The function of word-formation and the inflection-derivation distinction.

In
Henk Aertsen, Mike Hannay & Rod Lyall (eds), Words in their Places. A Festschrift
for J. Lachlan Mackenzie. Amsterdam: Vrije Universiteit. 2004, 283-292.

The Function of Word-Formation and the Inflection-Derivation

Distinction1
Laurie Bauer

Victoria University of Wellington

Lachlan has always been interested in the function of language, or the function of
constructions. In the context of a Festschrift for Lachlan, we might ask what the
function of word-formation is. The answer is easily available to us in the
literature (since this is far from the first time the question has been asked),
although the terminology may be rather variable: word-formation has two
functions, firstly a function of lexical enrichment, whereby new words are coined
to denote new, or newly salient, concepts, and secondly a transpositional
function, whereby lexemes (which we assume to have a fixed word-class) are
permitted to appear in a new word class so that the same meaning can be
transferred to a new function in a sentence.

Having answered that question, at least for the moment, we can go back
and ask a different question, though one which has worried linguists for some
time: what is the difference between inflectional morphology and derivational
morphology? Again this is not the first time the question has been asked, and
answers (plural) are easily available to us in the literature. For one thoroughly
exhaustive treatment, see Plank (1994). Plank allows us to deduce that the
categories of inflection and derivation are prototypical categories, from which
the categories of individual languages such as English and Dutch may differ
quite considerably. Such a view is also held by Dressler (1989) and by Bauer

1
I should like to thank Geert Booij for comments on an earlier version of this
paper. He is not to blame for the content, of course.

1
(1988). Other authorities give us rather more clear-cut answers. One such answer,
which has been extremely influential, is that given by Anderson (1982: 587):
‘inflectional morphology is what is relevant to the syntax’.

Now we are in the position of having answers to two questions which we

can put side by side. One function of word-formation is to allow words (in some
vague interpretation of the term) to appear in different word classes in different
sentence functions. Inflectional morphology is that morphology which is relevant
to syntax. The implication seems clear: word-formation is (or can be, or is at least
some of the time) a matter of inflectional morphology. The only difficulty with
this conclusion is that it flies in the face of all that we have been taught: word-
formation deals with derivational morphology, compounding and some rather
less central processes, but not with inflection; anything which changes word-
class is derivation.

Fortunately, we are saved from heresy at this point by the fact that two
scholars have already gone before to show the path. Booij (1996) argues that
inflection is not monolithic, but that it divides into two distinct classes. The first
of these classes is contextual inflection, inflection which is determined by
concord and government within the sentence. Thus contextual inflection includes
adjectival agreement with nouns in Romance languages for number and gender,
verbs agreeing with their subjects or objects in number, person, gender/noun
class in a wide range of languages, and subjunctive marking where it is
demanded by a matrix clause (as in French Il faut que j’aille à Paris ‘I have to go to
Paris’ where the subjunctive form of go is demanded by the construction il faut
que ‘it is necessary that’), though not where it is a free choice in a main clause (as
in English God save the Queen!). The second of these classes, Booij calls inherent
inflection. Inherent inflection is syntactically relevant but not constrained by the
sentence structure. Plurality on nouns and tense on verbs are two of Booij’s
examples. Whether a subject noun is plural or singular may well have an effect
on the form of the verb, but is determined not by the sentence structure but by
the choice of the speaker (corresponding to the real-world situation to be
described). This division is a useful, but it has the disadvantage that the same
formal category may sometimes be inherent inflection and sometimes contextual

2
inflection. For instance, the uses of the subjunctive mentioned above may be
different, and the comparative in (1) seems to be contextual inflection (unless its
presence is demanded by the semantics of the construction rather than by any
agreement), while the same form in (2) appears to be a case of inherent inflection.

(1) The red block is bigger than the blue one.

(2) I need a bigger block here.

The other pathfinder is Haspelmath (1996). He argues that it is possible to

have word-class changing inflection. Adverbial -ly in English is one of his
examples. This is, according to Haspelmath, inflectional in the sense that it is
regular, general and productive, but nonetheless transpositional.

