Clusters with nasals

The case where things do not quite work this way is where a nasal comes before a plosive consonant which shares the same place of articulation, in the clusters [mp nt nd ŋk ŋg], as in ‘lamp’, ‘rant’, ‘land’, ‘lank’ and ‘finger’. In these clusters, the oral closure is held and the velum is raised in order to produce oral airflow.

The three clusters with voiceless plosives deserve some comment. Recall that vowels are longer before voiced plosives, and before voiceless plosives, many speakers produce creaky voice during the vowel. The same is also true for nasal + plosive clusters: compare ‘lend’ and ‘lent’. For ‘lend’, you will probably produce something like  , but for ‘lent’, it will probably be something more like  . (The diacritic [˘] here is used to mark: shorter than expected.)

When it comes to nasal + voiced plosive clusters, the situation is more complex. There is a gap in English, in that there is no word-final [-mb] cluster. [-nd], on the other hand, is very common. The velar cluster [-ŋg] is more problematic. There are varieties of English where there is no final [ŋ], since it is always pronounced as [ŋg]. This means there are varieties of English where ‘singer’ and ‘finger’ rhyme, and have the same rhythm:  ; and others where they do not:  . The difference is occasioned by the different morphological structure of the words: ‘finger’ has one morpheme; but ‘singer’ has two: ‘sing’ (  ).

There is also a phonetic explanation for the origin of the [ŋ/ŋg] variation. To exit a nasal, the velum must be lowered, but the oral closure must also be released. This velar can be seen in Figure 9.4 above at around time 0.98. In this case, the velar release is produced simply as a percussive, that is, it is a transient noise that arises from the separation of the two articulators. We will transcribe this as  , using superscript to represent its shortness and low amplitude. But if it is preceded by more airflow, and there is more pressure behind the closure, it is a short step from  .