The Mental Lexicon, Psycholinguistics, and Neurolinguistics

Studies of the sort just discussed have goals beyond linguistic description and aim rather to describe what is actually going on inside the head of a language user when they use complex words. Research about what goes on inside people’s heads when they process language is usually called psycholinguistics, though when the research involves measuring neurological activity, as much modern research does, then it may also be termed neurolinguistics or cognitive neuroscience. The goal of psycholinguistic and neurolinguistic research on morphology is to understand exactly how the mental lexicon works, the lexicon that all language users store in their brains. Jarema and Libben (2007) is a collection of articles on the mental lexicon from a variety of perspectives. Pinker (1999) is a very lively and informal presentation of the major issues in the field. Aitchison (2003) is a more general introduction.

It is important to remember in discussions of the mental lexicon that we must distinguish the lexicon itself from how it is used by the speaker or hearer. Most studies of the mental lexicon are concerned with word recognition, a classic problem in psychology: How do people associate the speech or writing signal with the entries in their mental lexicon? Until very recently most word recognition studies dealt mostly with written words, not speech. Psychologists were interested in how people recognize words, both their sound and their meaning, when they read. Only in the last twenty years or so have psycholinguists turned their attention to spoken word recognition, which, for linguists at least, is a much more interesting and certainly more basic problem than reading. After all, as linguists have known for over a century, the primary form of language is spoken (or signed) and written language is secondary. The basic and universal task of word recognition is thus spoken word recognition. But from an experimental point of view, it has always been much easier to deal with written text than with speech, which is why it is only recently, with the advent of tools for regularizing the speech signal, that those psycholinguists have begun to turn to spoken language recognition.

The other side of accessing the mental lexicon, and in some sense the opposite or reverse of word recognition, is word production. When a person speaks, how does that person know what words to use, the words that will eventually come out of his or her mouth? This is a much more difficult problem than word recognition and has received much less attention from psycholinguists (the classic presentation is Levelt 1989). Most psycholinguists assume that there is only one mental lexicon and that both word recognition and word production access this same lexicon, though in different ways. From an experimental perspective, though, the mental lexicon is not something that we can study directly. We study it rather only through the intermediaries of production and recognition, both of which must access this dictionary hidden deeply within our heads.

To understand what sorts of issues might arise in the study of the mental lexicon, let’s look at some of the aspects of Kujamaat Jóola that we have discussed. Consider vowel harmony first. The vowels of Kujamaat Jóola fall into two sets, tense and lax, and all the vowels in a word must be either tense or lax.

Furthermore, when a suffix is added to a word, the vowel of the suffix may be either tense or lax, depending on the vowels of the stem. However, if a suffix contains a tense vowel, then the vowels of the stem must also become tense. Overall, tense vowels win: as long as a complex word contains one tense vowel, all its vowels must become tense.

There are several conceivable ways in which those stems and suffixes that alternate between tense and lax could be represented in a person’s mental lexicon. Since morphemes with only tense vowels never alternate, it is reasonable to assume that these morphemes have only one form in the mental lexicon. But other morphemes sometimes have lax vowels and sometimes tense vowels. We could say that each of these morphemes has two alternative forms (allomorphs) in the mental lexicon and that the form with lax vowels is selected just in case there are no tense vowels in the word. Or we could say that there is a mental operation that makes lax vowels tense if there are tense vowels in the same word. A psycholinguist or neurolinguist would ask how these different systems might be encoded in a person’s brain and would want to devise experimental tests and measures for determining whether a particular mechanism is actually at work. Remember also that speakers of Kujamaat Jóola are aware of the tense–lax distinction and differ in their use of tense and lax vowels. How could this awareness be represented in a mental lexicon?

What about inflection and derivation? As we noted, the description of Kujamaat Jóola verbs is the most difficult part. As outlined, a Kujamaat Jóola verb has many forms, with two prefix slots and three suffix slots. It must agree with the noun class of its subject, and there are 19 noun classes, so every verb has up to 19 different prefixed forms solely on account of noun class agreement, and it also has a number of distinct forms if the subject is a pronoun (some of which have a suffix as well as a prefix). Verbs can also include bound object pronouns, which are suffixes. And there are prefixes and suffixes that mark categories such as aspect, negation, emphasis, and various types of subordinate clauses. As a result, every single Kujamaat Jóola verb has several hundred forms in its paradigm.

