Structural ambiguity in English word-formation

Bogdan Szymanek

1. Introduction

It is well known that a number of phrase and sentence types in English may give rise
to cases of structural ambiguity. The phenomenon is exemplified by the following
sentences: The professor’s appointment was shocking, I found her an entertaining
partner, The chicken is ready to eat, etc. Structural ambiguity of this sort (also called
‘grammatical ambiguity’ or ‘constructional homonymy’) has been defined simply as
“[a]mbiguity explained by differences in syntax” (Matthews 2005: 151). In a more
elaborate formulation, the definition might read as follows: “An utterance is structur-
ally ambiguous when it can yield more than one syntactic interpretation or when it
implies more than one syntactic relationship between constituents within a structure”
(Oaks 2010: 15). These definitions imply that a structurally ambiguous sentence is to
be coupled with two (or more) alternative representations at a deeper level of analysis,
which account for the different semantics and paraphrases. A special case of ambiguous
structures within syntax are the so called “garden path sentences” like, for instance,
The man who whistles tunes pianos, Since John always walks a mile seems like a short
distance to him. But it can be argued that sentences of this type “do not allow more
than one structural interpretation when taken in their entirety, even though the initial
parts of their sentences do” (Oaks 2010: 19).
It must be emphasized, however, that the phenomenon in question is not limited
to syntactic categories (phrases, sentences). Some patterns of English word-formation,
notably within compounding and affixation, comprise complex lexemes that are also
characterized by structural ambiguity. It is the aim of this article to bring together and
to analyze some relevant examples of structural ambiguity in English morphology.
Section 2 presents a brief overview of structurally ambiguous compounds. This topic
has been addressed in the literature on several previous occasions. The parallel ques-
tion of structurally ambiguous affixal derivatives does not appear to have met with
comparable interest. Therefore, in Section 3 we examine the latter problem in some
300  Bogdan Szymanek

detail. Special emphasis will be laid here on ‘derivational homonyms’ – a category that
hitherto seems to have escaped the attention of students of lexical relations in English.
Section 4 concludes the discussion.

2. Structural ambiguity in English compounds

Textbooks on morphology often discuss parallel mechanisms of structural ambiguity

that are a feature of English compounds.1 For example, the following nominal com-
pounds are structurally ambiguous: California history teacher, World Trade Center,
student film society, etc. (Fromkin 2000: 68; Spencer 1991: 310). Accordingly, a com-
plex naming unit like California history teacher allows for two alternative interpreta-
tions paraphrasable as (a) ‘a teacher of California history’ or (b) ‘a history teacher who
happens to be from California’.2 The two interpretations correspond, respectively, to
the following tree-diagram representations (Fromkin 2000: 69):

(1) a.            N b.        N

          N      N       N      N

      N      N          N      N

  California   history teacher   California history   teacher

One might think that compounding, just like sentence formation, is not impervious to
structural ambiguity because it shares a number of properties with syntax. Indeed, when
we compare the abovementioned example with its close analogue, namely American
history teacher, the latter expression exhibits the same kind of ambiguity: (a) ‘a history
teacher who is American’ vs. (b) ‘a teacher of American history’ (Fabb 1998: 72), even
though its first constituent American is, strictly speaking, an attributive adjective – or

1
Of course, English compounds and noun-centered compounds in particular may be ‘ambiguous’,
out of context, in the sense of having unpredictable semantics, which is due to a variety of reasons
and not necessarily because of their inherent structural ambiguity. This is evidenced very well by
new compounds (Bauer and Huddleston 2002: 1647) or (compound) novel naming units (see e.g.
Štekauer 2005).
2
It is arguable that the semantic contrast between the two readings can be marked phonologically,
e.g. by different stress contours and/or pauses; cf. Fabb (1998: 79): “The stress pattern of compounds
may indicate the presence of hierarchical structure inside the compound.” Moreover, Spencer (1991:
320) points out that, in the case of English compounds, stress “can even disambiguate potentially
ambiguous strings”.
 Structural ambiguity in English word-formation  301

a (denominal) relational adjective, from the viewpoint of word-formation. Accordingly,

there are the following two simplified hierarchical representations (Fabb 1998: 72):

(2) a. b.

American     history teacher American history     teacher

Quite apart from the ambiguity evidenced here, there is a major structural problem
concerning the representation in (2a). Is the nonhead element American history there
a phrase or a compound? According to Fabb (1998: 72), it is a compound (within
a compound). However, alternative interpretations of such cases have been put for-
ward in the literature as well. For instance, Carstairs-McCarthy (2005: 37) points out
that, because of its stress pattern, American history as in our example (2a) “looks like
a phrase. We thus appear to have a phrase inside a compound word, that is a syntactic
unit inside a morphological one.” Accordingly, both structural options diagrammed
schematically in (2) receive the following representations in terms of labeled brackets
(cf. Carstairs-McCarthy 2005: 37):

(3) a. [[AmericanA historyN]Nˈ teacherN]N] b. [AmericanA [hístoryN teacherN]N]Nˈ

‘teacher of American history’ ‘American teacher of history’

As further pointed out in Carstairs-McCarthy (2005: 37), “it seems as if only

lexicalized or institutionalized phrases (clichés) can appear freely inside compounds”.
American history teacher is contrasted with superficially analogous but unacceptable
expressions like *glorious history teacher ‘teacher of glorious history’ or *dull history
teacher ‘teacher of dull history’. An alternative analysis of the case under discussion
sketched by Carstairs-McCarthy is based on the assumption that American history
teacher (and other such expressions) has only one structure, consistent with (3b), i.e.
the structure of a phrase, coupled with two possible semantic interpretations (for details,
see Carstairs-McCarthy 2002: 81; 2005: 38–39).
To sum up, expressions like California history teacher and American history teacher
demonstrate that the line separating compounds from phrases is sometimes hard to
draw in English (see also Berg 2011, Giegerich, 2005, 2009). The distinction between
compounds and phrases is also quite elusive in universal terms: “it is frequently difficult
to determine whether a sequence of two nouns is a morphological unit or a syntactic
one, for instance” (Bauer 2010: 138). Moreover, Carstairs-McCarthy, following Sad-
dock (1998), hypothesizes that “[p]erhaps compounding deserves to be regarded as
a third pattern of grammatical organization, distinct from both syntax and morphology”
(Carstairs-McCarthy 2005: 35).
302  Bogdan Szymanek

