Licensing Harmonization Working Group - Federal Aviation ...
Licensing of Consonant Clusters at Word Edges
Transcript of Licensing of Consonant Clusters at Word Edges
KATOLICKI UNIWERSYTET LUBELSKI JANA PAWŁA II
Wydział Nauk HumanistycznychFilologia Angielska
Radosław FencNr albumu: 116476
Licensing at Word Edges in English and Spanish
Praca licencjacka napisana pod kierunkiemdr Krzysztofa Jaskuły
Lublin 2011
Contents
Preface....................................................................................................................................4
Chapter I: Introduction Of the Government Phonology Framework.....................................5
1. Introduction........................................................................................................................5
2. Government Phonology......................................................................................................5
2.1 Basic information about GP.........................................................................................5
2.2 Government.................................................................................................................6
2.3 Licensing.....................................................................................................................7
3. Conclusion..........................................................................................................................9
Chapter II: Licensing of Edge Single Consonants and Consonant Clusters in Spanish.......10
1. Introduction......................................................................................................................10
2. Word-initial and word-final single consonants.................................................................10
2.1 GP analysis................................................................................................................13
3. Consonant clusters............................................................................................................18
4. Licensing of branching onsets..........................................................................................18
5. Conclusion........................................................................................................................23
Chapter III: Licensing of Wordedge Single Consonants and Clusters in English.
Spanish - English Comparison. ...........................................................................................24
1. Introduction......................................................................................................................24
2. Word-edge single consonants in English..........................................................................24
2.1 GP analysis................................................................................................................26
3. Branching onsets in English.............................................................................................28
3.1 GP analysis................................................................................................................30
4. Coda-onset sequences in English and Spanish.................................................................31
4.1 The Spanish case........................................................................................................32
4.2 Special status of s + C clusters..................................................................................33
4.2.1 (magic) licensing in Spanish..............................................................................36
2
5. Proper government...........................................................................................................37
6. Conclusion........................................................................................................................38
Conclusion............................................................................................................................40
Podsumowanie......................................................................................................................41
References............................................................................................................................42
3
Preface
The aim of this work is to provide a presentation and analysis of the phenomena linked to
the licensing of word-edge consonant clusters in English and Spanish. To deal with the
phonological issues, the Government Phonology framework, and its extension, the CVCV
model, will be employed in the analyses.
Chapter one is devoted to the introduction of the rudimentary notions of GP. The basic
principles are presented, and their application is laid out in a graphical manner. In
particular, chapter one is about Licensing and Government, the two most important notions
in GP.
The second chapter provides data and analysis for the Spanish phenomena. First, the
single consonants are scrutinised in terms of word-edge licensing. The theory of elements
is employed here to find a common denominator for consonants which behave in the same
manner. Later, being provided with the corpus, the same analysis is done on Spanish
branching onsets. Additionally, chapter two presents some phonetic alternations which may
have influence on the understanding of the phonological processes.
Chapter three is a mirror image of chapter two, it concerns English, in comparison with
Spanish. The first part of this chapter provides us with some essential data, i.e. licensing of
single consonants and their elemental structure for the better understanding of the
phenomena of licensing of the consonant clusters. Next, the licensing of branching onsets
is analysed and discussed. After that, there is an analysis and comparison between English
and Spanish coda-onset sequences. Here, the CVCV model is used to replace the old
'Magic Licensing', which concerns s + C sequences. In the last part of this chapter, the
notion of proper government is presented and analysed.
I would like to wholeheartedly thank my supervisor
for his advice, patience, and kind heart.
Though my eyes would see, I still was a blind man
Though my mind could think, I still was a mad man
4
Chapter I
Introduction to the Government Phonology framework
1. Introduction
This chapter attempts to introduce the theoretical framework of this paper, i.e. Government
Phonology KLV1 (1990), Harris (1990), Harris (1994), Gussmann (2002), Cyran (2003). It
is a minimalistic approach to phonology which is seen to be the most succinct one. Cyran
(1997) accurately notices that the invention of Government Phonology was a ground
breaking event in the old generative phonology theory. As he continues, plenty of rules
from the old theory were replaced by a handful of universal principles which provide
linguistics with complex tools to deal with diverse phonological problems.
2 Government Phonology
As can be found in Harris (1994), Government Phonology (GP) sees phonological
phenomena as a reflection of just several universal principles and language-specific
parameters. What is more, the name itself suggests that governing relations are crucial to
this approach. Trying to understand government, one may find out that government is
viewed as an asymmetric relation between two skeletal slots, i.e. units of phonological
timing.
2.1 Basic information about GP
Before presenting and explaining exact visualisation of government, the most rudimentary
principles of GP have to be mentioned. KLV (1990) argue that all syllabic constituents are
maximally binary (The Binarity Theorem). A syllabic constituent is to be perceived as an
onset, a nucleus, or a rhyme. In other words, this theorem disallows constituents to have
more than two branches.
Following Kaye (2000), one may encounter also notions of Strict Locality and Strict
1 Kaye, Lowenstamm, and Vergnaud (1990).
5
Directionality which are of utmost importance in GP. The Strict Locality principle states
that both positions in constituent government must be adjacent. The consequence of The
Strict Directionality postulation is that all syllabic constituents be head-initial, and the
direction of government is not subject to parametric variation (KLV 1990). Consequently,
ternary syllabic constituents are banned from Government Phonology.
2.2 Government
Given all the above-mentioned notions, it can be inferred that the possibilities of
government in GP are relatively limited. Only few combinations are possible. The
combinations of government in GP are as follows (the structures taken from Jaskuła 2010a:
211):
(1) a. constituent government
R syllabic tier
|
O N N
|
x x x x x x skeletal tier
| | | | | |
b. inter-constituent government
R
|
N O
| |
x x x
| |
In these representations governors are underlined. In (1a) there are formal graphic
6
representations of government within the same syllabic constituent. Namely, a branching
onset represented by the consonant cluster [, a nucleus in the form of a diphthong ,
and a rhyme exemplified by a sequence. In the inter-constituent government, a
governing relation exists between the onset (governor) and the rhymal complement [r]
(governee). Inter-constituent government (as in (1b)) is known to be a head-final one,
(Kaye 2000). The direction of this kind of government is from right to left. Overall, it can
be stated that in most cases obstruents are governors, and sonorants are governees. Still, it
is worth to state the obvious that non-branching constituents do not have governing
relations.
2.3 Licensing
In accordance with the Licensing Principle (KLV 1990: 306), every position in
phonological representation must be licensed (given power to exist or govern another
position) by the head of the domain which normally is the stressed nucleus. Then every
nucleus emanates its licensing power to the onsets in the phonological utterance, thus, all
nuclei have the power to license preceding onsets. These dependencies and their variants
may be graphically represented as:
(2)a. b. c. R
|
O1 N1 O2 N2 O1 N1 O2 N2 O1 N1 O2 N2 O3 N3
| | | | | | | | | | | | | |
x x x x x x x x x x x x x x x x
| | | | | | | | | | | | | | | |
glossary: traba 'an obstacle'
tobo 'a bucket'
tendero 'a salesman'
The Spanish data above present three variations of licensing possibilities. Point (2b) shows
7
licensing in the simplest combination, namely, two full syllables consisting of an onset and
a nucleus. As it can be seen, each nucleus licenses its preceding onset. Point (2a) illustrates
a slightly more complex situation. The first syllable contains a branching onset, the head of
which is underlined. Consequently, N1 licenses O1 which gives governing power to the first
part of the branching constituent. This type of licensing is called a licence to govern. Next,
graph (2c) represents how licensing works if the word contains a rhymal complement (or
'coda'). In this case, the coda [n] is governed by O2 [d] – this phenomenon is called inter-
constituent government.
