The aim of the present article is threefold: to examine certain problems inherent in dictionary defining; to discuss the most important changes that have been implemented as solutions to some of the problems; to evaluate the new problems which have arisen as side effects of the solutions. Finally, the historical precedents of a number of the alternative defining techniques are also considered, in an attempt to put the issue into perspective.

The Sorbian Seminary came into being in Prague at the beginning of the 18th century to educate Catholic clergymen. In 1846, the students at the Seminary founded the Serbowka association and began to keep journals as well as produce the handwritten Kwětki almanac. These two sources were used as the basis for an analysis of the language – to be more precise, of the lexicon – used by the members of the association. Pful’s dictionary, published in 1866, served as a point of reference for an analysis of the data collected.
The juxtaposition of the language material gathered in the study enables us to observe a great degree of conformity between the lexis used by the Serbowka members and the vocabulary recorded by Pful.
What is more, in the yearbooks of the Serbowka and in Pful’s dictionary we can notice a large proportion of loanwords from the Czech language, both older and more recent. This is connected with the attitude at that time towards the renascent language. In the lexicon of the Serbowka members the proportion of bohemisms (or interference from the Czech language) is much greater, which is a result of direct and close contact with the Czech language.
 

Tajik, as opposed to Fārsi and Dari, remained for a century strongly influenced by Russian. As a consequence, its lexicon abounds with borrowings from that language. The article deals with the problem of their pronunciation – are characteristic features of Russian phonology and phonetics preserved in these loanwords? Having analyzed a number of examples one notices that the pronunciation of such words is far from consistent and the idea of a fidelity level may be introduced to explain and classify the differences. This fidelity level depends on various factors, e.g. the education of a native-speaker.

The paper proposes a multi-dimensional, phonologically-aware numeric encoding of Turkish for use with neural networks. The system is evaluated and compared to PatPho (Li/MacWhinney 2002) in a test in which the network computes the shape of the past tense suffix.
 

In this study an analysis of the phonetic adaptation of Arabic and Persian loan-words in Ottoman Turkish is continued (for the vocalic part of the analysis see Stachowski M. [forthcoming]). Five phenomena are presented in the context of the general Turkic phonetic evolution. These are: [a] palatalization of (-)kE- > (-)čE-; [b] varying anlaut nasality: m- > b- and b- > m-; [c] despirantization of f > p; [d] epenthetic n; [e] shortening of geminates.