Defectiveness in Morphology
Summary and Keywords
Morphological defectiveness refers to situations where one or more paradigmatic forms of a lexeme are not realized, without plausible syntactic, semantic, or phonological causes. The phenomenon tends to be associated with low-frequency lexemes and loanwords. Typically, defectiveness is gradient, lexeme-specific, and sensitive to the internal structure of paradigms.
The existence of defectiveness is a challenge to acquisition models and morphological theories where there are elsewhere operations to materialize items. For this reason, defectiveness has become a rich field of research in recent years, with distinct approaches that view it as an item-specific idiosyncrasy, as an epiphenomenal result of rule competition, or as a normal morphological alternation within a paradigmatic space.
1. Concept Delimitation: Paradigmatic Gaps Versus Defectiveness in the Narrow Sense
Descriptively, defectiveness is the situation where a lexeme lacks one or more forms of the paradigm otherwise associated to words of the same class. For instance, in French verbs agree in person and number with their subject, and they have a distinct form for the first person plural (1a). However (Morin, 1987; Arrivé, 1997; Boyé & Cabredo-Hofherr, 2010), some verbs like clore ‘close’ lack that form (1b):
Importantly, the expected form in (1b) for 1st person plural, closons, would not violate any principle of French phonotactics, and there are no plausible syntactic or semantic reasons why this verb should not be used in combination with a subject nous. In this article, cases like (1) will be referred to as ‘defectiveness in the narrow sense’, while we will use the term ‘paradigmatic gap’ or ‘defectiveness in the wide sense’ (cf. also Sims, 2006) to refer to cases such as (2), from Hungarian, where there are plausible morphology-external causes for the gap. In Hungarian, there is a synthetic form for the subjunctive (2a), but some verbs lack it (2b).
In (2a) the form is impossible because it involves a phonotactically unacceptable cluster of three consonants (*-mlh-) (Lukács, Rebrus, & Törkenczy, 2010). The problem is, then, that the surface form exhibits an apparently irreparable sequence in phonology.
It is thus necessary not to equate defectiveness with any situation where there is a paradigmatic gap. There is a certain consensus in the literature (Sims, 2006, 2015; Baerman & Corbett, 2010; Stump, 2010; Nevins, Damulakis, & Freitas, 2014; Nevins, 2015), pace the distinct technical implementations, that defectiveness involves a gap in the operations that map the set of morphosyntactic features (the function) to a morphophonological representation (the form), as represented in (3).
This contrasts, definitionally, with situations where there is a gap because the combination of morphosyntactic features, or its interpretation, is ungrammatical for an item (4a), as in the cases discussed in §1.1. Defectiveness is also different from situations when the morphophonology of the resulting form would be ineffable at the phonological level (4b), as in example (2b).
These cases will be considered in turn with the goal of obtaining a better delimitation of what narrow defectiveness is.
1.1 Gaps Due to Syntactic Restrictions
In some cases, there is a gap because the morphosyntactic structure that would produce the form would violate a syntactic principle. Baerman and Corbett (2010, p. 4) mention one such case, from Georgian (cited from Tschenkeli): one of the three words for ‘nobody’, nuravin, lacks an ergative form that the other two forms (aravin and veravin) have. However, the explanation for this gap lies in the syntax: nuravin is used only under the scope of the prohibitive negative nu. The meaning associated to nu restricts its use to present or future contexts. However, Georgian restricts ergative case to past (aorist) contexts. The conclusion is that nuravin lacks an ergative form simply because the case value and the negation value required never appear in the same syntactic context. The problem is not that the morphosyntactic features fail to associate to a form, but that the morphosyntactic features required for this particular form cannot be put together to begin with.
Another relevant example in this sense is reflexive pronoun paradigms. It is very frequent that they lack a nominative form, as in Latvian (5) (Nau, 1998).
However, this gap is not surprising when one considers the syntactic conditions under which an anaphora can appear. Specifically, anaphors must take an antecedent that controls them from a higher hierarchical position within their own clause (Chomsky, 1981). As a consequence, in principle anaphors should not occupy the subject position (cf. *Himself bought the book for John).
As in Georgian, the lack of a nominative form for the Latvian anaphoric pronoun can be reduced to a situation where the form is missing because the morphosyntax that would need it is ungrammatical. Defectiveness in the narrow sense, in contrast, emerges in situations where the relevant morphosyntactic features should in principle be able to combine with each other.
1.2 Gaps Due to Semantic Restrictions
Similarly, it is important to distinguish defectiveness from situations where the morphosyntactic combination of features, while not ungrammatical per se, would give an impossible semantic interpretation. One obvious candidate for this is the fact that weather verbs (rain, snow, hail, etc.) lack first and second person forms. Speakers can produce without hesitation the first and second person form of these verbs. The problem is that the meaning of these verbs makes them incompatible in their literal sense with human subjects. However, if they are used in a metaphorical sense, the forms are acceptable and attested. In (6a), we have a verse by the Guatemalan poet Manuel José Arce where the Spanish verb llover ‘rain’ is used in first person singular.
1.3 Gaps Due to Phonological Restrictions: Ineffability
Similarly, defectiveness should be differentiated from situations where the morphosyntactic features are grammatical but the form is not attested because it would give rise to a sequence that fatally violates a number of phonological principles. In such cases, it is perhaps more appropriate to talk about ineffability. One relevant example comes from Norwegian (Rice, 2003). In Norwegian verbs, the imperative form is the verbal stem minus the final vowel -e, producing a form that is surface-identical to the root.
When the root ends in a vowel, a single consonant, a geminate, or a consonant cluster of falling sonority, there is an imperative form.