Yet, these are not the only possible combinations. Let us consider a situation where a
word ends not with a vowel but with a consonant. By way of illustration the following
examples should be analysed:
(3) a. b.
R R
| |
O1 N1 O2 N2 O1 N1 O2 N2
| | | | | | | |
x x x x x x x x x x
| | | | | | | | | |
glossary: tort 'cake'
'cake sing. Gen'
Above, three Polish words are presented. In (3a), the first nucleus [o] licenses the first
onset [t]. The last onset [t] has enough power to govern the coda [r]. A careful reader might
wonder where this ability of O2 comes from. This situation requires a more thorough
scrutiny. A linguistically-conscious Polish speaker will consider the word tort a one-
syllable utterance where is an onset and is a ternary rhyme comprising a nucleus
and a branching coda (ternary rhyme) The same speaker will probably take (3b)
as a two-syllable word; first is an onset, rhyme with a complement, and an
onset and a nucleus.
8
However, from the GP point of view, such a configuration is illegitimate word-finally
due to the violation of the Coda Licensing Principle and the Binary Theorem. The former
states that a well-formed coda needs an onset to follow and govern it, as presented in (2c)
and (1b), the latter, as it was mentioned above, disallows constituents to have more than
two branches. In other words, a cluster of consonants in word-final position cannot be
treated as a word-final coda in the Government Phonology framework. Consequently,
taking also the Licensing Principle (given in 1.3) into consideration, resyllabification is not
possible in GP approach.
As it can be inferred from the arguments above, the only legitimate configuration is
treating the last consonant as a new onset which is licensed by an empty nucleus at the end
of the word. Thus it has the power to govern the preceding rhymal complement. It can be
said that the empty nucleus in word-final position is able to license by a language-specific
parameter.
After saying that an empty nucleus licensing parameter is ON in Polish, a one more
closely related aspect should be considered. When analysing points (3a) and (3b) one may
notice that tortu is a paradigmatic oblique case of tort. The suffix is used in this Polish
word to form a singular Genitive case, e.g. smród 'stench', smrodu
'stench sing. Gen.'. As it was noted above, the word tort ends with an empty nucleus (in
bold) which licenses the preceding onset. After declension of (3a) it can be seen that the
final empty nucleus (N2) is now somehow realised in the form of (in bold). This
phenomenon is an outcome of the Projection Principle (Harris 1994) which states that the
relations among segments are perceived as stable, no matter which version of a single word
we deal with.
3. Conclusion
As we can see, the basics of GP are very succinct and straightforward. There is no place for
speculations and uncertainty. Next, having the rudimentary notions of GP explained, let us
proceed to the following chapter. Some additional remarks on the theory will be made if
necessary.
9
Chapter II
Licensing of edge single consonants and consonant clusters
in Spanish
1. Introduction
This chapter aims to present the phenomenon of the licensing of word-edge consonants and
consonant clusters in Spanish. The analysis will be conducted from the viewpoint of
Government Phonology. A consonant cluster is a configuration of two or more adjacent
non-vocalic segments. However, in order to understand their true nature, some beforehand
comments on the rudimentary essentials should be made.
2. Word-initial and word-final single consonants
To understand the licensing of single consonants in Spanish, each group of sounds present
in this tongue should be examined. To start with, let us scrutinize some hand-picked
Spanish data.
(1) plosives
sound
IPA
spelling polvo ebeo coto bañera dedo gordo
gloss 'dust' 'a comic book' a fence' 'a bath tub' 'a finger' 'fat (Adj)'
The data above show that the plosives [p, b, t, d, k, g] are present in Spanish. To add more,
they are able to occupy word-initial slots in a word. As the research has shown, it is
impossible to find words which end with a bare plosive such as English cat or
[nap. Instead in Spanish, phonetically similar words contain a full vowel
cata'a tasting' or napo 'a hawk' rendering word-final plosives an
impossibility in this language. What is more, there are no voiced stops even
intervocallicaly, despite of the ortography. Further investigation on this problem will be
10
made in the following paragraphs.
In opposition to the Spanish plosives, fricatives have even more limited distribution.
Rarely do they appear in word-initial position, and only few of them are to be found at the
end of the word. The following data exemplify the constraints in the licensing of fricatives
in Spanish. Sounds andwill be talked about in a separate paragraph.
(2) fricatives and approximants
word-initial
spelling zapatero fase segundo joya yo llaga
gloss 'shoemaker' 'phase' 'second' 'jewellery' 'I' 'a wound'
word-final ---
spelling abaz además abad ---
gloss 'a dresser' 'what is more' 'a dean' ---
Given that, it might be noticed that some fricatives such as have unlimited
distribution when appearing as a single consonant. This table shows that some of the
fricatives ( and) are not found in the initial position of the word in Spanish.
However, there are some contexts where these sounds are realized in word-initial position.
This issue is discussed in a greater scope in Oftedal (1985). To illustrate the problem, some
examples based on Oftedal’s ideas will be used. The Spanish word ganar in when it is
context-free begins with a [g]. However, when it is in intervocalic environment (not only
within a word) it may be realised as Example: ganar ['to earn money'but va a
ganar will win'and garras [but las garras
s' Asimilar situation takes place when [dbecomesandalternates within
similar contexts. Here, the bilabial sound marked by the symbol denotes a bilabial
voiced fricative (Oftedal, 1985), and not a bilabial approximant. These fricatives normally
do not exist at the beginning of the word, however, under some circumstances they are
phonetically realized. This phenomenon is subject to further research.
11
Let us scrutinize the sounds of Spanish that are left.
(3) affricates, liquids, nasals
word-initial
spelling cheo rabia leal mano nostro ñoclo ualabi ion
gloss 'drunk' 'rabies' 'loyal' 'hand' 'ours' 'doughnut' 'wallaby' 'ion'
word-final
spelling abajar real tren
gloss 'to lower' 'royal' 'train'
Regarding phonological analysis, the licensing is as follows. The word initial onset O1 is
licensed by the adjacent nucleus N1 in accordance with the GP principles laid out in
chapter one. The fricatives at the end of the word are licensed by the empty nucleus which,
as it was mentioned in the previous chapter, is able to share its power with the preceding
onsets by parameter.
As it is presented, one may observe that some sounds occur in all the aforementioned
positions, while the others do not. The affricate appears in a number of words in the
corpus, whereas is not considered an official part of the Spanish phonemic inventory.
However, it is included in this paragraph to show that the phenomena from (5) are present
also in other sounds. (in bold) may be an alternation of or in a velar context
such as in sin yoyo 'no yoyo', en llamas 'in flames' or inyection
'injection' where the trans-domain environment supports the affrication. It can
be seen that this change is similar to the word-initial alterations of plosives. It is worth to
mention that some borrowings genuinely starting with may be easily found in Spanish,
e.g. gigabyte 'gigabyte'.
The trill and the tap occur interchangeably, according to the table. However,
both can appear in word-medial position e.g. perro 'a dog', andpero 'but'.
From the nasals presented, only [n] occurs in both positions. However, there are some
loans which end with such as ohm 'ohm' or webcam 'webcam' but
these are deliberately omitted in this paper for the fact that they are not representatives of
this language. Semivowels andare also marginal in the language and occur in loan
12
words or in proper names.
