Justification of the twofold approach

By the twofold approach to the description of the significative system, a systematic distinction and equal consideration of the semasiological and the onomasiological perspective will be understood. The two approaches can also be called structural vs. functional, or somewhat more clearly, structure-based and function-based description, resp.

In the practice of the linguistic discipline, the twofold approach has played a very different role in lexicography and in grammaticography. Bilingual dictionaries have implemented the converse approaches for several millennia. Simultaneously, most monolingual dictionaries have been semasiological (essentially, alphabetic). However, onomasiological dictionaries (essentially, dictionaries of synonyms) have existed for two millennia, too, although the thesaurus as their consistent implementation was only invented in the 19th century. The usefulness of the twofold approach in lexicography has always been obvious, and nobody has ever doubted it.

Things are completely different with grammars. Traditional grammars have simply been unaware of the distinction between semasiology and onomasiology and mixed the two approaches unwittingly for two millennia. They typically start with the morphology of nouns and verbs in a semasiological perspective, just in order to arrive, towards the end of the grammar, at temporal clauses, questions and commands in an onomasiological perspective. Structural linguistics of the mid-twentieth century noticed the inconsistency and produced exclusively semasiological grammars. Some so-called discourse grammars published in the sequel pay attention to the functional perspective, though not consistently. The twofold approach was first postulated in the nineteenth century (Lehmann 1980). It was implemented in Nordhoff 2009, Ross & Sheng Ma 2014 and González Campos & Lehmann 2027. The latter work, in particular, illustrates the model proposed here.¹

A cleanly semasiological description is of no use if one compares languages with respect to the means they employ for the fulfillment of a given function of communication and cognition. If one wants to find out how one says, in this language, ‘X has a Y’ (something every language can say), one essentially has to peruse the entire grammar. Even worse, if the description shuns away from concepts of general-comparative grammar because it follows the doctrine of describing the language “in its own terms”, one will even look in vain for concepts such as prepositions or auxiliaries. An onomasiological grammar, on the other hand, does not answer these latter questions in principle.

A traditional grammar, and whichever grammar is not semasiological or onomasiological or both, cannot be consistent in its structure. The reason is that the grammar describes how the language maps expression onto content, but this is not a biunique mapping. A system following semantic structure is of necessity different from a system following expression structure. A description that only takes one of the two perspectives does not respond to half of the demands put by the user. The twofold approach is the only one allowing a comprehensive and systematic description.

Relationship of the two parts

In a systematic view, the onomasiological and semasiological description are complementary, symmetric and have equal weight in the overall description of a language. This equality is, however, limited, leading to an asymmetry between the two parts:

• The semasiological description takes the perspective of the hearer. It reflects passive knowledge of the language. Active knowledge of a language is never more comprehensive than passive knowledge; instead, it presupposes the latter.
• Although the object of both parts of the description are constructions, both the structural and the semantic properties of these are explained in the semasiological part. The onomasiological part describes fully the semantic properties of constructions, but refers to the semasiology for their structural properties.

General comparative grammar and the structure of a grammatical description

In principle, both the semasiological part and the onomasiological part of the description of a language are instantiations of a system of general comparative grammar. It is not necessary to describe the language “in its own terms”, inventing an ad hoc system for its description; the separation between a structure-based and a function-based system guarantees that justice can be done to the specific structure of this language.

However, it is not the task of a particular description to recapitulate the entire general comparative grammar.

• On the one hand, concepts underlying the structure of the description and used in it may require definitions to the extent they are not presupposed as known to the readership. Such definitions are taken from general comparative grammar. They may be specified according to the structure of the particular language. For instance, if it were necessary to define the concept of an enclitic, the definition would be universal. But if all enclitics of this language are verb enclitics, this may be added to the definition.
  At the level of academic writing, the complication here is that the description of a particular language is not an introduction to general comparative grammar. It may be advisable to shift definitions of terms to an appendix.
• General comparative grammar provides for many structural devices and many specific functions which are relevant to other languages but not in this one. If it is assumed that the reader may wonder about them, it may be mentioned that the language lacks them; but in general, it is unnecessary and impossible to enumerate them all.

The significative system of a language comprises lexicon and grammar. The boundary between these is fluid. Grammatical words such as Engl. have and to are both lexical entries and figure as operators in rules of grammar. Also, many grammatical regularities are limited to words of a certain class such as verbs of directed motion or inalienable nouns. Such lexical classes are therefore, at the same time, grammatical classes. As a result, there is, in the description of the significative system, no categorical division between lexicon and grammar. Both parts of the twofold description mention and illustrate classes such as those mentioned. Moreover, the onomasiological description mentions categories of general-comparative grammar such as recent past or non-specificity, in many cases only to say that the language under description has no grammatical means to code these and may instead code them lexically.

The role of complexity in the twofold description

Complexity is a structuring factor in both compartments of the grammatical description:

• In the semasiological grammar, each chapter devoted to a syntactic category consists of sections which climb the complexity hierarchy, from root and grammatical formative up to syntactic construction.
• In the onomasiological grammar, each chapter devoted to a notional category or operation starts with the most explicit, lexical-syntactic constructions and moves complexity levels downwards to the most grammaticalized constructions and formatives.

The place of grammaticalization in the twofold description

There are two kinds of places where grammaticalization becomes a topic of the description:

• In the semasiological grammar, the function of the formatives constituting a construction is described. This may fruitfully incorporate a diachronic perspective and, thus, an account of the formative's origin by grammaticalization. For instance, the account of the English prepositional phrase presents paradigms of prepositions of different degrees of grammaticalization. This is the place to mention that of is grammaticalized from off.
• In the onomasiological grammar, a function in a given functional domain may be fulfilled in different degrees of explicitness. The constructions affording this may be connected by grammaticalization. For instance, a left-dislocated topic may be integrated into the clause as a fronted topic.
  Since grammaticalization involves condensation of grammatical structure, leading from relatively variable high-level syntactic or even discourse constructions to tightly-knit morphological constructions, these are generally dealt with in different sections of a semasiological grammar. However, to the extent that the initial function is preserved, even in a diluted form, they can be treated together in a chapter of the onomasiological description.

In this way, the onomasiological grammar also makes it explicit to which extent a given notion or operation is grammaticalized in the language.

The place of semanto-syntactic operations in the twofold description

There are at least two topical areas which are hard to accommodate consistently in a grammatical description: Coordination and negation are operations that apply at different grammatical levels, in different clause types and obey constraints of all sorts. This situation produces the following alternative: either one chapter on negation which describes all constructions involving negation; or for each grammatical construction, an appendix on its negation. And the same for coordination.

Since onomasiological grammar is devoted to linguistic operations whereas semasiological grammar is devoted to linguistic constructions, the alternative is resolved by treating these topical areas in both compartments of the description:

• The onomasiological grammar contains a chapter each devoted to coordination and to negation, treating these as "logical" operations. Coordination is in the functional domain of junction.
• Consequently, the semasiological grammar contains no such chapters. Instead, each of its major chapters devoted to a grammatical category (adjectival, nominal, adverbial, verbal ...) may contain a section on negating a syntagma of this category and a section on coordinating syntagmas of this category, if this is possible in the language. Such a treatment may miss out possible generalizations in case coordination or negation work the same for syntagmas of different categories. Such generalizations aren then formulated in the onomasiological chapter.

¹ A twofold grammatical description comprising a “form driven approach” and a “function driven approach” is also postulated in Payne 2014. However, the conception there, especially as regards the onomasiological approach, differs substantially from the present one.