The article addresses Estonian syllabification as reflected in practical word inflection. According to the traditional view the nucleus of an Estonian syllable is invariably a vowel, while the next syllable begins with the last consonant of an intervocalic consonant sequence. There are, however, some foreign names like Google [ɡu:ɡl], for example, which despite being monosyllabic in terms of the traditional definition of the Estonian syllable can only be declined like a two-syllable consonant-final word, e.g. PartSg Google’it. In addition there are several disyllabic words of the second quantity degree, i.e. nouns of the C*VC-CV structure like ratsu ‘steed’, where in parallel with the expected PartSg ending -t (ratsut) a zero-ending form is used, as if the word had a C*VCCV structure (like taba ‘padlock’), e.g. PartSg käbla ‘sl. a fall’, mobla ‘sl. mobile phone’, sebra ‘zebra’, mügri ‘water vole’.
The article calls for a revision of the current rule of Estonian syllabification, arguing that Estonian syllabification practice is similar to the German one, where the syllable nucleus need not always be a vowel as the function can also be fulfilled by a voiced consonant, and it is not only a single consonant, but possibly also a stop-initial consonant sequence that can be considered as the onset of a non-initial syllable.
The arguments are supported by the available information on the declension of the relevant words in practical usage, some since the 19th century. The underlying logic of our approach is that language users tend to base their morphology or word inflection on phonetic facts. Their idea of how a word should be syllabified is one of the factors determining their choice of the declension of the word and thus those choices may a posteriori reveal the underlying language-specific syllabification rules.