Why should English adverbialisation be seen as inflectional but not English

nominalisation? We can go back to Chomsky (1970), where it is pointed out that
the forms of English nominalisations are unpredictable and that the semantics of
English nominalisations is neither regular nor constrained by the particular affix
or other morphological process used to create the nominalisation. Chomsky
gives examples such as belief, laughter, marriage to illustrate the formal
unpredictability of nominalisations, and examples such as belief, marriage,
revolution to illustrate the semantic problems.

The answer to the formal side of this problem was given in Bauer (1983: 79),
though the complete evidence was not available until Bauer (2001). That answer
is, in a word, lexicalisation. Nominalisations persist in a natural language like
English long after their coining (or long after they were borrowed, if that is the
relevant means of introduction of the new word into English). We are rarely in
the position of having to coin a nominalisation for a verb which we have not
inherited, ready-made, as part of our language’s norm. We use the established
nominalisation, whether or not that nominalisation could be produced in the
current state of the language system. When it comes to those few instances where
we are genuinely called upon to create a new nominalisation, we have two
possibilities: avoid the issue or create a new one. If the verb for which we wish to
create a new nominalisation uses the suffixes -ate, -ise or -ify, there is no choice of
nominalisation process. We must add -ion, -ation, -cation respectively. This is

3
important because these are the only affixes with which new verbs can be created
in current English (Plag 1999). This leaves us (a) with instances of new verbs
created by conversion and (b) with established verbs which have no established
nominalisation. It seems that we are adept at avoiding this issue. My favourite
example of continued avoidance is the case of the verb ignore. We would expect
to have a nominalisation for this verb, yet ignorance has become so specialised in
meaning that it is no longer obviously related to ignore. The Oxford English
Dictionary provides ignoration with the required meaning, but speakers do not
use it, and tend to find it impossible when it is suggested. Rejection of this form
could be because it does not fit with the productive patterns available to speakers
of current English. In other words, pace Chomsky, the creation of new
nominalisations to allow new verbs to be used as nouns is perfectly regular.

At one level, the problem of the lack of predictability of the meaning of a

nominalisation persists. For example, if we look in The Oxford English Dictionary
for relatively recent words using -ation, we find, among others, dollarization ‘The
action or process of basing the value of a national currency upon that of the U.S.
dollar’, myristoylation, ‘The process of becoming myristoylated; the state of being
myristoylated’, Saudiization, ‘The process or result of rendering (more) Saudi
Arabian in character’. Superficially, these entries suggest that nominalisations
can mean any of action, process, state or result. However, as is pointed out by
Bauer (1983: 188), similar vagueness can be found in the meaning of words which
are not derived nominalisations. Consider noise and farewell in (3):

(3) The noise continued for about twenty minutes. (State)

When I knocked over the saucepans, the noise was appalling. (Result)
His farewell was effusive. (Action)
Her farewell consisted of three major phases. (Process)

If these various readings are available for non-de-verbal nominalisations, then

they cannot derive from the process of nominalisation per se, and so cannot be
part of the meaning that these words hold qua nominalisations. Thus much of the
variation in the meaning of nominalisations is something inherent in nouns
rather than a feature of a morphological process. Accordingly, any of the
nominalisations cited above can be given a consistent meaning, which we can

4
gloss as ‘noun from verb’. The grammatically important feature is thus the
transposition, not that part of the meaning which might appear variable.

Of course, lexicalisation applies here, too. It cannot be predicted a priori that

residence can mean something concrete or that trial can be connected with the
legal system. But these meanings arise because there is a long tradition in the use
of words which are grammatically specified as ‘nominalisation of RESIDE’ and
‘nominalisation of TRY’. The grammatical specification is all that is needed in the
coining of new words.

Another side to this is that we do not appear to be able to distinguish the

various existing nominalisation suffixes from each other semantically. Given -age,
-al, -ance, -ation, -ence, -ery, -ment, -s, -t, -th, -ure, -y (see Bauer 2001: 177-83), we
cannot assign any meaning of nominalisations as a class to just one of these (or
even to some subset of these). Any nominalisation meaning may co-occur with
any of these forms (though some are so rare as to make a full range unlikely).
Similarly, though -esque seems to carry its own meaning, the adjectival suffixes
-al, -ar, -(i)an, -y do not appear to differ semantically, and Plag (1999) argues that
-ise and -ify are synonymous. We can thus argue that, while there are affixes like
-esque (more will be added later) which not only change word-class but also add
their own determinable meaning, there are those whose sole function is to effect
a transposition. The argument here is that we may wish to set these up as a
separate class of affixes (or word-formation processes), alongside contextual
inflection and inherent inflection. We can term this ‘transpositional2 morphology’
(and take no stance on whether such morphology is derivational or inflectional).

