Skip to main content


"You need memory to keep track of the flow of conversation" [1]

This chapter deals with the correlation of language and memory which is divided into two main parts, namely acquisition of natural language and speech production. Both are strongly connected to memory. Acquisition of language describes the act of learning not only the mother tongue (in childhood) but also additional languages, whereas speech production consists of the physical and mental processes. However, only the latter is part of our field of interest.

Basically, language is processed in the higher hemisphere of the brain (Broca Region). The dependencies between speech processes like recall, search and decision problems are localised within different memory specifications (long term memory, short term memory, working memory and sensory memory). Therefore if one of the these parts are affected by specific diseases (Anterior Aphasia) it causes problems of understanding language, building complex sentences or even remembering simple words. Although interacting of language and memory does not seem to be mentionable at first, by regarding it closer it exposed to be the most important interaction which is necessary of being conversational.



Language is one of the most important ability of human beings which allows them to communicate with each other. Language processing and comprehension occurs in the left hemisphere in the Broca's area (especially for processing/production) and in the Wernicke's area (especially for comprehension). The sounds and written language (based on phonemes, letters and symbols) need to be structured with the help of a specific grammar, which indicates the syntax and its semantics. That construction is the condition of sharing and conveying information about thoughts, knowledge and feelings with other human beings. Primitive species (other than human beings) do not have such a complex and mature communication system. On that score human's communication is unique. However, if the specific areas in the brain are damaged or suffer from a specific disease, it leads to a limitation of its function.

(For further information see also chapter : Comprehension)


Memory can basically be seen as a collection of a versatile memory system. That multilateral system has many-sided storage processes for each individual constituent. Every component varies
fundamentally in its specific functions. All of them play an important role in guiding human's life and behavior by encoding and storing information, which was received over sensory stimuli in the past, for later usage. For instance, if humans would not have such a storage system, communication (language processing, which is very important for this chapter) would not function to some extent. Due to the fact that memory and its abilities are very voluminous, the discriminative types are allocated in many regions of the brain. Nonetheless the temporal lobe is the most important area for memory storage.

(For further information see also the previous chapter : Memory)

Brain regions

Language-related areas in the human brain basically are located in the left hemisphere. The processing of language requires a more complex network of interacting brain areas as previously assumed. In addition to Broca's and Wernicke's areas, several cortical and subcortical regions seem to be involved in normal language processing. Broca's (adjacent to motor cortex) and Wernicke's (posterior superior temporal lobe) areas are joined by the arcuate fasciculus, a bidirectional pathway, and are part of the implementation system - one of three functional language systems (implementation, mediational, and conceptual system)

(For further information see also chapter : Comprehension)


Short term/long term Memory

The distinction between long-term memory (LTM) and short-term memory (STM) is necessary since experiments have shown that they are functioning independently. Experiments with patients having brain lesions and as a consequence either a poor LTM and a functioning STM, or vice versa have shown that there is a double dissociation of LTM and STM. This means that one can infer that they are independent and served by different mechanisms.

The STM holds information only 15-30 seconds after the initial stimulus and the informations get lost after this time span if they are not transferred into LTM. Therefore the STM is meant to provide the information that is just being worked with. The information in the STM is most commonly coded phonologically, which means that it is represented by its sound (in contrast to visual or semantic coding), this was shown by an experiment of Conrad (1964, Acoustic confusion in Immediate Memory).

The LTM in contrast, can hold information for long periods of time, up to decades. The information gets most commonly encoded semantically, with respect to its meaning, although as in STM all three types of coding occur. Within the LTM there are two different types of memories; declarative/conscious memories and implicit/unconscious memories. The declarative memory gets also divided into two different subtypes. The episodic memory, which is responsible for storing specific autobiographic events, and the semantic memory, which holds general world knowledge (not in an personal context). The implicit memory is, as said before unconscious. This means that experiences of the past may affect a persons future behaviour without explicit knowledge about the reasons. A good example for this kind of memory is shown by the propaganda effect, which states that a person is more likely to evaluate a statement as true if the statement was heard before, even if the person does not consciously remember. This class of phenomena is referred to as the priming effect. Practical skills, such as used for motor or cognitive tasks are also part of the implicit memory and stored in the procedural memory.