To summarise, more single consonants in Spanish can open the word rather than close
it. In word-initial position one can find a number of possibilities. Although some fricatives
are officially banned from this slot, they are still found in an appropriate context (vide the
behaviour of ). It can be said that Spanish phonology is quite liberal, allowing
almost every single consonant to take word-initial position.
On the other hand, Spanish does not allow just any consonant to occupy word-final
position. The phonological system of that language excludes all plosives, a few fricatives,
and other sounds from this slot. In a nutshell, only are possible single
consonants at the end of the word in genuine Spanish. Having given it a thought, one may
notice a regularity. The system allows only the 'weak' sounds which do not demand much
effort to produce. To put it more bluntly, these sounds which involve a kind of significant
closure (such as plosives, nasal plosives, affricates, or even the trill [r]) are too strong to
stand solitarily at the word-final position.
2.1 GP analysis
The conclusion from the previous section needs a further analysis in the Government
Phonology framework. However, word-initial single consonants will not be taken into
consideration by reason of being virtually unconstrained. The only limitation of a
consonant in this slot is, surprisingly, the unexceptional tap [which is never to be found
in the word-initial slot.
To complement the phonetic summary, a phonological analysis has to be done. In the
following paragraphs, the same data will be scrutinized by the use of the phonological
elements or primes which will be explained in the flow of the analysis. The theory of
elements and their application is used here on the basis of the works of Harris (1990),
KLV (1990), Harris (1994), Kaye (2000), and Cyran (2003).
In this part, each aforementioned Spanish sound will have its elemental description. It
will be done to have a deeper insight in the licensing of word-final consonants in Spanish.
To start with, the following table should be analysed. The examples from the previous part
still hold true. Some sounds are omitted due to their insignificant importance in this issue.
13
(4) elemental structure of Spanish consonants
a. obstruents
sound *
complexity U
h
U
h
L
A
h
A
h
L
@
h
@
h
L
U
h
L
@
h
L
I
h
L
A
+h
A
h
L
A
h
A-I
h
U
h
@
h
b. sonorants
sound
complexity N
A-I
A-I
A-I A-I N
U
N
I
I
h
U I
The tables show major Spanish consonants and their phonological interpretation in the
form of Primes or Elements. To start with, it is advisable to make a brief comment on their
internal structure in general. Each group of sounds has something in common when
regarding their complexity. Now, the similarities of the sound families will be laid out.
In the first place, there is a need to explain why the element (L) is used to mark lenis
consonants, and not the element (H) for fortis ones. Cyran (2011) suggests that in Romance
languages (such as Spanish) lenis consonants are fully voiced (their VOT2 is negative), but
the fortis obstruents are not totally voiceless (they have positive VOT). Bearing this
information in mind, it is more reasonable to mark voiced consonants rather than the
voiceless ones. It is also worth mentioning that sonorants need not to be marked as lenis as
they are such by their nature.
Next, the usage of individual primes will be made clear. At the outset, all labials are
described by the prime (U). It characterizes bilabials, labio-dentals, and bilabial
approximants as well. The prime (A) designates all dental sounds such as dental plosives,
2 Voicing Onset Time, see Cyran (2011) for thorough differentiation between fortis and lenis vs. voiced and voiceless consonants
14
affricate, and inter-dental fricatives.
The next element is a compound (A-I). It was introduced by Cyran (1997) for
coronals and employed in Jaskuła (2006) to describe dental non-vocalic elements.
However, this compound will be used here to specify alveolar sounds. Consequently, (A-I)
comprises alveolar fricatives, a nasal, a trill, a lateral, and a tap.
The symbol (@) is used to determine velar sounds, i.e. velar plosives and fricatives.
Next, (h) may be considered as a noise component present during the production of some
consonants. The stop element () stands for any kind of significant closure in the oral tract.
In this work, nasals also have stop element in their structures. It results from the fact that
each of them involves the aforementioned closure, simultaneously, they cannot be
perceived as non-sonorants. Next, the headedness element modification (_) describes
which of the elements in the structure is more prominent. As an example, let us consider
Spanish and case. Both are described here by the same elements, but here, the post-
dental part (A-I) part of is headed. It means that the articulatory component in is
stronger than its headless twin
The sound with an asterisk * needs a further comment on. The presentation in
brackets is the author's original idea. To the author's present knowledge, such a
presentation has not been done anywhere else and it should be considered highly
experimental and absolutely no references to this idea should be made. To make it clear,
the bracket presentation ((A.)+(h)) aims to put forward a contour structure first introduced
by Harris (1994) in a form of a tree. (A.) presents the common part of the structure, +(h)
is the release stage, in this case . What is more, this is the only significant affricate in
Spanish, and in addition, it does not occupy word-final position. Therefore, it will not take
more attention in this analysis.
The segments in bold are known to appear in the final slot of a word. (4a) is a
column devoted to obstruents. Column (4b) introduces sonorants. Such a division has been
made for it has been noted that obstruents act differently than sonorants when it comes to
their behaviour in final position.
As it may be observed, amongst obstruents, only a few fricatives are allowed by the
system to constitute a word-final consonant. Surprisingly, not all of them. The palatal
voiced fricative , the velar voiced fricative and the labial one are excluded from
15
this group. Yet, putting aside these two, it can be said that in Spanish there is a tendency for
the headless and stopless obstruents to occupy the word-final post. Consequently, the final
empty nucleus is unable to license (or to give power) to lenis obstruents which contain
head and stop elements in their compositions. To present it more graphically, the following
trees should be examined.
(5) licensing of single obstruents
a. x x x b. x x x
| | | | | |
V V A
L A-I
_ h
(L)
The graphs (5a) and (5b) aim to show the ability of the word-final empty nucleus to license
certain elements in the case of obstruents. In (5a) the empty nucleus is unable to license
any of the elements from the medial skeletal slot, both, together or as separate units. Graph
(5b) shows which elements the empty nucleus is able to give power to, here, each
separately or in the following combinations: (A.h.L), (A.h), (A-I.h). To make everything
clear, (V) stands for any phonetically realised vowel. As it was mentioned before, the
empty nucleus is unable to license stops and lenis obstruents, except for coronal sounds.
Tree (5a) presents graphically which elements may not be licensed in the word-final slot.
As it is stated, when regarding obstruents, the final empty nucleus is too weak to relay
power to sounds which contain in their structures elements of stopness, headedness, or
voice, except . On the contrary, (5b) makes it evident that fortis obstruents, namely,
voiceless fricatives, are the only sounds less vocalic than sonorants to be licensed by an
empty nucleus. Excluding the aforementioned inter-dental .
Yet, there are some exceptions. Three voiced fricatives are disallowed
from taking the last slot in the word, however, being a voiced fricative itself is a
common ending in Spanish.
16
Moving on to the other group, it has to be noted that sonorants are ruled by slightly
different principles. In this case, it is not the matter what the empty nucleus is able to
license, but rather what is preferred. It is secure to say that headed elements are not good
candidates to be licensed. What is more, the approximants are headed which, as it is
mentioned above, renders a segment unlicenseable. Among sonorants, only ,
andare supported.
Surprisingly, in sonorants, the element () is licensed. Gimson (2008: 206) explains
this phenomenon by showing the similarity between nasals and plosives where a total
occlusion in the oral cavity is made, with an exception of the soft palate being raised. It
allows the air to flow continuously with no audible friction. That is why nasals remind
vowels in many respects. When talking about Spanish nasals, only Despite
the similarities, andare not licensed for some reason.