Alternatively, one can speak here of the ‘lexicon-syntax continuum’: “Indeed, the
demarcation of compounds and phrases has been a major issue in theoretical linguistics
in the last decades and in particular within the Lexical Morphology tradition” (Masini
2009: 256). In order to remedy the situation, Masini (2009: 256) introduces a category
of multi-word units that is intermediate between syntactic objects like phrases and mor-
phological objects like proper compounds. They are designated as ‘phrasal lexemes’;
cf. Italian examples of ‘phrasal nouns’ involving adjectives like carta telefonica ‘phone
card’, disco rigido ‘hard disc’ or red tape in English.
Hence, it should come as no surprise that the boundary between cases of structural
ambiguity in syntax and morphology is not well defined, either. Consider, for example,
the following English sentence: I enjoy Indian summer holidays. This sentence is struc-
turally ambiguous as it can be paraphrased in two different ways: (a) ‘I enjoy summer
holidays in India’, and (b) ‘I enjoy holidays at the time of Indian summer’. Evidently,
the latter interpretation (b) is based on a parsing strategy that identifies Indian summer
as a fixed phrase or, indeed, a ‘phrasal noun’ as a lexical unit. Upon the alternative
interpretation (a), Indian summer (with Indian denoting ‘location’ or ‘destination’) is
neither a lexeme nor a syntactic constituent. When N + N compounds are premodified
by attributive adjectives (cf. Berg 2011), the resulting structures are often ambiguous,
particularly when the adjective is denominal. It can be relational (cf. Indian summer
holidays or American history teacher above) or, less frequently, qualitative; cf. dan-
gerous in the object NP of the sentence Congress passed a dangerous drug bill. In this
sense, derivational morphology, e.g. noun > adjective derivation, has its share in the
structural ambiguity of English compounds (or phrases).
In a somewhat different manner, the boundary between structural ambiguity in
syntax and compounding gets obliterated with expressions like Eye drops off shelf,
characteristic of newspaper headlines.3 Out of context, it is impossible to tell whether
the intended meaning is (a) ‘eye drops have been taken off of the shelves (of a drug
store)’ or (b) ‘an eye fell from a shelf’, even though the latter interpretation seems less
probable. Crucially, given that our attention is focused on only two words here, i.e.
eye and drop (off), the case is different from the ones presented so far, as it shows how
the placement of a constituent boundary alone may decide the status of a compound
(eye drops) or a phrase (drops off shelf). Additional factors like text condensation and
ellipsis evidently play a role in such expressions.
The problem is further complicated by the fact that, quite apart from structural
ambiguity, a given multi-word expression may also suffer from lexical ambiguity when
it contains polysemous items. Thus, for instance, the adjective American (cf. example
(2) above), depending on context, can mean ‘of or relating to the US’ or ‘of or relating
to North or South America’. Moreover, since the paraphrase ‘relating to’ is inherently
ambiguous, the adjective in question can really denote a wide variety of specific rela-

3
Source: http://ibatefl.com/?p=1440.
 Structural ambiguity in English word-formation  303

tions: origin, location, nationality, citizenship, etc. Similarly, the adjective Polish can
refer to any of those concepts, plus ‘language’. This sort of ambiguity sometimes causes
confusion. One fairly recent example is the controversial phrase Polish death camps
used by President Barack Obama by which he apparently meant ‘Nazi Germany’s
death camps operated in Poland’, i.e. he was referring to ‘location’ rather than ‘ori-
gin’ or ‘administration’.4 Incidentally, although the aforementioned phrase is, indeed,
ambiguous, this is not the same kind of ambiguity as in the case of American history
teacher. The ambiguity of President Obama’s phrase seems to be due exclusively to
the polysemy of the adjective Polish, since – structurally (or, in fact, pragmatically) –
the locution admits of only one plausible interpretation: cf. [NP [A Polish] [N [N death]
[N camps]]] vs *[NP [NP [A Polish] [N death]] [N camps]].
In the remainder of this paper we intend to focus on genuine examples of structural
ambiguity because, grammatically, they are more challenging than instances of lexical
ambiguity. Whereas the phenomenon of structural ambiguity has been addressed and
amply illustrated in the literature as regards English compounds, its scope in derivational
patterns has been somewhat neglected. Therefore, the latter topic is taken up below.

3. Structural ambiguity in English derivation

Indeed, when we look at products of derivational morphology, particularly certain

lexemes coined by affixation, there is, again, evidence for structural ambiguity. One
pattern that offers relevant examples in English are deverbal adjectives where the
verbal base is preceded by the prefix un- and followed by the suffix -able. Consider
the following representations, which correspond to two alternative readings of the
adjective unlockable:

(4) a.        A b.      A

      V      suff    pref      A

  pref     
V       V     suff

  un-    lock -able   un-  
lock    -able
   (i.e. ‘that can be unlocked’)    (i.e. ‘that cannot be locked’)

4
The phrase in question was used by President Barack Obama on May 30, 2012, in his address
during the ceremony to commemorate the Polish wartime hero Jan Karski. Source: “Polish” death
camps: Mind your language, Eastern Approaches (blog), May 29th, 2012, The Economist; http://www.
economist.com/blogs/ easternapproaches/2012/05/polish-death-camps.
304  Bogdan Szymanek

