Many grammatical constructions in the world’s languages contain (grammatical) markers, i.e. (generally quite short) forms that cannot stand on their own (i.e. they are bound forms), that cannot be focused (i.e. that cannot be the crucial aspect of an answer to a content question), and that are generally said to (help to) express grammatical meanings. For example, many languages have case-markers, tense markers, relative-clause markers (relativizers), person markers (indexes), number markers, definiteness markers (articles), and so on.
Now it is well-known, of course, that grammatical meanings are often expressed by the absence of a marker. For example, in the following pairs of English expressions, the first contains a marker (in boldface), while the second lacks a marker in the same “slot”, but it still seems that a grammatical meaning is intended (and can be safely inferred).
(1a) I will work here.
(1b) I work here ( > present-tense meaning)
(2a) We bought the flower-s.
(2b) We bought the flower (> singular meaning)
(3a) She help-s me.
(3b) They help me (> 3rd person plural “meaning”)
Linguists often say that there is a “zero marker” in such situations (a zero affix, a zero morph(eme), a zero vocabulary item, etc.), and this way of talking has a venerable tradition, going back at least to Jakobson’s (1939) paper on “zero signs”.
The purpose of this blogpost is not to argue that anything is fundamentally wrong with using zeroes in describing languages, but I want to make an important terminological point: Expressing a meaning through the absence of a form is best described as zero expression, rather than by invoking a “zero marker”.
The distinction may appear subtle (or even pedantic), but the basic idea is that every technical term of linguistics should be clearly definable (if it not a basic term) – I hope that this is not controversial. So here is a definition of the term grammatical marker (or simply marker, as there are no other kinds of markers):
A grammatical marker is a minimal linguistic form that cannot occur on its own and that expresses (or helps to express) a grammatical meaning.
Leaving aside the definition of “grammatical meaning”for now (cf. Boye & Harder (2012) for some insightful discussion), the definition contains the elements “minimal” and “form”. The former is intended to exclude elements that consist of two or more forms (i.e. compound elements), and the latter is perhaps the most basic term of linguistic analysis: A linguistic form is a recurrent combination of shape (sound shape in sound languages, or hand shape in sign languages) and meaning.
Given this definition of linguistic form, it follows that a marker cannot have no shape. One might consider changing the definition of “marker”, or the definition of “form”, in order to allow zero markers. But I cannot think of a good way of doing this. The problem is not only that the definition would end up being disjunctive (which is ugly and should always be avoided, especially for rather basic concepts).
The main problem is that if one removes the “shape” part of the definition of a linguistic form, there is no longer anything that the meaning can combine with – and the combination of meaning and shape is the most basic aspect of language.
It is not mysterious, of course, how semantic inferences can be made even in the absence of forms/shapes: from the context, specifically the grammatical (= constructional) context. If we hear someone say the flower, it’s clear that a single flower is meant, but not because there is a “zero shape” that is somehow paired with a singular meaning. Rather, it is because English possesses a highly specific construction [DEF N(count)], which contrasts with [DEF N(count)-s], and which means ‘singular count noun N’. Thus, the singular meaning is a constructional meaning, similar to the constructional meaning ‘patient’ of the postverbal NP in English, or the constructional meaning ‘extent’ of the pre-adjectival NP in expressions like three meters wide.
In many contexts, it may be useful to pretend that there is a slot that does not contain a pronounced element. I myself often use the symbol “Ø” in asymmetrical coding situations like (1)-(3) above, to make clear that there is a marker in the flower-s, but that there is no marker in the flower-Ø.
(One may also have zero notational elements for other constructional meanings, such as the strange little “v” in MGG notation, which seems to somehow stand for the agent role of the subject nominal. If the exact role of this notational device were clear, there would be no problem with it – we also use hyphens and brackets, which are not pronounced either, so nobody can claim that they don’t use “zero elements” in their notation.)
However, linguistic forms are different from notational devices that help us talk about the complexities of constructions. A form is something that can be pronounced and has a meaning, and a marker can only be defined as a special kind of form. Unfilled slots are not forms, and so they cannot be called “zero markers”. Meanings may be conveyed by not using any marker (so we may talk about “zero expression”), but forms always have a shape, so there are no “zero forms”.