To complete this scrutiny, one has to think what the licensed sonorants have in
common. Surely, they are headless, and are not of plosive-nature. What unifies them, is the
(A-I) compound which constitutes peculiar foundations for the licensing susceptibility. To
add more, it can be noted that these three sounds are in a succession of complexity. is
the most complex one, is less complicated, the last but not least is the tap consisting
of the solo compound. To show it more graphically, the following tree is to be seen.
(6) complexity of the licenseable sonorants
| | |
N | |
|
A-I A-I A-I
(6) shows that the only sonorants which are found in word-final position have the same
basis.
To sum this part up, let us draw the following observations. First, concerning all
consonants, the empty nucleus is under no condition able to license headed elements.
Second, obstruents are treated differently than sonorants. The conditions for licensing the
obstruents are more strict than for sonorants. To be precise, only unheaded coronal
17
fricatives are allowed to stand in the final position. Next, only three sonorants which share
the same foundations are licensed in word-final position. In general, it can be said that
consonants of dental-alveolar ((A) and (A-I) elements) origin are more welcome at the end
of the word. In a nutshell, coronals are special in Spanish.
3. Consonant clusters
There are three types of consonant clusters recognized by the linguists. The first is a
branching onset, which is discussed in 4. Next, for coda-onset sequences and proper
government see chapter III.
4. Licensing of branching onsets
To get familiar with the notion of a branching onset, let us accept the following definition.
A branching onset is a binary syllable constituent (The Binary Theorem) whose segments'
sonority rises. Sonority, in simple words, is the peak of the sound. Harris (1994: 56)
proposes the following sonority hierarchy:
(7) sonority hierarchy
Plosives Fricatives Nasals Liquids Glides
the least sonorant the most sonorant
Considering sonority in terms of phonology, one may observe the connection between
sonority and the complexity of a sound. Using an example, let us take two Spanish sounds,
the plosive and the liquidtheir complexity would run as follows: (@.h..L), [r]
(A-I). As we can see, is much more complex than [r]. In Spanish, there are many words
starting with cluster [gr], e.g. grabar 'to engrave', or gracia 'grace'. This
cluster is to be considered as a branching onset since it is true to the definition above. The
sonority of the components rises within a constituent.
From the phonological point of view, there is a government relation in this cluster (cf.
18
inter-constituent government, chapter I). Namely, is the governor, and is the
governee. In conclusion, we can say that sonority rises in inverse proportion to the
complexity, which is quite reasonable since the more 'obstacles' in the airflow, the less
vocalic the sound becomes. Thus, in phonology, a segment which is more complex governs
the more simplex one within a branching onset.
In order to see which combinations of Spanish consonant clusters in word-initial
position are possible, the following table has been prepared and analysed.
(8) possible word-initial consonant clusters
a.
+ - + + - + + (-) - - - -
+ + + + + + + - - - - -
+ + + + + + + + + + + +
+ + + + + + + + + + + +
Tables (8) presents which consonant combinations are found in word-initial position in
Spanish. Table (9) (below) is a repository of examples for the clusters given in (8).
Obviously, the vertical sounds constitute the first consonant, and the horizontal are the
second ones. It has been decided not to include to this list for they are just context-
dependent realisations of and behave in the same way in a branching onset.
Bearing in mind the definition of a branching onset, plosives make a perfect C1 (first
consonant) of a cluster. They are the most complex and the least sonorous sounds in
Spanish so there are virtually no limitations to their occurrence in this part of the syllable.
Of course, there are some exceptions. andare not usually found paired with [l].
Following Harris (1983), is allowed in some dialects, but is prohibited globally.
However, there are words starting with acluster, such as Tlaco 'proper name' or
kiosko 'kiosk' but they mostly loan words, thus they are of marginal importance to
this thesis.
However, it is the sound which is the most special one. In this table, only the non-
exceptional clusters are presented. [s] can be found in such words as suave 'gentle'
or suelo 'ground'. The coexistence of and is an issue for further discussion.
19
The Oxford Spanish Dictionary shows some loan words such as (e)slalom
'slalom' or [(e)s'lo slogan 'slogan' to have a parenthetic [e] which is not mandatory.3 As
a curiosity, it can be added that when is stressed, is obligatory e.g. [slot 'slot'.
The s+C sequences will be discussed later on.4
Being given some theoretical introduction, let us scrutinize the following example of
the consonant cluster [tr] occurring in a variety of positions. It can be said that [tr] is a
branching onset because [r] is more sonorous than [t]. And there are no homorganicity
factors preventing these to form a branching onset. The following table presents the
behaviour of a typical branching onset in a word.
(9) licensing of a branching onset at various word positions
a. b.
O1 N1 O2 N2 O1 N1 O2 N2
/\ | | | | | /\ |
x x x x x x x x x x
| | | | | | | | | |
glossary: [ tren 'a train'
cutre 'a stingy person'
In (9a) and (9b) Spanish data are presented. Starting with (9a), some phenomena from
chapter I can be observed. Namely, a full vowel N1 licenses the head of the preceding
branching onset O1, in which [t] governs [r] stating a well-formed consonant cluster in
Spanish. This type of licensing is named a license to govern. What is more, the empty
nucleus N2 licenses the onset O2 to be in accordance with the most rudimentary principles
of GP mentioned in chapter I.
Proceeding to (9b), it can be stated that the same consonant cluster, a branching onset
3 The examples are taken from The Oxford Spanish Dictionary, CD-ROM edition.4 According to Kaye (1996), s+C sequence cannot be treated as branching onset. This issue will be
discussed in chapter III.
20
occupying two skeletal slots whose phonetic realisation is [tr], in this word occurs in its
medial position. In this case, the licensing is not problematic. N2 is a full vowel which has
enough power to license O2. Yet again, license to govern may be observed.
One may ask why the cluster is not presented in word-final position. To answer
that question, it has to be stated that Spanish and English are not good data repositories for
such phenomena. In these languages a word-final branching onset simply does not exist.
Additionally, English as a non-rhotic language does not produce even if it is justified by
the spelling.
The following tables present examples of word-initial branching onsets in Spanish.
(10) word-initial branching onsets
a. sonorant + sonorant
IPA spelling gloss IPA spelling gloss
mueble 'moveable' miedo 'fear'
nueva 'new' 'granddaughter'
ruego 'a request' rielar 'to shine'
luego 'soon' liebre 'a hare'
b. obstruent + sonorant
IPA spelling gloss IPA spelling gloss IPA spelling gloss IPA spelling gloss
plata 'silver' pravo 'evil' pueblo 'a town' piedra 'a stone'
blanco 'white' brazo'shoulder
' bueno 'good (adj)' bien 'good (n)'
tres 'three' tueste 'toasting' tienda 'a tent'
droga'medicin
e' duelo 'duel' diabla 'she-devil'
clava 'a club' croar'to
chuckle' cuaco 'a nag' quien 'who'
glabro 'beardless' grifo'(of hair)
curly' guácharo 'frail' guiar 'to guide'
flaco 'thin' fregar 'to rub' fuego 'fire' fiebre 'fever'
suave 'gentle' siega 'to reap'
zueco 'a clog' cien 'one hundred'
The list above is an extension of the table (8). It has been prepared for the reader to have an
21
instant access to the word examples. What is observable is that, once again, obstruents
behave in a different manner than sonorants. As a matter of fact, there is only ban on
homorganic (for elements, see (4)) consonant clusters, That is why, there are no
combinationsin Spanish. When it comes to fricatives and sonorants, the restriction
is even more rigid. Only accepts all C2's, while and are paired with
glides exclusively. Similarly to and, all sonorants can mate only with glides.