This leaves us with that part of what has traditionally been called
‘derivation’ whose function is to expand the lexicon by modifying the meaning
of existing lexemes. Here we find such well-known instances as gender-marking
morphology (German Entericht ‘drake’ from Ente ‘duck’; English princess from
prince), morphology which marks agents, patients, instruments, locations (cf

2
The term ‘transposition’ is used more narrowly by Marchand (1969); this use
follows Haspelmath (1996).

5
English killer, interviewee, blender, diner, respectively), modal-marking on
adjectives (cf. eatable, payable), negation and reversative marking (cf. unequal,
inequality, undo, dethrone), as well as subtler meanings such as that shown by
-esque cited above.

Are there any other categories which we might want to distinguish on the
inflection-derivation cline? There may well be, and my provisional classification
here is not meant to exclude extra suggestions. There are two cases which may
have to be given particular consideration.

The first of these is the case of evaluative morphology, typified by the use
of augmentatives and diminutives. There are many ways in which diminutive
morphology, in particular, has been shown not to be typical of inflection nor of
derivation. For example, evaluative affixes are often transparent to the word-
class and/or gender of the base (see Scalise 1984: 131). Diminutives in many
languages are more productive than is typical of derivational morphology, yet
with a proliferation of possible markers which is otherwise more expected in
derivational morphology.

The second type is morphology which leads not to a change of word-class

but to a change of sub-type of word-class, in particular valency-changing
morphology for verbs (causatives, intransitivisers, transitivisers). These are rare
in English, where such changes are not usually morphologically marked (e.g. The
dog walked, I walked the dog; Flowers grow on this bank, I grow flowers). Things such
as the change from king to kingdom, kitchen to kitchenette, and biography to
biographee are excluded from this category since the affixes carry more meaning
than simply the grammatically specified output class of noun here, but it may
well be the case that some instances are difficult to classify. We can label this
type by is core members and call-it ‘valency-changing’, though a better label
would be preferable. I assume that changes of countability in English nouns
would fit into this category if they were morphologically marked.

If the suggestions above are accepted, we might wish to say we are dealing
with not two classes of morphology (inflectional versus derivational) but six:
contextual, inherent, valency-changing, transpositional, evaluative, lexicon-

6
expanding. Many of the criteria that have been traditionally taken to distinguish
inflection from derivation can be shown to be divided among these six
categories.

(a) Only contextual inflection is syntactic in the sense that it shows agreement
between various parts of syntactic structures or is introduced by
government.

(b) Both contextual and inherent inflection maintain the same lexeme (I make
the assumption here that no noun and verb can ever belong to the same
lexeme since they enter into different paradigms, and a fortiori that no
intransitive and transitive verb can belong to the same lexeme, since each
must be capable of entering into its own paradigm, and the two need not
match). The other types create new lexemes. Alternatively, in many
languages we can view this break as distinguishing those forms which
may appear within compounds or word-formation from those which must
be part of the syntax. The morphology which cannot occur word-internally
is precisely the inflectional morphology. Note that English is rather
insecure with regard to this particular point, since it does allow plurals
inside words, at least marginally on some analyses: admissions policy, teeth-
ridge, etc.

(c) The first three types are, in traditional terminology, class-maintaining

(they maintain the major word-class, e.g. noun, adjective, verb; the latter
three are class-changing, the involve shifts from one of noun, verb or
adjective (in some cases some other category) to another from the same
list. Evaluative morphology is awkwardly placed here, since it is typically
class-maintaining, though it can be class-changing

(d) The first four types share some grammatical function, while the last two
have a function which is more lexical in nature: categories such as
augmentative add new lexemes whose function is not determined by
grammatical needs, as do those categories which appear in the lexicon-
expanding set.