(For more detailed information see chapter 6 Memory)

Working Memory

Working memory consists of a number of parts which help human beings to manipulate information for solving complex cognitive tasks such as thinking (also decision making) , comprehension, learning and reasoning. It is a temporal storage and works parallel to the long term memory. Both share most of their cortical structures. It basically means that both structures communicate very closely with each other. If a sensory stimulus is retained in the working memory it activates a cortical network which encodes its past associative context from the storage of long term memory while the working memory contemporaneously encodes that stimulus also. Working memory could be seen as a system recalling explicit knowledge for a specific task which needs to be solved.

The activity of working memory is localized between the prefrontal cortex and associative areas of the posterior cortex.

Researchers such as Alan Baddeley leads to the assumption that working memory consists of three components (Baddeley's model of working memory). One of it is the phonological loop which is an important part for language. It holds information about verbal and auditory stimulus. Another one is the visuospatial sketch pad that upholds the visual and spatial information. The last and probably the most complex and important one of the three components is the central executive. It dovetails the information which is stored in the long term memory and the information of the other two components with regard to the specific parts of the task which has to be solved. Therefore the central executive part could be understood as being the responsible one for the attentional control to some specific portions of the task. Due to that fact long term memory and working memory should be seen as a parallel working system whereas the latter is the on-line processing part. It is called „on-line processing“ because the working memory is able to sustain information as long as it is needed to reach a goal, understand and comprehend a sentence or a complex story, or to be able to dail a telephone number after looking it up in a telephone book. This chapter will focus on the role which working memory plays in reading, learning and comprehension of language is the important one to focus on. Hence without working memory human beings would be incapable of recognizing words and consequently understand and comprehend the syntax of spoken language or a text. It is not possible to hold on phonemes, letters, sentences and numbers in another storage system long enough to comprehend the context of the information which was just been written, heard, spoken or read. A small limit of the capacity of working memory is already sufficient to give trouble in syntax comprehension.

Sensory Memory

Sensory memory functions as an unconscious, huge collection of all sensory impressions or stimuli a person gets exposed to. Sensory memory can register a very large amount of information but this
information gets lost very rapidly. It is only retrievable within seconds or sometimes only fractions of a second after the initial stimulus.

From today's point of view sensory memory is considered to be important for several purposes. First of all, it is important for collecting all the information that is potentially processed later on. Then it functions as a temporary storage while processing is going on. Additionally, sensory memory is important for filling in gaps if a stimulus gets interrupted, as will be discussed later on.

There are three types of sensory memory Iconic memory or visual icon, memory for visual stimuli, echoic memory, memory for auditory stimuli and haptic memory, memory for touch. The two types of sensory memory that are most investigated in are the iconic memory and the echoic memory.

A very common example for iconic memory is the light beam that for example a pocket lamp leaves behind when it is moved rapidly through the dark night. The apparently real light beam is only created by the images one retrieves and stores in the sensory memory while the pocket lamp moves (persistence of vision). Another, maybe more common example where iconic memory is important is watching a movie. The movie is broadcast with 25 to 30 images per second. This is sufficient to give the illusion of a fluently moving movie because the gaps or interruptions that occur between the images get filled in with the information from the sensory memory.

The time how long a visual stimulus is kept in iconic memory was first investigated in by George Sperling (1960). In his experiment Sperling flashed an array of 12 letters (in three rows) on a screen for a total time of 50 milliseconds. He distinguished between a whole-report procedure and a partial-report procedure. Participants of the first group had to recall as many words as possible from the array after the stimulus was presented. In the partial-report procedure, Sperling indicated the exact row participants had to recall by a tone that was presented right after the stimulus. The results of this experiment showed that participants from the whole-report group were able to report only 4 or 5 of the 12 letters, while participants from the partial-report group were able to recall an average of 3.3 of the 4 letters. Therefore the percentage of recalled letters was much higher for the partial-report group. Sperling concluded from these results that after reporting the first 4 or 5 letters from the array the rest had already faded away and was therefore no longer recallable for his participants. To determine the exact time-span it takes for a iconic memory to decay, Sperling repeated the experiment with the partial-report procedure only and delayed the tone after the stimulus representation. He found out that already after one second delay the participant could only recall 1 out of the 4 letters and therefore performed equivalent to the participants of the whole-report group. Sperling concluded that the information stored in the iconic memory fades away in less than one second.

