produces outcomes that are patently inadequate when it comes to the semantic-syntactic relations as a whole, erasing important differentiations occurring in a language. A consistently applied semantic-ontological criterion, which defines adjectives as the names of properties, would make us include in this class, for instance, abstract ad-
jectival nouns (zieleń etc.) and the corresponding verba essendi (zielenić się). Meanwhile, the status of referential and quantifying expressions (e.g. ta druga dziewczyna, różni ludzie, wie-
lokrotne zwycięstwo) as well as participles would depend on the clarification of the concept of property. If we disregard morphology and rely solely on the syntactic criterion, according to which adjectives are dependents in a nominal phrase, an adjective would be identified with an adnominal modifier; on top of that, there is also the issue of adjectives occurring in predicative expressions (which are not only possible in most cases, but even considered basic by some scholars). The use of the concepts of a primary (semantic) and a secondary (syntactic) function, implicitly suggests the existence of additional criteria. It is assumed, therefore, that a given lexeme is an adjective, even though it does not occur in the primary function (or it is not an adjective, even though it performs the function primary for this part of speech). Finally, the morphological criterion, according to which adjectives are defined as lexemes inflected for case, number and gender, extends to include pro-adjectives, ordinal
numerals, multiplicative, multiple and manifold numerals, as well as adjectival participles. In consequence, even though this classification is useful for the purpose of formal analysis, in a broader context it requires further specification from the semantic perspective. The scope of the class thus defined is broad enough to be rather labelled lexemes with adjectival
inflection, or, potentially, morphological adjectives – especially in view of the fact that not all of the “functional adjectives” are inflected lexemes (blond, bordo, khaki, mini, retro etc. cf. Kallas 1977). In view of these difficulties, classifications that rely on one criterion alone can hardly be found – the exception being the morphological division applied for specific formal-analytical purposes. In other cases, the terms semantic / syntactic classification refer to the prevailing criteria of classification.
To define an adjective as a name for a property of an object was a widespread practice in the grammars of the Polish language until the mid 20th century. Nevertheless, from the very outset, additional restrictions were put in place to narrow the scope of this class. In Stanisław Szober’s “Gramatyka języka polskiego” (1957 [1921]), apart from adjectives, the class of the names of properties of objects further includes numerals and pro-adjectives, while the adjective alone is identified as the so-called signifying name, in contrast to deictic names (proforms, pro-adjectives in this case) and signifying-deictic lexemes (corresponding numerals). In effect, this classification has at its core a combination of the two types of semantic criteria – characterizing the nature of an extralinguistic being on the one hand, and the manner in which linguistic expressions refer to that being on the other. Later on, the heterogeneous nature of criteria applied in such classifications came under criticism (cf. e.g. Laskowski 1998
[1984]) along with their subjectivity and the lack of precision. Also Jan Łoś (1923), Henryk Gaert ner (1938) or Zenon Klemensiewicz (1960 [1952]) relied primarily on the semantic understanding of this (and other) part of speech – the latter wrote about the qualities of objects listing various kinds thereof (shapes, dimensions, smell, material, etc.). Gaert ner also proposed a very extensive (albeit approximate and overlapping) division within the class of
“qualitatively differentiating words” on the basis of their word-formation properties. One might conclude that these detailed semantic distinctions not only aimed at organising the class internally, but also provided its indirect description when confronted with other classes in the context of the lack of unambiguous general criteria.
The relatively most precise classification of parts of speech, relying mostly on semantic principles, although not entirely, was proposed by Tadeusz Milewski (1965). In this approach, adjectives are viewed as autosemantic lexemes (in contrast to the synsemantic ones – prepositions, conjunctions and particles), and, within this category, are classified as symbolic (as opposed to expressive lexemes [interjections]), and then further as naming (rather than indicating [pro-forms] or ordering [numerals]) and eventually, within this final category – as naming a property of an object (rather than the object itself [nouns], a process [verbs] or a property of a process or of a property [adverbs]). This strictly semantic division is further overlaid with a variety of semantic-syntactic classifications, relying on the concept of categorial connotation (connotation of a syntactic head) – adjectives along with verbs are classified as secondary expressions, connoting (thus modifying) primary expressions referring to objects (nouns); tertiary expressions (adverbs), in turn, connote secondary expressions. So adjectives are words obligatorily connoting (the main noun) and optionally connoted (by the adverb that describes them). See also the classification by Jadwiga Wajszczuk (2010), based on semantic-syntactic connotations, which places adjectives in the class of independent syntactemes, i.e. words than open syntactic positions, but do not fill positions opened by other lexemes (unlike “terminals,” also referred to by Wajszczuk as terms, which fill open positions, but do not open such positions themselves, and dependent predicates, endowed with both of these properties).
When it comes to the classifications that attempt to systemically combine the semantic and syntactic criteria, we should also mention the one put forward by Stanisław Jodłowski (1976). Jodłowski viewed adjectives as lexemes with a denotative function (rather than expressive or impressive function, owing to the “stylizing attitude of the speaker”), providing content in a qualitative manner (as confronted with substantive, quantitative, action- or relationship-related – according to “epistemological categories”), combining a word with its referent on mnemonic basis (in contrast to the non-mnemonic – pro-form technique, relying on the “signifying technique”). Having restricted the class in this way, he considers adjectives as non-naming (but enumerating, i.e. conveying content in non-nominal categories) lexemes, which identify “a property inherent to an object,” and – from the syntactic perspective – capable of (actively) performing the function of a modifier and passively able to bind adverbials. Jodłowski explicitly stated that the criteria for the classification of parts of speech must be mixed, and, in addition, they do not unequivocally determine all the occurrences of specific lexemes (they are based on typical rather than universal properties). On top of this, the scholar extended his description in a general way by adding morphological properties “referred to as adjectival” (inflectional and word-formation-related).
