1. Introduction Not every historical linguist embraces the idea of Chomsky's syntactocentrism with enthusiasm. It may be untimely to say unkind things about it, but there are syntactic problems which cannot be resolved satisfactorily only by formal operations. Under the current psycholinguistic views there seem to be some chances of recognizing the old conceptual world of the speaker and thus contributing to a more appropriate understanding of the writings he has left. Following chiefly Jackendoff's ideas expressed in The architecture of the language faculty (1997) -- yet with due respect for other linguistic and psycholinguistic orientations -- I will discuss grammatical relations which involve word order, thematic roles and word-formation (compounding) and which by structural standards prove so intractable. A common trait of them all is that they are structurally ambiguous and consequently differ in meanning, or that they are simply semantically opaque. 2. Word order An example of how weakly significant word order in Old English can be is the first part of the following sentence: 1) Storm oft holm gebringep, geofen in grimmum selum (Maxims I 112/50) which has been understood as either 'The sea often brings a storm, the ocean in stormy seasons' (Gordon 1954: 342) 'The sea often brings a storm' (Bosworth, entry gebringan) or 'The often brings forth a flood' (Reszkiewicz 1971:35) 'storm oft brings ocean into a furious condition' (Bosworth, entry soel) The interpretative difficulty lies in the fact that the functions of a grammatical subject and a grammatical object are not clearly transparent: the nouns and holm are both singular and each can agree with the finite form of the verb, gebringep, which as a two (or even three) argument requires a subject and an object. This brings up a question: which is which? Structurally speaking each can perform either function. They are both masculine, singular, of a-inflection of which nominative/accusative syncretism is a norm. Besides, there is no adjectival or pronominal modifier to help, neither can alliteration be helpful. Reszkiewicz searched for a clue to the functional identification in the position of the noun with regard to the and came to the conclusion that: Older Old English, especially poetry, lacked both the definite and the indefinite articles; the object often preceded the governing verb (Reszkiewicz 1971: 35). Although the grounds on which such a decision is reached are formally defens ible, empirically they are less so as they can be falsified by a sentence, also a gnomic verse, which reads: (2) Moegen mon sceal mid mete fedan (Maxims I 118/44) in which it is the subject man and not moegen which is closer to the finite form of the verb, sceal (moegen and man also show inflectional syncretism in this respect); this sententious saying means: 'One shall nourish strength with meat' (food) (Gordon 1954: 344) 'A man must feed strength with meat' (Bosworth, entry fedan) The proponents of either of the two meanings of the gnomic storm verse would probably try to persuade us that their views are compatible with the formal grammatical relations. But which of the meanings would satisfy the pragmatics of the discourse? Although the senses of particular lexical items are clear, a real cognitive image is still concealed. As a historical linguist I am more comfortable asking questions than answering them, so my glimpse into the Old English cognitive mind will be based on the possible, we now try to see, life as it would have been over a millenium of years ago. Since the conceptual structure of our example is not immediately predictable from the syntactic structure, nor is it found in the lexical structures, I will try to consider the language context first and then to search for similar uses of and holm. …