Darwin et al. (1972) replicated the Sperling experiment in order to investigate the echoic memory. Their participants had to wear headphones and Darwin et al. simultaneously presented nine letters, three to the left ear, three to the right ear and three to both ears. Afterwards a bar on a screen indicated which letters the participants had to recall. The results this experiment showed were closely related to the results of Sperling's experiment. The differences they found is that echoic memory lasts a bit longer than iconic memory, two to four seconds, and that not as much information gets stored as in the iconic memory. One commonality echoic and iconic memory have is that there is a lot more information stored than could be retrieved in the shored time it is available.

Sensory memory is neither a kind of short term memory, since these memories last longer, nor any kind of long term memory. Because of its huge capacity and short duration Sensory Memory can be seen as a accumulator and filter for sensory information.

Semantic Memory

Semantic memory is a form of declarative long-term memory and stands in contrast to episodic memory (particular,personal time and place related events) as first suggested by Tulving (1972).

Semantic memory holds the general knowledge about the world. This can be for example in the form of facts, skills, concepts or vocabulary and is therefore not related to emotions. Wheeler, Stuss and Tulving (1997) specified the differences between episodic and semantic memory more concrete in terms of their retrieval. Whereas episodic memory depends on a special kind of awareness, autonoetic or self-knowing, which is experienced when people think back to a certain moment of their life-time and remember former states of time, semantic memory only involves knowing or noetic awareness, because people think without emotions or personal relation, and therefore objectively about what they know.

In addition Wheeler et al. (1997) pointed out that semantic and episodic memory are closely connected and that, due to the similar encoding, it is not possible to store some event (e.g. knowing a new vocabulary) in semantic memory without encoding some kind of subjective experience in the episodic memory. This holds also for the opposite direction.

There are different views concerning the brain regions that play an important role for semantic memory. The two major opinions are first, that semantic memory is processed or stored by the same brain regions as episodic memory (medial temporal lobes, hippocampal formation) and , opposing to this, that these brain regions do not plain a role for semantic memory. Researchers supporting the second opinion propose different alternatives. Some claim that the episodic memory gets encoded in the neocortex, and others claim that the different aspects of one fact or concept are represented in different brain, so for example sounds in the auditory cortex and visual representations in the visual cortex.

Correlation between Language and Memory

Acquisition of language

According to Chomsky (1959) a child possesses innate neural circuitry specifically dedicated to the acquisition of language. However, many psychologists and linguists believe that language is neither entirely innate nor only acquired by learning.

Children possess an innate capacity for language and they acquire the language without special training or feedback. Normally, babies start to speak the first words around their first birthday and produce fluent grammatical sentences at the age of two or three. In contrast, other species fail to learn at all. Children have the instinctive tendency to speak as babbling of babies show. In their first month they are even able to discriminate speech sounds that are not discriminated in their parent's language. Thereby, children perform a sophisticated acoustic and grammatical analysis of its parent's speech, rather than correlating sounds with meaning or merely imitating speech. Although language is more specific than general intelligence, it is not a specific system for language but rather a general capacity to learn patterns: Every child will learn any language it is exposed to. Thus, there seems to be neural system that analyzes communicative signal from other people according to the design of language.

Speech production

Speech production processing is a more complex activity than one might think and requires several skills. We have to think about what to say, then to select the right words, to order these words grammatically finally express the sentence in actual speech. The speaker eases for the listener to understand him by using prosodic clues as rhythm, stress and intonation. Generally, syntactic boundaries (e.g. the ends of sentences) or grammatical junctures (e.g. the ends of phrases) are signalled by hesitations or pauses.

