The terms superstrate and substrate seem to lose some of their predictive force in the earliest documented European contact language. While Lingua Franca (LF) is best known for its purported origins as a trade pidgin used across all areas of the Mediterranean, the vast majority of actually available LF documents come from slave colonies on North Africa’s “Barbary Coast’’. Here, the social and the linguistic data do not synchronically coincide according to the usual creolist framework, where lexifier and superstrate are largely treated as being synonymous. Algiers held the largest prison colony where, according to contemporary reports (Haedo 1612), some 25,000 people lived, all of them speaking LF with different proficiencies. Masters in these prisons were Arab and Turkish Moslems; slaves were the captured European Christians. Most of the lexicon of LF is derived from Romance languages, therefore the bulk of the lexifier is that of the oppressed, not of the oppressor. This situation is shown in the major documents of LF, such as the 1830 Dictionnaire de la Langue Franque ou Petit Mauresque. Bakker’s principle – pidgins often do not use the lexicon of the dominant group – being borne out by available LF data, the possible (social) causes for this difference to the majority of the well-known contact languages coming out of slavery contexts will be discussed.