Turning now to derivation, we discussed a number of productive verbal derivational affixes in Kujamaat Jóola, each of which can be used productively to form a new verb. For example, the suffix -εn is causative; we can take a verb root like jɔj ‘gather, assemble (intransitive)’ and add -εn to it to get the verb jɔj-εn ‘gather, assemble (transitive)’ as in the infinitive ka-jɔj-εn ‘to cause people to assemble’. We can then take this verb and add to it the nominalizer -a, resulting in the noun ka-jɔj-εn-a ‘gatherer of people, leader who brings people together by force of charisma’. How does a speaker produce these two forms, the complex verb and noun, and are they both in the mental lexicon of a speaker of Kujamaat Jóola?

We defined the lexicon as “a list of forms that you know” and said that it was “equivalent to your linguistic memory.” Psycholinguists take a somewhat broader view. For them, the question of interest is how people store and produce the forms of their language. What does this question mean for the inflectional and derivational forms of Kujamaat Jóola that we have just discussed? Given that every Kujamaat Jóola verb has several hundred potential forms, how are these forms stored in the mental lexicon? Does every speaker store several hundred different forms for every verb? If so, then the speaker must also have some mental mechanism for deciding which of these several hundred to choose on any given occasion. Or does a speaker store only the stem of the verb and then produce whatever form is needed at any given moment, on the fly? We certainly know that speakers and signers do not hesitate when speaking or signing, no matter how complex the morphology of their language. A Kujamaat Jóola speaker does not pause for a moment or even a microsecond before a particularly complex verb to either choose or compute the correct form. So whatever language users are doing, it is being done at lightning speed. Similarly, when a Kujamaat Jóola speaker uses the verb ka-jɔj-εn ‘to cause people to assemble’, is the speaker recalling this verb from memory or constructing it?

We can sometimes call on purely linguistic evidence to at least suggest what is the right answer to some of these questions. For example, the noun ka-jɔj-εn-a ‘gatherer of people, leader who brings people together by force of charisma’ is not purely compositional or predictable in meaning. It does not simply mean ‘gatherer of people’ but has the added meaning of ‘leader who brings people together by force of charisma’. The fact that this word has this particular idiosyncratic meaning tells us that the word must have its own entry along with its special meaning in a speaker’s mental lexicon, and cannot be produced on the fly. And if the noun ka-jɔj-εn is stored, then presumably so too must be the verb ka-jɔj-εn that it is formed from.

Semantics alone thus provides some information about which forms must be stored. But semantics does not tell us how they are stored. Let us return to the same three lexemes, the basic verb root jɔj, ‘gather, assemble (intransitive)’, the complex verb jɔj-εn ‘gather, assemble (transitive)’ and the noun ka-jɔj-εn-a ‘gatherer of people’. Are these stored as three separate lexical entries or are they stored together in one place? If they are separate entries, are the complex forms all stored as complex structures, in the way that we have represented them (by using hyphens to separate their parts), or are they stored without structure? Most linguists would probably feel that they are stored as complex structures, but there is no purely linguistic evidence to tell us whether that is so. Similar questions can be asked about all the inflectional forms of a single verb or noun. Are they all stored? And if so, are they stored together or separately and are they stored with or without structure?

As it happens, a fairly broad consensus has emerged among psycholinguists in the last decade that “morphological considerations need to be introduced into any model of the mental lexicon” (Frost et al. 2005: 1). In other words, regardless of whether any particular words are stored or not and regardless of whether related words are stored together or separately, they do have morphological structure. But as to whether related words are stored together or separately, we just don’t know.

Even if everyone agrees that all words in the mental lexicon have morphological structure, many other issues remain undecided. One of the most central is that of the number of different mechanisms speakers use to access their mental lexicons, either in recognition or in production. The most obvious answer is that there is only one mechanism, but in fact, of the theories of accessing the mental lexicon that are out there in the literature, we can identify two broad types, called the single-route and dual-route models. A very accessible account of both model types is provided by Pinker (1999).