As far as syntactic classifications are concerned, the critical point involves the decision as to which syntactic position should be considered as distinctive/unmarked for this group of lexemes – the attributive (a modifier) or the predicative (a predicative expression). According to the prevailing view, an adjective is (primarily) a dependent component of a nominal phrase, or, in order words, a modifier (e.g. Kuryłowicz 1948; Mirowicz 1949; Kurkowska 1954; Misz 1967; Szupryczyńska 1980; Laskowski 1998 [1984]; Karolak 2002; Szumska 2006), with reservation that this description leaves out expressions used purely as predicates (X jest rad/kontent v. *rad/kontent X; unlike gotów, łaskaw, świadom, etc. – they do not have full adjectival forms, and, as a result, cannot be considered as special values of the “ordinary” adjectival paradigms adopted in the predicative position). And the other way round: lexemes that occur only adnominally (cała grupa, żaden/ten/jakiś/pewien/każdy człowiek, istny/
prawdziwy geniusz, były/obecny szef, różni/wszyscy/niektórzy/poszczególni ludzie, cf. [in respective meanings] *grupa jest cała, *człowiek jest żaden/ten/jakiś/pewien/każdy, *geniusz jest istny/ prawdziwy, *szef jest były/obecny, *ludzie są różni/wszyscy/niektórzy/poszczególni) will remain outside the scope of the approach defining adjectives as (primarily) predicative descriptions of nouns (cf. Topolińska 1984; Nagórko 1987; Węgrzynek 1995; Danielewiczowa 2007), i.e. descriptions of originally non-restrictive, rhematic nature (ascribing a property to an independently designated referent). Indeed, one may conclude that the restriction of the predicative position typically testifies to the semantic autonomy of such lexemes – it is i.a. in this very way, that is using the (non-)predicativity test, that “adjective-like,” inherently thematic expressions are distinguished from the metatextual plan; these include, for instance, metapredicates introducing a value-adding commentary from the sender, such as zwyczajny głupiec, niezły cwaniak, dosłowny idiota, regularny wariat, istne piekło, kompletna klapa, czysta bzdura (cf. *głupiec jest zwyczajny, *cwaniak jest niezły, *idiota jest dosłowny, *wariat jest regularny etc.) or analogous adnumerative operators (marne sto złotych, dobry kwadrans, okrąg ły rok, bite dwie godziny, zgrabna sumka, cf. *sto złotych jest marne, *kwadrans jest dobry, *rok jest okrąg ły etc.); cf. Danielewiczowa 2007; Doboszyńska-Markiewicz 2013, but also e.g. Nagórko 1987; Maryn/Maryn-Stachurska 2009, 2010, 2019. The predicative filter further sieves out expressions with a referential and quantifying function, such as deictic and possessive pronouns, indefinite pro-forms and other lexical equivalents (specifications) of logical quantifiers (wszyscy/wszystkie, niektórzy/niektóre, każdy/każde, różni/różne, poszczególni/poszczególne (see Grochowski 2000, 2016; Karolak 2002; Rosalska 2021), numerals and such lexemes as obecny, bieżący, były, ubiegły, zeszły, miniony, przeszły, przyszły, powyższy, poniższy, tutejszy, tamtejszy, rzeczony, następny etc. (derived from adverbs – markers of relevant acts of reference, or being lexicalized forms of former participles (cf. Topolińska 1984; Grzegorczykowa 1994; Grochowski 2003; Danielewiczowa 2018). In consequence, when it comes to such – from the abovementioned – expressions as cały, obecny, prawdziwy, pewien, różny, we deal
with the issue of homonymy (x jest cały, obecny, prawdziwy…). A similar situation may be observed with regard to the adjectives of scope-related and restrictive meaning, which occur in obligatorily restrictive contexts, and are traditionally referred to as classifying modifiers (in contrast to descriptive modifiers), such as ogród zoologiczny – *ogród jest zoologiczny, błąd logiczny – *błąd jest logiczny; many of them have their predicative variants, cf. błąd logiczny
v. logiczny argument – argument jest (całkowicie, w pełni) logiczny. In its restrictive/scope-related version, with its typical adjectival postposition, both elements of a phrase combine to form a specific name and are inseparable from each other; an analogous mechanism applies to terms using primary (unmotivated) adjectives, cf. kości długie/płaskie etc. (see Wojtasie-
wicz 1972; Nagórko 1987). In extreme cases, we deal with – as Anatol Mirowicz (1949) put it – a complete disappearance of the “combinatorial moment” or, in other words, the logical articulation (logical parsing). As a result, the expression becomes fully idiomatic, cf. e.g. the relationship between demokracja and demokracja ludowa or krzesło and krzesło elektryczne (Mirowicz focused in particular on the dynamics of the shift from an active syntactic relationship to an erased one; from earlier studies see also Rysiewicz 1937). In consequence, insofar as błąd logiczny [logical fallacy] is still considered a fallacy, demokracja ludowa [popular democracy] is no longer democracy etc. Kazimierz Twardowski (1965 [1927]) referred to such combinations as abolishing-determining, characterised by “a partial removal of the content of the representation expressed by the noun.”
The proponents of primary predicativity (including Nagórko 1987; Danielewiczowa
2018) nevertheless admit that in many cases the superficial syntax criterion turns out to be too broad, as the problem of non-interchangeability of positions applies to the majority of denominal adjectives in general (e.g. dom ojcowski – *dom jest ojcowski, praca nauczycielska – *praca jest nauczycielska), and to leave them out of the class would not be sufficiently justified. The only exception concerns a specific type of usage indicating a material (denoting goods, cf. złoty pierścionek – pierścionek jest złoty, but not złoty kruszec – *kruszec jest złoty). In consequence, the criteria of semantic and syntactic predicativity diverge, and a different kind of derivational mechanism is postulated for this type of expressions – directly transposing the nominal derivational basis from the predicative position (or any other position within the predicate, e.g. object with an unexpressed predicate, cf. spotkanie odbyło się w niedzielę → spotkanie niedzielne v. *spotkanie było niedzielne) to the adjectival derivative in an attributive position (with no change of meaning). This corresponds directly to the assumptions of the semantic syntax, which operates with an abstract notion of predicate, pre-empting the division into functional classes (cf. in particular the arguments presented in Karolak [1984] and Szumska [2006]). This brief overview clearly shows that both the surface attributive and predicative position do not provide an unequivocal definition for the “proper” class of adjectives. What is more, both positions may also be filled by other traditionally defined parts of speech, and so in both cases the differentiation between adjectival modifiers/predicatives and non-adjectival ones requires taking account of the full congruence with the noun, that
is an analysis from the syntactic-morphological perspective.
The classification approach referred to as syntactic is diversified in itself, as it brings together the approaches oriented both towards the formal-distributive properties and the functional ones that set out the general model of directional syntactic dependencies. Henryk Misz (1967) was the first to put forward a comprehensive classification of parts of speech in the Polish language based on the formal criteria of syntactic distribution, and, in his approach, adjectives were described as syntactemes determined by nouns (dependents in nominal phrases). Importantly, the author emphasized the approximate nature of this classification,
which took account of certain general properties, while disregarding others; in view of this reservation, he classified e.g. participles as adjectives, while also noting that this group stands out from the entire class. The distributive cue with regard to adjectives was later taken up by, among other scholars, Maria Szupryczyńska (1980) in her analysis of syntactic requirements of a specific subclass of these expressions. On the other hand, in his classification, Roman Laskowski (1998 [1984]) describes syntactic relations from a broader functional perspective, taking account of directional accommodation relations. In this case, adjectives are classified as syntagmas (unlike non-syntagmatic components – appositions and interjections) that are syntactically independent (unlike dependent relative pronouns, conjunctions, prepositions and particles). Within this group, they are further placed in the set of accommodated (in contrast to non-accommodated adverbs and “modalizers”) dependents (as opposed to the heads of a clause – verbs). In the group of accommodated components, they are dependents in nominal groups (in contrast to heads, i.e. nouns), and, finally, within this set they are non-accommodating elements (unlike the accommodating numerals), cf. a slight modification to this approach by Wróbel (1996). Yet another functional classification, relying on the concept of categorial connotation and accommodation, as well as the set of inflectional properties – and thus explicitly based on a combination of criteria – was proposed by Zofia Zaron (2003). In this approach, adjectives are viewed as expressions that connote nouns or, if they fill the predicative position – verbs, without determining the value of any grammatical categories.
