UT

Cluster simplification


In cases of consonant cluster simplification, one or more consonants present in the underlying representation (UR) of a form are deleted. Some phonologists view simplication as a result of constraints on the ways that a language allows consonants to combine into syllable structures, while others attribute it to articulatory causes. These views are not necessarily incompatible. An example of cluster simplification in English occurs in a word like handball. At normal speaking rates, this word is pronounced as [hæmbal], so that the alveolar stop /d/ is deleted between consonants /n__b/, and the nasal is assimilated to the /b/ in ball.

To begin our discussion of cluster simplification in Central Catalan, we turn in (5) to a set of words whose roots end in a consonant cluster. In this table, orthographic representations for a set of words (nouns and one adjactive) are given in the leftmost column, followed by their IPA transcriptions. To the right, the diminutive form of each word is given (first the orthographic and then the phonetically transcribed form). The diminutive suffix, -ɛt (masc.) or -ɛtə (fem.), adds the meaning of little for nouns, and a little bit for adjectives.

(5) Set 1.
Noun and IPA
Diminutive form and IPA
English gloss
serp [seɾp] serpeta [səɾpɛtə] 'snake'
disc [disk] disquet [diskɛt] 'disk'
calb [kalb] calbet [kəlbɛt] 'bald'
remolc [ɾəmolk] remolquet [ɾəmulkɛt] 'trailer, tow'
colp [kolp] colpet [kulpɛt] 'blow, bang'
parc [paɾk] parquet [pəɾkɛt] 'park'
cresp [kɾesp] crespet [kɾəspɛtə] 'waves on surface
of liquid'

The root of any word is the content morpheme that gives the word its basic meaning. For example, the root of the English word unhappily is happy. A root can often be a free-standing word, as happy is - but not always: the root of the latinate verb prefer is -fer-, a bound root which cannot stand on its own. In Catalan, as in English, the root of any word will be the form of the word without any derivational or inflectional affixes. In (1), each diminutive has as its root the corresponding word to its left.


When we perform a basic morphological segmentation by factoring out the suffix from the diminutive forms, we see that in each case, the remainder of the diminutive matches the corresponding unaffixed form to the left. We assume quite straightforwardly that the corresponding string in each case is the morphological root. We assume that the underlying (phonemic) representations, or URs, for these roots are as shown in (6).

(6) Phonemic representations for roots in Set 1.
Root URs English gloss
/seɾp/ 'snake'
/disk/ 'disk'
/kalb/ 'bald'
/remolk/ 'trailer, tow'
/kolp/ 'blow, bang'
/paɾk/ 'park'
/kɾesp/ 'waves on surface
of liquid'

In (7) we see another set of diminutive words (from the noun, verb, and adjective classes). Parallel to the Set 1 forms in (5), each diminutive in Set 2 displays a consonant cluster before the diminutive suffix. However, these pairs are different from those in Set 1: the consonant cluster seen in the diminutive forms does not appear in the unaffixed word. The unaffixed words endin a single consonant.

(7) Set 2: diminutives.
Diminutive form and IPA
English gloss
campet [kəmpɛt] 'field'
tombet [tumbɛt] 'walk, stroll'
puntet [puntɛt] 'point'
banquet [bəŋkɛt] 'bank'
malaltet [mələltɛt] 'sick'

However, when we look at the Set 2 words in other contexts, we find that in fact they behave quite differently from those in (5). To begin, the chart in (8) shows that when these words appear without the diminutive suffix, they do not end in the consonant clusters seen in (7). Rather, only the first consonant in the cluster is present. That is, the diminutive campet in (7) contains the consonant cluster [mp], but camp has only [p] (even though this word is spelled with a cluster).

