One of the holy grails of cognitive science is the explanation of linguistic productivity: a model of the linguistic system that allows people to store what they hear and reuse it creatively for new purposes. There has been much debate about how to distinguish storage from combinatorial use, with some favoring a strict division between the concrete representation of lexical items and their combination in terms of abstract categories (e.g., Chomsky, 1965) and others who have argued that storage is of a more pervasive and associative nature from which abstract combinatorial patterns may initially arise (e.g., Langacker, 1987). Inflectional morphology has long been the testbed of choice for these competing views of language. Intensive research in this field offers a particularly rich set of explanatory alternatives that we explore here from an evolutionary perspective. Say no to plagiarism. Get a tailor-made essay on "Why Violent Video Games Shouldn't Be Banned"? Get an original essay Research on the acquisition of inflectional morphology tends to focus on the phenomenon of overregularization: the application of a regular inflection to an irregular or inflected root form, for example, mice or mice. Depending on one's point of view, such errors of commission have been considered the essential demonstration that children acquire abstract rules or generalize patterns. Much less attention has been paid to errors that occur when children simply don't say any words, such as when they say “mouse” to refer to five mice. However, these errors of omission can shed much light on the acquisition process and can potentially separate different theoretical accounts of inflectional morphology. The dual-path inflection model (Clahsen, 1999; Clahsen, Aveledo, & Roca, 2002; Clahsen, Rothweiler, & Woest, 1999; Marcus et al., 1992; Pinker & Prince, 1988) proposes that words inflected in a irregular (e.g., mice) are stored in an associative memory; , while the inflection of regular words (e.g., case) is calculated by a predefined rule (e.g., “add -s” for English plurals) that combines a symbol for a root with a symbol for a suffix. Before the default rule is acquired, if a child does not have an appropriate inflected form in memory, then he or she will be forced to pronounce a bare stem in its place (Pinker, 1999), and therefore, errors of omission may occur before the acquisition of a rule. After learning the rule (as evidenced by overregularization production), omission errors should disappear and the child should modulate consistently, albeit with overregularized forms for unknown irregulars (e.g., mice). However, the dual-route model allows analogies to be made between irregular nouns/verbs that form phonological subgroups, allowing for a certain degree of incorrect productivity (e.g., Bring-Brang by analogy with Sing-Sang, Ring-Rang). Therefore, any subsequent errors of omission could only be explained if one were to argue that the word in question was erroneously considered a “no change irregular” (e.g., cut) by virtue of being a close phonological neighbor of such irregulars. Otherwise, errors of omission are not expected. Usage- or pattern-based models (Bybee, 1985, 2001; Bybee & Slobin, 1982; Dabrowska, 2001, 2004; Köpcke, 1998) propose that both regulars and irregulars are handled by the same storage and processing mechanisms. In a sense, pattern models are also double-path models in that a given inflected form can be obtained by recovering the entireform from memory or by accessing a stem and adding appropriate affixes, which are represented as patterns (Bybee, 2001). . The difference is that the choice of path is determined by the frequency of the shape, not its regularity. The higher the frequency of the inflected form, the greater the probability that the entire memory form will be recovered (whether it is regular or irregular). Bybee and Slobin (1982) proposed that children form product-oriented schemas, which are generalizations about the properties of inflected forms (e.g., past tenses tend to end with –ed) as well as source-oriented schemas, which are generalizations about as an inflected form the form is made up of a root and an inflection (for example, to create a past tense, take the root of a verb and add –ed). Adult mastery is achieved by balancing product-oriented schemas with source-oriented generalizations about irregular suffixes and idiosyncrasies. Early development of product-oriented schemas predicts that children will make more omission errors with nouns and verbs whose stem already resembles the inflected schema (e.g., dress already ends in -s). This account does not predict that errors of omission should cease once errors of commission (productive suffix) begin. Rather, the developmental process is seen as a gradually refined balancing act that is sensitive to the symbolic frequency of the individual words that are inflected (whether regular or irregular) and the type frequency of the inflections that are schematized. Connectionist models have generally been compatible with schema models in that they propose the same storage and processing mechanisms for regulars and irregulars (e.g., Joanisse & Seidenberg, 1999; MacWhinney & Leinbach, 1991; Plunkett & Juola, 1999; Plunkett & Marchman, 1991, 1993, 1996; Rumelhart & McClelland, 1986; These patterns show signs of schema-induced omission errors (e.g., making more omission errors with the required /-id/ inflection for verbs whose stem ends already in d/t), although these errors are not very frequent and are often argued to be due to the analogy with irregulars without change (cut-cut) rather than to the implicit generalization of a product-oriented scheme. Connectionist models demonstrate more in general neighborhood effects whereby elements that fall within densely populated neighborhoods, for example drink-drink and sink-sink, tend to be similarly inflected by virtue of analogy (which may mean that they are particularly resistant to errors of excessive regularization). The type of model to consider is called a dual-path parallel running model (Baayen, Dijkstra, & Schreuder, 1997). In this model, both whole inflected forms and broken down roots and inflections are activated and compete for selection. The higher the frequency of the inflected form, the more likely it is that the entire form is used, not the split roots. This model is therefore similar to Bybee's (2001) model that we described earlier but has a more specified processing architecture that includes steps for activating segments, licensing their composition, and composing their meaning. This architecture is proposed to explain adult understanding. As a result, it is difficult to make precise predictions about omission errors in children's production. However, the pervasive memorization assumed by the model would presumably predict fewer omission errors for words whose inflected forms have a higher frequency. The logic here is that high-frequency inflected forms are more likely to be recovered as a whole rather than being formed from a root and a separate inflection..