However, when the root ends in a cluster of rising sonority, the imperative form is avoided by speakers because the resulting form violates the sonority hierarchy within a syllable (Clements, 1990).
Narrow defectiveness is different from such cases for two reasons. The first is that it is entirely predictable which bases will lack an imperative form: there is a well-defined property in the phonological shape of their roots that predicts it. The second is that some speakers produce and accept imperatives of these verbs in case different phonological processes avoid the fatal infraction of sonority (e.g., making the imperative disyllabic as in åp.n, or adding an extra vowel at the end, as in åp.na), showing that what fails here is not the mapping between function and form.
That said, depending on the analytical framework assumed, ineffability might be interpreted as narrow defectiveness due to theory-dependent assumptions. Such is the case in Optimality Theory (Prince & Smolensky, 2004), where there is a prevalence of surface representations and the violation of a principle does not automatically lead to ungrammaticality. In Optimality Theory, phonologically conditioned paradigmatic gaps and defectiveness are virtually indistinguishable, because they both involve situations where the grammar fails to produce an output for a form. We refer the reader to the collection of papers in Rice and Blaho (2009) for a detailed discussion of different approaches to paradigmatic gaps in OT.
1.4 Gaps Due to Lexical Specialization
Baerman and Corbett (2010, p. 3) also show that in other cases the paradigm of a word can be incomplete simply because the form—perhaps through historical accident—is only used inside a particular lexicalized expression. For instance, in Spanish we have a plural form creces ‘increase’ that speakers are unable to associate to any singular form (*crece, *crez). The reason is that for the vast majority of speakers this form is restricted to the idiomatic construction in (10) (cf. also Harley’s, 2014 proposal that some items only receive interpretation in very specific contexts).
1.5 Defectiveness in the Narrow Sense
Compare what has been said until now with a case of defectiveness in the narrow sense, such as the absence in Russian of a genitive plural form of some nouns; here we follow Sims’ (2015, pp. 82–86) detailed presentation of the facts. The noun in (11) is one example of noun that lacks a genitive plural form:
First, note that there is no plausible reason why this noun should not appear in a genitive plural context such as the nest of the crows, for syntax or for semantics. Second, there are no plausible phonological reasons for the genitive plural of (11) to be ill-formed. Genitive plural in this declension class is marked as ø, and the root ends in a cluster /rg/, which has lowering sonority. There are, however, ways in Russian to avoid this cluster, but perhaps even more tellingly, the accusative plural form karg, which would be homophonous with the expected genitive plural form, seems unproblematic for Zaliznjak (1977, p. 46) and is quite acceptable for the speakers consulted by Sims, which proves that there is no fatal infraction in the surface form karg per se. Finally, the word is not restricted only to a specific idiom. This is, therefore, a prototypical case of defectiveness where a specific lexical item lacks an otherwise expected form that the morphosyntactic system needs and whose phonological structure is acceptable. There is no identifiable infraction in the morphosyntactic feature combination, and no irreparable infraction in the morphophonological surface form, and yet the morphology fails to produce a form for the genitive plural.
1.6 Defectiveness and Syncretism
Taking defectiveness in the narrow sense as a failure to map a morphosyntactic representation into a morphophonological form allows also for a clear demarcation from syncretism. Consider Latin declensions. In the second declension, Latin marks differently genitive singular and dative singular (12a); however, in the fifth declension the form of the genitive is identical to that of the dative (12b).
In a sense, res ‘thing’ lacks a distinct form for the genitive, but the noun can be used in genitive contexts; it is just that a form belonging to a different cell in the paradigm is used. This situation, schematically, corresponds to (13) (contrast with 3).
The grammar, in such cases, does have a distinct form to map the morphosyntax to the morphophonology. See Sims (2006) and Stump (2010) for the interaction between syncretism and defectiveness, a rich area of study that we will not be able to cover here.
2. Challenges of Defectiveness for Morphological Theories
Defectiveness is problematic under the light of some theoretical assumptions. Some of the challenges presented by defectiveness are more general, and others are restricted to only some theories.
To start with the most general theoretical problem caused by defectiveness, in particular for widely accepted assumptions about acquisition, L1 Acquisition theory generally assumes that the child is not sensitive to negative evidence, that is, the evidence that a particular form is ungrammatical (Brown & Hanlon, 1970). Language acquisition would then proceed in accordance with the subset principle (Baker, 1979; Berwick, 1985); that is, the child would assume the most restrictive grammar compatible with the input perceived, and then positive evidence—utterances exhibiting other grammatical constructions in the language—would, if necessary, motivate the child to posit an incrementally more permissive grammar. In this context, the question is in what way a speaker learns that a particular word form does not exist.
Different approaches have been proposed here. One of them is to argue that the child gets indirect positive evidence of the absence of a form through the observation that in the relevant contexts the input uses an alternative word, or a periphrasis, instead of the expected word (Orgun & Sprouse, 1999; Johansson, 1999). Another is to challenge the claim that negative evidence is not available to children acquiring the language (Sokolov & Snow, 1994), or adopt the view that frequency of usage is crucial in the acquisition process (Kemmer & Israel, 1994), perhaps within a memory-rich network system (Sims, 2015). In these two approaches, failing to encounter a particular form can lead the learner to infer that a paradigm is defective.
