2.3 Limitations on which semantic units can be chosen as predicates
In a fully transparent language, only pragmatic and semantic information should determine the choice of formal units. Under such circumstances, all semantic units should be usable to form predicates, no matter whether they concern events, individuals or properties.
A few words of introduction into the Esperanto morphology may be helpful to understand this paragraph and 3.3 below. In Esperanto, lexemes as the basic semantic units are clearly discernible, but their status is still a debated issue. This is to some extent due to the fact that the first Esperanto vocabularies were defined as lexeme sets with word translations in five languages with differentiated parts-of-speech systems, i.e. English, French, German, Polish and Russian (see both Zamenhof 2004/1887 and 1963/1905). Most lexemes are bound lexical stems and cannot be used morphosyntactically without attaching to them the appropriate inflection that marks them for the slot in the clause they are destined to fill, i.e. the head or modifier position in a nominal or verbal phrase. Simply equating nouns, adjectives and verbs in the translations to their corresponding lexemes in Esperanto made it tempting for grammarians to classify these as specialized or categorial. According to rule 7 of the original grammar in Zamenhof (2004/1887), adverbs make use of the same lexemes as adjectives and differ from these only in their specific inflection, which becomes -e in lieu of -a. In other words, the Esperanto parts-of-speech (p.o.s.) system looks at first sight like a ‘type-3 flexible’ system, using specialized lexemes for nouns, verbs and modifiers (Hengeveld and Mackenzie 2008: 228).
However, this specialization is contradicted as early as in the language’s baseline documents themselves. The apparent verb-class specialization of certain lexemes would suggest that these are the only ones destined to fill predicate positions in a clause. The original presentation of the grammar points in this direction by showing the application of the lexemes far ‘do’ (see 4 below; the infinitive is fari) and kant ‘sing’ (see 5 below; the infinitive is kanti):
(Zamenhof 1994/1887: 37, 38)
Copula support is also evidenced among the first examples as illustrated by:
(Zamenhof 1963/1905: 85)
On the other hand, Zamenhof (1963/1905) shows that predicate positions are not obligatorily filled by lexemes that are translated as verbs. Among the 55 pages of practical exercise material in this baseline document we find reĝ-i, hero-i, sign-i, nom-i with the infinitive marker and nom-is with the past tense marker. They are associated with the lexemes reĝ ‘king’, hero ‘hero’, sign ‘sign’ and nom ‘name’, all given in these noun translations. We also find kuraĝ-as and san-i with the present tense and infinitive marker respectively, associated with the lexemes kuraĝ ‘brave’ and san ‘healthy’, both given in these adjectival translations (hyphens have been added in all cases for greater clarity). The indiscriminate use of lexemes in predicate positions, independent of what their word translations in other languages may be, has increased over the years, as may be illustrated by the following examples. In (7) the lexeme makul, translated as the noun ‘stain’, and in (8) the lexeme verd, translated as the adjective ‘green’, are both highlighted in bold print as they appear with the verbal tense endings as and is:
(Zamenhof 1925: 64)
(Waringhien 2002: 1224)
This trend extends to other lexical units, e.g. the temporal adverb baldaŭ ‘soon’ (which is a free lexical stem) in (9) below:
(Monato 2003/4: 14)
However, alternative solutions with copula support are often perceived as more normal by many speakers, probably because of their mother tongue habits. Examples of such constructions are provided in (10) with the predicative noun makulo contrasting with (7), and in (11) with the predicative adjective verda contrasting with (8):
(Waringhien 2002: 704)
(Zamenhof 1992: 85)
Frequently, predicates built on lexemes that are not translated as verbal words, have acquired a meaning that cannot be inferred from a straightforward conversion to a copula construction. Starting from lexemes with a nominal translation: the verb akvi from akv (akvo is ‘water’) appears to have the single definition ‘provide plants with the water they need’ (Waringhien 2002: 66). The verb loki from lok (loko is ‘place’) shows a variation on this theme: ‘to find a place for something or somebody’ (in other words ‘provide a place’ (Waringhien 2002: 689)), but also has a second meaning, which is ‘to put in a place’ (whereas akvi definitely does not mean ‘to put in the water’). Also the verb patri from patr (patro is ‘father’) has two quite distinct definitions: ‘to be a father’ and ‘to be like a father’ (Waringhien 2002: 849). The three verbs have acquired specific, more elaborate meanings, which are not automatically inferrable from a copula construction like ‘to be water/place/father’.
Speaking of lexemes with an adjectival translation: the verb beli from bel (bela is ‘beautiful’) is decribed as ‘to look beautiful’ (Waringhien 2002: 148). The verb rapidi from rapid (rapida is ‘fast’) shows three options: ‘to try and reach a destination in a short time’, ‘to complete an act in a short time’ and ‘to do something without delay’ (Waringhien 2002: 952). The verb verdi from verda ‘green’ (see also 8 above) means ‘to be intensely green’ (Waringhien 2002: 1224). The definitions of beli and verdi can be seen as intensifications of the more neutral copular constructions to be beautiful and to be green. Rapidi is different in that it displays three very specialized applications, without reference to a generic ‘to be fast’, which can, however, be inferred from the three as a kind of common denominator (though not explicitly mentioned in Waringhien 2002).
The observations above illustrate two trends, which point in opposite directions. On the one hand, the spontaneous and growing use of all kinds of lexemes in predicate positions, which can be seen as an attempt to exploit a transparency that was potentially present in the language, but initially hidden due to the extensive calquing of syntactic models by speakers of Esperanto’s Indo-European reference languages. On the other, the meaning that was assigned to many of the new verbs, which goes far beyond that of the paraphrase based on the associated copula construction. It is often unpredictable to the point of being idiomatic and only explainable as a semantic calque from one or more reference languages. This increase in semantic indeterminacy is, however, not the issue under debate in this chapter.
In allowing all kinds of semantic units to be used in predicate positions, Esperanto is transparent.