7
(e) The last type is much less automatic than the other types. If there is no
instrument in English called a retriever, it is not a matter of morphology
that the slot is vacant, it is a matter of societal needs. For the other
categories considered, it tends to be grammatically driven whether or not
particular transpositional classes exist or not. Where they exist, we expect
to find some form available in most slots, or some way of creating words
to fit the slots in a regular manner (though, of course, it may not be
predictable over the established vocabulary precisely what form will fit in
any given slot). In the lexicon-expanding set, while there may be some
well-established slots with regular exponence, it is frequently not true that
the number of slots is predetermined. Thus English has experimented over
the last half-century or so (with greater or lesser success) with slots which
can be filled by words in -nik, -(n)omics, -(t)eria and so on. This difference
can be summarised by saying that the lexical-expansion type of
morphology is less paradigmatic than other types.

A tabular summary of these major distinctions is presented in Table 1.

INSERT TABLE 1 ABOUT HERE

The creation of a new set of categories where there have previously been
only two or three inevitably brings problems with it. There are two obvious
problems with the suggestion made here: (1) it creates more borderlines, and
thus more borderline disputes; (2) it splits the outputs of more morphological
processes into different categories. The first of these problems is something that
has to be tested on real data: if the benefits of the categorisation outweigh the
borderline disputes, then the categorisation will have a value; otherwise it will be
discarded.

To illustrate the second problem, consider the outputs of -ery suffixation in

English. We find straightforward transpositional uses as in jugglery from juggle,
we find collective instances such as greenery from green or jewellery from jewel,
and these I take to be lexicon-expanding, and the clearly lexicon-expanding
locational nouns in rookery and swannery. The same morphological process feeds
— albeit not productively — two of the suggested categories. Either this suggests

8
that we should consider adopting a separationist view of morphology, following
Beard and others (see e.g. Beard 1994), or it suggests that we need to be much
stricter in distinguishing between productive and non-productive uses of
morphology. For example, we may take it, following Aronoff’s work, that -ity is
still used productively on certain bases to derive abstract nouns which we might
gloss, in Marchandian fashion as ‘state, quality, condition of —’, but once we
start talking about an atrocity, the muzzle velocity of a bullet, uttering profanities
and minor divinities we are no longer seeing productive uses of -ity affixation as
a morphological process, but the historical process of semantic change. It is not
entirely clear to me to what extent we can solve problems of this kind by
restricting our analysis to what happens in productive morphology, but I am
convinced that it will at least reduce the apparent size of the problem, if not
remove it entirely.

It is of some interest to consider other criteria that are standardly given for
distinguishing inflection from derivation, to see how they fare against this six-
way division. Such criteria include regularity of meaning, productivity,
obligatoriness, the relative size of the sets of affixes concerned and affix ordering
(see e.g. Bauer 1988, Dressler 1989).

(a) Regularity of meaning. The regularity of transpositional meanings was

argued for above, contra Chomsky (1970). It was noted, however, that
transpositional morphology is open to lexicalisation, which may render
this original regularity opaque. The same is true with lexicon-expanding
morphology. We cannot predict from the form of lover that it will mean a
person who has a sexual (rather than a purely emotional) relationship with
another. We can predict the meaning we find in music-lover. The sexual
meaning of lover is not morphological, but due to lexicalisation. That being
the case, we would expect to find similar lexicalised instances of
evaluative morphology. The Spanish use of the augmentative of the word
for ‘rat’ meaning ‘mouse’, ratón, must count here, as must the Klamath use
of the diminutive of ‘horse’ to mean ‘dog’, and the word for ‘sheep’ being
an augmentative of the word for ‘rabbit’ in Natchez and Tunica (Mithun
1999: 468, 532). English changeling, foundling and groundling also have more

9
 specific meanings than could be deduced from their form, and we could
argue that the specialisation of duckling, gosling, spiderling to mean ‘young
of the species’, while princeling focuses on negative connotations and
darling on positive ones are all indications of lexicalisation. (See also Bauer
1997: 551.) As was mentioned earlier, English tends not to have valency-
changing morphology, but in Diyari we find examples such as the
transitive form of a verb meaning ‘die’, palima-, can only take the noun
meaning ‘fire’ as its object and is specialised in the sense of ‘extinguish (a
fire)’ (Austin 1981: 160). Semantic lexicalisation (or idiomatisation) can also
affect inherent inflection. Beard (1982) cites examples of semantically
irregular plural forms such as feelings, greens, heavens, looks, waters.
Contextual inflection does not appear to lexicalise. If we look at semantic
regularity in terms of its inverse, lexicalisation, then it appears that, to a
large extent, the further left we are on the scale in Table 1, the less
lexicalisation there is likely to be. Evaluative morphology may be less
subject to lexicalisation that transpositional morphology, but we do not
have statistical information to support such a claim at the moment.