As may be seen, the trees in (4) above differ, first, in that (4a) is left-branching
while (4b) is right-branching. Secondly, the ambiguity is occasioned by the fact that
the prefix un- appears here in two functions: either as a reversative (deverbal) marker
in (4a) or as a negative (de-adjectival) formative in (4b).
Moreover, a closer scrutiny of the class of un-V-able derivatives in English reveals
that one more structural configuration must be considered.
It is hard to find an English pattern of the pref-X-suff type, let alone a single de-
rivative, that would illustrate alternative and systematic three-way structuring of the
three morphemes involved, i.e. that would be a case of triple structural ambiguity: left
branching, right branching, as well as parasynthesis (note that unlockable is compat-
ible with only two interpretations). Therefore, for the sake of argument and perhaps
counterfactually, let us assume that the adjective unbribable has no existing simpler
base of the form V-able, i.e. (*)bribable. This will justify the following tripartite (para-
synthetic) representation:

(5)         A

  pref  V  suf

   un-  bribe -able

This interpretation appears to make sense only if we are talking about actually at-
tested lexemes, rather than possible (potential) forms. The fact is that bribable is not
listed in any standard dictionary, including the OED.5 However, it is to be found in the
BNC corpus, just like unbribable, with only one citation (frequency 1): There are no
longer any bribable Vadinamian …, where bribable appears to be used as a neologism
/ nonce formation. On the other hand, bribable gets no hits at all on Google (Decem-
ber 20, 2012). Of course, one might argue that bribable has the status of a possible /
potential form. In fact, the situation seems to be somewhat more complicated here: the
unattested but (possibly) potential V-able forms are likely to occur (if at all) when there
is an element of negation in the sentence (cf. no longer in the example just cited). The
negative item (whatever it is) may be seen as functionally replacing the prefix un-. To
take another example of a similar kind: the pair believable > unbelievable demonstrates

5
The Oxford English Dictionary (version 4.0, 2009, on CD-ROM) lists 1018 matches for words
of the structure unXable, i.e. un*able, most of which are ultimately deverbal adjectives that we
are interested in here. Now, the vast majority of relevant un-V-able forms do have attested simpler
counterparts (bases) of the type V-able, e.g. unthinkable < thinkable. However, a few such adjectival
base-forms find no attestation in this dictionary; e.g. (*)benefitable, (*)bribable, (*)choosable are not
listed, even though the dictionary has separate entries for the corresponding negatives: unbenefitable,
unbribable, unchoosable.
 Structural ambiguity in English word-formation  305

that, intuitively, the base is much less common than the derivative.6 The OED offers
only four citations for the use of believable ‘capable of being believed; credible’ (time
range: 1382–1859), the most recent of which involves sentence negation: And that he
sinn’d is not believable. By contrast, there are 8 citations for its negative counterpart,
viz. unbelievable ‘capable of being believed, credible’ (time range: 1548–1895). The
pretty old last citation of believable in OED (1859) should not mislead us into thinking
that the word is obsolete now. The Cambridge Advanced Learner’s Dictionary (2008)
has an entry for it: “If something is believable, it seems possible, real or true: I didn’t
find any of the characters in the film believable”. Note, again, that the example sentence
characteristically involves negation.7
One may conclude by saying that the type of un-V-able adjectives is not at all
uniform, in terms of synchronic structure, which may produce the effect of structural
ambiguity with some forms: apart from the two well-attested alternative hierarchical or-
ganizations illustrated in (4a, b), there is the likelihood of flat parasynthetic constituency.
The parasynthetic option is quite real, anyway. The principle of parasynthetic deri-
vation has been brought up in analyses of several other patterns of English derivational
morphology. For instance, Adams (2001: 4) points out that “[…] in decaffeinate and
anti-bacterial, prefix and suffix operate together to derive a verb and an adjective from
the nouns caffeine and bacteria.” Another group of examples of parasynthetic forma-
tions involves adjectives with a noun base preceded by a locative or quantitative prefix
and followed by a (denominal) suffix; cf. mono-systemic, multi-racial, bicoloured,
sub-atomic, trans-global, etc. (Adams 2001: 49).8 Plag (2003: 40), who also points
to parasynthesis as a factor that may complicate structural analysis in some cases of
multiple affixation (like e.g. decaffeinate), considers nouns like reorganization or de-
centralization in order to demonstrate that they are inherently (structurally) ambiguous.
It is hard to decide whether reorganization involves right branching or left branching:
“Reorganization can refer to the organization being redone, or it can refer to the process
of reorganizing” (Plag 2003: 40). Both interpretations correspond, respectively, to the
following representations: [re-[organize-ation]] vs [[re-organize]-ation].
The sort of structural ambivalence evidenced here is then comparable, to some
extent, to the much discussed phenomenon of ‘bracketing paradoxes’ in English

6
The intuitive assessment as to their relative frequency is corroborated by a simple Google search:
unbelievable gets 108,000,000 Google hits while believable gets only 18,700. In the BNC, the token
frequency for unbelievable is 526, as opposed to 92 for believable.
7
Marchand (1969: 202), discussing negative adjectives in un-, points out that, within this formal
type, “[m]any adjectives are synthetic formations, i.e. their unprefixed counterparts do not exist”.
This is illustrated with participial adjectives like unassuming, unbending, etc.
8
In addition, Adams (2001: 109, footnote 1) employs the concept of parasynthesis in her analysis
of certain lexemes which involve both suffixation and compounding: “‘Syntactic’ noun compounds
like bicycle-repairing […] can be analysed as either [[bicycle repair]ing] or as [[bicycle] [repairing]],
since they are in effect parasynthetic formations […]”.
306  Bogdan Szymanek

morphology,9 invoked in analyses of forms like unhappier, ungrammaticality, or trans-