Second, specifically for competition-based theories of morphological exponence—such as Distributed Morphology (Halle & Marantz, 1993), Optimality Theory (Prince & Smolensky, 2004), or Paradigm Function Morphology (Stump, 2001)—defectiveness presents an additional problem. In competition-based theories of morphology, the speaker has a default rule or operation that applies to an element X in order to produce any required form (cf. Kiparsky’s Elsewhere Principle). The existence of defectiveness is at odds with the hypothesis that such default rules exist (see also Lignos & Yang, 2016). Several answers have been suggested here, again, and will be revisited in §6.1 and §6.2: a form might be defective because in its shape two possible rules clash, because it is not really parsed by morphology, or because the item the operation applies to is restricted in a way that in many cases the default rules will give undefined outputs.
Third, for Neo-Constructionist theories of morphology, which propose that syntactic operations build up words and morphology is reduced to spelling out those structures—such as Distributed Morphology, Nanosyntax (Starke, 2009; Caha, 2009), and the Exoskeletal approach of Borer (2013)—defectiveness is problematic because it seems to be a situation in which the absence of a morphological shape conditions the use of a syntactic combination of features. In principle, these theories would expect that the availability of a syntactic combination is blind to the morphological properties. One conceivable solution would be to propose that lexical insertion acts as a filter that can discard some syntactic derivations because there are no items to spell them out, as suggested sometimes in Nanosyntax, but the problem persists if the theory is additionally Item-and-Arrangement, because then words are treated as combinations of morphemes with individual exponents, and as has been seen there is no plausible reason that would block the use of a specific morpheme in the context.
Defectiveness is, in fact, one of the strongest arguments in favor of a paradigmatic approach to word formation and the storage of full word forms. As Mithun (2010) notes, the existence of defectiveness suggests that words, even inflected words, are stored in memory, because if speakers always built inflected words through operations, the absence of an expected form is at odds with the assumption that there are default rules. This does not mean, as Mithun notes, that speakers lack rules that let them compute inflected forms, but it strongly suggests that they should be stored.
Before moving to the next section, it should be noted that it is frequently said that defectiveness is a challenge to the claim that inflection is maximally productive. Leaving aside the problem of how to define productivity (see Bauer, 2001), Sims (2015) correctly points out that maximal productivity of inflection is a claim about the combination of morphosyntactic features, not about the combination of morphophonological objects. In this sense, defectiveness, which is about the mapping between morphosyntax and morphophonology, does not falsify the productivity of inflection.
3. Properties of Defectiveness
Four properties have typically been associated to defectiveness (cf. Hetzron, 1975; Hudson, 2001; Albright, 2003; Baerman & Corbett, 2010; Sims, 2006, 2015; Baerman, 2011; Maiden & O’Neill, 2010, among many others): lexical selectivity, morphological selectivity, sensitivity to inflectional patterns, and gradience.
Lexical selectivity refers to the fact that one specific item can be defective while phonologically or semantically close items in the same category can have full paradigms. For instance, in Spanish the verb pacer ‘to graze’ lacks a first person singular form and the present subjunctive (14). However, the phonologically similar placer ‘please’ and the virtual synonym pastar ‘to graze’ have such forms (15), (16):
Morphological selectivity means that defectiveness applies to some morphological forms but not others that should be related in terms of their morphosyntactic feature endowment. The same verb pacer ‘to graze’ that lacks all present subjunctive forms has full imperfective subjunctive forms.
There is no obvious morphosyntactic feature that both first person singular in the present indicative and all present subjunctive forms have, but other first person singular forms and imperfective subjunctive forms lack. The question is, then, what the first person singular present indicative and the present subjunctive have in common in Spanish. This takes us to the third property of defectiveness, sensitivity to inflectional patterns. The connection between the forms is that many verbs in Spanish display an irregularity pattern where a distinct lexeme allomorph is used in precisely these cells (the L-pattern: Maiden, 2004, 2005) (18).
The observation made by Boyé and Cabredo-Hofherr (2010) and Maiden and O’Neill (2010) is that when a lexical item is defective, one frequently finds situations where the missing forms would all have shared the same type of allomorph. The connection between the forms affected by defectiveness is—depending on the approach—morphophonological (the necessary allomorphic stem is missing, cf. Arregi & Nevins, 2014) or morphomic (the defectiveness pattern piggybacks on a morphological alternation that has meaning internal to the paradigmatic system).
The last typical property of defectiveness is gradience. Speakers might marginally accept and use some of the forms that are in principle missing from defective items, and some of the intended forms could be perceived as less deviant than others. Continuing with Spanish examples, the verbs agredir ‘to assault’ and abolir ‘to abolish’ are both defective for many speakers, who only allow forms where the root is followed by /i/ (Albright, 2003). However, for other speakers the first singular present indicative form agredo ‘I assault’ is marginally acceptable, but not perfect, while the forms abolo ‘I abolish’ is more deviant, but still more acceptable than abuelo ‘I abolish,’ with diphthongisation of the mid root vowel. In the case of pacer ‘to graze,’ some speakers reject both pazo and pazco as first singular forms, but report that pazo would be even less acceptable than pazco. Thus, there are degrees of defectiveness.
4. Causes of Defectiveness
The literature has also devoted some space to the question of what causes defectiveness. One frequently mentioned factor has to do with historical considerations (Hetzron, 1975; Sims, 2006; Baerman & Corbett, 2010; Baerman, 2011). Baerman and Corbett (2010) use the term ‘arrested development’ to refer to situations where the cause of defectiveness could be related to an item being of recent coinage or introduced at a later stage in the language. Many of the defective verbs mentioned above for Spanish are not documented until the 16th century and could be considered loanwords from Latin (Maiden & O’Neill, 2010). Baerman and Corbett (2010) also mention the case of Russian denominal verbs of recent coinage, such as pylesosit′ ‘to vacuum’ (from pylesos, ‘vacuum cleaner’), where speakers are reluctant to apply an expected final consonant alternation in the first person singular.
