Elena Even-Simkin & Yishai Tobin's book draws attention to itself by the extravagance of its claims. It announces a solution to the problem of ablaut, which English has inherited from Indo-European. While umlaut is at least partly amenable to an explanation, ablaut has remained relatively intractable. The apparent lack of a formal motivation for ablaut makes novel perspectives on this issue highly welcome.
Without much defence, Even-Simkin & Tobin (henceforth E&T) take as their starting point the iconicity hypothesis (what they vaguely call the sign-oriented approach): form follows function, i.e. form is fully motivated by meaning. They assume an exceptionally close relationship between the two sides of the linguistic sign. Thus, ablaut encodes a specific meaning by which it is brought about. Which meaning it expresses depends on its phonological realization. When it is instantiated by a backing process as in verbs (e.g. to forget – forgot), it metaphorically encodes a movement backward in time; if it is instantiated by a fronting process as in nouns (e.g. goose – geese), it metaphorically encodes a movement forward in space. The general meaning of verbal ablaut is claimed to be resultativeness, the general meaning of nominal ablaut is opaquely called ‘semantic integrality’ which may be translated into the more common and more appropriate term ‘collectivity’. At a deeper level, resultativeness and collectivity are viewed as different manifestations of the same underlying principle of integration. Collectivity integrates the tokens of a nominal type into a unified whole in much the same way as resultativeness combines the action and the result of the action into a single verbal concept. According to E&T, the connection between meaning and form is even closer. English ablaut verbs can be divided into a number of classes, depending on the kind of vowel change involved. The authors set up 14 such classes for Modern English, claiming that each of these formal classes can be semantically defined by a different shade of resultativeness. Consequently, if a verb has two different ablaut patterns (e.g. to shrink – shrank/shrunk), these must have two different semantic motivations. Moreover, the greater the phonological similarity between the base vowel and the ablaut vowel in two given ablaut classes, the greater the semantic similarity between the shades of resultativeness.
E&T's analysis is based on data from Old and Modern English which were taken from the Oxford English Dictionary. This double focus allows them to adopt both a synchronic and a diachronic approach and to offer explanations for certain historical developments. Paralleling their synchronic proposals, they explain the loss of ablaut (i.e. regularization) in terms of a loss of the semantic feature supporting it. Similarly, irregularization is understood as the addition of a semantic trait.
On the face of it, this looks like an all-embracing model with a wide explanatory potential. In the following, this model will be critically examined step by step. Let us begin with the distinction between ablaut and umlaut. Without much justification, E&T indiscriminately treat the vowel alternations in nouns and verbs as ablaut. When ablaut is defined as being synchronically unmotivated, this decision may be warranted. However, the diachronic perspective clearly identifies the vowel change in nouns as umlaut. While there may be good reasons to dissociate synchrony from diachrony, it seems unwise to do so in the present case. The fact that verbal and nominal vowel changes have different historical origins should not be ignored because these disparate origins are certainly part of the explanation for the different constraints observed in the two word classes. Subsuming both types of vowel change under the same category is misleading because it suggests more similarity between them than there actually is.
In view of the well-known historical fact that the vowel changes were all probably brought about by i-mutation, it is no wonder that the few surviving ablaut nouns in English exhibit a fronting process. By definition, i-mutation induces fronting. Note that, contrary to what E&T claim, not all ablaut nouns are adequately described as involving fronting. The two vowels in Modern English man and men are front vowels, hence there cannot be fronting. Again, E&T postulate more homogeneity than there actually is.
To my mind, the authors' claim that fronting in plural nouns iconically represents a process of moving forward is without foundation. It is impossible for me to forge a link between the notion of plurality and forward movement. As far as I can see, there is no way in which motion naturally entails plurality and, by extension, semantic integrality. Thus, the fronting process remains unexplained in their model. The historical perspective invites an equally sobering conclusion. E&T's model predicts that the regularization of plural nouns is occasioned by a loss of collectivity. Take the example of cow and friend, whose ablaut plurals in Old English underwent regularization. The interpretation of this change would be that cows and friends nowadays are no longer conceived of collectively as they used to be in the Anglo-Saxon period. The herds of cattle still grazing in our meadows bear silent testimony against this hypothesis.
At first sight, E&T's attempt to explain backing in verbs in terms of a movement backward in time appears more promising. There is an undeniable link between time and space in language. However, this link works in the direction opposite the one posited by E&T. Time is typically conceptualized in spatial terms (e.g. hind legs – hindsight) (see e.g. Radden Reference Radden, Brdar, Omazik, Takac, Gradecak-Erdelijic and Buljan2011). What E&T do instead is to interpret space as time, or to be more precise, moving backward in vowel space as moving backward in time. There is little independent support for this stance. In any case, E&T do not furnish any.
Assigning ablaut the meaning of resultativeness does not seem totally off the mark. Unfortunately, the authors miss a step in their chain of argument. The connection between ablaut and resultativeness is in fact an indirect one. Ablaut and resultativeness are linked through the intermediary of an aspectual dimension of the past tenses. Given our general focus on the present, past events may be less relevant as such, but more relevant in terms of the effects they have at the moment of speaking. The present perfect in English fulfils precisely this function. Since ablaut is specifically Indo-European, it may not be too far-fetched to call attention to the resultative perfect in Ancient Greek, which expressed a present state resulting from a past event (e.g. Wackernagel Reference Wackernagel and Wackernagel1904). The counterclaim would thus be that resultativeness is a particular conceptualization of past events rather than the semantic motivation of ablaut.