formational grammarian. Formal considerations enforce the following partial bracket-
ings: [un [happi-er]], [un [grammatical-ity]], [[transformational] [grammarian]]. The
bracketings based on meaning, on the other hand, are: [[un-happy] -er], [[un- grammati-
cal] -ity], [[transformational grammar] -ian]. However, there is a significant difference
between, say, reorganization as a case of structural ambiguity and ungrammaticality
as a bracketing paradox. In the former case, the two alternative structural representa-
tions are both motivated semantically,10 i.e. each of them receives a plausible semantic
interpretation (paraphrase). In the latter case, however, the two competing structural
representations are due to a ‘mismatch’ of form and meaning.
But still, among the putative instances of bracketing paradoxes in English, there
are expressions whose structural analysis may already involve semantic considerations,
which leads, again, to the question of ambiguity. Consider cases like French historian.
As opposed to, say, transformational grammarian which cannot be paraphrased as
‘grammarian who is transformational’, French historian is ambiguous as it can mean
either (a) ‘historian who is French’ or (b) ‘expert in French history (not necessarily
a French person)’ (Carstairs-McCarthy 2002: 80). Whereas the former interpretation is
compatible with the structure of an ordinary noun phrase in English and so it is unprob-
lematic, on the latter interpretation one might postulate a representation which reflects
the semantically motivated option of attaching a personal suffix to the institutionalized
phrase French history, thus: [[French histori-]-an] (Carstairs-McCarthy 2002: 80).11
Similarly, a phrase like Russian teacher is ambiguous: (a) ‘a teacher who is Russian;
(b) ‘someone who teaches Russian’. Expressions of this sort, involving derivational
affixes, are analysed as instances of ‘scope ambiguity’ in Beard (1991). Consider a few
more examples (adapted from Beard 1991: 196):

(6) nuclear physicist

a. [nuclear] [physicist]  ‘a physicist who is nuclear (to some project)’
b. [nuclear physic]ist  ‘someone who studies nuclear physics’
criminal lawyer
a. [criminal] [lawyer]  ‘a lawyer who is criminal’
b. [criminal law]yer  ‘someone who practices criminal law’
moral philosopher
a. [moral] [philosopher]  ‘a philosopher who is moral’
b. [moral philosoph]er  ‘someone who studies moral philosophy’

9
For an overview, see e.g. Spencer (1991: 397–422), Carstairs-McCarthy (1992: 92–97).
10
The traditional term ‘double motivation’ seems appropriate here.
11
Carstairs-McCarthy (2002: 80) contemplates one more solution here: “Or should we say that,
with both interpretations, the structure of the expression is the same (namely [[French] [historian]]),
but that for one of the interpretations this structure is a bad guide?”.
 Structural ambiguity in English word-formation   307

Evidently then, the expressions listed in (6) are ambiguous. Moreover, they are structur-
ally ambiguous, quite apart from the fact that, in each case, both readings may differ in
their relative frequency or plausibility. “The problem is that these constructions seem to
have, in addition to a wide scope reading [[Xx][Yy]], exemplified by (a), which parallels
syntactic structure, a narrow scope reading [[Xx Y]y], exemplified in (b), which does not”
(Beard 1991: 196). The case is interesting also in the sense that it demonstrates, again,
how structural ambiguity in English morphology may be entangled with the syntax, not
only in the case of compounding (cf. Section 2), but also as regards affixal derivation.
What is relevant from the viewpoint of the present discussion is that, based on the
examples given so far, in order to contemplate structural ambiguity in morphology,
we need to have a minimum of three constituents (stems, affixes) per lexeme, A+B+C
(as evidenced, for instance, by both the compound in (1)12 and the derivative in (4)).
Additionally, there are three basic structural configurations available, i.e., for right-
headed combinations: (a) a complex modifier followed by a simplex head; (b) a simplex
modifier followed by a complex head; or (c) a flat and symmetrical structure, where
the input form is flanked by two satellites (parasynthesis, as illustrated in (4)). When
viewed synchronically, such cases are often fairly indeterminate and may be described
as instances of double (multiple) motivation. From a diachronic viewpoint, they may
show signs of ‘reanalysis’, when, for instance, structure of type (b) replaces structure
of type (a); cf. oak-panelled paraphrasable as either ‘panelled in oak’ or as ‘having
oak panels’ (Adams 2001: 134). Another interesting question is this: could there be
any structural ambiguity in English morphology when a lexeme is composed of just
two meaningful elements? The following pairs of English deverbal adjectives in -able
appear to demonstrate that this is, indeed, possible, given a theory that is powerful
enough (Aronoff 1976: 123):

(7) a. cómparable b. compárable

réfutable refútable
préferable preférable
dísputable dispútable

The words in (7a) differ from those in (7b) phonologically,13 i.e. in stress placement,
as well as semantically: adjectives of type (7b) tend to have compositional meanings
while those of type (7a) often do not; cf. compárable ‘capable of being compared’ vs

12
Given the recursiveness of English N+N compounds, structural ambiguity may also arise when
there are four, five, or even more constituents. For example, because student film society is ambiguous,
i.e. [student [film society]] or [[student film] society] (cf. Spencer 1991: 310), it follows that student
film society committee is structurally ambiguous as well.
13
Because the words differ in stress, then they are ambiguous only in their orthographic form. Cf.
also Giegerich (1999: 29) on stress placement in the -able adjectives under discussion.
308  Bogdan Szymanek

cómparable ‘similar, equivalent’ (Aronoff 1976: 127). Within the model of Aronoff
(1976), two distinct -able suffixes are postulated in order to account for the observed
differences, viz. +abl and #abl, for the words in (7a) and (7b), respectively. Leaving
aside the lexical distinction that is thereby created (two suffixes rather than one), we
should note here, in particular, the positing of two boundary elements, i.e. # as a strong
boundary and + as a weak boundary. Crucially, as to the status of those boundaries,
Aronoff (1976: 122) argues as follows: “Boundaries have neither sound nor meaning.
They affect the two in parallel manners and are therefore not elements of linguistic
substance, but rather elements of linguistic structure.” Assuming the validity of this (or
any comparable) solution, evidently it enriches the scope of what is possibly meant by
‘structural ambiguity’: it turns out that two structurally ambiguous words might have
identical hierarchical structure (bracketing) but differ in (structural) boundary elements.
The following section will demonstrate that another special kind of structural
ambiguity may be attributed to the so-called derivational homonyms.