In a different sense, what Corbett & Baerman call ‘decay’, an otherwise productive rule or operation might be lost or progressively become more restricted in its application, resulting in incomplete paradigms. Hetzron (1975) observes that defectiveness for some Hungarian verbs could be related to the loss of a productive epenthesis rule to break triconsonantal clusters, which is now restricted to only a subset of verbs.
Similarly, the loss of productive phonological rules might lead to an otherwise complete paradigm being split in two when speakers stop perceiving two allomorphs as related. In that splitting, the new paradigms can be incomplete, as it is the case in Polish with the root dżdż- ‘drizzle,’ which lacks a nominative form. The explanation, according to Baerman and Corbett (2010), lies in the fact that dżdż- used to be an allomorph of deszcz- ‘rain’ for all forms except nominative; when the two forms separated and each produced its own paradigm, dżdż- did not develop a nominative form.
Besides diachronic reasons, avoidance of homophony is often cited as a cause of defectiveness. While this in many cases has the flavor of a post hoc rationalization (Lass, 1980), Baerman (2011) documents several cases where homophony avoidance seems to be at play, even if not as a sufficient condition. In his view, homophony can be a factor provided that other factors contribute, such as the homophonous word being taboo or the availability of alternative forms. In Tuvaluan, speakers whose demonstrative adjective in the plural is kinei avoid the third person form, which should be kilaa, because it is homophonous with an adjective for ‘hairless’ that has ‘improprietous undertones’ (Besnier, 2000, p. 419). Baerman notes that homophony plays a role here because there is an alternative form kolaa (used by younger speakers) that can replace the homophonous word.
Phonological identity also can trigger defectiveness at the affix level. Bat-El (2009) notes that Hebrew lacks a plural for forms such as sod-iy-út ‘secrecy.’ The reason is that the plural could be *sodiyut-ot, where there is a string of two affixes with /t/, or in order to avoid the two /t/, *sod-iy-uy-ot, which involves a string of two suffixes with /y/. Either form violates the Obligatory Contour Principle at the affix level, leading to defectiveness.
Other proposals mention low frequency of the lexical items subject to defectiveness as a factor that triggers speaker uncertainty in how some of the paradigmatic forms should be. This connects to the historical explanation to the extent that some borrowings or words that are introduced at later stages in the language tend to be of low frequency. Albright (2003) note that in an experimental task, speakers showed gradient levels of uncertainty with Spanish low-frequency verbs like despavorir ‘to frighten,’ hender ‘to split,’ empedernir ‘to harden,’ or aguerrir ‘to inure.’ Crucially, as Albright himself emphasizes, the low frequency is—like homophony avoidance—not a deterministic factor; it plays a role only if the verb, given its morphological shape, is a candidate to exhibit an allomorphic alternation, for instance one between a diphthongized and a non-diphthongized form depending on whether stress falls on the mid vowel of the root (/o/, /e/).
It is important to emphasize that the vast majority of published works on the topic agree that these factors play some role in making a paradigm defective, but none of them is a necessary or sufficient condition.
5. The Role of Attestations in the Study of Defectiveness
Starting from the assumption that a speaker feels hesitant with respect to what the form should be in a defective form, researchers discuss whether few attestations, or even one single attestation that fills a gap, is enough to argue that a lexeme is not defective. Raffelsiefen (2004, footnote 11) suggests that finding occasional examples in online corpora, or Google, should not be enough to argue that one form is not missing. A similar position is made by Sims (2006), who provides some criteria to determine when attestations should be enough to doubt that a form is defective. She discusses the example of the English verb forego, which is classified as defective in its past tense (*foregoed, *forewent). However, in Google searches both forms are attested. Importantly, the past tense forms of the verb constitute only 0.2% of all the attestations of the lexeme forego, which is much less than with a ‘normal’ verb (Sims estimates as 5% the average proportion of use of a past verb in English). In this view, what defines a gap is not the absolute number of attestations, but the deviation from the frequency of use in comparison to a lexeme with a full paradigm. The similar observation that occasional attestations do not make a form non-defective is made in Albright (2009), who further notes that it is to be expected that some ‘fix’ to use the missing form is tried by speakers (cf. also Kastner & Zu, 2015 and Nevins et al., 2014).
These attestations relate to the question of whether defectiveness is a problem of competence or performance—that is, whether speakers really lack a way to map the function to the form (as we have defined defectiveness here), or the problem is only that they do not use these forms, which otherwise are perfectly well defined in the grammar system. Sims (2015, pp. 54–55) points out that rarely used forms and defective forms are very different, in two senses. First, when a lexeme is non-defective, but simply infrequent, the forms are all attested rarely; in the case of a defective lexeme, the attested forms that fill the gap radically differ in frequency from the non-defective forms for the same lexeme. Second, if the lexeme is only infrequent, speakers normally do not hesitate to provide full paradigmatic forms even if they will never use them; for instance, the author of this article is a Spanish native speaker who had never heard the verb volitar ‘to fly while spinning,’ and can still produce without hesitation the second person plural imperfective subjunctive form volitarais.
6. Theoretical Approaches to Defectiveness
For expository convenience, a final short overview of the ways in which defectiveness has been analyzed in the literature divides the existing approaches into four classes, which is an oversimplification, but a useful one. For reasons of space it will concentrate on only some of the articles that have discussed these issues.
6.1 Defectiveness as Item-Specific Idiosyncrasy
In a first family of approaches, defectiveness is analyzed as an idiosyncratic property of some lexical items—words or morphemes—which plainly prevents them from being materialized in a real utterance.