(b) Productivity. It is tempting to equate productivity with paradigmaticity in

Table 1, on the view that each implies that there will be a form which will
fit into a particular slot or cell in the paradigm (even if the form will be
unpredictable in some instances). The example of the nominalisation of
ignore, cited above, suggests that this easy equation will not work, since
transpositional morphology, at least in English, appears to allow for a
greater number of gaps in the system that we might expect. This is
probably not accidental, and valency-changing morphology might well
show similar gaps depending on the degree of societal need for vocabulary
that we have already seen affecting lexicon-expanding morphology.
Perhaps a new criterion should emerge from this discussion: the degree to
which the coining of new forms is determined by grammatical imperatives
as opposed to societal needs, with those ruled mainly by societal needs
being at the less productive and less paradigmatic end of the scale. In this
context it should be noted that different languages may make use of

10
 different kinds of morphology in different ways. Turner (1973: 181)
comments on the nominal style of English scientific writing, and students
in translating to and from French are often taught to use a more verbal
style in French (see Ritchie 1963: 73 for a hint of such matters). If it is the
case that different styles or different language use more nominals than
others, there may be greater use of and greater regularity in
nominalisation markers in the one than in the other. This remains to be
proved, but seems a reasonable hypothesis.

(c) Obligatoriness. Obligatoriness is something of an awkward concept in

morphology under any circumstances: is it the category or the morph
which is obligatory, and under what set of circumstances? Let us take it
that it is the morphological category which may be obligatory in a
particular syntactic construction. This definition virtually defines
obligatoriness as a feature of the erstwhile inflectional categories,
contextual and inherent. Complex words from any of the other categories
are likely to be replaceable in context with morphologically simpler
(because not marked for valency-change, transposition, evaluation or
lexical-expansion, respectively) words.

(d) Relative sizes of affix sets. Doubt is cast on this potential criterion in Bauer
(1988), and without a serious attempt to count the comparable sets in a
number of languages nothing can be concluded here.

(e) Affix ordering. Again, without some cross-linguistic statistical survey a

sensible discussion of this is not possible. If we think in terms of Bybee’s
(1985) criterion of relevance, we might expect something like the reverse of
the order given in Table 1 (assuming suffixation), but such a hypothesis
would need to be tested against the findings in a range of languages.
Evaluatives are notorious for not showing the expected order with relation
to inherent inflection (see Bauer 1988, 1997: 546-7 for some examples and
references). On the other hand, examples like the following show the
expected patterning, if not every possible combination:

(4) Dutch: bakk·er·ij·tje ‘little bakery’ (bake·AGT·LOC·DIMIN)

11
 (5) French: mang·er·ons ‘we shall eat’ (eat·FUT·1PL)

(6) Telugu: tßa·mp·æ…·Îu ‘killed’ (die·TRANS·PAST·3SGMASC)

(Krishnamurti & Gwynn 1985: 203)

(7) Warekena: nu·kuÛua·ba·mia·Ûu ‘I drank too much’

(1SG·drink·AUG·PERF·EMPH) (Aikenvald 1998: 385)

Having said that, it must also be admitted that, even ignoring the cases
where evaluative morphology is apparently ordered in strange positions, it is
relatively easy to find apparent counterexamples, like that in (8) where valency-
changing and transpositional morphology come in the wrong order. Clearly
some kind of statistical survey would be needed to see what the preferred order
for these suggested categories is cross-linguistically.

(8) Macushi: te·es·enyak’ma·se·n ‘worker’

(ADVL·DETRANS·work·ADVL·NOML) (Abbott 1991)

We might also note things such as diminutives are often reported as

allowing the same suffix to apply to its own output (for Xhosa, Karok and
Spanish examples see Bauer 1997: 548-9), valency-changing morphology may
occasionally do the same, as in Kannada (but there apparently with a maximum
of two iterations; Sridhar 1990: 276), and lexicon-expanding morphology may do
the same when it is semantically appropriate, as in meta-meta-file or German Ur-
ur-grossvater ‘great great grandfather’. I am not aware of any instances of
recursive transpositional morphology without intervening transpositional
morphology of a different type. This discontinuity in the scheme in Table 1 may
or may not be important.