As far as the morphological criterion is concerned, it forms an explicit or implicit element of most of the proposed divisions (in syntactic criteria it is inseparable from accommodation, while in semantic-logical criteria the inflectional form of adjectives influences the manner in which they express content). In its pure form, independent of syntactic and semantic dependencies, it leaves out of description words that are uninflected at the surface level (all of them belong with the “uninflected” class). This is mostly the reason why this criterion as such is insufficient to account for the lexical resources of the Polish language as a whole. As shown above, this problem concerns also adjectives (such as bordo, khaki that may be described as adjectives with a syncretic paradigm only once we have applied the functional-syntactic criterion). Nevertheless, in the case of this class, the biggest difficulty lies in its broad range already discussed above. Having comprehensively analyzed the formal-semantic oppositions, there is no reason to include participles into the narrowly construed class of adjectives, as they retain their own syntactic and semantic properties, despite sharing with adjectives the same inflectional paradigm; cf. człowiek wycofany – coś wycofane z czegoś przez kogoś, coś zajmującego – ktoś zajmujący kogoś czymś, człowiek uznany – ktoś/coś uznany/e przez kogoś za kogoś/coś (cf. discussion on this topic e.g. in Szumska 2006: 57–58; Danielewiczowa 2018; on the adjectivization of participles see also Kurkowska 1954: 15–17; Bartnicka 1970). For this reason, many users of Polish perceive the joint spelling of nie with participles as inappropriate, erasing important distinctions suggested by the language system (including stress distinctions), cf. e.g. niezliczone tłumy – nie zliczona [jeszcze] kwota, nieodparta chęć – [jeszcze] nie odparty atak, niezmierzone przestrzenie – pole [dotychczas] nie zmierzone, nieopanowana żądza – [jeszcze] nie opanowany materiał, przedział dla [pasażerów] niepalących – [goście palący dogaszali papierosy na tarasie, a] nie palący weszli już do środka etc. The descriptions of the inflectional markers of the class of adjectives do not mention the category of degree traditionally included in the paradigm of this part of speech. This is a consequence, among other things, of the highly irregular nature of gradation, both in formal and semantic terms. Gradation applies only to the so-called qualitative adjectives (see later on in this article), cf. wysoki – wyższy – najwyższy, blady – bledszy – najbledszy (v. drewniany, niedzielny, miejscowy etc.), but not without exceptions. For instance, negated adjectives do not have comparative/ superlative forms (niejasny – *niejaśniejszy). The same applies to deadjectival derivatives, such as arcyciekawy, przezabawny or białawy, which in themselves contain gradation modifications (either intensifying or weakening a property); cf. also the use of adjectives in the function of classifying modifiers, e.g. muzyka poważna – *poważniejsza – *najpoważniejsza v. poważny człowiek – poważniejszy – najpoważniejszy. Gradable expressions are then divided into relative adjectives, which make the attribution of a specific property dependent on a specific (pragmatic) norm, and absolute adjectives, which indicate a property in an absolute manner (cf. X jest wyższy od Y-a, chociaż jest niski [in comparison to a norm] v. *X jest słodsze od
Y-a, chociaż jest gorzkie). In consequence, the positive degree of relative expressions entails tacit comparativus. From the formal perspective, only some adjectives are gradable synthetically, while others are gradable descriptively/analytically (ciekawy – ciekawszy – najciekawszy v. uroczy – bardziej uroczy – najbardziej uroczy); the synthetic forms of certain adjectives are suppletive (dobry – lepszy – najlepszy, zły – gorszy – najgorszy). Given this diversity, some scholars consider descriptive markers of gradation to be non-inflectional (i. e. syntactic; cf. Szupryczyńska 1980), while others completely exclude grade from inflectional categories, classifying it into the group of word-formation derivations (e.g. Kallas 1998 [1984]). Nonetheless, the latter solution erases the difference between the markers of the comparative / superlative degree and such expressions as starszy (człowiek), najświętszy, przenajświętszy (sakrament), najskrytsze (marzenia), najrozmaitsze, najróżniejsze, najprzeróżniejsze (rzeczy). On top of this, descriptive markers further encourage extending this category by including “negative” gradation (mniej_ – najmniej_), as well as other lexical markers of gradation – intensifiers such as bardzo, niezwykle, wyjątkowo, strasznie, okropnie, comparative structures, e.g. biedny jak
mysz kościelna, or weakening structures exemplified by dość, ledwo, mało, raczej etc. (see e.g.
Wierzbicka 1971; Laskowski 1977; Jurkowski 1976; Jurkowski et al. 1980; Orzechowska 1976– 1990; Janus 1981; Bogusławski 1994; Linde-Usiekniewicz 2000).
In Zygmunt Saloni’s strictly morphological classification (Saloni 1974, repeated with minor modifications in: Saloni, Świdziński 1981), adjectives are the only part of speech to be defined only positively, i.e. this class contains all the lexemes inflected for case (v. uninflected for case – verbs), gender (v. uninflected for gender – nouns) and number (v. uninflected for number – numerals). Apart from the previously listed expressions inflected according to this paradigm, the class of adjectives in this approach is also inclusive of deadjectival adverbs (as adverbial forms of adjectives, neutralized from the perspective of the abovementioned categories; the remaining adverbs and pro-adverbs are classified as “uninflected” and are further divided according to the syntactic criterion; the one exception is made for bardzo as a variant of wielce ← wielki) and, in consequence, adverbs derived from the enumerated types of numerals. What is more, in a strictly morphological classification, there is also a certain theoretical problem with a group of nouns with adjectival paradigms (cf. a monograph on this group, in diachronic perspective, Jodłowski 1964). However, since the classification takes into account the entire paradigm, nominalized adjectives will never be fully inflected in the same way as adjectives (schabowy, poborowy, prosta, wypadkowa, odstępne, wolne etc. are not inflected for gender, and blaszkowate, błonkoskrzydłe etc. are inflected neither for gender nor for number). Rather, their inflection will only be partially consistent with the paradigm of the relevant adjective, provided that such paradigm still operates in the language (cf. myśliwy, bliźni, czesne, komorne). In consequence, as regards double-gender personal names such as chory/chora, znajomy/znajoma, podróżny/podróżna, dyżurny/dyżurna etc., it has been postulated that the relevant masculine and feminine nouns exist independently. One should note, however, that when homonymy is in the process of becoming fixed (i.e. in the case of substantivization in statu nascendi), the only way to recognize it is to apply syntactic criteria.
The overview of the category in question in the Polish classifications of parts of speech showcases the conflict between the two different approaches – while one of them focuses on precision and distinctiveness of criteria, yielding potentially homogeneous but ‘coarse’ results, blurring subtler distinctions recognizable at other levels, the other relies immediately on a comprehensive, semantic-syntactic description of the category, which brings with itself an inherent risk of chaos. Most contemporary classifications balance somewhere between
these two extremes, proposing a clearly defined hierarchy of markers.
2. The subclassification of adjectives according to semantic and word-formation criteria
The attempts to classify adjectives into specific groups from the point of view of their actual meaning are primarily typical of grammars of the 19th and the first half of the 20th century – cf. the three main classes (subsequently subdivided into subclasses) in Feliks Żochowski’s study (1838; “depicting an attribute, matter or dependence of a thing”), as well as analogous, increasingly fragmented divisions in publications by Józef Muczkowski (1849), Antoni Krasnowolski (1906) or Henryk Ułaszyn (1915); these also include the above-mentioned indicative categories outlined by Zenon Klemensiewicz (1960 [1952]). Apart from the lack of precision typical of this criterion, they are characterised by a fundamental vagueness as to the interrelation of the semantic-ontological and the word-forming aspects, which makes them currently of little analytical value. When it comes to more recent studies, the criterion of “meaning-in-reality” is used as the prevailing one in detailed descriptions of selected lexical fields (implicitly assuming, but hardly ever listing other such fields), cf. e.g. the description of the field of colour (Tokarski 1995; Teodorowicz-Hellman 1999; Zych 1999), emotions (Buttler 1977, 1978; Nowakowska-Kempna 1998; Szumska 2000; Korytkowska 2017), dimensions
(Grzegorczykowa 1996; Linde-Usiekniewicz 1996; Nilsson 1997), mental qualities (Puzynina 1991) or evaluations (Termińska 1980; Nagórko 1982; Puzynina 1992; Jędrzejko 1993). If we disregard the early, general typological attempts and the subsequent analyses of semantic fields, the semantically-oriented classifications proposed in the literature of the subject are always subordinate to a certain, superior objective of a specific description. From this perspective,
we may distinguish between typologies oriented at: 1. a purely word-formation analysis; 2. an analysis of transformational mechanisms from the deep to the surface semantic level in approaches that assume the existence of an independent structure of meanings; 3. an analysis of surface syntactic connectivity and 4. an analysis of paradigmatic relations according to a specified, general criterion (e.g. semantic opposition).
