The lexicalisation of syncope
Syncope is a prosodically motivated process removing medial vowels in longstemmed disyllabic Old English adjectives e.g. <i>hālig</i> ‘holy’ when inflected: nom/acc.pl.neut. <i>hālgu *hāl</i><bi>i</bi><i>gu</i>. Syncope should not affect short-stemmed forms such as <i>hefig</i> ‘heavy’ e.g. nom/acc.pl.neut. <i>hef</i><bi>i</bi><i>g *hefgu</i> (Campbell 1959). However, forms such as <i>hālig</i> include derivational <i>-ig,</i> and produce surprising results. in relation to syncope, showing both over- and underapplication not attested in morphologically simple words. Two historical <i>-ig</i> suffixes exist: <i>ig¹</i> (vulnerable to syncope) and <i>-ig²</i> (immune to syncope). I will show how this interaction of syncope with <i>ig¹</i> results in a case of lexicalisation. I argue that cases of overapplication of syncope in forms containing <i>-ig¹</i> e.g. short-stemmed <i>hefgu</i> indicate the lexicalisation of syncope in forms exhibiting <i>-ig¹</i>. Syncope becomes associated with <i>-ig¹</i> to the extent that the affix itself becomes reanalysed as /j/ from /ij/ underlyingly. The original phonological conditions for syncope therefore cease to apply (Anderson 1989).