• Keine Ergebnisse gefunden

1.4.5.1. Graphematic structures in written language

Word division in writing necessarily involves graphematic units. Over the last twenty years there has been a growing appreciation that writing systems have their own internal structure which run parallel to, but may be independent of, those of prosody, morphosyntax and semantics (cf. Evertz 2018, 2–3). A graphematic hierarchy has been proposed by analogy with the prosodic hierarchy in spoken language (Evertz & Primus 2013; Evertz 2018). Thus, the graphematic word 〈ω〉 may be said to be made up of graphematic feet 〈Σ〉, which in turn comprise graphematic syllables 〈σ〉 etc., per (40):

(40) 〈ω〉 >〈Σ〉 >〈σ〉

1.4.5.2. Identifying graphematic words

The present study is primarily concerned with the denotation of the highest element of this hierarchy, the graphematic word. A prerequisite of investigating graphematic words is, of course, to identify them, that is, by specifying the means by which they are bounded.

Most recent discussions of the graphematic word have considered writing systems that separate graphematic words by means of spaces. Thus Evertz (2018, 21) defines the graphematic word as follows (for the same or similar defnitions, see Cook 2004, 42; Evertz & Primus 2013, 2; see also for Tiberian Hebrew Dresher 1994):

[A] g[raphematic]-word is a continuous sequence of letters bordered by spaces.

The Northwest Semitic texts considered in the present study do not, for the most part, make use of spaces to separate graphematic words, preferring various kinds of interpuncts, viz. graphemes comprising a number of dots. In the earliest linear alphabetic texts, word division was marked by a short vertical stroke (Naveh 1973b, 206), but dots and spaces are used in later texts (Naveh 1973b, 207; Lehmann 2016, 37–38*). In Ugaritic alphabetic cuneiform, word division is indicated by means of the small vertical wedge (Ellison 2002).

The fact that the same level of unit can in principle be demarcated either by dots or by spaces may be seen by comparison of the Hebrew Bible texts from the Dead Sea Scrolls (DSS), and the Siloam tunnel inscription. In these two sources, the same

word-level unit – as I will argue, the minimal prosodic word–is demarcated by means of spaces in the DSS, but dots in the Siloam inscription (see Chapter 12).

For the purpose of the present monograph, therefore, I take the graphematic word to be:

Any sequence of characters separated from surrounding characters by spaces or interpuncts.

1.4.5.3. Writing systems without graphematic words

Many texts from the ancient world do not separate word-level units at all. These texts are said to be written in scriptio continua. This is especially the case in many Phoenician texts, as well as later Greek inscriptions. Since these texts do not have units corresponding to the graphematic word, they do not fall within the scope of the present study.

It is worth pausing to make two observations. First, it is sometimes claimed that scriptio continua is possible only in the context of writing with vowels, and that writing without vowels necessitates word division. Thus Saenger (1997) in his seminal work on word division in the Middle Ages in Europe asserts (p. 9) that:

The uninterrupted writing of ancient scriptura continua was possible only in the context of a writing system that had a complete set of signs for the unambiguous transcription of pronounced speech. This occurred for the first time in Indo-European languages when the Greek adapted the Phoenician alphabet by adding symbols for vowels … Before the introduction of vowels to the Phoenician alphabet, all the ancient languages of the Mediterranean world – syllabic or alphabetical, Semitic or Indo-European – were written with word separation by either space, points, or both in conjunction. After the introduction of vowels, word separation was no longer necessary to eliminate an unacceptable level of ambiguity.

This is not in fact the case: scriptio continua is attested well before the arrival of vowel writing in the Greek alphabet. For instance, certain Ugaritic texts from the late 2nd millennium BCE are written without word separation (Tropper 2012, 69). Many Phoenician texts are also written in scriptio continua (Steiner 2016, 330).

It is worth adding an additional caveat. Lehmann (2005; 2016) has shown that Phoenician and Aramaic inscriptions that had been thought of as being written in scriptio continua are, in fact written with spaces separating graphematic words. It is entirely possible that other inscriptions previously thought to have been written in scriptio continua are in fact written with words divided by spaces. This is a topic that will no doubt be pursued in future research.

1.4.5.4. Minimal graphematic words

In addition to defining the graphematic word in terms of its boundaries, it may also be defined according to its subcomponents under the graphematic hierarchy (§1.4.5.1):

(41) A graphematic word ‘consists of at least one graphematic foot, which in turn consists of at least one graphematic syllable’ (trans. from Evertz 2016, 392).

Under this definition, a graphematic word must minimally consist of at least one graphematic syllable. What constitutes a graphematic syllable will obviously be writing-system specific. For English and German, Evertz (2018, 47, with references) adopts the following definition:

(42) Every g[raphematic]-syllable has a v[owel]-letter in its peak.

For English and German, therefore, the definition of the minimal graphematic word is dependent on the definition of the vowel letter: if a grapheme is a vowel letter, it will satisfy the definition of graphematic wordhood. Vowel letters can be defined in several ways:

• Phonologically, that is, with reference to correspondences phonological vowels (Evertz 2018, 47);

• Definitionally, simply by listing the set of vowel letters, e.g. 〈a, e, i, o, u〉 ;

• Distributionally, by comparing which graphemes may be substituted for one another;

• Graphetically, by comparing the features of letter shapes.

The particular definition of the vowel letter adopted for German and English is not important for present purposes. However one chooses to define the vowel letter, the corollary of the definitions (41) and (42) above is that a minimal graphematic word is a vowel letter. This is borne out by the fact that both 〈a〉 and 〈I〉 (albeit capitalised) are valid graphematic words. Conversely, the formulations make the prediction that a single consonant letter does not constitute a valid graphematic word, a result borne out by the absence of any such words in English or German.19

Of course, simply meeting the definition of minimal graphematic wordhood does not guarantee existence as a graphematic word: for existence as a graphematic word, a morpheme corresponding to the grapheme sequence must exist in the morpheme inventory of the language represented.

This definition of minimal graphematic wordhood need not apply to writing systems in general. Indeed in English it has been argued (Evertz 2016) that minimal lexical words can be defined in terms of graphematic weight. A writing system could in principle adopt such a definition for all graphematic words. For example, a minimal graphematic word could consist of at least one graphematic syllable.

19 This excludes, of course, abbreviations marked by a final period. These, of course, cannot stand on their own between spaces, and so are not graphematic words in the same sense that a word such as a is.

We will explore a purely graphematic explanation for graphematic wordhood in Tiberian Hebrew at §10.4 below, exploiting the notion of the minimal graphematic word, before ultimately rejecting such an explanation in favour of a prosodic one.