PGmc. *-ārjaz was borrowed from Latin, and became a productive masculine agentive suffix in the older Germanic languages, gradually displacing various native agentive suffixes. Despite some assertions to the contrary, OE -estre can be shown to have been a feminine agentive suffix of Germanic origin that was also applied to certain feminine animal names. During the Old English period, a pattern obtained whereby masculine agent nouns were frequently derived using -ere (< PGmc. *-ārjaz), while the corresponding feminine agent nouns used -estre. This system was partially preserved in Modern Dutch, but died out in English as -er was increasingly used to form agent nouns of both genders.*

Although it has long been recognized that the i of OHG gi-could be lost before vowels and certain consonants, the handbooks recognize no loss of i in bi-. This is one reason why many scholars are reluctant to derive OHG irbarmên from *ir-bi-armên and irbunnan from *ir-bi-unnan, instead positing an original prefix ab- and comparing semantically similar forms such as OE ofearmian. In fact, however, loss of i in hi-, though highly sporadic, is attested before vowels and before r and l from the 8th century on, and provides a basis for the etymology of two hitherto unexplained words in Notker: ferbrasên and prázelig. There are thus clear parallels for a similar explanation of irbarmên and irbunnan.

This paper examines four dative reflexive verbs of motion in Old Saxon:faran, gangan, kuman and giuuitan. The linguistic universals of GB syntax and the universal phenomenological categories of Peircean semiotics provide the theoretical framework for this paper. Various syntactic and semantic aspects of the dative reflexive verbs are targeted in the analysis: binding of the reflexive, word order patterns and subject types. The article suggests that a link exists between subject types, V1 syntactic configurations and the appearance of the dative reflexive — only human, animate subjects can license a dative reflexive and the reflexivized verbs exhibit a strong tendency to appear in the narratively visible V1 structure.

The verbal particle durch shows two stress patterns: stressed and unstressed. These are correlated with two different construals of the relation of a Trajector to a Landmark. When the relation is construed to be sequential from one side of the Landmark to the other, the particle is stressed. Non-sequentiality is correlated with unstressed particles.