Although discussing derivation rather than inflection, Halle (1973, p. 5) famously proposed that expected, but missing, words such as those in (19) were marked with a [−Lexical Insertion] feature that, as the label suggests, would prevent them from materializing any node in a structure. These words would be generated by the rule component of the lexicon, and the Filter that intermediates between this generative engine and the dictionary of words would mark them as unable to appear in any real sentence of the language.
Halle’s (1973) approach has been criticized extensively, as early as Hetzron (1975). The main critique is that defectiveness is less idiosyncratic than Halle’s form-by-form treatment suggests; as we have seen, the forms involved in defectiveness tend to follow well-attested inflectional patterns, for instance. However, it is interesting to note that, under different shapes, the main idea is preserved in some modern analyses, particularly within the Distributed Morphology framework.
For instance, Embick (2000) discusses the defectiveness of Latin deponent verbs in the perfect. In such verbs, there is no morphological realization and a syntactic periphrastic form is used:
Embick proposes that deponent roots are marked with a feature [pass](ive), which conditions the choice of exponents. The operation that builds one single constituent (word) combining aspect and tense is blocked by a [pass] feature. To the extent that only some roots have (arbitrarily) this feature, and that the blocking itself cannot be reduced to a more general principle but is treated as an idiosyncrasy, this approach is not radically different in spirit from Halle’s. Ultimately, what prevents a set of features from being mapped to a realization is in both cases an idiosyncratic, lexically specific property.
Arregi and Nevins (2014) present a related take where a defective form is not one marked with a feature that prevents insertion of the whole word, but a form that contains a morpheme that cannot be spelled out in a particular context. They start from Harley (2014), who makes the proposal that some roots lack an elsewhere rule that interprets them. In such cases, the root must be in a specific context in order to be assigned a meaning. For instance, cran- can only be assigned a meaning in the context of berry (cran-berry). Arregi and Nevins (2014) propose to extend the proposal from the semantic to the morphophonological component. In the morphophonological side, absence of an elsewhere materialization rule triggers defectiveness when the root is in a context distinct from the one it is specified for (see also Spencer, 2016 for a similar approach using lexical indexes and functions). In the specific case of the Spanish verb abolir ‘abolish,’ which lacks any form where the root is not followed by the vowel /i/, defectiveness follows from the following vocabulary entry (where the number is just an index to identify the root):
This means that the root only gets a morphophonological form in the context where it is followed by /i/ (assuming the vowel inventory of Spanish). This leads to a view of defectiveness as some form of ‘frustrated allomorphy’ whereby the spelling out of the root defines less allomorphs than would have been necessary in order to spell out the form in the whole paradigm. Nevins (2015) extends the analysis to Portuguese, claiming that in some verbs allomorphs for contexts where stress falls on the root are missing.
The inconvenience with these approaches, as has been noted repeatedly (Hetzron, 1975; Kiparsky, 2005; Sims, 2015, among many others), is that the lexical specification of the form is at the same time too strong and too weak. It is too strong in the sense that it predicts that any form could be defective, but we have seen that typically defective forms are related to irregular paradigms, would exhibit marked phonological shapes, and follow well-attested allomorphy patterns. At the same time, it is too weak because it is unclear how it could be extended to cases such as the absence of a genitive plural form for the Russian word for ‘crow,’ karga, when the surface identical accusative plural is accepted, unless the features are assigned specifically to whole word forms and not morphemes.
6.2 Defectiveness as the Result of Rule Conflict
Other approaches treat defectiveness as a situation where the materialization rule is undefined for a particular form due to competing pressures. This situation may occur when there are conflicting rules that could be equally applied to the form in order to materialize it. One crucial ingredient that all these approaches share is that defectiveness is epiphenomenal, which avoids the access to negative evidence problem: speakers learn the rules, and the problem is that the interaction between these rules might be undetermined in some cases.
Kiparsky (2005) is an example of this kind of approach. Kiparsky’s data are, as in Embick (2000), those related to the absence of a perfect form for deponent verbs in Latin (cf. 20). In his view, the absence of such forms is related to their unique combination of features: in addition to [Passive], they contain both [Present] and [Past] in their internal makeup. The perfect of a passive is defined by Kiparsky as follows, where the capitalized letters are functions taking the following object as an argument.
Both present and past are, thus, functions that would take an item and produce a distinct outcome. Normally, past can be applied to present or vice versa without producing an ungrammatical output, but in the case of the passive there is a third function applying to the same form. This, according to Kiparsky, produces an increase in complexity that might lead grammar to substitute that form with a periphrastic (syntactic) construction in order to avoid a portmanteau word that is at the same time present, past, and passive. Kiparsky hastens to add that having three functors inside one form does not, in itself, make a morphological expression ungrammatical, but takes this competition under complexity as a fair predictor for potential cases of defectiveness.
Albright (2003) approaches the problem in a related sense. He, through a study of defective verbal paradigms in Spanish, notes that speakers are more hesitant to produce those inflectional forms that could in principle display morphological irregularity if the shape of the infinitive suggests that the verb should undergo allomorphic alternations. Defective verbs in Spanish tend to belong to the third conjugation, where it is estimated that only 8% of verbs belong. Moreover, defective verbs in Spanish involve roots with mid vowels (/o/ and /e/). Verbs of that shape in the third conjugation always show some kind of alternation, either raising or diphthongization, without clear rules to determine which of the two processes the verb will undergo.
The claim is that defectiveness is favored in such cases precisely because the speaker is aware that the defective verb form should undergo an alternation, but there are different alternations that are equally possible, and the specific verbs are not frequent enough to decide. In other words: defectiveness is an extreme effect of speaker uncertainty in situations where there are conflicting possible outcomes and no reliable data to choose one.