Since speech is normally way too fast it is hard to identify processes involved in speech production. Therefore, research focuses on speech errors in spoken language that can reveal how this complex system might work. There exist several types of speech errors while selecting the correct word. One kind of this lexical selection is semantic substitution where a word is replaced by another with a similar meaning, and normally of the same form class (e.g. "week" instead of "day"). Blending is the joining part of a word (or sentence) on to part of another (e.g. "breakfast" and "lunch" becomes "brunch"). In the case of the word-exchange error two words switch their places. If inflections or suffixes are attached to the wrong word, it is call morpheme-exchange error (e.g. "buyed"). Spoonerism is switching the initial letters of two or more words. Consonants are always exchanged with consonants and vowels with vowels, and often similar phonemes are switched. Mostly, letters within the same clause are switched which shows that a clause is an important unit in a sentence.

On the base of speech errors several theories have been developed. There is a strong similarity among these theories and most of them agree on the following points: Pre-production planning of speech, series of processing stages and procedure from the general to the specific. The spreading-activation theory by Dell et al. is based on the assumption that a representation is formed at the semantic, the syntactic, the morphological and the phonological level. Processing occurs at all four levels, and is both parallel and interactive. So-called categorical rules define categories at each level and dictate the required word. Nodes of the lexicon (network containing concepts, words, morphemes and phonemes) become activated. The most activated node of the appropriate category is then selected by insertion rules. A further approach is the theory by Levelt assumes that there is a network containing three levels. The levels represent lexical concepts, lemmas or abstract words, and morphemes. Activation proceeds only forwards and, the speech production involves a series of six processing stages: Conceptual preparation (potential concepts are activated), lexical selection (lemma is selected), morphological encoding (basic word form activated), phonological encoding (syllables of the word are computed), phonetic encoding (speech sounds are prepared) and articulation. The theory is to show that the word production proceeds from meaning to sound.


The research on patients with brain lesions gives important evidence to the structure of the brain and thus, to the function of certain brain regions. By these dysfunctions existing theories about memory or language production can be tested or new hypothesises developed. Amnesia describes the loss of memory and can among others be caused by a bilateral stroke, closed head injury or the Korsakoff's syndrome (chronic alcohol abuse). Since brain damage is often widespread the function of a certain area in the brain is problematic to determine. A further result of brain damage is aphasia, which is the impairment of language abilities. There are several forms of aphasia, e.g. patients with Wernicke's (or fluent) aphasia suffer from impaired language comprehension while patients with Broca's (or non-fluent) aphasia are not able to speak properly.

(For further information read the chapter Neuroscience of Language comprehension)


1. ↑ E. G. Goldstein, "Cognitive Psychology - Connecting Mind, Research, and Everyday Experience", page 137, THOMSON WADSWORTH TM 2005

External resources


• “Cognitive Psychology: A Student's Handbook”, fourth edition, M. Eysenck, 2000

• “Cognitive Psychology – Connecting Mind, Research, and Everyday Experience”, E. Bruce Goldstein (University of Pittsburgh), Thomson Wadsworth, 2005

• “Neuropsychology - The Neural Bases of Mental Function”, Marie T. Banich (University of Illinois at Urbana-Champaign), Houghton Mifflin Company, 1997

• “PRINCIPLES OF NEURAL SCIENCE”, fourth Edition, international Edition, Erik R. Kandel, James H. Schwartz, Thomas M Jessell, McGraw-Hill, 2000

Links ( Almaden Institute; Joaquin Fuster, UCLA: Cortical Dynamics of Working Memory, 2006) (Maxine L. Young, 2000) (University of Winnipeg; Prof. Evan Pritchard, PhD Attention & Memory, 2006) (ThinkQuest Team)

• of Oxford, Department of Physiology, Anatomy and Genetics; Kristofer Kinsey PhD)

• Tech, College of Computing; Harish Kotbagi, Human Capabilities(Memory), 1997)

• Tennessee State University; Stephen R. Schmidt, Copgnitive Psychology)


Syndicate content