3.1. English derivational homonyms

Before we proceed, let us introduce informal definitions of the key terms to be used
in this section:
• Complete homonyms – words which have identical pronunciation as well as
spelling
• Homophones – words which have identical pronunciation but different spelling
• Homographs – words which have identical spelling but different pronunciation
Of course, the phenomenon of homonymy in the English lexicon has been discussed
and amply illustrated in numerous studies and textbooks, including in particular the
classificatory controversies that it engenders, notably concerning the lack of a clear-
cut dividing line between homonymy and polysemy (cf., for instance, Jackson and Zé
Amvela 2000: 73; Liberman 2005: 205; Lipka 1992: 135; Pinker 2008: 108). Mat-
thews (2005: 164) defines homonymy as “[t]he relation between words whose forms
are the same but whose meanings are different and cannot be connected: e.g. between
pen ‘writing instrument’ and pen ‘enclosure’.” Allan (2001: 42) offers the following
definition of homonymous listemes: “Two listemes of the same form are homonymous
if they warrant separate lexicon entries because the identity of form is coincidental.”
The special term “derivational homonyms” is not to be found anywhere in such
discussions as, in fact, it does not seem to be used in English linguistic jargon. The
available accounts are not sensitive to the question of morphological complexity in
English homonyms, perhaps because the vast majority of English homonyms are not
morphologically complex, anyway. Nevertheless, it is precisely an aim of this paper to
demonstrate that (a) derivational homonyms do exist in English; (b) they do have inter-
esting properties of their own, and so (c) the term (and concept) ought to be incorporated
 Structural ambiguity in English word-formation  309

into current grammatical parlance.14 We propose the following tentative characterization

of the notion in question, i.e. derivational homonyms as a special kind of homonym:
• Derivational homonyms (or homophones/homographs) differ in morphological
structure and complexity, i.e. one is simplex (monomorphemic) while the other is
complex (polymorphemic) or both are polymorphemic but they involve distinct
affixes and/or morphological boundaries in different positions.
Derivational homonyms arise due to the operation of word-formation processes, no-
tably affixation. Consider, as our first example, the following pair of unrelated words: the
noun entrance ['entrəns] and the verb entrance [ɪn 'trɑ¦ns]. Since the two forms differ in
pronunciation but not in orthography, they illustrate the particular concept of derivational
homography. Both words are morphologically complex, but their structure is different:
the noun entrance is a deverbal coinage with the suffix -ance (thus enter + ance) while
the verb entrance is a denominal formation featuring the prefix en- (en + trance).
At first glance, it might appear that, apart from affixation, conversion in English
regularly yields comparable pairs, i.e. cases like (total) homonymy (illustrated in (8a))
or just homography, when, for instance, the members of a pair differ in stress and/or
vowel quality (cf. (8b)).15 Consider the following examples:16

(8) a. Noun Verb

pilot ['paɪlət] pilot ['paɪlət]
balance ['bæləns] balance ['bæləns]
feature ['fi¦tʃə] feature ['fi¦tʃə]
partition [pɑ¦ 'tɪʃən] partition [pɑ¦ 'tɪʃən]

b. Noun/Adjective Verb
absent ['æbsənt] absent [æb 'sent]
perfect ['pɜ¦fɪkt] perfect [pə 'fekt]
permit ['pɜ¦mɪt] permit [pə 'mɪt]
attribute ['ætrɪbju¦t] attribute [ə 'trɪbju¦t]
moderate ['mɒdərət] moderate ['mɒdəreɪt]

However, the sort of structural ambiguity evidenced by conversion pairs is of

a different kind. Even though it can be argued that both members of such a pair differ
in morphological structure or complexity (particularly when a zero suffix is postulated),
14
The category of “derivational homonyms” is not a new idea, cross-linguistically. It is successfully
employed, for instance, in Polish lexicology; cf. Polish homonimy słowotwórcze ‘derivational hom-
onyms’, as opposed to homonimy leksykalne właściwe ‘lexical homonyms proper’ (see e.g. Buttler
1988: 7–9); cf. also Szymanek (2004), (2010: 265).
15
See, for instance, Gimson (1980: 233) for more examples of pairs with stress-shift and concom-
itant vocalic alternations.
16
The phonetic transcriptions employed here and elsewhere in this text are according to Wells
(1990), though occasionally simplified.
310  Bogdan Szymanek

it is crucial to note that they are morphologically related and hence those cases fall be-
yond the scope of homonymy as it is normally understood (i.e. as a relationship holding
between two lexically distinct and often etymologically unrelated items).17 It would
make more sense to speak of derivational homonymy in cases like to process ['prəᴜsəs]
(conversion based on the noun process) vs. to process [prəᴜ' ses] (a back-formation
from procession, OED); at least, as one can see, two distinct derivational processes are
involved here.18 On some accounts, regular conversion-related pairs, like for instance
hammer (noun) > hammer (verb), represent so called complementary polysemy, as
opposed to contrastive ambiguity evidenced by homonyms (Pustejovsky 1995: 27–28).
On the other hand, the phenomenon of derivational homonymy (-phony, -graphy) and
the ambiguity involved therein bears some resemblance to the phrasal pairs like a name
/ an aim, ice cream / I scream, grey tape / great ape, my turn / might earn, etc., which are
usually discussed by phonologists (see, for instance, Roach 2000: 144).19 The possible
ambiguity of such expressions in spoken language concerns the placement of “internal
open juncture”, i.e. the word boundary that separates two neighbouring constituents.
This is comparable to the problem of positing the morpheme boundary in the word(s)
entrance, as mentioned above. Actually, there is some overlapping between phenomena
of both types, i.e. the syntactic and morphological ambiguity under discussion, since
pairs can be given where one member is a phrase while the other is a complex word; cf.
a head vs. ahead. Specifically, ‘minimal pairs’ like nitrate and night rate demonstrate
that the ambiguity in question may also concern derivatives and compounds, respectively
(quite apart from the fact that in some dialects or idiolects they are not pronounced iden-
tically).20 Oaks (2000: 22) mentions another relevant example where a whole sentence
can be homophonous with just one word: Did you make her? / Jamaica.
As regards the narrowly defined topic of English derivational homonyms, Table 1
given below offers a background typology with examples. It lists exemplary pairs of
English lexemes arranged according to two significant parameters: firstly, the type of
homonymy involved, i.e. (a) complete homonyms, (b) homophones, and (c) homographs;
and, secondly and more importantly, morphological complexity. The latter factor gives
us, a priori, a threefold division of homonymous pairs: (A) simplex / simplex, i.e. cases
where both items are morphologically simple or underived, (B) simplex / complex,