It is easy to imagine how a single-route model might work but what, we might ask, are the two routes in the dual-route model? They are whole-word recognition (sometimes called rote recognition) and recognition that uses the morphological system of the language (sometimes called rule-based recognition). The basic idea is that, while unanalyzed words like be and irregular words like am are recognized as whole words, morphologically regular words like being whose internal organization is determined by the morphological system of the language (in this case English) are recognized by a psychological analogue of this morphological system. The model thus instantiates the idea that people make a sharp distinction between regular and irregular (rule and rote) in accessing their mental lexicon. There is a twist, though: the dual-route model claims that a person uses both methods (rule and rote) at the same time when recognizing a word. There is a race between the two and whichever method is faster wins.

For a word like be or am there is no contest. The rote method will always win, because the word must be stored as an unanalyzable whole and no rule is applicable. At the other extreme, with a word like unregisterability, which is totally new to us as we write it and hence could not be in our mental lexicon, the only way we could recognize it as a word of English would be to use the morphology of English to analyze its structure and accept it as valid (which is precisely what we must have done in making the word up just now). So the rule mechanism must be at work. For words in between, though, matters begin to get interesting. These in-between words are morphologically complex but not unknown to us. Here, frequency or familiarity begins to play a role.

It is well known that the more frequent or familiar a word is, the faster a speaker can recognize it (Aitchison 2003). An English speaker will recognize clean more quickly than glean, even though the two differ phonologically in only one distinctive feature, because the first is about thirty times more frequent than the second. Most of the 500 most frequent words in English are simple, but even if a word is morphologically complex and perfectly regular, it may still be common. Francis and Kučera (1982) put the word development at number 250 in their list, ahead of develop at 311. All else being equal, on the basis of frequency alone, one might expect a speaker to respond more quickly to development than to develop, even though the first is technically derived from the second. According to the dual-route model, this is because, even though development is a regular, morphologically complex word, it may be stored in the mental lexicon and, because it is frequent enough, the rote method of lexical access, which is sensitive to frequency, will get to it faster than the rule-based method. In this way, a derived word may be detected more quickly than its base.

The single-route model can also accommodate the difference between rule and rote along with frequency effects, but it does so by assuming that the rule–rote difference is more gradual. The two models both have strong adherents, and their adherents use more and more sophisticated methods to argue their respective sides, including a variety of types of brain-scanning techniques, but the jury is still out. Whichever model wins in the end, we hope that we have shown our readers that the problem of the mental lexicon is fascinating.

Unfortunately, almost all the psycholinguistic and neurolinguistic research that has been done to date is based on English or other European languages whose inflectional and derivational morphology is fairly simple. Let us hope that researchers can broaden the scope of their work in the future to include languages like Kujamaat Jóola, whose complex morphological systems may help to find answers more quickly and more certainly than the relatively simple systems that most research on the mental lexicon is devoted to.

اخر الاخبار

اخبار العتبة العباسية المقدسة

استمرار جهود مطبخها المركزي في لبنان.. العتبة العباسية المقدسة تعلن إيصال وجبات الطعام إلى المتضررين ومراكز الإيواء

متحف الكفيل يُنجز أعمال تركيب الزجاج في عارضات مقتنيات العتبة العسكرية المقدسة

سماحة السيد الصافي يؤكّد استعداد العتبة العباسية المقدسة لدعم القطاع البيطري والتعاون في تعزيز الأمن الصحي

المجمع العلمي يستأنف برنامج قمر العشيرة في محافظة المثنى

المزيد

الآخبار الصحية

المشمش المجفف.. كنز صحي للقلب والعظام مع تحذيرات مهمة

خطر خفي في أكياس الشاي يثير مخاوف صحية

نظام غذائي شائع يظهر نتائج واعدة لمرضى السكري

تقنية نانوية روسية تشخص النوبة القلبية خلال 6 دقائق فقط

المزيد

الآخبار التكنلوجية

دراسة جديدة تكشف التركيب الحقيقي لكوكب الأرض

علماء يكتشفون انخفاض صلابة الألماس عند تحويله إلى مقياس نانوي

ثورة في علاج الصمم.. علاج جيني يحقق نتائج استثنائية في استعادة السمع

منهج جديد باستخدام الساعات الذرية يعيد طرح سؤال: ما طبيعة الزمن؟

المزيد

مواضيع ذات صلة

Definition and substitutability

Definition by genus and differentia

non-compositionality

Compositionality

Meanings above the word level: idioms

Meanings below the morpheme: sound symbolism

Semantics and lexicography

Meaning and the dictionary

Meaning and explanation

Meanings as referents/denotations

Lexemes

Meaning in Chinese