Pitting these rival claims against each other brings to the fore what I regard as the most blatant omission in E&T's argumentation. The authors fail to test their hypotheses against non-ablaut cases, i.e. so-called regular plural and past tense/past participle formations. The predictions generated by their model are quite clear. Regular past tenses should not implicate resultativeness. By the same token, regular plurals should not implicate collectivity. Neither of these predictions is borne out. Consider the sentence I have written/composed/drafted a letter. Arguably, the three different verbs (or rather VPs) convey the same amount of resultativeness even though the first verb is an ablaut verb while the other two are regular. This identical amount of resultativeness is inconsistent with E&T's claim. A similar argument can be advanced for nouns. Regular plurals are fully compatible with the notion of collectivity. Insects such as ants and bees live in colonies and are usually experienced as a collective. In this respect, they do not differ much from geese or mice. Therefore, collectivity cannot be invoked as an explanation for irregular plural formation.
Rather than carrying the critical analysis any further, it seems more rewarding to take stock of the patterns of vowel change in English and thereby transcend the limitations inherent in E&T's study. In this way, the scope of a theory of ablaut could be more comprehensively defined. In addition to the verbal and nominal patterns analysed by E&T, the following types need to be taken account of: (a) adjective comparison (e.g. old – elder), (b) transitive vs. intransitive verbs (e.g. to lay – to lie), (c) verb/noun alternations (e.g. to sing – song), (d) adjective/noun alternations (e.g. hot – heat), (e) noun pairs (e.g. thief – theft), (f) partial reduplication (e.g. zigzag) and (g) binomials (e.g. rock and roll). While English has only a few members in some of these groups, it should be borne in mind that other Germanic languages, such as German and Icelandic, make heavy use of ablaut (and umlaut, for that matter).
It is more than doubtful that a semantic explanation can be adduced which is capable of accommodating all these cases. In particular, it is completely unclear how vowel alternations which cross word-class boundaries can be semantically motivated. German, for instance, abounds in verb/noun pairs. The vowel of the noun may be identical to the (tonic) vowel of a conjugated present tense form (e.g. helfen – Hilfe ‘to help – the help’), to that of the past tense (e.g. trinken – Trank ‘to drink – the drink’), to that of the past participle (e.g. springen – Sprung ‘to jump – the jump’) or different from either of these (e.g. fliegen – Flug ‘to fly – the flight’). Any effort to come up with a semantic explanation à la E&T for these various patterns appears misguided and futile.
On a more constructive note, let us briefly sketch where a viable solution to the problem of ablaut and umlaut might be found. I see no reason for being discontented with the historical explanation of umlaut as distant and partial assimilation. The fact that it was usually triggered by a high front vowel satisfactorily explains why it involves both fronting and raising. Of course, this does not explain why umlaut occurred in the first place and why it eventually lost its productivity.
The explanation of ablaut that I would like to submit relies on lexical statistics and has a psycholinguistic slant. Sereno & Jongman (Reference Sereno and Jongman1990) demonstrated that high-frequency verbs in English have a significantly higher rate of front than back vowels. Given that ablaut usually involves backing (see above), we may assume an association between front vowels and non-ablaut forms, in particular the infinitive and inflected non-past forms.
This effect naturally raises the question of how a front vowel may be changed in order to generate a past tense form. The logic of a possible answer is that there are various ways in which a front vowel may be changed, that these ways may differ in terms of the mental cost that they incur and that the way that is less costly is the one that tends to be chosen by ablaut.
Evidence for which phonological path is preferably taken by vowels comes from slips of the tongue, in particular vowel substitution errors like the following:
(1) the cleaning fluid kind of lukes – leaks(from Shattuck-Hufnagel Reference Shattuck-Hufnagel1986)
This error may be described as involving a change along the backness dimension. The (intended) front vowel /iː/ is replaced with the (inadvertent) back vowel /uː/. No other phonological parameter is implicated. Shattuck-Hufnagel (Reference Shattuck-Hufnagel1986) makes the pertinent observation that backness is the dimension that is most frequently altered in speech errors. Stemberger (Reference Stemberger1992) takes us a step further. He examines whether a change from [front] to [back] is as likely as the inverse change and finds that the former change is much more likely than the latter.
These results may be interpreted as follows. A switch in backness is the easiest switch for the language production system. Furthermore, a switch from [front] to [back] happens more easily than one from [back] to [front]. It is striking that this set of results predicts the ablaut patterns quite nicely. It may now be argued that the preference for backing in ablaut is rooted in processing biases.
Let us briefly note two points on which the psycholinguistic account compares favourably to the semantic account. Backing is not only involved in verbal ablaut but also in other areas such as partial reduplication (e.g. flip-flop, category (f) above). The psycholinguistic account offers a unified explanation for the phonological basis of ablaut in different domains whereas E&T would probably have to develop different semantic motivations for the same phonological pattern and thereby introduce such obscure concepts as iconic homonymy.
Moreover, the psycholinguistic account is better able to do justice to the phonological diversity of ablaut. E&T are ready to concede that not all English ablaut verbs involve backing. Their semantic approach coerces them into making the uncomfortable claim that there are disparate semantic motivations for verbs like to get – got and verbs like to fall – fell for example. By contrast, the psycholinguistic account basically predicts the diversity in ablaut patterns that we find in vocalic speech errors. To be more specific, it allows changes from [back] to [front], changes of other phonological dimensions as well as changes of two phonological parameters simultaneously. This is exactly what we find in ablaut and umlaut patterns.
To conclude, while I am generally sympathetic to efforts that try to bridge the gap between form and meaning, I believe that E&T have grossly overstated their case. Their attempt at explaining ablaut in semantic terms fails. It is doubtful that Louden's (Reference Louden2000) account of ablaut and umlaut in pragmatic terms is any more successful than E&T's.