In contrast to the previous set of theories, in this view defectiveness is semi-predictable within a language: some combinations of morphological rules, or some forms whose shape makes them likely candidates for irregularity, are expected to exhibit defectiveness if other factors, such as a low frequency, contribute. Defectiveness is not an idiosyncratic property, then, but a reflection of speakers’ uncertainty about how to solve a conflict between potential outcomes. In contrast, in the next approach we will present, defectiveness does not need to be necessarily related to situations where there is a complex interaction between morphological or morphophonological rules.
6.3 Defectiveness as Morphological Indefinition Within a Paradigm
Stump (2010) presents a theory where defectiveness is the result of morphology failing to define a particular form for paradigmatic-internal reasons. His initial observation is that defectiveness does not always appear within the more marked and more complex morphological forms; sometimes only the nominative of a paradigm is missing, which contradicts the theories presented in §6.2. Within the general theory of Paradigm Function Morphology, Stump builds on previous results (Stump, 2002) and starts from the proposal that a language’s inflectional morphology involves a content paradigm, a form paradigm, and rules of paradigm linkage. The content paradigm consists of a set of paradigm functions (PF) that realize different pairings of the same lexeme (L) with a morphosyntactic property set (τ).
The form paradigm consists of paradigm functions that realize the pairings of the root form of the lexeme and the set of morphosyntactic properties for which that root (R) may inflect (σ), as in (25).
The rule of paradigm linkage defines the realization of a content paradigm cell as the realization of a form paradigm cell, which is then its form correspondent. In (26), the content cell paradigm function is defined as a form cell paradigm function.
Defectiveness in the narrow sense—that is, defectiveness that does not directly follow from syntactic, semantic, or phonological causes—is a situation where there are idiosyncratic restrictions on which values of τ are definable as paradigm functions of form cells. Thus, the lack of a genitive plural form for some Russian nouns is accounted for, from this perspective, as those particular lexemes having an undefined paradigm linkage rule when the content form is set to the values of genitive and plural. The theory allows for an explicit formalization of the relation between syncretism and defectiveness: syncretism emerges, in contrast to defectiveness, when the form correspondent is not undefined, but defined as a form paradigm function that is specified disjunctively for more than one set of morphosyntactic properties. It is impossible to cover all details of this rich proposal in this short overview, so we refer the reader to the original work for further information.
This kind of approach to defectiveness contrasts with the previous ones in the sense that here defectiveness is a property of relations between paradigms of the same item. However, as in the first account, the paradigmatic rule has to be sensitive to an idiosyncratic restriction on the linkage rule for specific cells and specific items. In contrast, the final set of approaches we will review here treat defectiveness closer to a fully integrated morphological object within a paradigm.
6.4 Defectiveness as a Morphological Object
The last set of approaches we will consider, in a sense, extend the spirit of Stump’s approach, but treat it as a morphological property that follows from the complex internal structure of paradigms. A common property of these approaches is that they highlight that, in contrast to the proposals presented in §6.2, a form can be defective even if its morphological shape could be in principle predictable. In such approaches, rule conflict or speaker uncertainty can play a role as a historical cause of defectiveness, but synchronically defectiveness is grammaticalized as a normal morphological property, and is thus able to extend to additional forms determined by the internal structure of the paradigm.
Boyé and Cabredo-Hofherr (2010) and Maiden and O’Neill (2010) notice that, contra Albright (2003), the whole set of defective forms of verbs like abolir ‘abolish’ in Spanish include some forms which should be free from the conflicting alternations that caused speaker uncertainty in Albright’s approach. Specifically, in the first and second person subjunctive forms the word stress does not fall on the mid vowel of the root (27).
When confronted with these forms for verbs like abolir, speakers should not hesitate, because the mid vowel could in principle remain unaltered when stress does not fall in it. However, abolir is also defective in such forms. Even if insecurity played a role, in other situations speakers allow two forms for the same paradigmatic cell, so uncertainty due to rule conflict cannot be the whole story. What determines the whole set of defective forms within a paradigm, for these authors, is the fact discussed in §3 (cf. 18): defectiveness extends to forms that would in principle share the same lexeme allomorph. Specifically, defectiveness in abolir extends to all subjunctive forms because all subjunctive forms and the root-stressed present indicative forms share the same form (as illustrated in 28).
Irrespective of its original historical causes, where uncertainty might have played a role, defectiveness has become grammaticalized as a morphomic alternation within the paradigm (cf. Aronoff, 1994).
This property of defectiveness is central in Sims’ (2006, 2015) theory. For her, defectiveness is an expected property of paradigms that have a low degree of internal cohesion. Cohesion is here defined as the extent to which all cells in the paradigm are internally related to each other in their morphophonological realization. Cases of defectiveness first emerge in paradigms where the cohesion is lower, and might be originally due to the lexeme being a loanword or having low frequency, making the speaker hesitant about how the non-cohesive paradigm cells should be realized. However, through time, the gaps become morphologized and are treated like any other morphological alternation: they can extend to other lexemes with a family resemblance to the originally defective form. This explains that defectiveness in a language tends to cluster around forms belonging to the same inflectional class, and with similar phonological properties, as in the case of Russian anti-egotistic verbs lacking a first person singular form (cf. Sims, 2015, chapter 7).
Sims (2015, pp. 253−257), in fact, suggests that defective patterns should be viewed as another form of allomorphy. Just like standard allomorphy, defectivity is organized in inflection classes and sensitive to paradigmatic structure, as we saw; like allomorphy, defectivity involves competition between forms—potential realized forms for the same cell, potential syncretism extension to that cell. Also, like allomorphy, defectiveness extends to forms which have a particular shape or structure, and result in non-isomorphic mappings between structure and form.