Most Polish studies containing a classification of adjectives fall into category 1. The most extensive classification was presented by Henryk Gaert ner (1938), subsequently often criticized for the atomistic nature of his approach and the lack of sufficient generalizations (nevertheless, many appreciate this scholar for the abundance of collected linguistic material). In this study, derivatives (in the author’s terminology – indirect adjectives, in contrast to unmotivated direct adjectives) were grouped according to suffixes, identifying the origin, meaning and degree of productivity of each of them. As already mentioned before, all earlier grammars contain references to the word-formation structure, but the first systematic study of derivational mechanisms may be found only in the monograph by Halina Kurkowska (1954). In that study, adjectives are divided according to their derivational bases (verbal, nominal – substantive and adjectival – as well as adverbial). Compound derivatives (also classified according to the nature of their bases) and borrowings were discussed separately. The internal division within these groups is based on “semantic functions” (relying on case), and only then does the author analyse specific derivatives according to affixes, using extensive illustrative material grouped by periods: Old Polish, New Polish and the most recent one (19th–20th century). The study presents the historic process of semantic changes, including lexicalization, typical of adjectives, as a result of which words cease to be divisible into their formative constituents (względny ← wzglądać, wstrętny ← wstrącać, świetny ← świecić etc.). This involves the “qualitatization” of nominal adjectives (śmierć męża → śmierć mężna → mężny charakter), and often also the repartition of meanings (serce → sercowy, serdeczny; koniec → konieczny,
końcowy). Both the “qualitatization” and the repartition of meanings may also apply to adjectives that today are still divisible from word-formation perspective, cf. przetwory z mleka
→ przetwory mleczne → mleczny kolor, as well as owoc → owocowy, owocny or dwór → dworski, dworny. These diachronic analyses are then reflected in the identification of the main tendencies in the development of the formation of adjectives, which include mainly the pursuit of the economic use of linguistic measures, manifesting itself, among other things, in the abandonment of co-functional forms, along with the pursuit of monofunctionality related
to the semantic specialization of forms.
It may be said that Kurkowska’s study set the direction for the word-formation research for the decades to come, as it was the first one to analyse denominal adjectives in the context of nouns that they describe, which – paradoxically – over time resulted in identifying a unique multifunctionality (ambiguity) of words belonging to this class, cf. e.g. the position
presented by Renata Grzegorczykowa (1982) and Krystyna Kallas (in the so-called “academic grammar” 1998 [1984]), who, attempting to rein in this ambiguity, proposed to distinguish 18 groups from the perspective of the semantic roles expressed in adjective-noun combinations (e.g. subjective, objective, possessive, instrumental, material, product-oriented, genetic, singular, partitive, part-whole-relationship-oriented, characteristic specific, characteristic abstract, similative, scope-related, restrictive, locative, temporal as well as action-/process-/ state-oriented). Derivatives from verbs (predispositional and non-predispositional), adjectives (negation, gradation, expressive and other), prepositional phrases, adverbs and compounds are analysed separately. For instance, in this model, the adjective rowerowy would represent at least three categories: object (sklep rowerowy), instrument (wycieczka rowerowa) and a relationship between a part and a whole (koło rowerowe). The division relying on the semantic roles is based on the translatability of the nominal group containing the derivative into a corresponding simple sentence. In Polish linguistics, this topic was already taken up by Witold Doroszewski (1946). Also, in their analysis of adjectives, Renata Grzegorczykowa (1972) and
Jadwiga Puzynina (1976) referred to the concept of role. In the category of adjectives derived from verbs, Puzynina distinguishes the COMPL. class – comprising the functions that cannot be included in the distinguished groups, which brings to light the difficulties inherent to this description strategy. See also detailed accounts of specific structural subclasses inspired by this approach, e.g. in: Kaproń-Charzyńska 2005; Urban 2006.
Later on, this differentiation strategy was challenged, in particular by Dorota Szumska
(2006) who identified specific pragmatic mechanisms that govern the saturation of expressions with content in particular contexts (textual intra- and extraphrasal, as well as situational), cf. also Karolak 1984; Nowakowska 1998; Szymanek 1985. According to Szumska, the opinion that the meaning of an adjectival lexeme may depend on the noun it modifies originates from the erroneous transposition “of a textual description of a sign into its systemic-linguistic description” (2006: 94), which eventually leads to an impasse in semantic studies, as one cannot predict all the semantic variations that may appear in different contexts. The difference between typical combinations such as jesienne przymrozki and atypical ones, such as jesienne paznokcie, lies only in “the degree of predictability of inference as regards the removed designators” (2006: 101); if the interpretative context is insufficient or whenever there is no context whatsoever, such predictability depends on the degree of consistency
with the notional standards in a specific linguistic picture of the world (cf. ‘frosts occurring in autumn’ – ‘nails painted in a colour fashionable in autumn’ – ‘nails painted in a colour that resembles the colours of autumn’…). Meanwhile, the opposite, “extremely atomistic”
view posits the issue of description of the lexical meaning of adjectival derivatives (not only derived from nouns) “in a distorted mirror of their realization variants” (2006: 100). This applies to an equal extent to – in Szumska’s view unjustified – polisemization of evaluating adjectives (Puzynina 1992; Jędrzejko 1993).
Apart from the abovementioned studies, the classifications of adjectival derivatives largely based on “atomistic” principles, but applying a different degree of granularity, have been proposed – with respect to Polish derivatives from nouns (Szlifersztejnowa 1960; Jadac ka
1995) and in comparative studies (Sieczkowski 1957; Gawełko 1976), see also studies on the productivity of affixes relying on analogous distinctions (Satkiewicz 1969; Tekiel 1977; Jadacka 2001). One common thread typical of such classifications is the basic differentiation of adjective formations into qualitative, i.e. naming the inherent properties of an object designated by a noun (in which they are similar to non-motivated adjectives) and relational – specifying the relation of the derivative to the noun they occur with as semantically identical
with the relation between the nominal base to that noun (dom drewniany ← dom z drewna, jesienne przymrozki ← przymrozki jesienią, klimat morski ← klimat nad morzem etc.); cf. Szupryczyńska 1980; Markowski 1986; Szumska 2006. The former are gradable (or, more broadly: intensifiable) and may occur in a predicative position. They typically also create adverbial forms. The latter are not gradable (i.e. are attributed on a binary basis) and typically occur only adnominally; they also do not create any adverbial forms. This is not, however, a rule
without exceptions, cf. on the one hand bladawy, supernowoczesny, which are “qualitative” but not gradable, or zdolny which does not have an adverbial variant, and, on the other hand, structures with limiting adverbials, e.g. mieszkaniowo stali dobrze. What is more, many adjectives are bi-functional, cf. logiczny and mleczny above, as well as e.g. szmaragdowy naszyjnik
(‘made of emeralds’) – szmaragdowe oczy (‘of the same colour as emeralds’). In fact, most of such combinations are inherently ambiguous, although only sometimes does it break through into the interpretative order, e.g. pomarańczowy kwiat (‘the flower of the orange tree’ – ‘of the same colour as oranges’). Scholars have emphasized the productivity of this word-formation mechanism, cf. kraje europejskie (located in Europe) → europejskie standardy (typical of Europe, satisfying the criteria of “Europeanness”).
It is usually assumed (cf. Kallas 1998 [1984] et al.) that relational adjectives are a product of purely transpositional derivation (in Kuryłowicz’s terminology [1936/1979] referred to as syntactic), which transposes a noun into the class of adjectives without changing its meaning. The very same mechanism applies to the formation of derivatives from adverbs and prepositional expressions (tamtejsze zwyczaje, dzisiejszy obiad ← zwyczaje (które obowiązują) tam, obiad (który jest/był) dzisiaj, bezludna wyspa ← wyspa bez ludzi), as well as structures derived from verbs that do not introduce any additional meanings (e.g. wędrowne ptaki ←
ptaki, które wędrują, przenikliwy chłód ← chłód, który przenika). Qualitative derivatives, on the other hand, always involve semantic changes (they are a product of semantic derivation). In structures derived from adjectives, there is no change in terms of the syntactic category (only the meaning of the base is modified), while the derivatives from nouns and verbs (so-called predispositional ones, e.g. rozsuwany stół ← stół, który można rozsunąć, chorowite dziecko ← dziecko, które często choruje) involve syntactic derivation (typically referred to as mutation). Although in theory this distinction seems clear, in practice the decision regarding the transpositionality/mutability frequently depends on the relatively free structure of the paraphrase. Often the very proponents of this approach emphasize that, in many cases, it is impossible to unequivocally determine the derivation type, i.e. conclude whether the derivation does or does not involve semantic modifications.
Two radical solutions representing opposite poles have been proposed in the literature. According to the first, derivation from any expressions other than adjectives is substantially transpositional – this is the view adopted in the semantic syntax which assumes that derivational bases and derivatives refer to the same predicate; cf. Szumska (2006) who argued – as already mentioned above – that the alleged “added” meaning of adjectives comes from contextual enrichment, and Karolak (1984) who emphasized that the term relational adjectives is misleading, since the type of relation may only be deduced from the content of both predicates (according to this author, the sole difficulty arising from this approach concerns certain deverbal derivatives). This means, at the same time, that the division of adjectives into classes on the basis of the nature of relation is bound to fail from the very start. According to the opposite approach, the shift from a derivational basis to a derived adjective always involves a semantic modification. Paradoxically, this position is espoused by Adam Heinz (1956, 1957), often quoted by the researchers who diligently distinguish between the semantic qualitative and the asemantic relational derivatives with regard to derivatives from nouns. The distinction between qualitative and relational determination itself was inherited from linguistic tradition (when it comes to Polish linguists, it may be found – albeit in other terms – in Gaert ner [1938]), and was then adopted and elaborated by Heinz, who went deep into the analysis of the diversity of semantic relations within nominal groups. The starting point is the following: the concept of an object consists of two aspects – the substance (reference to the object as a whole) and the set of related properties. Noun modifiers mostly convey the “objective moment,” i.e. they describe the noun in terms of contact with the substance (relational determination, as in książka brata), but may also extract the “qualitative moment,” i.e. determine the noun in terms of equivalence or similarity of properties (qualitative determination which occurs in the case of apposition [człowiek mucha] and in other combinations, e.g. chłop jak dąb). More on conditions where noun modifiers become “qualitalized” and then adjectivized see Mirowicz 1949. Meanwhile, adjectival modifiers, as a rule, name a qualitative property – non-derivative expressions are purely qualitative, while adjectives derived from nouns may approximate the meaning of relevant concrete cases, but – being adjectives – they never lose their “qualitative moment.” At the same time, adjectives derived from nouns always retain their reference to the base in the background and as such they also contain – unlike non-derived adjectives – the “objective moment.” (As such, this solution is completely different from the subsequent approaches that associate relationality with nouns and their semantic adjectival equivalent, reserving qualitativeness only for semantically modified adjectives). Heinz admits that while the essence of the “qualitative moment” is difficult to define, it nevertheless guarantees the distinctiveness of adjectives derived from nouns from the corresponding case form, with the
“objective moment” defining the distinctiveness of such adjectives from purely qualitative (non-derivative) expressions. The basic semantic effect of the shift from the nominal base to an adjective is “blurring the relational content,” already mentioned above, i.e. blurring the specific relation between the describing and the described nouns (which results from the removal of the formal relational markers – prepositions or/and inflectional endings) – the speaker substantially uses the nominal form if “they may and want to name a certain relation in a specific and unequivocal way” (1957: 64), and the adjectival form if they are not ready for it. However, it is not the case that all case forms point without exception to a “specific relation,” while all adjectival forms to unspecific ones (which, in extreme cases,
mean in general “having a relationship to what the base suggests”). These are only the two extremes, while the specification of the relation is a matter of degree – both the former may express a relation with a lower degree of precision (e.g. dziecko miasta), and the latter may be characterised by a higher degree of specificity (although a certain indeterminacy of relation is inherent to the nature of adjective – absolute precision is not achievable). The meaning of an adjective is made more specific (as a result of the narrowing down of the scope of reference) in a specific textual setting, while with regard to a lexical unit without any context it may be established only indicatively, based on inference from the most frequent uses. In consequence, the greater the proportion of contextual factors, the higher the degree of precision, and the other way round. When it comes to the involvement of contextual factors and
the substantially unlimited interpretative possibilities, this approach – if we disregard the basic disagreement as to the semantic/asemantic nature of adjective derivation – runs basically parallel to the position espoused by Szumska (2006). The subsequent word-formation approaches, aimed at classifying derivatives in the possibly most systemic way, by necessity simplify this complex picture of dependencies at the semantic-pragmatic level. Yet another chapter in the research on adjectival formations focuses on the morphemic structure of ad-
jectives, cf. in particular Kowalik 1977, but also e.g. Strutyński 1979.
Transformational approaches (2.), assuming the existence of deep semantic structures formalized on the surface, primarily include the studies by Nagórko (1987), Karolak (2002) and Szumska (2006). In her approach relying on the assumptions of generative semantics, Alicja Nagórko proposes a multi-level model of adjective derivation – from the semantic level, to the syntactic and morphological, and finally the surface one (in this case, then, derivation is not limited to word-formation). The representations of the semantic plane have the form of a hierarchical system of semantic markers, which include: +/-GRAD (gradability),
+/-NORM (reference to norm, important for distinguishing – within the group of gradable adjectives – parametric and evaluative expressions [+NORM] from other expressions),
+/-MES (measurability), +/-INTENS (intensity; its surface exponents are adjectives corresponding to the expression bardzo, as in: głęboka cisza), +/-MAX (in gradable expressions: approaching a limit, e.g. absolutna cisza, wysokie ideały), +/-ABSOLUT (absolute v. relative
properties; all non-gradable expressions are absolute, but this group also includes certain gradable expressions, e.g. names of colours), +/-INHERENT (internal and external properties typical of objects, which corresponds to the traditional opposition between qualitative and relational properties), +/-DISTANCE (distance in space; in combination with marker
+INHERENT allows for describing such adjectives as gęsty, rzadki), +/-QUANT (quantitative v. qualitative properties; quantitative include parametric adjectives, adjectives that name physical properties of objects and temporal properties, as well as expressions related to the concept of abstract quantity [trudny – łatwy] and expressions with added quantification
[bogaty → który ma dużo dóbr]; qualitative adjectives include mental predicates), +/-HABIT (habitual v. current properties) and +/-EVAL (evaluations v. descriptive properties). This is the initial set of markers which groups the class of adjectives into basic categories; as the description gets more specific, it may be supplemented by other markers as well, e.g. +/-LOC
(differentiating between +MES usage such as wysoki dom [-LOC] and wysoki sufit [+LOC]),
+/-LIKE (designating metaphor and simile) or +/-ACTIVE (designating active v. passive attitude of the participant). A sample derivation for a polysemic adjective pełny is as follows: pełny kielich [-GRAD, -INHERENT], pełny tekst [+INHERENT, +QUANT], pełna czekolada [+INHERENT, -QUANT], pełna wiosna [+MAX, +INHERENT], pełne szczęście [+MAX,
-INHERENT], pełny kwiat [-MAX, +QUANT], pełna twarz [-MAX, -QUANT]. The essence of this approach is that these qualifications pertain both to non-derived and derived adjectives – with regard to the latter, the stage of semantic derivation, analogous to the deriva-
tion of non-derived adjectives, precedes the rules of morphological affixation, which operate on already semantically interpreted structures; as a result, there is no separate word-formation semantics. For the majority of adjectives, the level of syntactic derivation is the intermediate stage between the semantic and morphological levels. It involves transferring the predicate from the position inherent to the semantically basic structure (in the case of non-derivative expressions and expressions derived from adjectives – from the predicative position) to the attributive position; in doing so, it is important to distinguish between the nominalizing and
non-nominalizing (objective) nature of the noun being described. Such derivation does not apply to adjectives derived from nouns, as the transformation of dom ojca into dom ojcowski does not involve any change to the syntactic properties (or any semantic change). Rather, it is a purely morphological derivation (the concept of syntactic derivation in this approach is narrower, then, than in the approach postulated by Kuryłowicz).
In Stanisław Karolak’s approach (1984, 2002), the immanent properties of predicates are identified with the abstract meaning of concepts, while the relational properties arising therefrom (based on extralinguistic reality) determine the basic form of the structures they constitute. As a result, syntactic relations concern the deep semantic level (syntax of concepts). In language, predicates may be represented by predicative expressions that belong to various functional categories, differing in terms of their distribution within a clause. For instance, the predicates NUDA / ZGODA are expressed in Polish by nudny, nudzić się, nuda / zgodny, zgodzić się, zgoda, obtained via purely syntactic derivation (independent from formal derivation, cf. e.g. the postulated equivalence of śmierć and umrzeć). Predicative expressions, including adjectives, are subdivided according to the number and nature of complements; in this respect, adjectives and adverbs are so-called adjunctive expressions (the former are adjuncts to nouns, while the latter to verbs), in contrast to sentence-forming expressions (finite verbal forms) and the non-sentence-forming ones (nouns). The requirements as to the valency of semantically corresponding lexemes may differ from the connotative (according to author’s
terminology – implicative) structure of the base predicate, and, in addition, they may differ from each other. More specifically, the number of surface complements may be lower than (or equal to) the number of arguments at the conceptual level, but cannot be higher, except for the cases of disjunction (argument splitting). In this interpretation, NUDA is a two-argument predicate (implying the subject of the state and the object triggering the state), the
verb nudzić się obligatorily implies the subject and optionally the object (ktoś nudzi się czymś, ktoś się nudzi), nudny obligatorily implies the object and optionally the subject (coś jest nudne, coś jest nudne dla kogoś), meanwhile the noun nuda obligatorily cancels out these arguments (*nuda kogo, *nuda czym). Thus, adjectives are essentially equivalent to their corresponding verbs and abstract nouns, and, likewise, adjectives derived from nouns are equivalent to nouns (the derivation mechanism for restrictive and non-restrictive structures is different, however, because only the latter – as semantically independent – can be regarded as the equivalents of relative sentences).
These are the general theoretical bases on which Dorota Szumska (2006) constructed her original, insightful description of this part of speech, supplementing it with an overview of the semantic-connotative and implicature-based mechanisms necessary for the interpretation of combinations with adjectives (involving the ‘information slack’ typical of them, see above). What is more, the economization of nominal structures compared to sentence predication and their non-negatability also play a role. While it is not a priori assumed that surface formalizations do not trigger a change in meaning (or that they do so), the analysis leads to such a conclusion anyway. The author classified transpositional derivatives into indexical expressions (dzisiejszy, tamtejszy, possessive pronouns, etc.) and attached predicative expressions, while among the latter she distinguished constitutive adjuncts (which are correlates of the founding predicate of the adjunctive proposition) and non-constitutive adjuncts (typically undercoded as for the marker of the founding predicate). Constitutive (traditionally: qualitative) adjuncts are characterised by the symmetry of the sentence distribution between the adnominal and predicative positions, while non-constitutive (traditionally: relational) adjuncts occur only adnominally (sharing this feature with indexical expressions; only certain material-related uses are an exception). Nevertheless, one should emphasize that the author rejects the established division into qualitative and relational derivatives, along with the concept of the “qualitatization” of the latter. Adjectives such as europejski, męski in the qualitative sense are not derived from relational adjectival bases, but from nouns created either in the process of the neosemantization of the base noun (Europa1 → Europa2 as a communicatively fixed, unspecified set of properties associated with Europa1) or from abstract nouns expressing such properties (mężczyzna → męskość → męski2) (cf. Przybylska 2003). As a result, in each of these cases we are dealing with a transposition rather than with a qualitative change. Subsequently, the scholar analyses certain textual phenomena, especially the types of undercoding and overcoding of the structural plane in comparison to the content plane (absolute and relative pleonasm, condensation, compression, contextual gaps) as well as the relationships within poly-propositional sequences.
When it comes to authors who have proposed a semantic classification of adjectives from the perspective of their syntactic connectivity (3.), one should mention in particular Władysław Śliwiński (1990, 1993) and Krzysztof Jassem (2002). The aim of Śliwiński’s work is to register and catalogue typical combinations of adjectives with nouns. His study (1990) presents the assumptions of a “multilevel, hierarchical, bidirectional and reflexive” model, as well as lists of adjectives grouped according to particular categories. His subsequent work (1993), complementing the previous one, is an alphabetically arranged connotational dictionary (inspired by K. Polański’s “Słownik syntaktyczno-generatywny czasowników polskich”), which the author intended especially for language teaching. For the purpose of the analysis, the scholar selected non-derived adjectives transformable to the predicative position (also referred to as simple adjectives). For each lexeme described, its semantic (categorial and singular) and grammatical connotation is determined, and definitions, phraseologisms, synonyms, antonyms, frequency, possibly proverbs and comparative structures are provided. To determine the categorial connotation, it is necessary to identify the semantic classes of both nouns and adjectives, which allows for appropriate generalizations (supplemented with detailed information). Nouns are grouped according to the following classes: Hum, Anim, Sens (sensorial), Ment (mental) and Sens/Ment, while adjectives are divided into: Qual Phys
(qualitative designating physical properties), Qual Psych (analogous for mental properties), Qual Eval (qualitative evaluating), Qual Relat (qualitative designating relations), Quant Phys
(quantitative designating physical properties), Quant Psych (analogous for mental properties), Quant Eval (quantitative evaluating), Quant Relat (quantitative designating relations) and Intens (designating the degree of intensity of a property). An analysis of combinations produces an outcome in the form of a matrix – it shows, for instance, that Quant Relat adjectives connote categorially Hum and Sens/Ment nouns, while Intens adjectives – Ment only. Also the study by Krzysztof Jassem (2002) is an attempt at modelling the collocational properties of adjectives for lexicographic purposes (in machine translation from Polish into English). This time, however, the focus is on the right-sided connectivity (syntactic-semantic combinations with dependents). For each lexeme, a meaning or set of meanings is determined; each designated meaning has its own set of collocations (extracted automatically); expressions with analogous collocations group themselves into natural classes that – according to Jassem’s hypothesis – yield to semantic characterisation. There are as many as 68 such classes (combined into 8 supercategories), e.g. in the category “attitude towards people” –
“high requirements” (bezwzględny, groźny, ostry, rygorystyczny, srogi, stanowczy dla / wobec / względem / w stosunku do _), “anger” (wściekły, zagniewany, zły, rozzłoszczony na _), “cold negative attitude” (krytyczny, sceptyczny, wymagający dla /wobec / względem / w stosunku do _), “resilience, sensitivity” – czuły, nieczuły, obojętny, odporny, nieodporny, podatny, uodporniony, wrażliwy, niewrażliwy, wytrzymały na _ etc.; cf. Danielowiczowa’s (2018) critical comments on this classification, including her own suggestions on the correlation between the semantic properties and collocational properties of adjectives.
When it comes to the proposed general classification of the class of adjectives based on their paradigmatic properties (4.), one should note in particular the study by Andrzej Markowski (1986). It presents the linguistic material on the basis of different types of semantic opposition (also A. Nagórko [1986] recognizes the importance of a paradigmatic structuralization of this kind, emphasizing additionally its likely high psychological veracity). As regards opposite names, the author distinguishes contradictory and contrary names. The former, filling the entire logical space of a specific main concept, are divided into equivalent (two expressions with their own positive meanings, e.g. żywy – martwy, męski – żeński) and complementary (one of the expressions is the basic one, while the other, derived from
it, forms its logical complement, e.g. polski – niepolski). By contrast, contrary names do not fill the logical space completely, i.e. they assume the existence of at least one more name of the type; they may be unorganised (e.g. lądowy – morski) or organised, with the latter group being further divided into expressions ordered on the basis of the intensity of the property (wysoki – niski), spatial-temporal relations (przedni – tylny), or organised in a different way (czerwony – czarny). The description focuses on the pairs of strictly antonymous adjectives – opposite expressions ordered in accordance to the degree of intensity of a property. To be more precise, antonyms are placed on the two ends of a scale, at equal distance from its middle (as a result, adjectives wysoki – niski form an antonymous pair, but wysoki – niewysoki do not). This definition of antonyms assumes their gradeability with regard to the specific norm
(“a small elephant is a large animal”) and strict semantic-syntactic parallelism. In particular, it assumes that they share the scope of connectivity and the repertoire of semantic properties – apart from the sole differentiating property: the one that determines specifically the antonymy (in consequence, the relationship is very close to that of synonymy; one may say,
therefore, that the expressions synonimiczny – antonimiczny in themselves are not antonymous to each other, i.e. there is no “everything in common” – “nothing in common” relation between them). Markowski subdivided antonyms on the basis of them having the same or different roots, the type of opposition, sectionality v. location on the opposite poles of a scale and the specific semantic properties of the pairs. Within the group of antonyms with different roots, he distinguished adjectives describing physical properties (7 groups) and mental properties, including evaluations; antonyms sharing the same root mean exclusively mental properties and evaluations. Not all derivatives created using the prefix nie– are antonymous; this applies only to such derivatives that have a specific semantic surplus in comparison to the base (nieprzyjemny ≠ taki, który nie jest przyjemny), with this surplus (causing a semantic irregularity arising from – according to the author – the presence of the volitive predicate in the semantic structure of the base [chcieć]) allowing the derivative to enter into sensu stricto antonymous relationships (for more on negated derivatives see also Oliva 1967; Tokarski 1976). Otherwise, lexemes are contradictory to each other (nienormalny = the one who/which is not normal; partially lexicalized forms such as nieczęsty [= rather rare], niewysoki [= rather short], as well as the euphemistic uses such as niemały [= big] and nieduży [= small] represent yet another, separate phenomenon). The “regular” properties of negation create a specific methodological problem related to explications, as it is not – except for the predicates based on the opposition chcieć – nie chcieć (i.e. evaluative) – a good tool to construct antonymy. In consequence, the types based on the opposition większy – mniejszy, skłonny do_ – nieskłonny do_ and z – bez_ are distinguished in parallel. Moreover, it has been argued that antonyms create oppositions belonging to the privative type, with one unmarked element, which is also designating the main concept, and the other marked one (cf. Jak zdolny / miły / wysoki jest
X? v. ? Jak tępy / niemiły / niski jest X?). In consequence, according to Markowski, the traditional differentiation between qualitative and relational adjectives is imprecise and does not provide a sufficient indication for establishing antonymous dependencies.
3. The syntax of adjectives: positions in a clause, the order of the nominal group and connotative requirements
In addition to the modifier and predicative positions commented on in section 1, adjectives also occur in a secondary predicative position; they thus form a collateral, non-sentence forming predicative phrase, characterised by the fact that the adjective is either attached to the noun by way of an apposition (Wszyscy mieszkańcy, ciągle senni, marzą o niedzieli), or is at the same time a determiner of the nominal head and the predicate (e.g. Ojciec wraca bosy
[ojciec bosy – wraca bosy], Brat leży chory, Dumna jechała obok niego, Zastałam dziecko głodne, Staś ubrany leżał na łóżku; cf. leżeć chorym, wracać bosym). Such occurrences were referred to in traditional Polish grammars − in the first case − as appositional modifiers (Szober 1957; Jodłowski 1976) or predicative modifiers (Klemensiewicz 1937), and as adverbial modifiers in the second case (Szober 1957); in more recent studies – following the literature in the English language – they are called depictives (cf. Szumska 2006). They were the topic of an early study by Krystyna Pisarkowa (1965; the quoted sentences come from this source). The author pointed out to the affinity of predicative adjectives in the function of adverbials and structures with adverbs referring to the subject (wraca bosy – wraca boso, leży cichy – leży cicho). She also analysed the relationship between the combinations in which the ancillary predicative element is connoted (wydaje się zdrowy) and the structures where it is not a constituent connoted by the verb (mąż lubi żonę zdrową). An important marker of structures with ancillary predication is their typical, two-peak intonation (also noted by Klemensiewicz). With regards to semantics, prediactivity is manifested – as discussed in detail in the second part of the study – in the relative temporal nature of the ancillary element; the study further includes non-adjectival material (in the function of complements and predicative adverbials) and presents changes from a diachronic perspective (e.g. suggests that the preterite origi-
nated from predicative structures: mam dzieło napisane → mam to napisane → mam napisane
→ napisałem). For more on predicative, non-sentence forming uses of adjectives see also Zofia Czapiga (1993, 1994, 1997), who examined i.a. the differences between occurrences in the initial position (Piękna i dystyngowana, budziła podziw) and in postposition (Zobaczył góry, groźne i majestatyczne). The intermediate uses between predication and attribution include adjectives in the function of the so-called semantic predicate, describing a noun that is se-
mantically empty, e.g. Choroba jest przykrą rzeczą (Mirowicz 1949; cf. also Topolińska 1984, as well as the concept of a non-functional pleonasm in: Grochowski 1999, as in the sentence Sukienka była koloru niebieskiego).