17
However, opinions differ on this issue. For example, according to Lipka (1992: 138), a conver-
sion-related pair like a can > to can ‘put into a can’ ought to be regarded as a case of homonymy.
18
Verbal outputs of denominal conversion occasionally result in two stress-differentiated and
semantically distinct forms as well; cf. ábstract A/N > abstráct ‘remove’ vs. ábstract ‘summarize’,
súrvey N > survéy ‘look at or examine, etc.’ vs súrvey ‘carry out a survey’.
19
Also etymologists deal with similar phenomena from a diachronic viewpoint. Cf. the misdivision
(metanalysis) or redistribution of boundaries involved in changes like a napron > an apron or mine
uncle > my nuncle (Liberman 2005: 99).
20
Cf., for instance, the pertinent remarks on this issue, concerning juncture, in John Wells’s phonetic
blog (1st October 2012) at: http://phonetic-blog.blogspot.com/2012/10/the-wardrobe-in-bedroom.html.
 Structural ambiguity in English word-formation  311

where one member of the pair is morphologically simple while the other is derived, and
(C) complex / complex, i.e. instances where both members of the pair have the status
of derivatives (although they may differ in the details of their morphological structure).
It will be seen that, from the viewpoint of the topic under discussion, the types desig-
nated as (B) and (C) deserve closer scrutiny whereas all kinds of homonymy involving
underived forms (type (A) – simplex / simplex) will be ignored in what follows, simply
because morphology is not at stake here. In other words, we will have nothing more to
say below about pairs like: (a) bear N / bear V, bank N / bank N, etc. (underived complete
homonyms); (b) sea N [si¦] / see V [si¦], sweet A [swi¦t] / suite N [swi¦t], bow V [baᴜ] /
bough N [baᴜ], etc. (underived homophones); or (c) bow N [bəᴜ] / bow V [baᴜ], lead V
[li¦d], lead N [led], etc. (underived homographs). Incidentally, such items account for the
vast majority of homonymous pairs in English. According to McArthur (1992: 483), “[t]
here are over 3,000 homographs in the Concise Oxford Dictionary (8th edition, 1990)”.
Today, lists of English homonyms are available on the Internet, as well as article-length
discussions of their various aspects; not to mention analyses in printed publications. So it
might seem that the topic has been thoroughly exhausted. Indeed, it would be all too easy
to amass and replicate the evidence concerning English homonyms in general. However,
one aspect seems to be missing in the discussions so far. This is the morphology of the
existing homonyms and, in particular, cases where the homonymy relation is due to the
operation of a word-formation process. Cases of this sort are relatively infrequent but
they are worth investigating because, as a class, they pose some interesting questions
about the interplay of phonology, morphology, semantics, and the lexicon.

Table 1. A typology of English homonyms according to the parameter of morphological

complexity
(a) (b) (c)
Morphological Complete Homonyms Homophones Homographs
complexity (Heterographs) (Heterophones)
A. (A.a) (A.b) (A.c)
Simplex / Simplex bear N [beə] (animal) see V [si¦] lead V [li¦d]
bear V [beə] (carry) sea N [si¦] lead N [led]
B. (B.a) (B.b) (B.c)
Simplex / Complex ? wait V [weɪt] irony N ['aɪərǝni]
weight N [weɪt] irony A ['aɪəni]
weigh + t iron + -y
C. (C.a) (C.b) (C.c)
Complex / Complex finer A ['faɪnə] stationery N ['steɪʃənəri] entrance N ['entrəns]
fine + -er Comp. stationer + y enter + -ance
finer N ['faɪnə] stationary A['steɪʃənəri] entrance V [ɪn 'trɑ¦ns]
fine + -er Ag ‘one who station (+)ary en- + trance
fines’ (obsolete)
312  Bogdan Szymanek

Table 1 gives one example for each type of homonymy, unless there is a gap (i.e.
no relevant examples have been found) – cf. the cell marked as (B.a). Disregarding
the trivial case of simplex pairs of type (A), let us now elaborate on the relevant data,
by giving further examples of derivational homonyms which represent types (B) and
(C), i.e. the classes which are of our immediate concern. In particular, the following
three varieties find attestation in dictionaries of English:

(9) Homophones of type (B.b) – simplex / complex

cellar N ['selə] / seller N ['selə] sell + -er
finish V ['fɪnɪʃ] / Finnish A ['fɪnɪʃ] Finn + -ish
forth Adv [fɔ¦θ] / fourth Num [fɔ¦θ] four + -th
hangar N ['hæŋə] / hanger N ['hæŋə] hang + -er
minor A ['maɪnə] / miner N ['maɪnə] mine + -er
navel N ['neɪvəl] / naval A ['neɪvəl] navy + -al21