To narrow down this thought, the following statement may be made. The more a sound
is complex, the more it can govern. That is why almost all stops can take all possible
second consonants to form a cluster. It is easy for them to govern such sounds as
Let us now consider a special case in Spanish, namely, coronal sounds. As we
know so far, they are the only ones to be licensed word-finally as single consonants. Here,
however, they do not constitute potent governors. For the sake of consistency, let us remind
that coronal sounds comprise dentals (A), and alveolars (A-I). From the observation that
has been made, it can be said that they are somehow 'weak', and behave in a different way
than other sounds.
It might be noted that the 'coronality' elements make sounds 'weaker', consequently they
are easier to be governed by the FEN5, and simultaneously they are not good governors in
branching onsets as they have to much in common. For the same reason, all sounds can
constitute an onset with a semi-vowel because it has the simplest structure of all
consonants, so it is not a demanding governee.
In order to summarise, it has to be said that a branching onset is not a favoured
consonant arrangement in Spanish. It is hardly surprising that a language which does not
license the whole consonant inventory in all possible positions would license some more
complicated structures word-finally.
5 Final Empty Nucleus
22
5 Conclusion
To sum up this chapter, let us recall several issues tackled here. First, Spanish is a quite
liberal language when it comes to word-initial single consonants. However, at the end of
the word, only headless coronals are allowed. Word initial branching onset comprise
obstruent + liquid/glide, and sonorant + glide sequences, provided homorganicity
principles are preserved. In general, the Spanish language system prefers coronal sounds.
This is the only group of single segments which can occupy the end of a word, when
licensed by a final empty nucleus. Presumably, their being 'coronal' makes the
licensing/government easier for the nuclei in Spanish. What is more, coronal sounds of
speech can also constitute branching onsets together with glides, and each other if the place
of articulation varies in both cases. As a conclusion, it can be said that Spanish copes with
clusters by by licensing clusters/single consonants with a full vowel word-finally.
23
Chapter III
Licensing of word-edge single consonants and clusters in
English.
Spanish - English comparison.
1. Introduction
Chapter III aims to briefly show how single consonants and consonant clusters behave in
English in comparison with Spanish. The pattern will stay the same as in the chapter above.
First, single consonants will be discussed, then word-initial and word final clusters will be
scrutinized. Unlike Spanish, there is an abundance of literature on English modern
phonology to draw on. To keep this paper coherent, only a non-rhotic variant of English
will be presented.
2. Word-edge single consonants in English
As it was done in chapter II, here, all possible word-final and word-initial consonants will
be shown altogether with word examples. It will be presented in a form of a table.
(1) single consonants
a.
t k b d g
word-initial IPA
spelling put take car bat dear get
word-final IPA
spelling grope pit tick crab kid gig
24
b.
word-initial IPA
spelling fought vet say zebra think that
word-final IPA
spelling tough give mess his north smooth
c.
word-initial IPA
spelling should gigolo huge chair jail man
word-final IPA
spelling polish rouge ------------ patch badge trim
d.
word-initial IPA ------------
spelling no ------------ leak roar yes one
word-final IPA (----------) (----------) (----------)
spelling bean thing bowl (----------) (----------) (----------)
The data in (1) show that English is a very liberal language when it comes to single
consonants. It allows almost every one of them to occupy word-initial and word-final
position. Because of that, we can expect that there are also fewer limitations to the
distribution of consonant clusters in this language. In a nutshell, English excludes only
from word-initial position, and end of a word.
In table (1c) the distribution of word-final ,, andis in parenthesis because of
the fact that they can be found at that post in connected speech. Referring to Gimson
(2008: 305), linking (or intrusive ) is common in fast speech. e.g. far off in normal,
careful speech will be pronounced as but in fast, careless manner of speech the
production looks like . A similar thing happens to and , e.g. my arms
, window open . For more information, see Gimson (2008).
25
2.1 GP analysis
This part is devoted to the presentation of English sounds in terms of GP. Once again, the
element theory will be implemented here. Knowing the internal structure of English
sounds, we may be able to tell more about the licensing of the clusters.
(2) complexity of English consonants
a.
sound
complexity
U
H
A
H
@
H
U
A
@
U
h
H
U
h
A
h
H
A
h
A
h
H
I
A
h
I
b.
sound
complexity A
h
H
A
h
@
h
H
N
U
N
A
N
@
A A
U I A
+h
H
A
+h
Table (2) clearly shows that English has a completely different sound structure than
Spanish. First of all, the element (H) denotes voicelessness. Unlike Spanish, where the
voiced sounds are marked by (L). English consonants, in opposition to Spanish, are not
fully voiced, thus it is more reasonable to mark the voiceless ones. Cyran (2011) states that
in general, Romance and Slavic languages are characterised by (L), while Germanic
tongues take (H).
Next, the plosives stock is quite the same as in Spanish, there are three voiced, and three
voiceless stops in the repository. However, the changes which concern English stops are of
26
different nature than the Spanish ones. In Spanish, voiced
becomeintervocalically, and in various morphological contexts. On the
other hand, there is an alternation of English voiceless stops. English speakers tend to
aspirate voiceless stops in word-initial position before vowels, thus, in such words as
take, the detailed relisation would be . However, Gimson (2008: 168) claims that in
some dialects of English, a similar situation as in Spanish takes place. Namely,
may be followed by or altered to a brief fricative ,
especially among educated peoplee.g. becomes imported. Despite
its being intriguing, it is of marginal importance to this thesis. Coming back to Spanish
affrication and English aspiration, it is a good idea to take a deeper look into this in terms
of GP. According to Watson (2006), spirantisation is a form of lenition. Thus, it can be
explained by the use of the elements. Let us take the case of Spanish as an example.
becomes intervocalically, and in morphologically dependent environments. As a
reminder, the complexity of is (@.h..L), when it alters to , its structure also changes
to (@.h.L). As we can see, the stop element is deleted, making this sound a fricative. Yet,
there is something different going on with the English fortis plosives. When aspirated, their
elementary structure remains intact. Consequently, we can state that the alternations of the
Spanish plosives are of phonological nature because their complexity changes, whereas
English plosives just undergo some phonetic changes, which are not considered important.
In 1.1 it is stated that English excludes from word-initial and from word-final
position. Let us make an attempt at explaining it in terms of GP. As we know, all sounds,
except for these two, are allowed to occupy every position in the word. Following
Gussmann (2002: 46-52), the velar nasal is an extraordinary type of a consonant which
always shares its place of articulation with the following velar plosive and may only occur
after a short vowel. Next, Gussmann compares to a specific type of consonant cluster,
where, word-finally, the second constituent (velar plosive) is inaudible, but often present in
the orthographic representation. Thus, the velar nasal cannot stand in the first slot of the
word because it has to be followed by a velar plosive, which is impossible word-initially in
English.
27
3. Branching onsets in English
As it is stated above, English accepts almost all solo consonants to constitute a non-
branching onset at in word-initial and final position. The following paragraphs will include
a table of possible branching onsets word-initially in English and a short analysis.