When it comes to the syntax of the nominal phrase, the attention of grammarians has focused from the outset on the difference between the prepositional and postpositional uses of adjectives. Gaert ner (1924) noted that nouns are preceded by “primary” adjectives that designate accidental properties, while adjectives following nouns are derivative adjectives and denote fixed, species-specific properties. He further associates this order with the order of perception and psychological expression, and even the “intensity and mood.” Doroszewski (1948) records the predicative function in postposition (Światło księżycowe oświetla gnijące kłody na dnie wód v. Światło oświetla kłody gnijące na dnie wód), associating it with “highlighting a specific feature in the context of a more general view”; for more on the order of phrases containing participles see e.g. Orzechowska-Zielicz 1954, Wróbel 1975. When it comes to the studies focusing on the distribution of noun dependents with regard to each other, cf. in particular Sankowska 1962, Topolińska 1981, 1984, Śliwiński 1984, Gębka-Wolak 2000.
Zuzanna Topolińska (1984) links the attributive position of adjectives with their restrictive usage, narrowing down the reference of the noun they describe (which produces a description, i.e. a group designating the referent, which is either complete [designating such referent unequivocally] or incomplete), while linking the predicative position (including apposition) with non-restrictive usage, where the adjective provides an additional description of an independently designated referent. Insofar as there are no doubts as to the latter, the former does stir some controversy. For instance, Szumska (2006: 79) provided the ambiguous example Znalazłam czerwony notes, which may either mean that X found a notebook that was red (non-restrictive use) or that she found the red notebook – the one she was looking for (restrictive use). In free (syntactically analysable) utterances, expressions with attributive function are typically prepositional and together with markers of reference and of quantitative assessment they form the so-called basic (canonical) nominal sequence. The constituents of that sequence are placed sequentially according to the established rules (Topolińska 1984): the markers of reference and quantitative assessment (I) are placed before predicative expressions and argumentative expressions in the function of attributes (II), which in turn precede the constitutive component of the group. The order within I is the following: referential operator + numerical operator + name of the measurement unit (te dwa kilo jabłek etc.). As regards II (which, together with the constitutive component, makes up the nuclear group), the so-called type A relational (non-intensifiable) attributes precede intensifiable attributes,
which are then followed by type B relational attributes; group A, similar to I, is made of expressions such as taki, jakiś, wszelki, różny, possessive pronouns and adjectives, numerals with adjectival paradigm of inflection as well as previously discussed deadverbal derivatives (dzisiejszy, wczorajszy) and metapredicates (rzeczywisty, autentyczny, prawdopodobny); group B collects attributes ascribed on binary terms. It is assumed that, in the case of poly-segmental sequences, the linear order of the sequence reflects the history of its derivation, i.e. the order of absorption of specific predicates into the sequence (the attributes that are the closest to the noun are “sucked in” as first). Thus, in the sequence niski drewniany stół, the adjective niski designates a subset from the set of drewniane stoły etc. Sometimes certain attributes in a poly-segmental sequence are moved to postposition (without implying the inclusion of the attribute into the constitutive segment of the group). In such cases, the postpositional segment is perceived as the closest one, cf. wisząca lampa cepeliowska → ‘such a lamp from CePeLia that you can hang’. In preposition, the elements of the sequence are joint concatenatively (without conjunctions), by a conjunction (usually i) or are separated with commas (in the latter case this might, but does not have to, signify apposition). The specific semantic dependencies between the segments of adjectival sequences have been studied by Dorota Szumska (2006), who demonstrated that their nature is, to a certain extent, independent from the marker of fusion and that the number of surface segments does not translate into the number of propositional structures at the semantic level. On the one hand, the descriptions, as the phrases identifying the referent, are non-propositional structures (they do not predicatively assign any properties to the referent), while pleonastic structures such as małe, ciasne mieszkanko or the inclusive krótka, kilkunastoletnia historia are monopropositional. Contrary to that, combinations such as wiosenne spacery are derived from poly-propositional structures (ktoś spaceruje wiosną).
Małgorzata Gębka-Wolak (2000) in turn examines − on the grounds of formal syntax and using digital tools − the exact distribution of nominal phrase elements, including the co-occurring adjectival attributes. It is assumed that two adjectives separated by a comma account for a single adnominal position, while without a comma – for two different posi-
tions (though it is a simplification of actual syntactic relationships, it seems necessary to carry out a formal analysis). The left- and right-handed expressions are never in a coordinate relationship to each other, while coordinate attributes (linked with or without a conjunction) are always juxtaposed with each other against the background of their common semantic
properties. A distributive analysis – just like the semantic-syntactic analysis discussed above – has shown that the order of specific elements of the nominal group is not free. Rather, such elements form a strict linear order, i.e. either form local phrases whose components have a specific position with regard to the noun they determine and with regard to each other, or form quasi-local relations with elements that are free with regard to each other but localized with regard to the noun. Within the former we may distinguish four prepositional and one postpositional locations with the reservation that – unlike in Topolińska’s approach – classes of lexemes assigned specific positions are defined by the enumeration of sample specimens, relatively exhaustive only with regard to limited sets. As far as quasi-local relations, typical of developed adjectival phrases, are concerned, Gębka-Wolak distinguished three postpositional locations (after the class locally related to the noun), which are by definition interchangeable. When it comes to the factors that influence the location of NP components, the author lists: the type of dependent (part of speech), the manner and degree of its development, the type of head (nominalizing/non-nominalizing), the number of dependents and their length, the tendency towards balance between the left- and right-handed strings, as well as semantic and pragmatic determinants, to which, however, the author dedicates less attention (the latter are strong enough to trigger a change of any linear order). All these factors cooperate
with each other in a complex way, which either reinforces or weakens the strength of influence of any of them independently.
Maria Szupryczyńska (1980) also dedicated a study to the formal-distributive analysis of adjectives, focusing in particular on their connectivity to dependents (disregarding the collocations of participles, secondary with respect to verbs). She divided the connected segments into connoted (including those governed by an adjective [with case government], e.g. znany komuś, and not governed by it, e.g. znany u kogoś, wśród kogoś) as well as free ones (neither governed nor connoted, e.g. znany pod [nazwą]). For most adjectives complements are deemed optional (X jest miły dla kogoś – X jest miły), unlike in certain derivative forma-
tions which retain the case government of the base words, such as pochodny / zależny od czegoś, zgodny z czymś etc., as well as those for which these options entail a semantic difference głodny – głodny czegoś, ślepy – ślepy na coś. The analysis produced 19 classes of adjectives distinguished on the basis of case government; the government of the forms of comparative and superlative degree is analysed separately (comparative and elective schemes). Yet another description of adjective connectivity, applying the tools of the generative-transformative grammar, was proposed by Katarzyna Węgrzynek (1995). The basic rule for adjectival phrase is AP → (ModP) (AdvP) A (N7) (PP) (PP) (S2), where digits symbolize the degree of indirect category between the terminal expression (lexeme) and the phrase; the combination of all options gives a total of 70 possible models (e.g. AP → A S2 – lepszy, niż był wczoraj, AP → ModP AdvP A N7 PP – naprawdę bardzo wdzięczny ojcu za pomoc etc.). Due to the specific nature of the Polish language, horizontal trees, describing grammatical categories and accommodation dependencies, have been added to the vertical trees illustrating the syntactic structure.