(10) Homographs of type (B.c) – simplex / complex

invalid N ['ɪnvəli¦d] / invalid A [ɪn 'vælɪd] in- + valid
number N ['nʌmbə] / number A ['nʌmə] numb + -er
polish V ['pɒlɪʃ] / Polish A ['pəᴜlɪʃ] Pole + -ish
sewer N ['su¦ə] ‘drain’ / sewer N ['səᴜə] sew + -er ‘one that sews’
statist N ['stætɪst] ‘statistician’ / statist N ['steɪtɪst] state + -ist
supply V [sə 'plaɪ] / supply Adv ['sʌpəlli] supple + -ly
shower N ['ʃaᴜə] / shower N ['ʃəᴜə] show + -er ‘one that shows’
tier N [tɪə] / tier N [taɪə] tie + -er ‘one that ties’

(11) Homographs of type (C.c) – complex / complex

bustier A ['bʌstiə] busty+ -er / bustier N ['bʌstɪeɪ] bust + -ier
furrier A ['fɜ¦riə] furry + -er / furrier N ['fʌriə] fur + -ier
multiply V ['mʌltɪplaɪ] multi- + ply / multiply Adv ['mʌltəpli] multiple + ly
periodic A [ˌpɪəri 'ɒdɪk] period + -ic / periodic A [ˌpɜ¦raɪ 'ɒdɪk] per- + iod + -ic
resign V [rɪ 'zaɪn] re1- (+) sign / resign V [ri¦ 'saɪn] re2- + sign

It appears that the suffix that is most commonly found in English derivational homonyms
is -er. This is partly due to the fact that, strictly speaking, there are two homophonous
suffixes of that form, i.e. the comparative -er1 (in adjectives and adverbs) and the agen-
tive/instrumental -er2 (in nouns).22

21
Although the adjective naval is, ultimately, a borrowing from Latin, it appears to be bimorphemic
synchronically, being motivated by the noun navy.
22
Cf. also the pair better (adjective, comparative) vs. better / bettor (noun; ‘one who bets’). Here,
however, the bimorphemic status of the suppletive comparative better is rather dubious.
 Structural ambiguity in English word-formation  313

4. Comments and conclusion

A few comments are now in order, concerning the distinct groups of derivational
homonyms given above. Firstly, the lists may appear uncharacteristically short. How-
ever, no claim is being made here as to their exhaustiveness. Rather, in each case we
are dealing with a selection of representative and relatively straightforward pairs even
though, admittedly, the number of all relevant examples illustrating each case is not
high. Morphological complexity is a gradable concept, even when it is understood in
purely synchronic terms. Therefore, the list in (9), for instance, does not include the pair
site (or: cite) / sight, even though some native speakers may perceive the relationship
between the verb see and the noun sight as transparent enough for the latter form to
be treated as morphologically complex. Moreover, the fact that the present analysis is
focused on derivational morphology prevents us from discussing numerous instances
of homophones where one member of a pair is an inflected form; cf. the regular noun
plural or the third person present tense endings in, for instance, bruise / brews, choose
/ chews, clause / claws, cox / cocks, cruise / crews, nose / knows, phlox / flocks, raise
/ rays, sax / sacks, size / sighes, tax / tacks.23 Consider as well the regular past tense
(past participle) endings in bard / barred, board / bored, fold / foaled, hold / holed,
least / leased, mind / mined, mist / missed, mode / mowed, pact / packed, road / rowed,
side / sighed, staid / stayed, sword / soared, tact / tacked, told / tolled, wade / weighed,
etc. Some irregular verbs also contribute to the stock of attested homophones; cf. herd
/ heard, maid / made, red / read, etc.
Secondly, it is worth emphasizing the apparent lack of complete derivational homo-
nyms in English, if we disregard cases of conversion-related pairs. Based on a casual
search of the data, I have not been able to find any relevant examples for type (B.a) and
only one instance for type (C.a); cf. Table 1. Although the evidence for all six types
is scanty, indeed, and so generalizations are hard to arrive at, it appears that there is
a relative preference for pairs whose members preserve their formal identity at least
partially, either at the level of speech or writing, i.e. (derivational) homophones or
homographs. If this generalization is correct, it would amount to saying that complete
homonyms in English tend to be represented only by pairs of underived lexemes (usu-
ally monosyllables, like the noun bear and the verb bear). This feature of the English
lexicon appears to be language specific and seems to be largely due to the complex
nature of the phonetic/graphemic correspondences. By contrast, it ought to be pointed
out that, for instance, Polish offers quite an impressive set of lexical pairs which are
complete derivational homonyms, indistinguishable in both written and spoken form

23
Additionally, a pair like rights (noun) and writes (verb) demonstrates that, since the two inflec-
tional markers (plural -s on the noun and 3rd person sg. present tense -s on the verb) are homophonous,
when they are combined with homophonous stems we get two homophonous inflected forms.
314  Bogdan Szymanek

(for instance: rannyA I ‘wounded’ < rana ‘wound’ vs. rannyA II ‘morning attr.’ < rano
‘morning’; podrobićV I ‘to forge, countefeit’ < robić ‘to do/make’ vs podrobićV II ‘to
crumble, perfective’ < drobić ‘id., imperfective’, etc.).24
Thirdly, it has occasionally been pointed out in the morphological literature that
a certain type of blocking prevents the formation of complex words that otherwise might
be totally homonymous with particular, well established lexemes. Thus, for instance,
Bauer (1983: 289), while discussing subject (agentive) nominalizations in English,
notes the following: “In four cases blocking appears to prevent the use of -er: adaptee
(adaptor is an instrument), knockee (knocker is an instrument), mergee (merger is what
the mergees are a party to) and waitee (waiter is a profession)”. Moreover, Bauer adds
that (*)meeter ‘sb. who meets’, a conceivable subject nominalization of the verb to
meet, “might have been blocked by the homophonous meter/metre” (Bauer 1983: 289).
These remarks may lend further support to the claim that complete (total) derivational
homonymy in English is extremely rare.
Overall, it turns out that cases of structural ambiguity in English word-formation
that are solely due to derivational processes are no less challenging and diversified than
their counterparts in the realm of compounding.