(3) branching onset
+ - + + - + + - - -
+ + + + + + + + - -
- + + - + + - + - -
+ + + + + - + - + +
IPA gloss IPA gloss IPA gloss IPA gloss
please prey pure
blow bridge beauty
trench tube twig
drive dew dwell
class crane cue choir
glow grow ------------ guano
fly fry few
thrive thwart
mute
new
Although English is more liberal when considering single consonants, word-initial
branching onsets remain Spanish-like. Only obstruent/nasal + liquid/glide compositions are
permitted. What is more, Spanish allows also more sonorants () to state C1 in a
branching onset which in English case these are only and. What is more, we can
see that English plosives are less flexible when it comes to C2 acquisition. English forbids
28
homorganic branching onsets, including not only , but also 6, and
In fact, English is more limited when it comes to the possibility of forming branching
onsets. Spanish allows 38 combinations of word-initial branching onsets, while in English
only 25 instances can be found. The next peculiarity is that English plosives are weaker
than the Spanish ones and do not accept just any possible second consonant. On the other
hand, it this can be explained by saying that English phonological system just rigidly obeys
the homorganicity ban in branching onsets. In terms of GP, both constituents in a
branching onset cannot have the same place of articulation element in their structures.
Consequently, we will not find clusters such as ((U.h.H) (U)) or ((A.h.H) (A)),
though, in rapid speech, only the release stage changes. e.g. keep walking
cut log , but these changes are purely phonetic events. Next, there is also the
problematic case of cluster, which according to the observation above should be
permitted, because the place elements differ, but in fact no English words begin with such
combiantion. However, occurs in connected speech: beg you Gimson (2008:
226) causing palatalisation (but not the change of the release stage) of this velar stop.
Relying on this data, it can be said that the case is a phonological issue because the
elemental quality of these sounds changes (the (I) element is added as a secondary place of
articulation), while in word-initial it is sheer phonetic effect having no
influence on the phonological structure of the sound. Thus, the cluster may be treated
as a well-formed word-initial, though exceptional, branching onset. In conclusion, there is
enough data to confirm the statement that homorganic clusters are dissallowed in English.
There is one thing which Spanish and English have in common. Namely, they do not
have a branching onset in word-final position. Consequently, words such as Polish
Piotr 'Peter' or łagr 'labour-camp' simply do not exist in the phonological
systems of English or Spanish. However, there is one slight difference. As it is mentioned
in chapter I, a full vowel is substantial to license a branching onset in Spanish. Thus, a
considerable number of words, which hypothetically would end with a branching onset
plus a phonetically realised nucleus, as an example, litro 'litre', while in Polish the
same word is realised as litr 'litre'. On the other hand, English, as a non-rhotic
6 Although, Longman Pronunciation Dictionary (2005) shows one word starting with : Fuegian, however, it is a proper name and it is not considered here as a prominent exception.
29
language, acquired a different solution to this issue. Namely, it just deletes the liquid ,
prolonging the vowel, or altering it to the , and rendering a word-final branching onset
an impossibility, e.g. litre. From the orthographical point of view, the grapheme 'e' is
here to close the second syllable (in a grammatical way) to prevent the vowel from
being realised as a diphthong , as in lighter, where the final syllable is open from
the orthographical point of view.
In conclusion, it can be said that the English branching onset is quite similar to the
Spanish one despite the poorer inventory. These clusters are licensed word-initially by a
full nucleus, potent enough to emanate its strength over the C2 to license the first
consonant. Gussmann (2002: 72) gives the following condition: 'English branching onsets
must consist of an obstruent followed by non-homorganic, and non-nasal sonorant'.
Similarly to the Spanish FEN, the English one is not able to license word-final branching
onsets.
3.1 GP analysis
As it is already known, the first consonant of a branching onset must be a true obstruent
(preferably a plosive or a fricative), and the second one a non-homorganic, non-nasal
sonorant. All attested branching onsets which occur word-initially, are also found word-
medially within one phonological word. That excludes prospective homorganic clusters to
be perceived as branching onsets. For the sake of illustration, let us take the word deadly
, phonetically, a cluster in this word is possible, however, in phonology, this
cannot be taken into consideration because the cluster is between two morphemes which is
not phonologically approved; .7
Knowing these facts, we may now state what is happening to the attested branching
onset in various word positions. As we know, in every possible position, such an onset is
licensed by an audible nucleus, either non-branching or branching, and never by an empty
nucleus. Thus, the following tree showing the overall word-initial licensing preferences
should be drawn.
7 Data taken from Gussmann (2002: 74).
30
(4) possible licensing configurations of branching onsets
a. b. c.
O N O N O N
/\ | /\ /\ /\ /\
x x x x x x x x x x x
| | | | | | | | | \/
T R V T R V1V2 T R V:
All graphs (4a,b,c) present the licensing of branching onsets in every possible word
position in English. To make everything clear, let us explain the terminology used here. T
stands for a true obstruent, such as a plosive or a fricative. R is a sonorant, the second
sound in branching onsets. And the V combinations as follows: V is a single vowel, V 1V2
stands for a diphthong, and V: denotes a long vowel, e.g. (4a) twig, (4b)
thrive, and (4c) freeze. As we can see, the nucleus licenses the preceding
onset, relaying power to the head of the branching onset, the true consonant, which finally
governs R. There are no examples of a well-formed branching onset being licensed by an
empty nucleus in any word position. Consequently, it can be stated, that the English empty
nucleus is somehow stronger than the Spanish one because it licenses almost all single-
consonants word-finally, but it also has the same weakness of not being able to handle a
branching onset.
4. Coda-onset sequences in English and Spanish
Having thought over the conclusion from 3, one may notice that it does not solve all
possibilities of the language to form consonant clusters. So far, we have concluded that
branching onsets must consist of a true consonant (T), and a sonorant (R), looking from
left to right. However, it is not hard to find words which comprise clusters in which the
order is opposite. Let us examine such words as help and alba8. The word-
final consonant cluster cannot be named a branching onset due to the lack of accordance
with the principles which state that in a branching onset, the sonority of the constituents
must rise from left to right. As we can see, here the sonority rises in the opposite way. This
8 Structures taken partially from Gussmann (2002)
31
fact denotes that a different type of a cluster is being dealt with.
To start with, the notion of the coda should be explained. the Coda is a traditional name
of a rhymal complement. The Coda exists only if there is a less sonorous consonant in an
adjacent syllable following and governing it. What is more, it is worth to define what a
coda-onset sequence is in terms of GP. Harris (1994: 67) proposes the following definition:
'In an optimal coda-onset cluster, the first consonant is no less sonorous than the second.'
Let us see, how it may be presented by the use of a phonological tree.
(5) coda-onset sequence in English
a. R b. R
| |
O1 N1 O2 N2 O1 N1 O2 N2
| | | | | | |
x x x x x x x x x
| | | | | | | | |
Examples (5a) and (5b) present the most classic representations of coda-onset sequences.
Both of them include a consonant cluster consisting of two consonants, which, looking
from left to right, have decreasing sonority. What is more, in front of every rhymal
complement, there is a nucleus. What is worth to observe is the direction of government. In
this arrangement, the following constituent O2 governs the preceding coda.
This type of cluster is much more 'digestible' for the licensor. The nucleus does not have
to 'jump over' a consonant in order to license the less sonorous one as it takes place in a
branching onset. The direction of the licensing/government is the same so it can be said
that combining these two factors (licensing adjacency and one direction of the
relationships), coda-onset sequence is easier for the phonological system to be put within a
word. Thus, the FEN is able to license coda-onset sequences word-finally in English.
4.1 the Spanish case
Although the English Final Empty Nucleus behaves similarly to the Spanish one when it
comes to licensing word-final branching onsets, it is mightier in the licensing of the coda-
32
onset sequences. Let us take a brief look at the following examples of the Spanish codas.