In this final set of approaches, defectiveness in fully integrated among the different ways in which language can define contrast and opposition within a paradigm; its exceptional character is denied, and while the lexemes subject to it still need to be idiosyncratically specified, the way in which defectiveness operates inside a paradigm follows the standard rules of morphological alternations.
7. Defectiveness in the Future of Morphological Theory
Defectiveness defined in the narrow sense has very specific properties: it is sensitive to the internal structure of a paradigm, it is a gradient phenomenon, and it requires some level of lexical specification. While some theories treat defectiveness either as a lexical idiosyncrasy of individual items, others propose that it is a type of allomorphic alternation within a paradigm, with several intermediate views where defectiveness is more or less epiphenomenal. All these approaches open up new lines of research relating to how operations are codified in grammars, how specific items relate to their allomorphs, the nature of paradigms with low internal cohesion, and the relation between frequency, processing, and speaker competence. Defectiveness has been a rich field of study in the last years, and the interest of the phenomena related to it suggests that it will continue to be so in the future. In further developing the study of this phenomenon, experimental approaches (such as those used in Albright, 2003; Nevins et al., 2014; or Sims, 2015) are increasingly important in order to determine to what extent speakers allow forms to fill defective paradigms, which types of defectiveness pattern together by virtue of their properties, and which factors are central in triggering defectiveness.
Links to Digital Materials
The Surrey Morphology Group has made available two databases on defectiveness, one typological (27 languages), and one cross-linguistic (100 languages), as well as an annotated bibliography on the topic.
Albright, A. (2003). A quantitative study of Spanish paradigm gaps. In G. Garging & M. Tsujimura (Eds.), Proceedings of the West Coast Conference on Formal Linguistics, 22 (pp. 1–14). Somerville, MA: Cascadilla Press.Find this resource:
Albright, A. (2009). Lexical and morphological conditioning of paradigm gaps. In C. Rice & S. Blaho (Eds.), Modeling ungrammaticality in Optimality Theory (pp. 117–164). London: Equinox.Find this resource:
Aronoff, M. (1994). Morphology by itself: Stems and inflectional classes. Cambridge, MA: MIT Press.Find this resource:
Arregi, K., & Nevins, A. (2014). A monoradical approach to some cases of disuppletion. Theoretical Linguistics, 40(3–4), 311–330.Find this resource:
Arrivé, M. (1997). La conjugation pour tous. Paris: Bescherelle.Find this resource:
Baerman, M. (2011). Defectiveness and homophony avoidance. Journal of Linguistics, 47(1), 1–29.Find this resource:
Baerman, M., & Corbett, G. (2010). Introduction: Defectiveness, typology and diachrony. In M. Baerman, G. C. Corbett, & D. Brown (Eds.), Defective paradigms: Missing forms and what they tell us (pp. 1–18). London: Oxford University Press.Find this resource:
Baerman, M., Corbett, G. C., & Brown, D. (2010). Defective paradigms: Missing forms and what they tell us. London: Oxford University Press.Find this resource:
Baker, C. L. (1979). Syntactic theory and the projection problem. Linguistic Inquiry, 10, 533–581.Find this resource:
Bat-El, O. (2009). A gap in the feminine paradigm of Hebrew: A consequence of identity avoidance in the suffix domain. In C. Rice & S. Blaho (Eds.), Modeling ungrammaticality in Optimality Theory (pp. 165–195). London: Equinox.Find this resource:
Bauer, L. (2001). Morphological productivity. Cambridge, U.K.: Cambridge University Press.Find this resource:
Berwick, R. (1985). The acquisition of syntactic knowledge. Cambridge, MA: MIT Press.Find this resource:
Besnier, N. (2000). Tuvaluan: A Polynesian language of the central Pacific. London: Routledge.Find this resource:
Borer, H. (2013). Taking form. Oxford: Oxford University Press.Find this resource:
Boyé, G., & Cabredo-Hofherr, P. (2010). Defectiveness as stem suppletion in French and Spanish verbs. In M. Baerman, G. C. Corbett, & D. Brown (Eds.), Defective paradigms: Missing forms and what they tell us (pp. 35–52). London: Oxford University Press.Find this resource:
Brown, R., & Hanlon, C. (1970). Derivational complexity and order of acquisition in child speech. In J. R. Hayer (Ed.), Cognition and the development of language (pp. 11–53). New York: Wiley.Find this resource:
Caha, P. (2009). The nanosyntax of case. (Unpublished doctoral dissertation). University of Tromsø, Tromsø.Find this resource:
Chomsky, N. (1981). Lectures on government and binding. Berlin: De Gruyter.Find this resource:
Clements, G. N. (1990). The role of the sonority cycle in core syllabification. In J. Kingston & M. Beckman (Eds.), Papers in laboratory phonology 1: Between the grammar and physics of speech (pp. 283–333). Cambridge, U.K.: Cambridge University Press.Find this resource:
Embick, D. (2000). Features, syntax and categories in the Latin perfect. Linguistic Inquiry, 31(2), 185–230.Find this resource:
Halle, M. (1973). Prolegomena to a theory of word formation. Linguistic Inquiry, 4, 3–16.Find this resource:
Halle, M., & Marantz, A. (1993). Distributed morphology and the pieces of inflection. In K. Hale & S. J. Keyser (Eds.), The view from Building 20 (pp. 111–176). Cambridge, MA: MIT Press.Find this resource:
Harley, H. (2014). On the identity of roots. Theoretical Linguistics, 40(3–4), 225–276.Find this resource:
Hetzron, R. (1975). Where the grammar fails. Language, 51, 859–872.Find this resource:
Hudson, R. (2001). *I amn’t. Language, 76(2), 297–323.Find this resource:
Johansson, C. (1999). Learning what cannot be by failing expectations. Nordic Journal of Linguistics, 22, 61–76.Find this resource:
Kastner, I., & Zu, V. (2015). Blocking and paradigm gaps. Unpublished manuscript, Humboldt University Berlin. Available as lingbuzz/002394.Find this resource:
Kemmer, S., & Israel, M. (1994). Variation and the usage-based model. In K. Beals et al. (Eds.), Papers from the 30th regional meeting of the Chicago Linguistic Society (pp. 165–179). Chicago: CLS.Find this resource:
Kiparsky, P. (2005). Blocking and periphrasis in inflectional paradigms. In G. Booij & J. van Marle (Eds.), Yearbook of morphology 2004 (pp. 113–135). Dordrecht: Springer.Find this resource:
Lass, R. (1980). On explaining language change. Cambridge, U.K.: Cambridge University Press.Find this resource:
Lignos, C., & Yang, C. (2016). Morphology and language acquisition. In A. Hippisley & G. T. Stump (Eds.), The Cambridge handbook of morphology (pp. 765–792). Cambridge, U.K.: Cambridge University Press.Find this resource:
Lukács, Á., Rebrus, P., & Törkenczy, M. (2010). Paradigmatic space and defectiveness in Hungarian: Description and experimental study. In M. Baerman, G. C. Corbett, & D. Brown (Eds.), Defective paradigms: Missing forms and what they tell us (pp. 85–102). London: Oxford University Press.Find this resource:
Maiden, M. (2004). When lexemes become allomorphs: On the genesis of suppletion. Folia Linguistica, 38(3–4), 227–256.Find this resource:
Maiden, M. (2005). Morphological autonomy and diachrony. In G. Booij & J. van Marle (Eds.), Yearbook of morphology 2004 (pp. 137–175). Dordrecht: Springer.Find this resource:
Maiden, M., & O’Neill, P. (2010). On morphomic defectiveness: Evidence from the Romance languages of the Iberian Peninsula. In M. Baerman, G. C. Corbett, & D. Brown (Eds.), Defective paradigms: Missing forms and what they tell us (pp. 103–124). London: Oxford University Press.Find this resource:
Mithun, M. (2010). The search for regularity in irregularity: Defectiveness and its implications for our knowledge of words. In M. Baerman, G. C. Corbett, & D. Brown (Eds.), Defective paradigms: Missing forms and what they tell us (pp. 125–149). London: Oxford University Press.Find this resource:
Morin, Y-Ch. (1987). Remarques sur l’organisation de la flexion des verbes français. ITL: Review of Applied Linguistics, 77–78, 13–91.Find this resource:
Nau, N. (1998). Latvian. Munich: Lincom Europa.Find this resource:
Nevins, A. (2015). Book review: Defective paradigms. Revista Linguística, 10, 29–37.Find this resource:
Nevins, A., Damulakis, G., & Freitas, M. L. (2014). Phonological regularities among defective verbs. Cadernos de Estudos Lingüísticos, 56(1), 1–11.Find this resource:
Orgun, C. O., & Sprouse, R. L. (1999). From MParse to Control: Deriving ungrammaticality. Phonology, 16, 191–224.Find this resource:
Prince, A., & Smolensky, P. (2004 ). Optimality Theory: Constraint interaction in generative grammar. London: Wiley.Find this resource:
Raffelsiefen, R. (2004). Absolute ill-formedness and other morphophonological effects. Phonology, 2, 91–142.Find this resource:
Rice, C. (2003). Dialectal variation in Norwegian imperatives. Nordlyd, 31, 372–384.Find this resource:
Rice, C., & Blaho, S. (2009). Modeling ungrammaticality in Optimality Theory. London: Equinox.Find this resource:
Sims, A. (2006). Minding the gaps: Inflectional defectiveness in paradigmatic morphology. (Unpublished doctoral dissertation). The Ohio State University, Columbus.Find this resource:
Sims, A. (2015). Inflectional defectiveness. Cambridge, U.K.: Cambridge University Press.Find this resource:
Sokolov, J., & Snow, C. E. (1994). The changing role of negative evidence in theories of language acquisition. In C. Gallaway & B. Richards (Eds.), Input and interaction in language acquisition (pp. 38–55). Cambridge, U.K.: Cambridge University Press.Find this resource:
Spencer, A. (2016). How are words related? In D. Siddiqi & H. Harley (Eds.), Morphological Metatheory (pp. 1–26). Amsterdam: John Benjamins.Find this resource:
Starke, M. (2009). Nanosyntax: A short primer to a new approach to language. Nordlyd, 39, 1–9.Find this resource:
Stump, G. (2001). Inflectional morphology: A theory of paradigmatic structure. Cambridge, U.K.: Cambridge University Press.Find this resource:
Stump, G. (2002). Morphological and syntactic paradigms: Arguments for a theory of paradigm linkage. In G. Booij & J. van Marle (Eds.), Yearbook of morphology 2001 (pp. 147–180). Dordrecht: Kluwer.Find this resource:
Stump, G. (2010). Interactions between defectiveness and syncretism. In M. Baerman, G. C. Corbett, & D. Brown (Eds.), Defective paradigms: Missing forms and what they tell us (pp. 181–210). London: Oxford University Press.Find this resource:
Zaliznjak, A. A. (1977). Grammaticheskij slovar’ russkogo jazyka: Slovoizmenenie. Moscow: Russkij jazyk.Find this resource: