Recursivity, Derivational Depth and the Search for Old English Lexical Primes

Abstract

The aims of this article are to coin the term of derivational depth and to assess the role of zero derivation in the search for the lexical primes of Old English. Derivational depth is characterized diachronically as making reference to the (un)productive derivational processes that function as input to productive derivational processes. As regards the related concept of lexical recursivity, it is defined as the derivation of derived bases, which, in Old English, can be the case with the output of processes of zero derivation, affixation and compounding. In the search for the lexical primes of Old English, evidence has been found for three types of zero derivation, including zero derivation with zero inflection, with full inflection and with formative. The main conclusions of the article are that lexical recursivity is a linguistic property, whereas derivational depth represents a property of lexical representations; that zero derivation constitutes a principled formal criterion for defining lexical primes; and that nominal zero derivatives of strong verbs play a role central to lexical derivation in Old English.

Acknowledgement

This research has been funded through the project FFI2011-29532.

Notes

1. Although substantially revised and modified, the information gathered in the lexical database of Old English Nerthus draws on Bosworth and Toller's An Anglo-Saxon Dictionary, Clark Hall's A Concise Anglo-Saxon Dictionary and Sweet's The Student´s Dictionary of Anglo-Saxon. On specific queries, The Dictionary of Old English has been consulted. Overall, Nerthus contains a full lexical inventory of the language consisting of around 30,000 lexemes (with more than 33,000 alternative spellings), all of which are provided with information on inflectional morphology, derivational morphology and gloss.

2. Numbered predicates are used throughout the article to stress the categorial or morphological contrasts holding between homonymous predicates with the same lexemic root. For instance, ābūtan 1 ‘on, about, around, on the outside, round about’ is an adposition and ābūtan 2 ‘about, nearly’, an adverb; andfenge 1 ‘acceptable, agreeable, approved, fit, suitable’, while besēon 1 ‘to see, look, look round’, is a class V strong verb, and besēon 2 ‘to suffuse’ a class I strong verb. See Martín Arista (2010a) on this question.

3. Derivation by inflectional means is a type of zero derivation in this analysis. The lexicalized meaning of byrstende ‘roaring’ indicates that we are not dealing with the productive inflection of the present participle of (ge)berstan ‘to break, burst, fail, fall; break away from, escape; break to pieces, crash, resound’.

4. Compounds, which are not discussed in this article, belong in at least two derivational paradigms. This is the case with the feminine noun mūðberstung (MŪÐ-BERSTAN) ‘eruption of the mouth’, the masculine tēmbyrst ‘failure to secure a voucher’ (TĒON 1-BERSTAN) and the adjective feðorbyrste ‘split into four’ (FĒOWER-BERSTAN).

5. Several scholars, including Kastovsky (1992), Lass (1994) and Quirk and Wrenn (1994), stress the difficulty of assessing the productivity of derivational processes in a historical language like Old English. Quirk and Wrenn (1994: 104), for instance, remark that “it is often impossible for us to distinguish processes that were active and flourishing during the Old English period from those which had ceased to be formative before the Anglo-Saxons left the continent of Europe.” For this reason but also because the methods proposed for measuring the productivity of the processes active in the current language are aimed at deciding on the likelihood of new creations (Baayen and Lieber 1991; Baayen 1993), productivity is assessed indirectly on the basis of type frequency.

6. It follows that for such a choice to exist wuldor is of necessity a derivative of (ge)wealdan.

7. According to Williams (1981: 261), X can be related to Y if Y and X differ only in a head position or in the nonhead position. Pounder (2000) focuses on the paradigmatic effects of relatedness and defines a lexical paradigm as consisting of relations between word-formations that share a lexemic root.

8. While the seminal work by van Marle (1985) demonstrates the relevance of relatedness and paradigmatic structure for morphological theory, the psychological basis for relatedness and the stepwise processing of complex words are discussed, among others, by Marslen-Wilson et al. (1994) and Wurm (1997). Processing restrictions on the complexity of derived terms are identified by Hay and Plag (2004). Other types of restrictions compatible with the ones imposed by processing include the pragmatic restrictions put forward by Lieber (2004) and the general principles of semantic organization that require that core meanings take up more central positions and peripheral meanings occupy less central positions (Hay 2002, 2003).

9. For a detailed explanation of the linking meaning-form in word-formation, see Martín Arista (2008, 2009, 2011a).

10. For Schuldt (1905), the strong verb hrēowan constitutes the source of the weak verb hrēowsian ‘to be sorry’ and the feminine noun hrēow ‘sorrow’. Cf. Hallander (1966: 374), who holds that the adjective hrēow ‘sorrowful’ is the base of derivation of hrēowsian but admits that the evidence for the adjective hrēow ‘sorrowful’ is scarce.

11. This article is not meant as a theory-internal discussion. There are, nevertheless, two aspects of current debate in functional theories of language to which the LSW contributes and which deserve mention at this point. The first has to do with the ability of systems of lexical representation to capture the productivity of derivational processes and the second with the Completeness Contraint (Van Valin and LaPolla 1997: 325), which requires that all external variables in an underlying representation must be realized syntactically whereas internal variables may not be coded explicitly. See Butler (2003: 219ff) for more information.

12. Whereas Giegerich (1999) speaks of lexical strata, his approach is strictly synchronic, given that different strata account for restrictions on affixation due to the properties of the bases of derivation. Kastovsky (1992) also uses the term lexical strata to refer to external variation, but it refers to internal variation in this research. Kastovsky (1968 and subsequent work cited throughout this article) deals with some of the phenomena discussed here in terms of a typological change occurring in English from variable base morphology to invariable base morphology.

13. More quantitative evidence is offered in section 3, which discusses specific formations.

14. The data provided by Haselow (2011), although restricted to nouns, also indicate that zero derivation is outnumbered by affixation in Old English.

15. The demise of inflectional endings represents the end of zero derivation as understood in this work. Cf. Haselow (2011: 281) on the reduction of the number of derivational morphemes.

16. See, for instance, Brinton (1986) and Brinton and Traugott (2005), who explain the recursive attachement of preverbs as a case of grammaticalization of locative adverbs and prepositions motivated by the loss of semantic weight and subsequent lexicalization of the Germanic prefixes in the series ā-, be-, for-, ge-, of-, on-, and tō-.

17. This stratum might also be seen as the destination of lexicalized items, but this aspect needs further research. See Martín Arista (2010a, 2010b, 2011a, 2011b, 2011c, 2012) on derivational paradigms in the lexicon of Old English.

18. Torre Alonso (2009, 2010, 2011a, 2011b) distinguishes between recursivity proper, occurring when a certain morphological process is fed by the same process, from process feeding, holding if a process is fed by a different process, as in zero derivation providing the input to affixation, for instance.

19. Lexical paradigms are labelled after their exponents. For instance, the exponent of the lexical paradigm (GE)RĪSAN is the strong verb (ge)rīsan. Strong verbs, in turn, are preceded by the prefix ge- between brackets to indicate that this prefix is attached strong verbs, unlike weak verbs, on an inconsistent basis (see Martín Arista 2012). When another pure prefix like ā- is attached, ge- occupies the leftmost position. This means that the following strong verbs are listed in the lexical paradigm of (GE)RĪSAN: rīsan, gerīsan (< rīsan), ārīsan (< rīsan) and geārīsan (< arīsan).

20. Zero derivatives with suffixed bases, as in hīwcūðlic ‘known’ > hīwcūðlician ‘to make known’, are not required if weak verbs have basic status, at least exceptionally. A derivation such as hīwcūðlician ‘to make known’ > hīwcūðlic ‘known’ seems more plausible, but forms the adjective on the weak verb. See Kastovsky (1992: 392) on what he calls misanalysis of zero-derived forms.

21. A good example of this stance is Ettmüller´s (1968) Lexicon Anglosaxonicum.

22. This typology has the advantage of insisting on the continuity between inflection and derivation that is, to some extent, typical of a language with variable base morphology. See further Kastovsky (2006) and Haselow (2011).