(8) Set 2: Basic (nondiminutive) forms of the words in (7).
Basic word and IPA
English gloss
camp [kam] 'field'
tomb [tom] 'walk, stroll'
punt [pun] 'point'
banc [baŋ] 'bank'
malalt [məlal] 'sick'

The words in (8) were pronounced in isolation (before a pause, with nothing following). However, they have the same pronunciation - with no final cluster - when they occur before another word. Moreover, this is true regardless of whether the following word begins with a consonant or a vowel. In (9), we see examples of each word before the vowel initial word es 'is' (2nd column), and before the consonant initial word sigi 'following, next' (3rd column).

(9) Set 2.
(Glosses for the basic words are those in (8). Forms in the 2nd, 3rd, and 4th columns are transcribed in IPA.)
Basic word
Before V-initial word es 'is'
Before C-initial word sigi 'next'
camp [kam es] [kam siγi]
tomb [tom es] [tom siγi]
punt [pun es] [pun siγi]
bank [baŋk es] (see note) [baŋ siγi]
malalt [məlal es] [məlal siγi]

Notice in the chart above that banc, unlike the other words in the chart, has final [k] before the word es. We should note that the final consonants of the words in (8) and (9) are normally dropped in this context (before a vowel initial word), but are pronounced when speakers are being careful. In the case of banc, we've noticed that our speaker has a tendency to prounounce the final [k] in this particular word, although she deletes the final C regularly in other cases. We have kept [baŋk es] in the chart as an acknowledgement of the fact that even regular linguistic patterns can have exceptions.


Based on the charts for Set 2 words, we see that the roots of these words have an allomorph with the form CVC1C2 which appears before the diminutive suffix (see (7)). This allomorph alternates with an allomorph of the form CVC1 which surfaces in other environments (see (8) and (9)).

(10) Set 2: alternating allomorphs
Basic word
C-final allomorph
C-final allomorph
camp
[kam]
[kəmp]
tomb
[tom]
[tumb]
punt
[pun]
[punt]
bank
[baŋ]
[bəŋk]
malalt
[məlal]
[mələlt]

The point we are examining in this section has to do with the presence or absence of a root-final consonant cluster. However, another difference can be seen in the qualities of corresponding vowels. These differences are due to differences in the way stressed is assigned. We will look more closely at this issue in our section on vowel reduction.


Given this alternation, there are two available analyses as to the UR for these roots: (i) a UR with a final clusters, e.g. /kamp/, in which case, the second C is deleted in certain contexts, or (ii) a root with only a single final consonant, e.g. /kam/, in which case the final consonant is added in derived contexts, before a suffix (assuming that it is the morphological context that is important). is The UR we assume for these root morphemes end in a consonant cluster, parallel to the roots in Set 1.

Note that a difference between the Set 1 and Set 2 roots is that the final clusters in Set 2 are homorganic while those in Set 1 are not. The distribution of the clusters in generalization here is that the homorganic consonant cluster [mp] occurs before a vowel, but not word finally. A similar observation can be made about the forms in (1b)-(1e). The clusters [nt], [ŋk], [lt], and [st] occur word internally, but not word finally.

Note that the forms in (1f) and (1g) do not undergo cluster simplification. The underived form for 'snake' is [sərp], and the form with the diminutive suffix is [sərpɛt]; the cluster [rp] occurs both word internally and word finally. The difference between the forms in (1a)-(1e) and the forms (1f) and (1g) is that the consonants forming clusters in (1a)-(1e) are homorganic, whereas the consonants forming clusters in (1f) and (1g) are not. We can capture this generalization with a derivational rule that deletes the second consonant in a cluster of homorganic consonants, just in case the cluster occurs word finally.

(8) Cluster Simplification: [-son, αplace] → Ø / [-son, αplace] ___ #

The rule in (8) states that whenever two obstruents occur in sequence and share the same place features, the second of the two obstruents is deleted. The symbol alpha is a variable that stands in for the place features of the segment in question, whatever they may be.

The process of nasal place assimilation described in the last section and cluster simplification interact in an interesting way. We turn in the next section to a discussion of this interaction.


prev | top | next