(6) coda-onset sequence in Spanish
a. R b. R
| |
O1 N1 O2 N2 O3 N3 O1 N1 O2 N2
| | | | | | | | | |
x x x x x x x x x x x x
| | | | | | | | | | |
k
gloss artero 'a postman'
ante 'elk'
The Spanish data present coda-onset sequences word-medially. As we can see, these
sequences are licensed in the same manner as in (5) with a full vowel. However, there is a
serious limitation in Spanish concerning the licensing of a coda-onset sequence word-
finally. In opposition to English, the Spanish FEN is not able to license a final coda-onset
sequence, thus, a full vowel is required. Let us compare two phonetically similar words,
the first is English, the second Spanish: , and 'elf'. Here, the difference
is obvious. Despite the same semantic meaning, their phonological structure is different
because both languages have different phonologies. In brief, the English FEN licenses
coda-onsets word-finally, thus it can be said that it is stronger than Spanish, which needs a
full vowel to license such a cluster.
4.2 Special status of s+C clusters
The research conducted by the most recognised linguists KLV (1990), Harris (1994), Kaye
(1996) claim that clusters consisting of and a consonant cannot state a branching onset
due to a range of reasons.9 Basically, words such as stop do not begin with a
branching onset, but are rather perceived as a special type of coda-onset sequence which
can occur word-initially. Of course, this kind of consonant cluster is easy to find in every
word position, e.g. vest, text, street. Let us see, how it may be presented graphically.
9 For more information on s+C clusters, consult the aforementioned positions.
33
(7) licensing of s+C clusters
a. b. c.
R R R
| | |
N0 O1 N1 O2 N2 O1 N1 O2 N2 O1 N1 O2 N2 O3 N3
| | | | | | | | | | | | | |
x x x x x x x x x x x x x x x x x
| | | | | | | | | | | | | | | | |
R
|
N0 O1 N1 O2 N2
/\ /\ | |
x x xxx x x
| | | \/ | |
The graphs in (7) present the behaviour of s+C clusters in various word positions. (7a)
shows word-initially, (7b) presents word-final position, and (7c) depicts what is
happening word-medially in an even more complicated cluster . Tree (7d) is an
interesting example of a word starting with a s + C1C2 sequence, so after [s] there is a
branching onset.
To start with, it has been stated above that s+C sequences are treated as a special kind of
coda-onset configuration. As we can see in (7a), when such a cluster occurs word-initially,
the preceding nucleus N0 is not realised, being licensed by magic.10 is governed by as
in a standard coda-onset sequence. Next, (7b) presents a usual situation, with a full rhyme,
where the nucleus N1 is realised. For more comments, see 4. Example (7c) shows a
situation where there are two adjacent empty nuclei, which is somewhat controversial due
to the problematic licensing of the word-medial empty nucleus. (7d) depicts a situation,
where after the fricative, there is a branching onset. What is interesting in this
10 The notion of 'Magic Licensing' is to be found in Kaye (1996).
34
configuration is the licensing abilities of the closest nucleus. It has to, at one time, license
the true obstruent in the branching onset and the preceding coda. Kaye (1996) names such
configuration 'Magic Licensing'.
However, there is a newer theory which may disprove magic licensing. In terms of
CVCV, we can explain the problematic licensing of the s+C clusters. The CVCV theory is
an expansion of the Government Phonology, introduced by Lowenstamm (1996),
developed by Scheer (1998), Cyran (2003), Jaskuła (2006), Jaskuła (2010b). It assumes
that there are no branching syllable constituents whatsoever. Thus, the aforementioned
problem may be solved by treating each consonant as a separate onset with an nucleus
(empty or full) in between. Let us scrutinize the following proposal:
(8) s+C sequences in CVCV
a. b. c.
O1 N1 O2 N2 O3 N3 O1 N1 O2 N2 O3 N3 O1 N1 O2 N2 O3 N3 O4 N4
| | | | | | | | | | | | | | | | | | | |
x x x x x x x x x x x x x x x x x x x x
| | | | | | | | | | | | | | | | | | | |
d.
O1 N1 O2 N2 O3 N3 O4 N4 O5 N5 licensing
| | | | | | | | | proper government
x x x x x x x x x left interonset government (LIO)
| | | | | | | | | right interonset government (RIO)
Here, the same words are presented in CVCV framework. In a nutshell, it states that there
35
LIO
LIORIO
PGLIO
are no branching syllable segments and there are no closed syllables (codas).11 In (8a), N3 is
licensed by parameter, and N1 is licensed to be mute by LIO. (8b) shows quite a different
situation. O3 is licensed by the FEN, and governs O2 using Left Inter Onset government
introduced by Cyran (2003). (8d) is the CVCV variation on the cluster licensed by a
long vowel. As we see, the long vowel licenses O2 to perform RIO and LIO. Next, (8c)
shows the weakness of CVCV to present the licensing of this type of clusters. However,
Jaskuła (2010b: 404) proposes the following solution:
(9) affrication
O1 N1 O2 N2 O3 N3
| | | | | |
x x x x x x
| | | | |
Jaskuła (2010b) treats the plosive + s sequence as an affricate, which is LIO governed by
the next onset. As we see here, there is license to govern relationship between N3 and O3.
Subsequently, O3 becomes a governor of the preceding onset O2. Thus, the problem of
three consecutive empty nuclei and their licensing is solved.
4.2.1 (magic) licensing in Spanish
There is a parenthetic 'magic' in the title of this part because there are no such phenomena
in Spanish. What is more, the lack of magic licensing in Spanish, makes CVCV a strong
argument to accept its validity. Words which would start with a s+C cluster, obligatory
have an epenthetic before the cluster, filling up the preceding empty nucleus, rendering
it a conventional coda-onset sequence. e.g. (e)slalom 'slalom',
escuela 'school'.
11 For further reading on CVCV, see Lowenstamm (1996).
36
LIO
5. Proper Government
Considering all the possibilities, it can be found out that the configurations above do not
reflect the whole system. So far, the following types of consonant clusters have been
discussed. First, the branching onset where the sonority of two consonants must rise
looking from left to right, and heterogeneity is demanded. The second one is a coda-onset
sequence in which the government direction is from right to left. However, there is one
more type of a consonant cluster in English, Spanish and other tongues. Now, let us
consider a situation where we cannot differentiate between the sonority of the two
consonants (or the sonority is hardly distinguishable), or/and what is more, they are
homogeneous (either by the place of articulation or the membership to the same sound
family, for instance, are both homogeneous. There are no data in the corpus to state that
this type of clusters occurs word-initially (as in for example in Polish). The graphs below
depict this problem.
(10) proper government in English
*a. b.
O1 N1 O2 N2 O3 N3 O1 N1 O2 N2 O3 N3 O4 N4
| | | | | | | | | | | apt
x x x x x x x x x x x atlas
| | | | | | | | | | |
Graph (10b) shows a typical environment for proper government, while graph (10a) shows
a situation in PG cannot be contracted (see, and, help, text). In (10a), the consonant cluster
occurs word-finally and its licensing is as follows. Though the cluster consists of two
plosives of similar sonority, it cannot be properly governed, because there is no proper
governor (a full nucleus). Instead, a coda-onset sequence should be recognized as in 7b and
8b. (10b) Shows a situation, where the proper governor is a full nucleus N3. It also governs
the empty nucleus N2 by 'jumping over' O3, simultaneously licensing it. O2 is licensed by
the following empty nucleus.
As it may be observed, yet again, the FEN, which itself is licensed by parameter, is able
37
to license another preceding empty nucleus. Now, let us see how does proper government
look in Spanish.
(11) proper government in Spanish
a. O1 N1 O2 N2 O3 N3 O4 N4 b. O1 N1 O2 N2 O3 N3 O4 N4
| | | | | | | | | | | | | | | |
x x x x x x x x x x x x x x x x
| | | | | | | | | | | | | | |
glossary: garlar 'to chatter'
acepto 'well accepted'
Table (11) presents a situation where a consonant cluster cannot be classified as a coda-
onset sequence or a branching onset. Such a situation happens due to their similar sonority
and/or their place of articulation. To solve this problem, an empty nucleus which is
properly governed by the next full nucleus may be placed between the consonants. This
type of inter-constituent relationship is named proper government in GP. In Spanish, this
type of government is common in Spanish in word-medial position. What is more, there is
no significant evidence in the Spanish corpus to state that this configuration can be found
word-initially, let alone, word-finally due to the lack of a proper governor which must
follow the cluster.
6. Conclusion
Generally, this chapter shows the main differences and similarities between various
phonological behaviours of different types of consonant clusters in English and Spanish.
To sum up, it has to be said that the behaviour of the English FEN is similar to Spanish
when taking into consideration word-final branching onsets. However, it is also attested
that the English FEN is able to license other types of consonant clusters word-finally,
unlike Spanish, which does not cope with anything more than single coronals in this
38
position.
What is more, it has been shown that the 'magic licensing' model may be abandoned in
a more modern approach to GP. CVCV is used here to show that some problematic
licensing issues may be solved in a different way than the classic one. It has also be said
that in Spanish, there is no need to employing 'magic licensing' due to its syllable structure
which does not allow s + C structures word-initially. All the more, coda-onset groups will
not be found word-finally due to the Spanish FEN inability to license any consonant
clusters. Finally, divided constituents which are understood as two separate onsets with a
properly governed empty nucleus between them exist and are licensed both in English and
Spanish.
39
Conclusion
This thesis presents an attempt to show the behaviour of different types of consonant
clusters at word-edges in Spanish and English. First, some theoretical account on GP was
given, showing crucial pillars of this approach.
Next, the second chapter aimed to discuss the issues of the licensing of single
consonants, and consequently, consonant clusters as branching onsets in Spanish. The
Element Theory was employed in this analysis. What is more, some phonetic digressions
were also made to account for better understanding of phonological phenomena in Spanish.
As a result, we concluded that the Spanish final empty nucleus licenses only single,
unheaded coronals in word-final position.
Chapter three provided an account of the licensing of consonant clusters in English in
comparison to Spanish. As well as in chapter two, the Element Theory was used to indicate
the complexity of English consonants. The analysis showed that English allows all possible
consonants word-initially and word-finally, with some exceptions. Subsequently, the
Elements were used to explain the possible word-initial branching onsets. As well as in
Spanish, branching onsets are too hard to be licensed by the English final empty nucleus.
However, the next type of consonant clusters, coda-onset sequences, occur to be easily
licensed by the English FEN, but not by the Spanish one. The clusters with
undistinguishable sonority are licensed in both languages.
To summarise, it can be said that in some respects the licensors/governors in Spanish
and English behave in the same way, but other facts show that the English is more flexible
when it comes to the licensing of various types of consonant clusters at word-edges.
40
Podsumowanie
Celem niniejszej pracy jest przedstawienie i zanalizowanie zjawiska licencjonowania
różnych typów zbitek spółgłoskowych na obrzeżach słów w językach hiszpańskim i
angielskim. Aby osiągnąć powyższe cele, wprowadzono założenia Fonologii Rządu i jej
pochodnych. Pierwszy rozdział ma na celu zaprezentowanie i usystematyzowanie
najważniejszych zagadnień teoretycznych z dziedziny Fonologii Rządu. Są to relacje rządu
między rozgałęzionymi nagłosami i licencjonowanie zbitek poprzez końcowy pusty
ośrodek sylaby.
Rozdział drugi zajmuje się przedstawieniem licencjonowania zbitek spółgłoskowych w
języku hiszpańskim. Celem wprowadzenia, zostały przedstawione i zanalizowane
pojedyncze spółgłoski, które mogą znaleźć się na obu krańcach słowa. Następnie,
zastosowana została Teoria Elementów, która mówi, iż u podstawy wszystkich dźwięków
w różnych językach leżą te same składowe – elementy. Jako dopełnienie tej wstępnej
prezentacji i analizy, zaprezentowane zostało licencjonowanie rozgałęziających się
nagłosów. Rezultatem tej analizy było stwierdzenie, że w języku hiszpańskim, końcowy
pusty ośrodek sylaby nie jest w stanie licencjonować zbitek spółgłoskowych będących
rozgałęzionymi nagłosami.
Rozdział trzeci, ostatni, ma na celu zanalizowanie języka angielskiego pod kątem
licencjonowania i relacji rządu na krańcach słowa i porównanie go do języka
hiszpańskiego. Wnioskując z danych, okazuje się, że angielski końcowy pusty ośrodek
sylaby mimo tego, że tak samo jak hiszpański, nie radzi sobie z licencjonowaniem
rozgałęzionych nagłosów, doskonale robi to w przypadku sekwencji o malejącej
sonorności i zbitkach homorganicznych.
41
References
Cyran, E. (1997) Resonance elements in phonology. A study in Munster Irish. Lublin:
Folium.
Cyran, E. (2003) Complexity scales and licensing strength in phonology. Lublin: Redakcja
Wydawnictw KUL.
Cyran, E. (2011) Laryngeal Realism and Laryngeal Relativism. Two Voicing Systems in
Polish. Unpublished, Ms.
Durand, J., and B. Laks (eds.) (1996) Current Trends in Phonology. Models and Methods.
Salford, Manchester: European Studies Research Institute, University of Salford.
Galimberti, J. B., R. Russell, C. S. Carvajal, and J. Horwood. (2003) The Oxford Spanish
Dictionary: Spanish-English/English-Spanish. Oxford: Oxford University Press.
Gussmann, E. (2002) Phonology: Analysis and Theory. Cambridge: Cambridge University
Press.
Harris, J. W. (1983) Syllable Structure and Stress in Spanish: a Nonlinear Analysis.
Cambridge: MIT.
Harris, J. (1994) English Sound Structure. Oxford: Blackwell.
Harris, J. (1990) "Segmental Complexity and Phonological Government." Phonology 7,
255-300
Jaskuła, K. (2006) Ancient Sound Changes and Old Irish Phonology. Lublin:
Wydawnictwo KUL.
Jaskuła, K. (2010a) "Polish Place-names and Word-final Consonant Groups: More
Affricates and Diphthongs in Polish?" Poznań Studies in Contemporary
Linguistics, 46(4)
42
Jaskuła, K. (2010b) "From Generators to Mirrors – a Comparison of Phonological
Theories." Facta Simonidis 1, 205-222
Kaye, J. (2000) A User's Guide to Government Phonology (GP). Unpublished, Ms.
Kaye, J. (1992) "Do You Believe in Magic? The Story of S+C Sequences." SOAS Working
Papers in Linguistics and Phonetics 2
Kaye, J., J. Lowenstamm, and J.-R. Vergnaud. (1990) "Constituent Structure and
Government in Phonology." Phonology 7, 193-231
Lowenstamm, J. (1996) "CV as the Only Syllable Type." In J. Durand and B. Laks (eds),
419-441
Scheer, T. (1998) "Governing Domains Are Head-final." In E. Cyran (ed.)
Watson, K. (2006) "Lenition and Segmental Interaction: Evidence from Liverpool English
(and Spanish)." Glossa - An Ambilingual Interdisciplinary Journal 1.1
Wells, J. C. (2003) Longman Pronunciation Dictionary. Harlow: Longman.
43