References

Adams, V. 2001. Complex words in English. Harlow: Longman.

Allan, K. 2001. Natural language semantics. Oxford: Blackwell.
Aronoff, M. 1976. Word formation in generative grammar. Cambridge, MA: MIT Press.
Bauer, L. 1983. English word-formation. Cambridge: Cambridge University Press.
Bauer, L. 2010. An overview of morphological universals. Word Structure 3 (2): 131–140.
Bauer, L., and R. Huddleston. 2002. Lexical word-formation. In R. Huddleston, and G. K. Pul-
lum (eds.), The Cambridge grammar of the English language, 1621–1721. Cambridge:
Cambridge University Press.
Beard, R. 1991. Decompositional composition: the semantics of scope ambiguities and ‘bracket-
ing paradoxes’. Natural Language and Linguistic Theory 9: 195–229.
Berg, T. 2011. The modification of compounds by attributive adjectives. Language Sciences
33 (5): 725–737.
Buttler, D. 1998. Słownik polskich homonimów całkowitych. Wrocław: Ossolineum.
Carstairs-McCarthy, A. 1992. Current morphology. London and New York: Routledge.
Carstairs-McCarthy, A. 2002. An introduction to English morphology. Edinburgh: Edinburgh
University Press.
Carstairs-McCarthy, A. 2005. Phrases inside compounds: a puzzle for lexicon-free morphology.
SKASE Journal of Theoretical Linguistics 2 (3): 34–42.

24
See Szymanek (2010: 265–8) for more examples and discussion.
 Structural ambiguity in English word-formation  315

Fabb, N. 1998. Compounding. In A. Spencer, and A. M. Zwicky (eds.), The handbook of mor-
phology, 66–83. Oxford: Blackwell.
Fromkin, V. A. (ed.). 2000. Linguistics. An introduction to linguistic theory. Oxford: Blackwell.
Giegerich, H. J. 1999. Lexical strata in English. Morphological causes, phonological effects.
Cambridge: Cambridge University Press.
Giegerich, H. J. 2005. Lexicalism and modular overlap in English. SKASE Journal of Theoreti-
cal Linguistics 2 (2): 43–62.
Giegerich, H. J. 2009. The English compound stress myth. Word Structure 2 (1): 1–17.
Gimson, A. C. 1980. An introduction to the pronunciation of English. London: Edward Arnold.
[3rd edition]
Jackson, H., and E. Zé Amvela. 2000. Words, meaning and vocabulary. An introduction to
modern English lexicology. London: Continuum.
Liberman, A. 2005. Word origins … and how we know them. Etymology for everyone. Oxford:
Oxford University Press.
Lipka, L. 1992. An outline of English lexicology. Lexical structure, word semantics, and word-
formation. Tübingen: Max Niemeyer.
Marchand, H. 1969. The categories and types of present-day English word-formation. München:
C. H. Beck. [second revised edition]
Masini, F. 2009. Phrasal lexemes, compounds and phrases: A constructionist perspective. Word
Structure 2 (2): 254–271.
Matthews, P. H. 2005. Oxford concise dictionary of linguistics. Oxford: Oxford University Press.
McArthur, T. (ed.). 1992. The Oxford companion to the English language. London: BCA.
Oaks, D. D. 2010. Structural ambiguity in English: An applied grammatical inventory. London:
Continuum.
Pinker, S. 2008. The stuff of thought. Language as a window into human nature. London:
Penguin Books.
Plag, I. 2003. Word-formation in English. Cambridge: Cambridge University Press.
Pustejovsky, J. 1995. The generative lexicon. Cambridge, MA: MIT Press.
Roach, P. 2000. English phonetics and phonology. A practical course. Cambridge: Cambridge
University Press. [third edition]
Saddock, J. M. 1998. On the autonomy of compounding morphology. In S. G. Lapointe,
D. K. Brentari, and P. M. Farrell (eds.), Morphology and its relation to phonology and
syntax, 161–187. Stanford: CSLI.
Spencer, A. 1991. Morphological theory. An introduction to word structure in generative gram-
mar. Oxford: Blackwell.
Štekauer, P. 2005. Meaning predictability in word formation. Amsterdam/Philadelphia: John
Benjamins.
Szymanek, B. 2004. Morphophonological markers of “partial” lexical polysemy/homonymy
in English and Polish. In H. Kardela, W. J. Sullivan, and A. Głaz (eds.), Perspectives on
language, 281–292. Lublin: Wydawnictwo Uniwersytetu Marii Curie-Skłodowskiej.
Szymanek, B. 2010. A panorama of Polish word-formation. Lublin: Wydawnictwo KUL.
Wells, J. C. 1990. Longman pronunciation dictionary. Harlow: Longman.
316  Bogdan Szymanek

Abstract

The phenomenon of structural ambiguity, which commonly appears at the level of syntax, re-
veals itself in the lexicon as well. It is the aim of this article to bring together relevant cases of
structural ambiguity in English word-formation, i.e. complex lexemes that arise as ambiguous
structures due to the operation of processes like compounding or affixation. A brief overview
of structurally ambiguous compounds is presented first. Next, the discussion focuses on the
ambiguity of English affixal derivatives, a topic which has been somewhat neglected in the
literature. In particular, special emphasis will be laid on ‘derivational homonyms’ – a category
that hitherto seems to have escaped the attention of students of lexical relations in English. For
example: entrance (noun) vs. entrance (verb). The data under analysis demonstrate that several
patterns of English affixation as well as compounding contribute to the stock of structurally
ambiguous lexemes. The sources and kinds of this ambiguity are manifold, partly depending
on what is meant by ‘morphological structure’.