Even with the extended schema suggested here, the position of evaluative
morphology may not be completely secure. Nevertheless, it is of interest that the
other types do seem to provide some support for the ordering that is laid out in
Table 1.

By starting out with a functional question, we have come back to a new

look at the formal system of morphology, with a solution to an old problem
arising because we have approached the problem from a functional angle.

12
Whether the solution has lasting value or not, the general principle ought to be
one of which Lachlan would approve.

References

Abbott, Miriam 1991. Macushi. In Desmond C. Derbyshire & Geoffrey K. Pullum

(eds), Handbook of Amazonian Languages Vol 3, Berlin and New York:
Mouton de Gruyter, 23-160.

Aikhenvald, Alexandra Y. 1998. Warekena. In Desmond C. Derbyshire &

Geoffrey K. Pullum (eds), Handbook of Amazonian Languages Vol 4, Berlin
and New York: Mouton de Gruyter, 225-439.

Anderson, Stephen R. 1982. Where's morphology? Linguistic Inquiry 13: 571-612.

Austin, Peter 1981. A Grammar of Diyari, South Australia. Cambridge: Cambridge

University Press.

Bauer, Laurie 1983. English Word-formation. Cambridge: Cambridge University

Press.

Bauer, Laurie 1988. Introducing Linguistic Morphology. Edinburgh: Edinburgh

University Press.

Bauer, Laurie 1997. Evaluative morphology: in search of universals. Studies in

Language 21: 533-575.

Bauer, Laurie 2001. Morphological Productivity. Cambridge: Cambridge University

Press.

Beard, Robert 1982. The plural as lexical derivation. Glossa 16: 133-148.

Beard R. 1994. Lexeme-morpheme base morphology. In R.E. Asher (ed.), The

Encyclopedia of Language and Linguistics, Oxford, etc.: Pergamon, vol. 4,
2137-2140.

Booij, Geert 1996. Inherent versus contextual inflection and the split morphology
hypothesis. In Geert Booij & Jaap van Marle (eds), Yearbook of Morphology
1995, Dordrecht, etc.: Kluwer, 1-16.

13
Bybee, Joan L. 1985. Morphology. Amsterdam/Philadelphia: Benjamins.

Chomsky, Noam 1970. Remarks on nominalization. In Roderick A. Jacobs &

Peter S. Rosenbaum (eds), Readings in English Transformational Grammar,
Waltham, Mass., etc: Ginn, 184-221.

Dressler, Wolfgang U. 1989. Prototypical differences between inflection and

derivation. Zeitschrift für Phonetik, Sprachwissenschaft und
Kommunikationsforschungen 42: 3-10.

Krishnamurti, Bh. & J.P.L. Gwynn 1985. A Grammar of Modern Telugu. Delhi:
Oxford University Press.

Haspelmath, Martin 1996. Word-class-changing inflection and morphological

theory. In Geert Booij & Jaap van Marle (eds), Yearbook of Morphology 1995,
Dordrecht, etc.: Kluwer, 43-66.

Marchand, Hans 1969. The Categories and Types of Present-Day English Word-
Formation. Second edition. Munich: Beck.

Mithun, Marianne 1999. The Languages of Native North America. Cambridge:

Cambridge University Press.

Plag, Ingo 1999. Morphological Productivity. Berlin and New York: Mouton de
Gruyter.

Plank, F. 1994. Inflection and derivation. In R.E. Asher (ed.), The Encyclopedia of
Language and Linguistics, Oxford, etc.: Pergamon, vol. 3, 1671-1678.

Ritchie, R.L. Graeme 1963. A New Manual of French Composition. Cambridge;

Cambridge University Press.

Scalise, Sergio 1984. Generative Morphology. Dordrecht: Foris.

Sridhar, S.N. 1990. Kannada. London and New York: Routledge.

Turner, G.W. 1973. Stylistics. Harmondsworth: Pelican.

14
Table 1: A representation of the differences between the 6 suggested categories

Contextual Inherent Valency-Changing Transpositional Evaluative Lexicon-expanding

Agreement No agreement

Lexeme-maintaining Creating new lexemes

Class-maintaining Class-changing

Grammatical Lexical

Paradigmatic Non-paradigmatic

15