Preaspiration in the Nordic Languages: Synchronic and Diachronic ...

(although there are several tokens of each). The words are mot ‘against,’
åt ‘(un)to’ and utan ‘without, instead’ and will henceforth be referred to
as mot type words.
The third group (exemplified by vatten ‘water, lake’) contains
content words in which a fortis stop occurs in a VC syllable with primary
or secondary lexical stress. Such conditions are met in words such
as upp ‘up,’ vatten, fortsätter ‘continues’ and bukt ‘bay.’ These words
will be referred to as vatten type words.
The fourth group (exemplified by att ‘that, to (inf. marker)’) contains
function words in which a fortis stop occurs in a VC syllable. There are
five words (each with several tokens) in the present data that have been
assigned to this group: att, ett ‘a (indef. article),’ också ‘also,’ nåt ‘something,’
and liksom ‘like.’ Such words will be referred to as att type
words.
The fifth, and last, group (exemplified by norrut ‘northwards’) contains
words in which a fortis stop occurs in a lexically unstressed syllable.
There are both content and function words in this group and the
syllable type may be either VC or VC. Examples of such words include:
norrut, havet ‘the sea,’ kommit ‘(have) come’ and inåt ‘inwards.’ Such
words will henceforth be referred to as norrut type words.
In some cases, a stop that occurs word-medially in a postvocalic
context is excluded from the analysis. This applies to contexts in which
the stop occurs after a morpheme boundary (e.g. in påtår ‘second helping’
and otaliga ‘many, uncountable’) and in which the stop occurs before
a stressed vowel (e.g. hotell ‘hotel’ and afrikanskt ‘African’). These
types of context trigger the production of the type of stop that occurs
word-initially rather than word-medially.
4.2.3.1 Data overview
In total, the five word types discussed in the preceding section
yielded 1051 sequences of CSw word-medial and word-final vowel +
fortis stops (see Table 4–4). In 960 cases, the subjects produced a full
stop closure, allowing measurements of preaspiration and, if a vowel followed,
postaspiration. In 48 cases, the subjects did not produce a full closure,
which made it impossible to obtain measurements of pre- and postaspiration.
In a further 29 cases, no identifiable stop correlates were
– 119 –